Sar1 Agar

AD-A255 6S7
ADVISORY GROUP FOR AEROSPACE~ RESEARCH 8t DEVELOPMENT

7 RUE ANCELLE 92200 NEUILLY SUR SEINE FRANCE
Fundamentals and
Special Problems-of
Synthetic Aperture Radar (SAR)
LsAspects Fondamentaux et
Thismatria
pulictio
inthi wa asembed o sppot aLectureSeries~
udrtesponsorshipof the Avionics Panel ofAGYARD and the Consultant
and Exchange Programme ofAGARD presented on Sth-6th October 1992
in Bad Neuenahr, Germany, Sth-9th October 1992 in Gebze-Kocaeli
(near Istanbul), Tu rkev and 26th-27th October 1992 in Ottawa, Canada.
- - NORTH ATLANTIC TREATY ORGANIZATION
Published August 1992

Distribution and Availability on Back Cover
AGARD-LS-182
ADVISORY GROUP FOR AEROSPACE RESEARCH & DEVELOPMENT

7 RUE ANCELLE 92200 NEUILLY SUR SEINE FRANCE
. . - '- - - -. . . ..
AGARD LECTURE SERIES 182
Fundamentals and
Special Problems of ....-
Synthetic Aperture Radar (SAR)
Les Aspects Fondamentaux et
les Probl~mes Splcifiques aux
Radars AOuverture Synth6tique (SAR)
rrT(' QUALITY INSPMD 5
This material in this publication was assembled to support a Lecture Series

under the sponsorship of the Avionics Panel of AGARD and the Consultant
and Exchange Programme of AGARD presented on 5th-6th October 1992 in
Bad Neuenahr, Germany, 8th-9th October 1992 in Gebze-Kocaeli
(near Istanbul), Turkey and 26th-27th October 1992 in Ottawa, Canada.
- _ North Atlantic Treaty Organization

-- -- Organisationdu Trait de I'Atlantique Nord
92-257 i
....... U3U3U.t.....
The Mission of AGARD
According to its Charter, the mission of AGARD is to bring together the leading personalities of the NATO nations in the fields
of science and technology relating to aerospace for the following purposes:
- Recommending effective ways for the member nations to use their research and development capabilities for the
common benefit of the NATO community;
- Providing scientific and technical advice and assistance to the Military Committee in the field of aerospace research and
development (with particular regard to its military application);
- Continuously stimulating advances in the aerospace sciences relevant to strengthening the common defe,ce posture;
- Improving the co-operation among member nations in aerospace research and development;
- Exchange of scientific and technical information;
- Providing assistance to member nations for the purpose of increasing their scientific and technical potential;
- Rendering scientific and technical assistance, as requested, to other NATO bodies and to member nations in connection
with research and development problems in the aerospace field.
The highest authority within AGARD is the National Delegates Board consisting of officially appointed senior representatives
from each member nation. The mission of AGARD is carried out through the Panels which are composed of experts appointed
by the National Delegates, the Consultant and Exchange Programme and the Aerospace Applications Studies Programme. The
results of AGARD work are reported to the member nations and the NATO Authorities through the AGARD series of
publications of which this is one.
Participation in AGARD activities is by invitation only and is normally limited to citizens of the NATO nations.
The content of this publication has been reproduced

directly from material supplied by AGARD or the authors.
Published August 1992

Copyright 0 AGARD 1992
All Rights Reserved
ISBN 92-835-0683-9
HI
Printed by Specialised Printing Services Limited
40 ChigwellLane, Loughton, Essex IGiO 37Z
Abstract
The Lecture Series will cover the field of airborne and spaceborne SAR with respect to its technical realization in order to
# convey the participants' ideas and know-how on SAR, on its capabilities and on the technology necessary for the successful
construction and application of airborne and spaceborne SAR systems.
The basic principles of SAR will be explained and its peculiarities, ambiguities and special effects will be highlighted especially
in comparison with airborne and spaceborne Radar with Real Aperture (RAR). The influence of speed and altitude variations,
lateral motions on respective compensation possibilities will be presented.
The antenna is a system related SAR component. Therefore, the influence of the antenna parameters on specification and
capabilities of SAR will be considered. Polarization effects and multi-polarization SAR are presently key-points for SAR
development and SAR application as well as questions on absolute SAR calibration. The advantages, necessities and limits of
these topics will be included.
Digital SAR processing is for SAR indispensible. Theories and special algorithms will be given along with basic processor
configurations and different processing techniques on hardware and software bases.
The simulation of SAR-systems as well as SAR-products will also be a topic of the Lecture Series. This includes the simulation of
SAR-techniques and SAR-components as well as the simulation of SAR images. Special SAR-methods like squint-, stretch- and
spotlight-techniques for example will be presented in addition to the inverse SAR-techniques using the motion of targets instead
of the motion of the radar. A presentation of the state of the art giving examples of presently planned and up to now realized
airborne and spaceborne SAR with its application foreseen will conclude the Lecture Series.
The Lecture Series should appeal to technically and technologically oriented engineers concerned with development of SAR
and to scientists, who have to work with SAR for different applications as well as to students of both specialities who have
already attained a high degree of knowledge in techniques and of remote sensing applications.
This Lecture Series, sponsored by the Avionics Panel of AGARD, has been implemented by the Consultant and Exchange
Programme.
iii
Abrege
Ce cycle de conferences traitera du domaine des SAR aeroportes et spatioportes du point de vue de leur rtiaisation technique.
Les conferenciers mettront a profit leur competence POUr presenter leurs Wes concernant les SARl. leurs possibilit~s, et les
technologies qui sont Autiliser pour r~ussir l'industrialisation et la misc en oeuvre des systemes SARl airoportis et spatioportes.
Les principes de base de ces; syst~mes seront exposes, ainsi que leurs particularites, ambiguitis et sp~cificitas, en particulier par
rapport aux radars aeroportes et spatioportes a ouverture reite (RAR). Linfiuence des diplacements lateraux ainsi que des
variations de vitesse et d'altitude sur les possibilitis respectives de compensation sera 6galement presentee.
L'antenne est un composant du systime des SAR. Par consequent, l'infiuence des parametres d'antenne sur les sp~cifications
techniques et les capacites des SAR sera prise en consid~ration. A l'eure actuelie, les effets de polarisation et de
multipolarisation SAR doivent &tre consid~r~s comme des points c16 pour le developpement et les applications des SAR. de
m~me que leur 6talonnage absolu. Les avantages, les nicessitis et lea limites de ces donn~es seront examines.
Le traitement num~rique est indispensable aux SAP,. Des theories et des algorithmes sp~cifiques; seront proposes, ainsi que des
configurations de processeur de base et diff~rentes techniques de traitement sous les aspects matiriel et logiciel.
La simulation des systemes SARl et des produits SARl constitue un autre sujet de ce cycle de conferences. Ce sujet comprend la
simulation des techniques et des composants SAJl, ainsi que la simulation de l'imagerie SARl Des methodes sp~cifiques au SAR,
telles que le deport antenne (squint) lea impulsions itaI~es (stretch) et le mode tlI~scope par exemple, seront pr~sentis en
complement des techniques SAR inverses faisant appel aux mouvement des cibles au lieu du mouvement du radar. Le cycle de
conferences; se terminera par une prisentation de N'tat de l'art dans ce domaine, avec des exemples de syst~mes SARl
aeroportes, spatioportis existants et projetis et des applications pr~visibles pour lesquelles ils ont &6econcus.
Ce cycle de conferences est susceptible d'intlresser les,ingenieurs travaillant sur le d~veloppement des SAR et lea scientifiques
appeles a travailler avec les SAR pour diverses applications ainsi que les etudiants ayant deja des connaissances avancees dans
les deux domarnes des techniques en question ainsi que des applications en ulledtection.
Ce cycle de conferences est pr~sent6 par le Panel AGARD dAvionique; et organis6 dans le cadre du programme des
Consultants et des Echanges.
iv
List of Authors/Speakers
Lecture Series Director. Dr Wolfgang Keydel

DLI1
Institut fiir Hochfrequenztechnik
8031 Oberpfaffenhofen
Germany
AUTHORS/SPEAKERS
Dr John C. Curlander
Jet Propulsion Laboratory
4800 Oak Grove Drive
Pasadena, CA 91109
United States
Dr Anthony Freeman
Pasadena, CA 91109
United States
Mr Jean-Philippe Hardange
Thomson-CSF
178, Boulevard Gabriel Peni
92242 Malakoff Cedex
France
DLR
Institut ftir H-ochfrequenztcch.-.:k
Germany
4 Dr Herwig OttI
DLR
Institut ftir Hochfrequenztechnik
Germay
Dr R. Keith Raney
RADARSAT Project Office
Canadian Space Agency
110 O'Connor St, Suite 200
Ottawa, Ontario K IA IA I
Canada
V
i if Contents
Page
Abstract i
Abrfti iv
List of Authors/Speakers v
Refrrene
Basic Principles of SAR I
by W. Keydel
SAR Peculiarities, Ambiguities and Constraints 2

by W.Keydel
Motion Errors and Compensation Possibilities 3

by D. Hounam
The Real Aperture Antenna of SAR, A Key Element for Performance 4

by H. OWt
Polarization Effects and Multipolarization SAR 5

by A.Freeman
Radiometric Calibration of SAR Systems 6
by H. Otti
SAR Simulation7
by D. Hounani
Multi-Frequency Multi-Polarization Processing for Spaceborne SAR 8
by J.C. Curlander and C.Y. Chang
Inverse Synthetic Aperture Radar 9

J.-P. Hardange
Special SAR Techniques and Applications 10

by R.K. Raney
Review of Spaceborne and Airborne SAR Systems I1I

by R.K. Raney
Bibliography B
- - Vi
*I-I
BASIC PRINCIPLES OF SAR

by
W. Keydel
Deutsche Forschungsanstalt f~r Luft- und Raumfahrt e.V.
Institut fUr Hochfrequenztechnik
Germany
SUMMARY overcome by the here considered Synthetic Aper-

ture Radar (SAR) techniques.
The basic principles of SAR will be explained.
Equations for geometric and radiometric reso- 2. The SYNTHESIS OF AN APERTURE [1 to 5]
lution and their inter-relations will be gi-
ven in addition to a range equation. The dif- Radar techniques are principally one dimensional
ference between focussed and unfocussed SAR measurements. For image-construction the second di-
and the conception of beam sharpening will be mension has to be added. This is done principally
explained. by moving the radar platform of side looking sy-
1. RESOLUTION, KEY TO REMOTE SENSING stems in aircrafts or satellites. The geometric
range resolution of such systems is dependent on
For remote sensing purposes the resolution of the bandwidth which estimates the shortest proces-
the respective sensor is one of the main fac- sed pulselength. The azimuth resolution is depen-
tors. Resolution in the wide sense is defined dent on the antenna aperture 1 which determines the
as the degree to which a sensor can discrimina- half power beamwidth of the radar beam.
te two closely spaced targets, having similar The basic idea of the synthetic aperture radar
properties (geometry, colour, velocity, etc.)
However, angular resolution is a matter of do- (SAR) is the construction after data collection of
minant concern. It is defined as the minimum a very long antenna along the flight path (here as-
anlar sepaationn.
angular separation twefe two itemicn
between to items which can sumed as strongly linear) by means of aurg
si.Alnthfigtphaete data proces-
be distinguished by a system, sing. Along
points the flight phase
for amplitude, path and
are frequency
the measuring
of the
Note that for all systems using elecromagnetic backscattered signal. In this way the real antenna
Notes, thla oall
s stes using heragnetc acts as individual elements of the large (synthe-
waves, the laws of optics apply. The angular tic) array antenna. The stored echos are combined
resolution of an optical system is principally truhdt rcsig n h A mg spo
limited by both the aperture diameter and the through data processing, and the SAR image is pro-
turbulence of the wave propagation medium (i.e. duced.
the earth's atmosphere). Quantitatively the an-
gular resolution
1 is found by the ofratio
an aperture of a given
of wavelength size
X over setv
The Fig. 1emnlg fSRgeometry and the re-
and 2 show the
this size. ... , ,
The human eye, for example, is an optical sy-

stem, the resolution r ofwhich is limited by
the pupil diameter. Under mean illumination
with a wavelength of about 5 x 10- 4 mm the pu-
pil diameter varies between 1 mm and 8 mm and
an angular resolution power between about 2 arc
minutes and 10 arc seconds results. The mean
resolution power of 1 arc minute corresponds to
a ground resolution of 3 m for a distance of 10 IZ-
km. (This value can be experienced (approxima-
tely!) by looking down to the ground from an
aircraft). For optics with diameters roughly a -Ulm ,
factor 100 larger and at altitude of about 100
km, a resolution power in the order of 10 cm
results. This corresponds excellently to values A.
known or guessed for military satellites which
are now operational.
Diffraction limiting laws of optics apply to .. L-, AM UAW
microwave remote sensing systems, too. Here,

the antenna is the lens of the system and the -
antenna aperture, the diameter or length deter- -..
mines the aperture size. The antenna diameter ,., .-.
I determines the so-called antenna pattern. The
halfpower beamwidth of this pattern determines
of conventional radars the angular resolution Fig. 1 Geometry and terminology for SAR in
following in good approximation equation (1). 3-dimensional representation.
In order to obtain the same resolution as the
human eye in the visible region tremendous an- Due to the inherent velocity u of the radar, the
tenna diameters are required. A wavelength of frequency of the received signal is Doppler shif-
5 cm for instance (corresponding to 5 GHz) ted against the transmitted frequency. For a target
would require an antenna diameter of 175 m. seen under small angles A against the crosstrack
This illustrates the principal disadvantage of direction (less about 300) these Doppler frequency
conventional microwave systems in orbital ap- is:
plication. However, these disadvantage can be
1-2
z
Bmin12! (6)
; d Oepression Angel This T., is simply the time necessary to shift the
Doppler frequency from the absolute highest value
Nadi p, to the absolute lowest value within the bandwidth
Sof
\\\ the filter.
change Fromtime
of fD with (3) and (4) results for the
I ' AD 2u dx 2 u2
1 "R dt I - R (7)
Rn \AT--
I \ For AfD = BmIn and AT = Tmx results from (6) and

.\.. \ (7) by multiplication
2 2
B 2 U
-
ry Y B > ---
X (8)
- - -Swath Width- - -
With AfD = Bm.n results from (5)
(B)2= 2U )2 (9)
In connection with (9) results herefrom:

Li , I (10)
4T. T' amin 2
.' /
x. ' j.eI Rano. Bin This is the optimum resolution obtainable by use of
a fixed filter for a certain distance R (slant
Rrange). Remarkable is the angular resolution which
is dependent on the square root of R and not R as
is the case for a normal SLAR. The resolution is
Laet Rang. Bin independent of the antenna dimensions.
Principally, for image construction it is necessary
RealAntenna
Footpant to use for each range bin an other bandwidth (Fig. 1,
Pfozaessd Pixel in Y 2). The use of a filterbank following these equa-
theRangeBin tions within the swath from near range to far range
Mn S R S Rf) allows a relative simple real time
processing.
If a tracking filter will be used than a so-called

matched transmit receive system results. Here, the
Fig. 2 Twodimensional scheme of SAR. total time of target illumination can be used; this
is the total data acquisition time of the system
2u = 2u TD . During this time the aircraft flies a length
fD
- =--si
-- (2) Lp, which is the largest obtainable synthetic aper-
ture. This time is also called dwell time,
If a filter is used with the bandwidth AfD at the
output of this filter a singal will be showing up u
which is obtained from a small angle range Af with LP = v • TD = . (11)
the bandwidth AfD. To this Af belongs an azimuthal
distance Ax which is the azimuth resolution raf for Together with (3) follows
this simple case. In formulas written holds:
2t u Ax = 2v . (3 rap= (12)
AfD =FAX = X R X R
By use of the well known general valid relations
X2 Rbetween real aperture length 1, half power beam-
raf = .. Afp .(4) width Phr, slant range and wavelength I (Fig. 1,
2), (Lp = nr"R, Phr = 1/i) follows:
a syn-
This is the first and simplest method to built
thetic aperture by means of a filter with the band- 1
width Afn, it is called Doppler beam sharpening. It rap = . (13)
is remarkable, that (4) is independent on the real
antenna aperture. This is the rearkable formula for the optimum
Following (1) to this resolution corresponds a syn- theoretical reolution of so-called focussed SAR.
Folloianta at re Ld
thetic antenna aperture Lfp: o aRemarkable, that it becomes as better as smaller
the real antenna size 1 is. This is opposite to
2u real aperture radars and most optical systems. The
Lf (5) remarkable fact is that for SAR the theoretical li-
mit of azimuth ground resolution is given by the
half antenna length in the flight direction of thet
In order to optimize this method, the smallest pos- radar.
sible bandwidth B,n has to be estimated. Bm. n de- r
pends on the maximum observation time Tmax availab- The independence from range and wavelength is a
le. further remarkable attribute of SAR resolution. A
look on Fig. 1 and 2 shows: For each distance, each
1-3
range bin, holds an other synthetic aperture and these tude signals must be stored as well as in the unfo-
anetc cussed case. But before the addition of the various
synthetic aperture increases with increasing distancethe phase
as well as the geometric
apruedecreases with increasing of any optical
resolutiondistance, d n a eR0 -R
The range difference made. Fig. 3th
hashas totobe eqalizig phe
shows the
aperture of th incresin isae . The impulse response for a point target for different.
independence of the resolution is a reason for the phase errors (30*, 900, 1500). The signal degrada-
possibility to extend SAR results gained with airborne tion with increasing error evidently can be seen.
systems to spaceborne systems.
As allready mentioned the construction of the syn- The consideration of SAR as a synthesized array leads
to the same equations and results with respect to res-
thetic aperture for a SAR can be considered as the olution, synthetic aperture length etc. as the Doppler
artificial construction of an array by means of consideration does. However, the meaning of focussed
computer techniques. The signals will be stored and unfocussed becomes more understandable in the case
correctly with respect to amplitude, phase and the of the synthetic aperture consideration. Also the name
appropriate positions. During the image processing SAR becomes here more evident.
procedure the stored signals will be added up cor-
rectly and processed to a SAR-image. However, it 3 THE PHASE INFORMATION. KEY TO SAR [1 to 5]
will be a difference between a real array aperture
length and a synthesized aperture of the same 3.1 THE PHASE REFERENCE
length. Whilst for a conventional array the one way
beamwidth Ph estimates the resolution, for a syn- The range R between the platform and a point target
thetic array due to the coherent illumination the can be written:
two way beamwidth has to be taken into account,
since the beamforming phaseshift is introduced on
both the paths to and from the target. Therefore, R2 = R + (u-t)2 (20)
for the azimuth beamwidth of a synthetic array, a
factor 1/2 must be included in the formula (1)for
the one way pattern. u = flight velocity and t = observation time (the
maximum observation time is the dwelltime), R. is the
The time during which a point target is illuminated shortest distance between platform and target. With
by the radar beam is called the dwelltime TD (11). R + Ro - 2Ro follows:
The required dwelltime for a specified azimuth re- 1 u2 t2

solution is: R - 2 Po (21)
(14) To this oneway range difference corresponds a

T
UX Ru
r.- phaseshift
2
An illumination time T D can be provided with an 2 - o U t (22)
oneway antenna beamwidth X R,
= P.
1ha• (15) The two way phase change will be twice:
u2 2
t
There is principally a need for some phase compen- = 2x - . (23)
sation during the whole dwelltime. The distances Ro
of the far end of the synthetic aperture to a point This equation is very important as a reference
target are larger as the distance of the middle Ro. -function for digital SAR processing. The phase
(Fig. 1). A so-called unfocussed SAR ignores these shift is a quadratic function of time. This is
differences. This case limits the observation angle the so-called phase history of a point target,
to the area, where the differences in distance are the phaseshift versus time is a parabola.
smaller than X/16, which is identical with x/4 ra-
dian in phase To the quadratic phase function of time belongs a
linear frequency shift. The Doppler shift fD of the
IR. - Rol S - (16) signal due to the inherent constant platform motion
16 is given by
If this focus condition is fullfilled the occuring 2
degradations and the reduction of the result is not OD = 21 fD d 4u (24)
very large in comparison with the ideal case. Howe- -
ver, the dwelltime is not as large as it could be For a transmitted signal Vt = V, sin(2x ft) the re-
and the resolution is degraded as well as the maxi- ceived signal has the form
mum aperture length. From fig. 1 results with c
P, + Ro - 2Ro:
Vr = V2 sin(2% ft+ 2x fD - t) ,
2
L~)2 = Rj - P. - 2R,,(R - R). 7)Vr=Vsi2xf+2 U t (25)
2 Vr = V 2 sin[2(f + 7-R. -)t]
The limitation (16) leads to the maximum aperture
length of a unfocussed SA: This is strongly equivalent to a linear frequency
modulation. The received signal of a SAR is linear-
(18) ly frequency modulated. This modulation will be
2 .considered as a code which designes all points with
respect to their azimuth angle during flight time.
This aperture is responsible for the resolution The steepness of the so-called chirp is
=ii;__
obtainable and leads to
2
2v
rau, (19) R,
The bandwidth required for this linear modulation
In the focussed case the incoming phase and ampli- results from the dwell time T0 :
__
iI
1-4
Bf = 2 TD (26)
3.2 DEPTH OF FOCUS
Under the assumption that a maximum phase error of

X/8 may be allowed at the end of the aperture the
so-called "Depth of Focus" (DOF) can be defined.
DOF is the accuracy with which a given quadratic
phase reference function must be matched to the
considered range, it defines the number of diffe-
rent reference functions required over the whole
swath considered. It holds per definitionem:
DOF = 2AR . (27)
The maximum allowed one way phase error caused by

improper positioning of the phase reference func-
tion can be written as JAR - ARoI = 1/8. From (20)
follows:
2
dR R° AR R0 1 L
R-1 (-.) ,(28)
0 2
AR AR 2
Fig. 4 Example for focus influence of an image quali-
ty. The lower image is unfocussed (errors re-
sult from velocity) the upper image is focus-
2 (29a) sed.
DOF = 2AR0 = 2X L
4. HIGH RESOLUTION IMAGING BY USING THE PHASE
With (12) results REFERENCE FUNCTION
8 r. The phase of the received signal (25) contents the
DOF (29b) required information on an observed target. These
information can be extracted by a quadratic demodu-
The depth of focus becomes smaller as ra is made lator which eliminates the terms with the carrier
smaller. This is a very important fact for image frequency and leaves the socalled "inphase compo-
processing. For example, if X = 5 cm, ra = 3 m the nent I" and the "quadrature component Q" and a
DOF is about 1.44 km and, therefore, the processing joint amplitude factor A containing amplitude or
of a 4.3 km swath requires 3 different reference signal to noise informations on the target.
functions. For a resolution of 50 cm under the saae
conditions the DOF is 40 m and 10 references are u2 t
required. This increases the processing complexity. I(t) = Acos(2x( -)t)
Fig. 3 and Fig. 4 show the influence of focussing (30)
effects on SAR images exemplarically. u2t (
Q(t) = Asin(2x( )t).
1.6.
Therefore, the complex signal relevant for the in-

1.4 - formation on the target is (with A = 1):
1.2 ...... u2 t
S, = I(t)+jQ(t) = exp(j 2x(-)t) (31)
0.8 Herein, the target is positioned at t = 0 (where fD

- 0!) on the variable t, the dwelltime runs from
0.6 -TD/2 to +TD/2. The further information can be ob-
tained by correlation with a known reference func-
0.4 tion Sr for a point target. The appropriate refe-
..
--....-.. rence function for a point target at this location
0.2 is the same as (31). The output of the correlator
is:
:rM../
,AU +T/2
So = J S.(t)'Sr(t+tr) dt , (32)
Fig. 3 Point target impulse response degradation for -T/2
unfocussed prozessing with different phase de-
I
viations (300, 90@, 180").
2 2 +T/2
Soe 2x X e,
2 f e- 2I
2x
2 U2 Rtr t dt,
-T/2
! 1-5
2 u S sinc{ U t) . 3) zes the vector character and the signal delay time
X. 2 " The received radar signal is of the kind of

N
The sinc function in (33) delivers the image of the k = E j(i t + 9) (35)
point target. Remarkable are the sidelobes. The
=
correlator output has a maximurm for tr 0, when i=l
the reference function and the target aligned with Here is N the total number of scatterers contribu-
each other. Fig. 5 shows the unprocessed amplitude of ting to the signal, &± the vector-field scattered
a point target as well as it gives the processed image by the ith scatterer, r the totally received
of a point target in azimuth direction. field, mi = 2 sfi the circular frequency and ti =
t + Ti the observation time with the respective

delay time Ti. The radar receiver in connection
received signal with the data processing part has the task to
I. l IItransform this complex signal into an observable
signal proportional to IMyf2 and to extract the in-
A formation on distance, velocity, benaviour, shape,
quality and other quantitative aspects of the ob-
served target. IJr2 is equivalent to the observed
2power.
E real antenna A very important part plays the term "coherency"
synthetic which will be used for techniques, instruments and
antenna .- for scattering mechanisms as we. Coherency means
the phase stability and the statistical behaviour
I of the phase of electromagnetic waves over the ob-
servation time. For complete incoherent scattering -
0.0 0.2 0.4 0.6 0.8 this means if there are many independent scatterers
time > within the beam and for a pure statistical phase di-
se stribution between the scattered signals - the perfor-
Fig. 5 Amplitude of a pulse response (upper curve) mance of radar can be described with good validations
mncofrdranbderidwthgdvldtos
from a point target and SAR-prozessed impuls by the radar equation.
response of a point target (lower curve)
following (33) in comparison with point tar- 6.1 THE RADAR EQUATION FOR SAR
get response for a real aperture radar.
The iadar equation is given
by Hall for a point
5. MODULATION PRINCIPLES FOR RANGING target (which consists per definitionem of one
The main considerations up to now went into direc- single scatterer) [Il:
tion of the azimuth resolution capability of SAR, Pave G2 X2 a " T,
which is obviously the main advantage: The azimuth =
SIN 4 . (36)
resolution of SAR can be equivalent to the resolu- (4s)' (k T% F) R • L 0to
t'on of optical systems, despite of the fact that
small antenna apertures will be used. However, the Herein is: Pave = average transmitted power, G =
range resolution should have the same order in or- antenna gain, X = wavelength, a = radar cross-sec-
der to produce high quality imagery. The range reso- tion, R = distance radar-target, k = Boltzmann con-
lution ry of a pulsed radar is equal to half of the stant, T. = noise reference temperature usually 290
pulslength ip. K, F = receiver noise figure, Lto all losses.
For the dwelltime TD Results from (11) and (12):

ry 2 (34)
TI R (37)
High resolution would require extremely short pul- TD = 2 r. u
ses and this would require extremely small band-
width. However, this would entail considerable This leads to the SAR equation for point targets:
technological problems and would rapidly lead to
the border of SAR possibilities. A solution is here = avN X3 a
to use pulse compression-methods [1]. There are main- SIN = (4 x)J R3 (kTF)2ur.Ltt 138)
ly two methods: the commonly used frequency chirp mo-
dulation and the digital pulse code modulation with For area targets the radar cross section per unit
Barker or Pseudo-Noise codes respectively. Both me- area a0 has to be taken into account:
thods require large time bandwidth products and are
well developed for RAR with ranging tasks. However,
it should be mentioned here, that the principles of a = Oo • ra • ry F
SAR can be considered without special ranging pro- 2 X3I0, Y _
cedures. But looking at the results of the previous Pay. G2 )A0o *y
Doppler shift considerations during the azimuth S/N = (4x). R (kToF)'2u'Ltot (39)
measurements a SAR uses principally a linear fre-
; quency for the
modulationwith
mally identical azimuth coding,
frequency which isprin-
modulation for- Remarkable is here the dependence of the cube of the
ciple for ranging it th r ncy an ri - range even for point targets. This is a significant
For both, ranging and azimuth difference to real aperture radar. This equation holds
measuring the same receiver unit (and the same if all scatterers within the resolution cell ra'r are
processor) can be used and this is the main reason i th
why always the FM-chirp for ranging will be consi- indepenn at means for incoherent summation o the
dered in connection with SAR in the literature. backscattered power. Fig. 6 shows as an example the
image of an oilspill on the seasurface taken with a
6. POWER CONSIDERATIONS AND FADING STASTISTICS
Principally, radar uses information on electroma- Fig. 7 shows the signal to noise ratio3 as a func-
gnetic waves characterized with amplitude E, phase tion of range [7]. The dependence of R and of the
9, frequency f, the polarization which characteri- antenna diagram becomes evident as well as the so-
called amplitude line-echo (ALE) which results from
1-6
40
30
20
10
0 100 200 00 50 W
0
Fig. 8 Signal intensity curve of the image in Fig.

6 after radiometric correction using the
curve in Fig. 7b. (660 m flight altitude,
30 deg depression angle.)
Fig. 6 SAR image of an oilspill on the sea sur- but many scattering centers which contribute to the
face taken with DLR E-SAR, C-band (7]. radar response signal; individual scattering cen-
ters may be seen with very high resolution systems
20 (exeptions are very simple targets only like sphe-
res, corners, zylinders, plates). Fig. 9 shows as
an example the distribution of such centers on a
car [8].
_ I
00 20 0 400 WO
Fig. 7a Signal to noise ratio over the whole scene in

Fig. 6 as a function of range. (660 m flight
altitude, 30 deg depression angle (7].) The
Altitude line echo (ALE) clearly can be iden-
tified.
Fig. 9 Distribution of reflection centers on a car

25 in a twodimensional representation. Resolu-
tion is about 15 cm x 15 cm (measurement Graf,
20 DLR [2]).
Under the assumption of statistical distributed

amplitudes Ej and phases 91 (35) becomes a stati-
stical phasor the quality of which is dependent on
the respective distribution functions. For uniformly
distributed phases vi and Rayleigh disributed am-
plitudes results an exponential distribution for
the amplitude square of (35). That means in general
if homogeneous areas of the earth are illuminated
3Woo
- o
40 o o by a coherent radar the backscattered signal ampli-
s.,tI
"W" -, tudes of the single observations (pixels) are sta-
Fig. 7b Signal intensity curve of Fig. 7a as a func- tistically distributed and this is the reason for
tion of range integrated over 50 S. (660 m the speckle
images typical
of very for each
homogeneous radarThere
areas. image is
also for
no pos-
flight altitude, 30 deg depression angle.) sibility to characterize a radar target by measure-
reflections directly underneath the aircraft. This ment of one pixel only. Principally this speckle
can be reduced by image averaging with the use of
curve can be used for image correction. multi looks for instance, this reduces the standard
Fig. 8 gives the signal intensity curve of Fig. af- deviation proportional to the root of the look num-
ber and smoothes the speckle. Speckle can be seen
ter correction. The image in Fig. 6 is corrected ( in all the radar images presented here.
by using the results of Fig. 7.
6.2 SPECKLE STATISTICS The Rayleigh distribution for the envelope Vr be-
longing to (35) is
However, equation (35) shows the statistical nature V2
of the radar response signal due to the target and =
scatter statistics. Normally complexe radar targets P(Vr) expf- -) for 0 f Vr . (40a)
a 2o
like cars, trees and areas etc. have not only one
t 1-7
The phase is uniformly distributed
P(9) = for 0 ! 9 5 2n . (40b)
Mean value Vr and the second moment V2 of (40a) - M OIL

are: ....M
The 2=2 (41) -
The variance is by definition:
Var(vr) Vj - V 0.429 .2 V2 %
The variance represents the fluctuation of the re-

gistered voltage around the mean value, this is
principally an inherent noise and it can be written zntmity/1 -
as a mean squared AC component Va. Together with Fig. 10 Histogram of the filtered SAR image of Fig.
(41) one obtaines the inherent signal to noise ra- 6. (Oil slick A2, 660 m flight altitude.)
rio Sn for Rayleigh fading Both curves follow a Rayleigh distribution.
Sn, - = 3.66 ' 5.6 dB
This occurs, even in the absence of any other ad- VL = N 1 Vrl -

ditional noise. That means, the Rayleigh fading is i=l
equivalent to addition of noise that results in a
signal-to-noise ratio of 5.6 dB, even if the ratio For the mean holds principally
of the Thus,
high. average
the signal level to actual
best equivalent noise is very
signal-to-noise ra- VL =
V 2 a, i
tio that can be obtained when Rayleigh fading
exists is 5.6 dB. This is the reason that fading But for the variance one obtaines
distributions are so important in radar problems.
One often would like to have a high signal-to-noise V 0.429 . 02
ratio, but the effective signal-to-noise ratio can- VL-- =- =-
N N
not exceed 5.6 dB unless multiple independent fa-
ding samples are added together, regardless of the With these results for the optimum equivalent
ratio of the mean received power to the thermal- Snr results instead of (41)
noise power in the system [5]. r
The measured signal Pr which is considered in (36) Snr = 3.66 • N . (44)
is the power, which can be written asI
The maximum obtainable signal to
noise ratio in-
P
=
V2 = S creases rapidly with increasing N.
The time T during which a point target remains

From distribution (40) for Vr results the exponen- within the real beam is the time for observing
tial distribution for Pr: this fading signal. This time can be obtained from
the consideration of the extension of the footprint
S 0in flight direction which is the azimuth groundre-
2 e 2 fo solution ra of the real antenna.
2ov 2ov
For the mean value P. and the second moment FT holds
(45)
2 ra h R = vT.
r=2U2, F7 = .
The number Naz of independent samples in azimuth

With these results the Rayleigh disribution (40) direction is the product of bandwidth and observa-
can be rewritten:
Vr V8 Naz AfD • T ,
P(Vr) = expi- -- )-} (43) (46)
P, 2Pr r R
Fig. 10 shows as an example the histogrammes of N,, 2 .

Fig. 7 [7].
This leads to the simple equation
If N independent pixel values are added than fol-
lows from the laws of probability theory that the ra
mean value of the sum is identical with the mean N52 = i7 " (47) ,
value of the single element whereas the variance
for the sum of N independent samples is the varian- The equations (46) and (47) are very fundamental.
ce of the single sample divided by N. That means The a to (46) and (47) a vey fuatal.
for the case of a linear detector if N numbers of They apply to SLAR and to SAR as well. Equation
samples are averaged and if Vr± is the envelope (47)
bewe shows
ag1eouin n ml which
the possible tradeoff can be made
5 adhg
voltage of the ith sample than is the average re- specle reduction (large N).
ceived voltage VL: pcerdtin(agN)
The number N of independent samples in range di-
rection results from considerations of range re-
solution. The slant range resolution for a pulse
1-8
radar ry is given by:

c • p c
Reolutfon: 8.94 m
If NY returns with that resolution will be averaged "PSLR:-S.O1dB

than a poorer resolution will result. Therefore,
the number of independent rangepixels is proportio- 4 1. ISLR :-6.9dB
nal to the reduced range resolution ryred.
ryred
NY =
The equivalent pulselength Tp.q associated with NY

averaged returns is: a
pq = N ,y . 0 iam acm aco

Time (ms)
Using these results, one can express the number of
independent range samples:
ryred tpeq B R
NY = = P B (48) PSLR -28.1 dB
e ISLR : - 13.9 dB
In (48) is B.q the equivalent bandwidth which be-
longs to Tpq.
The total number of independent samples is given

by:
N = Naz •Ny. (49)
Equation (48) shows as well as equation (47) in 0 ,o0 Tim (ma

connection with (45) the possible tradeoff between
high geometrical resolution and high speckle reduc- Fig. 11 Impulse responses (2dimensional image) of a
tion which corresponds to high radiometric resolu-
tion. Radiometric resolution describes the possibi- resolution and siderlobe level.
lity to consider two areas with two different
speckles (or medium grey tones) as different. One
possibility is to take the mean values X and X2 of - , , O O
the probability density functions of the pixels of
those two fields and use the ratio of both mean va-
lues for a measure of the separability:
x2
rrad = - , (50a)
R1 + Ax
rrad = (50b)
This ratio is called radiometric resolution. For

SAR the angular resolution is closely connected to
the radiometric resolution. The radiometric reso-
lution is the minimum brightness contrast necessary
for the discrimination of two targets. The radiome-
tric resolution in radar images is dependent on the
image statistics, the speckle. Therefore, by in-
creasing the integration time (i.e. the observation
time of a certain area represented through the num-
ber of looks on it) the speckle will decrease, the
radiometric resolution increases (the image becomes
sharper), but the angular resolution decreases.
Fig. 11 shows the processed radar answer of a point

target with different geometric resolution (8.94 m
and 11.4 m). The decrease of the sidelobes with in-
creasing resolution is evident. A quality measure is
the peaksidelobe ratio PSLR as well as the integra-
ted sidelobe ratio ISLR which are defined as fol-
lows: Fig. 12 SAR image of DLR-Research Center and airfield
in Oberpfaffenhofen taken at 5 cm wavelengths
PSLR = power within the sidelobes from an aircraft at 914 m altitude. Scene di-
power within the main beam mension 2700 m x 3510 m, resolution 2 m x 3 m
(azimuth x range), 8 looks.
ISLR = power within the sidelobes
power within the total diagramm
~I-9
In Fig. 12, 13 and 14 this tradeoff between geome-

tric and radiometric resolution as well as the in-
fluence of the speckle is illustrated. The images
from the DLR-Dornier airport at Oberpfaffenhofen
have been taken by the airborne SAR of the DLR at 5
cm wavelength [9, 10, 11]. Fig. 12 shows an 8 look
image with low speckle and high contrasts. This
image with its dimensions of 2.700 m x 3.510 m
%range x azimuth) has a resolution of 2 m in range
and 3 m in azimuth; with its many details it is
comparable to a real photography. Fig. 13 shows en-
larged a part of Fig. 12: the scene dimension is
828 m in range an 1.079 m in azimuth. This one look
image has an azimuth resolution of 0.5 m and a ran-
ge resolution of 2 m. (The resolution in range here
can not be increased due to the limited bandwidth
of the system.) The increase of the speckle against
the 8 look image in Fig. 12 is evident. Impact of
high resolution becomes clear by looking at the
large black platform in the upper left corner. The-
re an aircraft clearly can be seen. This aircraft
together with its background is shown enlarged in
Fig. 14 with the same resolution. The image size
corresponds here to 60 m x 60 m. The differences
in range and azimuth resolution are evident. This
same image was used for estimation of the dimen-
sions of the aircraft. A comparison with the known
dimensions of a DO 228 aircraft leads to the con-
clusion, that the image shows with a high degree of
probability a DO 228. The estimated dimensions
(with the real values of a DO 228 in paranthesis)
are: Total length 14.3 m (15.04), total wingspread
Fig. 13 SAR image, part of the airport in Oberpfaffen- 16.3 m (16.97 m), wingspread of the elevator unit
hofen. The scene corresponds to a section of 7.7 m (6.45 m). This points out, that for airborne
the image in Fig. 12. Scene dimension is 826 m SAR a resolution in the order of dm is the present
x 1079 m, ground resolution 0.5 m x 2 m (azi- state of the art [12, 13] However, for spaceborne
muth x range), 1 look. The aircraft on the SAR the present state of the art is poorer due to
platform in the middle will be enlarged re- different reasons (atmospheric turbulences, power-
presented in Fig. 14 (Moreira, DLR). limitations, data-rates). Present spaceborne SAR
4L-.. E-, ( - - P E-SO.
O principally can have a ground resolution power in
the order of one or a few meters.
! ... . . ....... .. ..... .I 7. SAR IMAGES
7.1 IMAGE GEOMETRY
SAR-pictures are different from normal photographs

in many details. They look like an aerial photo-
graph although taken from a sidelooking perspecti-
ve. This side looking perspective leads to shadow-
effects and to the well-known characteristic pla-
sticity of SAR-pictures. The shadows can be used
for height estimation of trees, rocks, and other
shadow producing elements in a SAR image as in Fig.
22 for example.
SAR
RADAR BEAM
,N RADAR-IMAGE PLANE
N ... RADAR-IMAGE
- FORMAT
Fig. 14 Enlargement of the aircraft on the dark plat- "I

, II
form in the upper left corner of Fig.13. Scene, ,
dimension is 60 m x 60 m, ground resolution is NEAR RANGE
\ FAR RANGE
0.5 m x 2 m (azimuth x range). The unsymmetry a b C
in the resolution pixels caused by the unsym- LEVEL SURFACE
metrical ground resolution area clearly can be
seen. The aircraft is probably a Do 228; this
is a conclusion of a geometrical evaluation of Fig. 15 Radar-image format showing slant-range pre-
the aircraft image in comparison with the well sentation a, b, cl relative to ground-range
known real dimensions. The different reflec- abc, assuming a level surface long-dashed
tion centers typical for radar imaging as well line represents radar-image plane. Look ang-
as the speckle clearly can be seen (Moreira, le is inclination of the radar beam off ver-
DLR). tical (14].
I-I10
An incorrected SAR image is allways a so called

slant range image it represents basically the di-
stance from the radar to each of the respective
surface elements in the scene. Therefore, in its
raw stage it shows some geometric distortion due to
the differences between slantrange and groundrange
explained in Fig. 15 (14]. Fig. 16, 17 and 18 show
raw data a slant range image as well as a finally
processed image of the same scene.
S!AlT PAJCS----
Fig. 18 Azimuth processed image - only the totally

focussed region. Processed from raw data in
Fig. 16 and slant range image in Fig. 17.
Foreshortening [14]
Slopes inclined toward the radar appear compressed

relative to slopes inclined away from the radar.
The effect is illustrated in Fig. 19. The foreshor-
tening factor F is approximately:
F = sin(O-a) ,
Fig. 16 Raw data: RCM corrected, range compressed where the look angle 0 is the angle between the
image before the azimuth processing. vertical plane and a line that links the imaging-
radar antenna to a feature on the ground, and is
the slope angle of the surface. Alpha is positive
(a) where the slope is inclined toward the radar
(foreslope), and negative (W) where the slope is
inclined away from it (backslope).
SAR
* RADAR BEAM
*,,9 RADAR-IMAGE PLANE
a RADAR - MAGE
/ IO
NEAR RANGE FA RANGE
SLOPING SURFACES
Fig. 19 Radar foreshortening of slope ab, which is

projected as a, bl, relative to slope bc,
- which is projected as b, cl, showing look
angle 0, foreslope angle a+, backslope angle
a-. Long-dashed line represents radar image
Fig. 17 Slant range image processed from raw data plane.
in Fig. 16. Azimuth processed image inclu-
ding the partially processed region. Layover [141
A further set of geometric relief dispalcements Layover is an extreme case of foreshortening that
that is characteristic of all imaging radars is the occurs when the look angle 0 is smaller than the
apparent variation in length of equal terrain slo- foreslope a+ (6< * ). This is illustrated in Fig.
pes when the respective slope lengths are imaged at 20. In this case, a mountain is laid over on its
different incidence angles. The displacements re- side.
sult in foreshortening, layover, and shadowing.
These effects are feature-dependent and cannot be
easily corrected.
w t1-11
SAR
RADAR BEAM
* RADAR-IM~'kGE PLANE
'N Nb RADAR-IMAGE
FOII
/, / C
NNEAR
RANGE ~ .- FARRANGE
,,
SLOPING SURFACES
Fig. 20 Radar .,:ycver of slope ab projected as b, a,

on radar image, where look angle @ is smaller
than slope angle a'. Long-dashed line repre-
sents radar-image plane (14].
SAR
RADAR BEAM
RADAR-IMAGE PLANE
o f. Fig. 22 Iller river and a channel taken by DLR-SAR

(C-band) from 1820 m altitude. Resolution
I RADAR 2 m x 2 m, scene dimension: 768 m x 768 m.
b,"'
1 SHADOW The altitude of the trees beside the road in
_the
/+ upper left corner can be estimated from
'F dI knowledge about the flight geometry and the
NANGE shadow length's. For the first 4 trees from
RNE I, right to left just left beside the left brid-
nRA.
dge result: shadow length's between 35 m and
a C d40 m and tree altitudes between 20 m and 24 m
Fig. 21 Radar shadow of surface bcd projected as (image: Moreira, DLR) [12].
b, d, on radar image. Long-dashed line re-
presents radar-image plane [14].
Shadowing [14] NA-
Shadowing is essentially the opposite of foreshor-

tening. Slopes inclined away from the radar are in n u
shadow when the look angle 0 plus the backslope 2D.
angle a- are greater than 900 [(O-) > x/2]. Sha-

d-ws are caused by ground features that obstruct
the radar beam and prevent illumination of the area
behind them. This effect occurs on Seasat SAR ima-
ges whenever the backslope in the radar viewing di- r N,-
rection exceeds about 700. It is shown diagrammati- iiA

cally in Fig. 21. Fig. 22 shows clearly the shadows
of different features. u
8. BASIC HARDWARE CONSIDERATIONS "
8.1 PRINCIPAL CONSIDERATIONS
A simplified block diagramin for a SAR is shown in

Fig. 23. Fig. 23 Block diagram of a SAR (schematically).
The SAR is mounted on a platform moving at a con- A typical SAR transmitter is designed to overcome
stant velocity. The PRF must be sufficiently high limitations of peak power in components and to sa-
to avoid azimuth ambiguities. This criterion requi- tisfy stringent azimuth and range resolution requi-
res that the radar platform displacement cannot ex- rements. High resolution in azimuth requires strin-
ceed one-half the antenna size between successive gent phase stabilities from pulse-to-pulse and over
transmit pulses. the integration time. High range resolution requi-
res wide RF bandwidths. To meet the signal-to-noise
In a SAR, phase stability is exceedingly important. ratio and target detection requirements correlation
The prime oscillator which provides the signal for
the transmitter as well as the reference for the
receiver must be very stable. The timing of the
transmit pulses must be very precise with respect
principles and pulse code modulations can be used
respectively with special wave forms.
Electronic circuits using a voltage controlled os-
I
to the prime oscillator. If the radar platform ve- cillator can provide the desired transmit pulse.
locity is not constant, the deviations must be mea- The output from the high power amplifier passes
Sincoming sured andsignal
this data used toon compensate
or passed either
to the signal the
proces- through
antenna. aThe
circulator
receivedand is radiated
signal by the radar
passes through the sa-
sor as a correction. me circulator, is amplified (and normally pulse
1-12
compressed). (the pulse compression circuit can use

a frequency dispersive delay line which converts a
wideband linearly swept FM long pulse into a short N -
pulse signal with same bandwidth.) The phase of the
short pulse is measured by the phase coherent de- .... -:
tector, and the resultant signals are delivered to
the signal processor. -
8.2 THE DLR E-SAR AS AN REALISATION EXAMPLE[R - -

The DLR airborne experimental synthetic aperture
radar system E-SAR, designed and manufactured. ..
at the DLR Institut fuer Hochfrequenztechnik,
is a research tool to elaborate SAR related prob-
lems concerning both system performance and data
analysis. The instrument is installed on board a
DLR Dornier DO 228 aircraft, which is a small .
STOL aircraft (STOL: short take-off and landing),
offering the advantages of low costs and operation
from airstrips in any part of the world. Fig. 24 E-SAR sensor system blockdiagram (RF-elec-
obtained with
All SAR results shown here have been
this instrument.
RF centre frequency, L-band: 1.3 GHz
Since the beginning of 1989 the E-SAR system has C-band: 5.3 GHz
been flown many times in preparatory campaigns for X-band: 9.6 GHz
the European Remote Sensing Satellite ERS-l. The IF centre frequency: 300 MHz
German/Italian X-SAR, which will fly with SIR-C System bandwidth: 120 MHz
on three Shuttle Radar Lab missions, and the French SAW chirp, signal bandwidth: 100 MHz
Radar 2000, both spaceborne SAR projects, are sup- expanded pulse length: 4.98 ps
ported with E-SAR image data. compressed (analogue) pulse
length: 17 ns
Applications Digital chirp, signal bandwidth,
narrow swath mode: 90 MHz
The E-SAR is a high-resolution SAR operating in L-, wide swath mode: 50 MHz
C- and X-band with either horizontal or vertical super wide swath mode: 18 MHz
polarization. Although being developed mainly for expanded pulse length: 5.0 l's
use by the research community, commercial lease Antenna gain, L-band: 14 dBi
opportunities are as well anticipated. The sensor C-band: 17 dBi
is versatile, with many options for flight and ra- X-band: 17.5 dBi
dar configurations and image products. It provides Antenna 3 dB beamwidth,
the opportunity to image areas, wether flat or azimuth, L-band: 18 Deg
mountaineous terrain, ocean or ice, with excellent C-band: 19 Deg
image quality. It can be used for monitoring resour- X-band: 17 Deg
ces, renewable such as agriculture and forestry, or elevation, L-band: 35 Deg
nonrenewable such as geological resources. Changing C-band: 33 Deg
characteristics such as urban growth, deforestation X-band: 30 Deg
or ocean waves also can be monitored. Transmit peak power, L-band: 500 W
C-band: 90 W
The System Platform X-band: 2500 W
Receiver noise figure, L-band: 8.5 dB
A Dornier DO 228 aircraft equipped with modern na- C-band: 4.0 dB
vigation systems like a laser inertial reference X-band: 4.5 dB
system (IRS) and a GPS receiver carries the E-SAR Receiver dynamic range with
sensor. Its maximum take-off weight equals 5980 kg. AGC/STC: ? 40 dB
The maximum operating altitude above mean sea level Nominal pulse repetition fre-
(MSL) is 8000 m. The maximum cruising speed is quency (PRF): 952.38 Hz
about 440 km/h. For SAR operation the nominal ground Variable PRF range: +/-30 %
speed of the aircraft is 70 m/s, which corresponds Quantization (I or Q): 6 bit
to 252 km/h. A/D converter dynamic range
(at 35 MHz): 25 dB
The On-board Segment Sampling rate, narrow swath mode: 100 MHz
wide swath mode: 60 MHz
The radar sensor is a modular designed system, which super wide swath
contains three different RF-segments in L-, C- and mode: 20 MHz
X-band. Pulse generation and I/O-detection are loca- Echo buffer memory capacity
ted in the IF-section. A single digital conversion (I or Q): 2560 words
and recording system is used to store the SAR raw da- Nominal data rate on high density
ta on high density digital tape (HDDT) formatted in tape: 28 MBPS
the SAR 580 HDDT format. Maximum recording time per tape
(14 inch tape reel): 15 min
The On-ground Segment Spatial resolution, range and azi-
muth, narrow swath mode: 2.5 m x 2.5 m
The ground segment consists of the following wide swath mode: 4.5 m x 4.5 m
units: super wide swath mode: 11.5 m x 11.5 m
Number of statistically inde-
- Radar faw Data Tramskziptiom pendent looks: 8
A SAR 580 High Density Tape Transcription Radiometric resolution (8 looks): < 2 dB
system (HTS) transcribes the raw data from Geometric distortion: les than one re-
HDDT to computer compatible EXABYTE (Video solution cell
8) tapes. A further data transfer to con-
ventional CCTs is possible. This operation Table 1 E-SAR specifications.
1-13
[6]Elachi, Ch. Spaceborne Radar Remote Sensing
Applications and Techniques.
2!r IEEE, New York, 1987.
[7] Moreira, A. Entwurf und Ergebnisse des DLR-
Echtzeit-Azimutprozessors fdr
das E-SAR-System.
DLR-FB 89-30.
[8] Graf, G. High Resolutin Imaging of Radar
Targets with Microwaves.
:' IConf. Proc. Military Microwaves

'78, London/Engl., 25.-27.10.78.
Microwave Exhibitions and Publ.
Ltd., pp. 295-302.
Fig. 25 E-SAR sensor system blockdiagramm (digital [9] Horn, R. C-Band SAR Results Obtained by
electronics). an Experimental Airborne SAR
Sensor.
also converts the SAR 580 HDDT format into Proc. IGARSS, IEEE, 1989, pp.
the SAR 580 video signal CCT format and pro- 2213-2216.
vides the full data rate in a single channel
transcription mode. [10] Moreira, J. Estimating the Residual Error
of the Reflectivity Displace-
- E-SAR Standard SAR Processing ment Method for Aircraft Mo-
The E-SAR standard SAR processor consists of thre tion Error Extraction from SAR
basic modules, auxiliary data processing, off- Raw Data.
line motion compensation and focused multi- ICCC Intern. Radar Conf., May
look SAR processing. The development was carried 7-10, Arlington, USA, 1990,
out in DLR. The processor output is calibrated pp. 70-75.
and available in form of a standard image pro-
duct. [11] Moreira, J. A New Method of Aircraft Motion
Error Extraction from Radar Raw
Motion Compensation Data for Real Time SAR Motion
Compensation.
The DO 228 aircraft is fairly sensitive to air turbu- 12th Canadian Symposium on Re-
lence and, therefore, not very well suited for car- mote Sensing, IGARSS '89, Van-
rying a SAR sensor. A SAR is a coherent system and cover, Canada, 10-14 July 1989.
flight instability causes phase errors, which, in Proc. IGARSS, IEEE, 89, pp.
turn, defocus and distort the image geometrically. 2217-2220.
This problem can be overcome by measuring the dyna-
mic behaviour of the platform and correcting the SAR [12] Keydel, W. Microwave Sensors for Remote
data, either on board the aircraft or on ground. Two Sensing of Land and Sea Sur-
methods for compensating platform motions are imple- faces.
mented with the E-SAR, one using an inertial Measu- Geo Journal 24.1, 1991, pp.
rement Unit (IMU), the other, the "Reflectivity Dis- 7-25.
placement Method (RDM)", extracting true forward ve-
locity and Line-of-Sight (LOS) accelerations out of [13] Keydel, W. Verification Using Spaceborne
the radar raw data. This guarantees that the E-SAR Microwave Imaging.
achieves good image quality with high spatial and ra- IEEE, Technology and Society
diometric resolution. Magazine, Dec. 1990/Jan. 1991,
pp. 53-61. i
p 3
9. REFERENCES
[14] Ford, J.P. Seasat Views North America. the
[1] Skolnik, M.I. Radar Handbook. Blom, B. Caribean and Western Europa
McGraw Hill Book Company, 1970. Daily, D. with Imaging Radar.
Elachi, X. JPL-Publication 80-67, 1980.
[2] Hovanessian, S.A. Introduction to Synthetic Array
and Imaging Radars.
Artech House, Inc., 1980.
[3] Tomiyasu, K. Tutorial Review of Synthetic
Aperture Radar (SAR) with Ap-
plications to Imaging of the
Ocean Surface.
Proc. of the IEEE, Vol. 66,
No. 5, May 1970, pp. 563-583.
[4] Kowaly, J.J. Synthetic aperture radar. The

Artech radar library.
(5] Ulaby, F. Microwave Remote Sensing Active

Moore, R.K. and Passive.
Fung, A.K. Vol. II Radar Remote Sensing
and Surface Scattering and
Emission Theory.
Addison-Wesley Publ. Comp. Ad-
vanced Book Program/World
Science Devision Reading, Has.,
1982.
2-1
SAR PECULIARITIES, AMBIGUITIES AND CONSTRAINTS

by
W. Keydel
Deutsche Forschungsanstalt fdr Luft- und R3umfahrt e.V.
Germany
ABSTRACT These facts entail peculiarities of SAR and special

ambiguities which are different from those arising
A synthetic aperture radar (SAR) is basically a with real aperture radar (RAR). The objective of
coherent scatterometer that employs a coherent this paper is to point out the special peculiari-
real aperture radar with highly sophisticated data ties and ambiguities of ZAR in comparison with si-
evaluation and image processing capabilities. The- tieanfamgt RR Bof i copariso ith si-
refore, the coherence of the system is very impor- milar facts of RAR. Basic peculiarities like range
tant te
tant; furthermore, futhemor,
the kypoits or SAR
keypoints for AR are
re data
ata lution and of
dependence signal of
influence noise ratio,
to platform velocity. reso-
azimuthFur-
storage, evaluation and processing. These facts ltor and o
azimuh amig its pur-
azimuth ambiguities,
range and limitations,
thermore, frequency pulse re-
entail
which are fromSAR
different of
peculiarities those arising ambiguities
and special with real petition velocity effects
and phase errors influence SAR-images and cause
aperture radar (RAR). The objective of this paper motion compensation problems.
is to point out the special peculiarities and ambi-
guities of SAR in comparison to the corresponding 2. BASIC SAR-PECULIARITIES IN COMPARISON WITH REAL
properties of RAR. Main topics in this connection APERTURE RADAR
are: basic peculiarities like range dependency of
signal to noise ratio, azimuth resolution and in- For complete incoherent scattering, this means if
fluence of platform velocity. Furthermore, range there are many independent scatterers within the
and azimuth ambiguities, pulse repetition frequen- antenna beam and for a pure statistical phase
cy limitations, velocity effects and phase errors distribution between the scattered signals the
influence on SAR-image that cause motion compensa- performance of radar can be described with a good
tion problems. All these effects will be explained validation by the radar equation. The decisive
together with different contrast-equations between factor is here the signal to noise ratio. For the
the target and clutter signals of SAR and RAR. detection of a point target not only the radar
1. INTRODUCTION signal itself but also its background, the clut-
ter, and the respective contrast between signal
A Synthetic Aperture Radar is essentially a cohe- and clutter is most important. However, each ima-
rent scatterometer or real aperture radar with so- ging radar measures and uses the clutter as prima-
ry signal. Therefore, the clutter to noise ratio,
phisticated data evaluation and image processing. C/N, is for imaging radars decisive, whereas for
Important is the coherence of the system. However, point target estimation the contrast between si-
storage, evaluation and processing of the data are gnal and clutter, S/C as well as the signal to
the key points for SAR. The use of a synthetic noise contrast S/N is essential.
aperture in connection with extremely high range
resolution methods like pulsecompression requires Table 1 shows the different equations for area and
a high degree of coherency and frequency stabili- point-targets and for real and synthetic apertures
ty. This holds also if pulse compression techni- and the respective contrasts. For contrast compu-
ques for scatterometer will be used. However, the tation the losses in all equations have been assu-
combination of synthetic apertures with pulse com- med as identical and equal. However, this is nor-
pression entails requirements for frequency adju- mally not the case and it shall be remarked that
stable oscillators for example and other high so- normally special system losses are dependent on
phisticated components and this makes evident: SAR the special applied techniques; but for simplicity
requires extreme effort not only with respect to it shall be allowed here to take all losses out of
software but also to hardware, consideration in order to reach the simple compa-
Point Target Area Target Contrast
Real Apertur (S/N)reai = (C/N)reai = (S/Clreai =

(RAR) P., G2 12 a 2 3
Pave G 1 o0 • r
y ol
(4")3 R4 (kToF)'B fp 1pLtot (4X) 3 R3 (kToF) fp Ltot °'R'1'rY'B'tP
Synthetic (S/N)sAu =
(C/N)sAR (S/C)sA
Aperture Pav* G2 13 * Pave GX
3
o ry
(SAR) P.,G aP, .r
3 3
(4X)3 R (kT.F)'2ur.Ltot (4x)3 R (kT.F)2uL,., *o ry * r,
Improvement (S/N)sA/(S/N) real (C/N)s R/(C/N)r.i = (S/C)sAi/(S/C),.. =

Factor B.tp X R fp f- 1i_ B.T .R.1
ry 2u r. 1
Table 1 Radar- and contrast equations for real aperture radars and SAR.
2-2
rison and contrast equations in table 1. 1. The maximum-azimuth resolution is independent
(C/N) = received clutter to noise ratio, (S/N) =of wavelength and distance.
received signal to noise ratio, Pav* = average 2. A better azimuth resolution can be reached with
transmitter power, G = antenna gain, I = wave- smaller real antennas and not with larger an-
length, R = distance, a = radar cross section, 00 tennas as it is the case with RAR and optical
= radar cross section per unit area, k = Boltzmann systems respectively.
constant, T. = 290 K, B = receiver bandwidth, F =
receiver noisequeny,
figure,= fp =t~t=
ulseengh,
quency, Vp = pulselength,
pulse repetition
Lto t = loses
losses, u ~-
fre-
u = SAR- 3. The S/N for a SAR is inversely proportional to
tepafr eoiy
velocity, 1 = real azimuth diameter of SAR anten-
na, ry = ground range resolution, ra = prozessed 4. The S/N for point target detection is inversely
azimuth resolution. proportional to the third power of range and
inversely proportional to the prozessed azimuth
Table 1 shows for point and area targets as well relyopr
ra-
the peculiarities of SAR against conventional
dar. It shows evidently the importance of the pul- 5. The S/C for point targets is independent of di-
se repetition frequency PRF and platform velocity. stance R.
For the point target case the dependence of azi-
muth, resolution, bandwidth, pulselength, wave- 3. AMBIGUITIES
length and distance becomes evident while for area
targets the real aperture length plays an impor- Ambiguities play an important part in all radar consi-
tant part. Especially the expressions in the derations. For all pulse-Doppler radars ambiguities
"Improvement-Factor-Line" and in the "Contrast- exist due to the periodical structures of the signals.
Row" show factors, which are most important for a This illuminates the fact, that the pulse repetition
comparison between RAR and SAR. frequency PRF, fp, will be the decisive factor. Howe-
ver, ambiguities can also be responsible for the choi-
In order to increase the understanding of table 1 se of the basic radar frequency. The principal Dopp-
it should be mentioned, that there are basically ler frequency system received with periodically pulsed
two SAR-techniques called the focussed and the un- Doppler radar for moving configurations is shown in
focussed techniques. Fig. 2 schematically.
Table 2 compares the maximum azimuth resolution

and maximum synthetic aperture length for imaging fO 2V f w
radars with real aperture, focussed and unfocussed -
synthetic aperture for a real azimuth antenna- )
length 1. Afif
Real Aperture Radar Sid I.
max. synthetic aperture length 1 -- MMovkw

azimuth resolution IR/I
Unfocussed SAR I-PRF f-to "D I .pfF
max. synthetic aperture length 7(XR) Desed OoppW-FreQncy-Ba .)
azimuth resolution 1/2 F 1-R)
Focussed SAR
max. synthetic aperture length X-R/l f-2PRF f-PRF f t-PRF t.2PRF
azimuth resolution 1/2 Ab.0y Le'o

f,"
Table 2 Comparison of RAR, focussed and unfocussed f-PR AfNf Le.'Mo=,I,"

SAR.
Fig. 1 represents the maximum azimuth resolution f

of different radar systems versus height for a 10 f-Pff
m antenna length 1 typically used for spaceborne
SAR as it is installed on ERS-1 for instance. Azimut Ambiguity. Pulsespectra
...... - Fig. 2 Scheme of a pulse spektrum and of azimuth
, -- I 1 ambiguities caused by PRF.
3.1 BASIC AMBIGUITY CONSIDERATIONS FOR RAR

The distance between main spectral lines corres-
1,
I _P9ponds
/ the unambiguous PRF range
strongly toDoppler and, therefore, holds for
Fig. 1 Azimuth-resolution of radar systems versus - "

height, real antenna length 10 m, incidence Th
angle: 45 . The Doppler frequency ft results from the relative ve-
locity u. between radar and radar target following
the
The following basic peculiarities of a SAR, especially relation fD - 2 u./c f.
of a focussed SAR, in comparison with a conventional This leads to the unambiguous range for the velocity
real aperture radar (RAR) result from table 1:
2-3
2
C fp sampling. With bandwidth B = AD fD follows from
Uu 5 (2) (1) the so-called oversampling ratio
The unambiguous range for the distance measurement fP

Ru., is given by 1 . (7)
c1
R 5 2.
iT . (3) Figure 2 shows the spectral parts coming from the
first and second ambiguous band and contributing to
(This holds for a periodically pulsed radar and for the desired SAR band schematically [5].
small t.)
A combination of (2) and (3) leads to the ambiguous The described ambiguity is mainly PR? conditioned.
product However, a main influence has the antenna pattern
which is partly responsible for the spectrum shape,
C2 1 Fig. 2. Especially improper (that means higher) side-
Uunam unam 5 8 " f (4) lobes enable the SAR to receive power from positive
and negative squint angles, which is within the de-
This equation shows that the choice of frequency for a sired frequency band. These ambiguities are called
pulse doppler radar limits principally the possibility dopler squint angle ambiguities.
for the simultaneous measurement of distance and velo-
city of a radar target. On the other side a frequency Fig. 3, 4, 5 and 6 show the influence of PR? on azi-
limit is fixed if the unambiguous values of Uun. and muth ambiguities. Fig. 3 shows a 2-dimensional re-
R... are given. Example: presentation of a radar image of 5 corner reflectors
and Fig. 4 gives the same image in 3-dimensional
uuna, = 300 m s-', = 40 km, representation taken in X-band with a PRF of 952 kHz.
Fig. 5 and 6 show the same scene taken with a PR? of
frequency requirement: f !51 GHz. 238 kHz. The ambiguities can be seen evidently.
These relations derived for a simple RAR hold princi-

pally for SAR also. The equations show the importance
of the PR? for all radar considerations. In fact, the
PR? is the central radar parameter for SAR. The ap-
pointment of PR? has deep going consequences for the
whole SAR, and the discussion of PR? limitations is
widely identical with the discussion of the ambigui-
ties. Range ambiguities will result if the PR? is too
high, following in principle equation (3). Azimuth am-
biguities will result if the PRF is to low, so that
the reflected signal phase changes by 2s radians ore
more between two successive pulses. However, the SAR
ambiguities are not only controlled by the waveform Fig. 3 S imge of 5 corner reflectors (X-band), PR
(represented by PR? for instance) but also by the an- Fig 3or
tenna pattern.
3.2 AZIMUTH AMBIUITIES FOR SAR
Basic equation for azimuth ambiguity consideration is

equation (1), which gives a lower limit for the PR?,0
and which is identical with the requirement of the ()
(B
sampling theorem. fD is the maximum considered doppler
frequency:
=
fD fD 2u sino - 2u .(5)
u is the platform velocity of the SAR and the angle __ 4
against the antenna mainlobe direction. (Normally the

mainlobe of a SAR antenna is vertical to the velo-
city-vector; however, squint angles are possible but
not considered here for simplicity reasons.) From (1) &ZIMUth
and (5) follows for small antenna beams (P < 300): Fig. 4 3dimensional representation of the SAR image in
Fig. 3, PR? 952 kHz. The contours on top repre-
sent -3 dB values.
4u f .
For the half power beamwidth Phr = 20 and because of

the validity of Ph = A/ follows:
2
u
This equation defines the lower limit of PR?, it im-

plies that the transmitter must be pulsed before the
radar platform moves a distance equal to one half the
real antenna length. The basis for the sampling theo-
rem represented by (1) is the use of ideal low pass
limitation. However, this condition is normally not Fig. 5 SAR image of the corner reflector configuration
fullfilled and this entails ambiguity levels within of Fig. 3 and 4 taken in X-band with 238 kiz
the used doppler band. This level can be minimized by PR?, the azimuth ambiguities clearly can be
increasing the Pr, that means by a so-called over- seen.
-." ... ..
. . .;
.....
...
.. ...
.. .. ._...
i - . ._...._ . "_ _... _...
2-4
3.3 RANGE AMBIGUITIES

In order to guarantee a range unambiguity an equation
similar to (3) holds:
Run 1
Irp+ 2 "-- 5 fp
1
fp 1 Run (
TP + 2--
This equation defines principally an upper limit for

the PRF. However, this condition cannot be fullfilled
-, _b(A) Fig. 7a SAR image of ZUrich airport disturbed by si-

delobe ambiguities caused by an antenna squint
angle of 6.80 (Phr = 12*); near range 2805 m,
-
far range 5355 m, u = 70 m s 1, X-VV (DLR-
E-SAR measurement Horn, Moreira).
azimuth [mJ
Fig. 6 3dimensional SAR image of Fig. 5. The con- -t
tours on top represent the 3 dB values for
the mainlobes and the -20 dB values of the
azimuthal ambiguity sidelobes.
at any time, it can be incompatible with the condition
for the azimuth ambiguity and in this case one makes
allowance for range ambiguities and tries to suppress
these ambiguous signals with proper antenna pattern
design or with special processing procedures respecti-
vely. Fig. 7a shows exemplarically an X-band SAR image
with ambiguities. The ambiguities are eliminated in
Fig. 7b. Fig. 7b SAR image of Fig. 7a, correctly prozessed to
4. PRF-CONSIDERATIONS the squint angle 6.80; all ambiguities are
eliminated (measurement Horn, Moreira).
4 ~~~sive
It hasfactor
been already mentioned, that
for SAR-ambiguities. the PRFfrom
However, is a deci- viewing
other geometry is fixed by other reasons (given or-
viwngemtysfxdbyohreans(vno-
sienftofor SAN-aiguis.deep owever omus otr bit height of a satellite, fixed incidence angle etc.)
point of view the
the
ofaPRF hasaso.
ffetivs
SR deep going consequences
he efiitio ofPRFfor tePFms etndt
the PRF must be eetdvle.Fg
tuned to selected hw
values. Fig. 8 shows
the effectives of a SAN also. The definition of PRF allowed PRF-bands for a satelliteborne system with a
becomes difficult due
andwich to different other conditions
aveto b fulfiled.fixed depression angle, taking into account the varia-
imiatios
tion of the local orbit height [7]. The white ranges
are the allowed PRF-ranges, m is the number of the
There are principal limitations due to azimuth ambi- respective range ambiguities. The lower limit for the
guities, range ambiguities, swathwidth, complete co- PRF with respect to azimuth ambiguity is also given in
verage. There are also unallowed PRP-bands due to geo- Fig. 8.
metrical variations like earth curvature, orbit ex-
centricity, height variations, and altitude line The result is:
echos.
The ambiguity equations (6) and (8) lead to the follo- - with increasing range ambiguity decreases the re-
wing m iiteations nspective width of the allowed PRF-bands.
This can lead to the request of switchable PRF for
2u 1 different purposes (shown in Fig. 9 as necessary due
S Run "to ALE-influences). Substituting (9) into (6) (consi-
+ - dering the equality in both relations) one obtaines
together with the maximum azimuth resolution relation
The choice of PRF estimates the maximum swath width R9 an important relation between swathwidth, azimuth re-
or vice versa. Principally, an impulse needs for cros- solution and SAR velocity:
sing the swath the time i = 2R.1c. Herefrom results:
2u P,
(9 -r, = c . (10)
f1 2R9
..
--
9
A pulsed radar using a single antenna is normally un-
able to receive during the duration of the transmissi-
on pulse, it is blind at certain slant ranges. If the
2-5
i
r ~platform.
PR
z
PRF-Bands Satellite X-Sandpltom
S For a pulsed radar ALE considerations will be identi-
s0o cal with altimeter-considerations. An example for PRF-
4limitation due to ALE shows Fig. 9 in comparison with
-Fig. 8. For m = 12, the influence of ALE leads to the
request of PRF-switching as mentioned in the previous
2500 section.
5. ANTENNA PECULIARITIES
rn.l6 Equation (6) gives a relation between the real azimuth
2000 diameter of the SAR-antenna, flight-velocity and the
1500. '-m -12 2u !
-m1 A minimum vertical dimension of the antenna 1vrt will

-
!be estimated due to the necessity to focus the beam
1000 AZut Ambim 1tv into the swath width Rg. From Fig. 2 in lesson 1 and
- Ain connection with (10) the following relation can be
-M-S derived for a small half power beamwidth:
70 0 . - -
..........
1 20 30 40 50 60 70 80 90 RP R - X
Ah/km - = inODi - sini lyert (Ii)
Fig. 8 Allowed PRF-bands versus altitude variations From (6) and (10) follows in connection with (1.1):
for a satellite, X-band.
c sinb i 5 lvert laz " (12)
SATELLITE
X-Band Equation (12) gives the lower bound for the antenna
area.
SPRF 6. PECULIARITIES AND CONSTRAINTS CAUSED BY PHASE ERRORS
m-14 200 Z Essential for the effectivity of a SAR system is its
210 phase coherency. High quality images can be produced
only if the targets phase history is observed along a
M-13[ precisely known radar translation. However, these pha-
o se history is normally not known exactly. Phase errors
917M usually occur. Principally there are two different ty-
m-12 pes of phase errors:
1 More or less deterministic errors, caused by well
m-11 known geometry effects and defined instrumental in-
fluences.
M-10 Statistical errors caused by instabilities of the ra-

dar itself, of the platform or of the propagation path
m-e Ifor example.
130 H6.1 DETERMINISTIC GEOMETRY EFFECTS, RANGE CURVATURE
SAmftgmity-nift 1Ig4H In lesson 1, Cap. 3.2 the "Depth of Focus" has been
l MW =considered and in Fig. 3 an example for the influence
XnI ,' of defocussing on a SAR image has been shown there.
-20 -10 10 20 The same considerations can be used for the computa-
LAh/km -- w tion of the so-called "Range Curvature Effect". The
range curvature (RC) is given by
Fig. 9 PRF limitations due to altitude variations
Ah, example for PRF switching. RC = - . (13)
*The Altitude Line Echo
For high azimuth resolution application a long in-
The allready mentioned altitude line echo (ALE) is the tegration time is required and image degradation may
radar signal coming from Nadir. The time duration of be caused if the time delay variation corresponding
ALE is with sufficient accuracy identical with the to (13) reaches the same order or exceeds the range
2
transmitter pulse duration %p. ALE appears earlier resolution cell (Cp/ ) for a pulse radar. From (13)
than the desired swath echo. On the one side it will follows
more attenuated than the desired signal proportio- )2
nal to the sidelobe level of the antenna pointing in Cu TD)2
(
Nadir direction and on the other side it will be less RC - 2 (14)
attenuated due to the shorter distance, where ALE is
resulting from. o has normally a higher value for Na- 12 R
dir direction than in any other directions and this RC - - . (15)
will increare the ALE-signal level. Principally, the 16 r.
ALE can be used as a reference for geometrical cali-
bration as well as for altitude estimation of the
2-6
By this the importance of ra is evident. In compari- Other so-called deterministic phase changes can occur
son to ry follows: due to orbit excentricity of satellites and each rota-
X2 Rtion effect, which can normally be estimated exactly
R (16) or due to antenna influences like diagramme deforma-
ry 16 ry r2 tion or angle switching. Both effects can be measured
and estimated exactly.
This equation gives the number of resolution cells
through which a point migrates during the formation 6.2 PLATFORM INSTABILITY EFFECTS
of the synthetic aperture. Ratio values in excess
of 0.3 will normally cause image degradation and Fig. 11 shows measured displacements in the line of
this must either be avoided by design or compensa- sight of an airborne SAR, extracted by a RDM motion
ted during signal processing (2]. compensation equipment.
In Fig. 10a the range curvature clearly can be seen.
However, here the curvature is less enough to avoid
image degradation.
Fig. 10b shows the influence of a range curvature ef-

fect together with ambiguities in the sidelobes on
the pulse answer of a point target. Here, a degrada-
tion clearly can be seen.
processed image
range Fig. 11 SAR platform line of sight displacements

compressed extraction from SAR raw data by RDM (DLR-
raw data E-SAR, L-band, measurement Horn, Moreira).
Fig. 12 shows as an example the degradation of a

point target- imppulse response due to turbulences
azimuth > up to 2 m s 1 (RMS 1 m s - 1) for an airborne SAR.
"
Sideal impuls
-20
. raw da 'a _40 ! d
SAR
raw data -40
A 0
4 01
-0.5 0-. 0.5 15
Tim [,]
Fiq. 10a The raw data received from a point target
(lower image) show as well as the range com- Fig. 12 C-band impulse response of a point target
pressed data the influence of range curva- for DLR-E-SAR degraded by phase errors due
ture. to- displacements and turbulence (RMS = 1 m
s ' , peak value 2 m s - 1, displacements be-
tween -2 m upto 4.5 m).
AIn principle these effects can be neutralized
by
motion compensation by measuring the dynamic beha-
---viour of the platform and correcting the SAR data
0() eighter on board of the aircraft or on ground. Two
methods are principally possible, eighter the mea-
a 0surement using an Inertial Measurement Unit (IMU)
ore a procedure, which extracts the true forward
velocity and line of sight changes and accelera-
tion out of raw data called 'Reflectivity Displa-
cement Method" RDM. RDM, recently has been deve-
S/ loped at DLR [9, 10). This procedure has been used
to exclude the errors caused by aircraft motion and
velocity instabilities during the processing of the
images shown here. In the DLR approach only the in-
formation in the radar signal has been used for mo-
Fig. 10b Impulse response of a point target disturbed tion compensation; principally no intertial platform
due to range migration caused by range curva- or other equipment for measuring details of the
ture. The sidelobes are caused by ambiguity. aircraft motion is necessary. This method enables
The 2dimensional representation on top shows the estimation of deviations of the aircraft from
evidently a smearing effect (DLR-E-SAR, Horn, an ideal path as shown in Fig. 11.
_Moreira)
2-7
Obviously, phase errors caused by atmospheric and PLO O-* F5 R

ionospheric turbulences in principle cause the P PSG
4
same motion errors as the aircraft instabilities.
Therefore, the same method can be applied for the - , .... ,* . .
correction of phase errors due to propagation. This " " -
implies that for spaceborne SAR the negative in-
fluence of atmospheric and ionospheric turbulences
to a large extend can be neutralised when they
occur [10]. This results in an increase of the reso-
lution power and in image quality. Therefore, it
seems in principle that for spaceborne SAR a dm
resolution can be reached. However, special tech-
niques are required in order to fulfill power and
data requirements in this case.
The tolerable platform velocity variation Au can
be computed from (18) in lesson 1. The radar range
R at the ends of a synthetic aperture (aft, the
dwelltime TO) is:
uTD )2
R : R2 + (_-
0 2
The differentiation with respect to u delivers

uT
dR _ __ AR
du 4R Au'
AU =4. aR . Fig. 14 Example for errors in SAR images caused by

u T2 phase distortions and the effect of motion
compensation. The upper image is distorted
Fcr a tolerable range error of 1116 follows a tolerab- due to a 12 % velocity variation, the lower
le velocity component Au (for R - Ro): image shows the same scene after a motion
= 4 R u r2 compensation. Scene dimensions are 1254 m x
AU -- o (17) 1344 m, ground resolution 3 m x 3 m (measure-
u T= 6 7
R ( ment Moreira, Horn, DI.R-SAR).
Remarkable is: The tolerable au becomes more critical for a moving target at a distance R the displacement
if the azimuth resolution is improved. Ax
Fig. 13 gives the measured forward velocity variations Utr

of an airborne SAR during a flight time of about 60 Ax = ± R- sint . (18)
seconds.
The sign in (18) depends on the direction of the
target velocity with respect to the radar. An exam-
ple for image shift due to the radial component of
target motion is shown in Fig. 15. Ships with veloci-
ty components radial to the Seasat SAR orbit plane
have an image displacement. This effect can be used
for the estimation of target velocities if the di-
stance between the SAR and the targets as well as the
velocity of the SAR is known.
Fig. 13 Changes in forward velocity of E-SAR extracted

from DLR-E-SAR raw data, L-band, by RDM (mea-
surement Horn, Moreira).
Fig. 14 shows as an example errors in a SAR image due

to a 12 % velocity variation and its corrections with
a motion compensation procedure.
6.3 TARGET MOTION EFFECTS
Usually a SAR signal processor locates the position of ig. 15 SAR image of a sea surface with 2 ships taken
a non-moving target in the image place where its dopp- by SEASAT. The displacement of the ships
ler frequency is zero. If the target is moving with a against the satellite clearly can be seen. The
radial velocity component Utr, then it imposes a Dopp- opposite displacement against the wakes cor-
ler shift on the signal. Thus a target moving towards respondings to opposite velocity directions
the radar will cause a shift Ax in target location in (processing: DLR-T-DA). Mean displacement is
the flight direction of the radar (and vice versa).
if the radar is moving with a velocity U than results In addition to radial velocity, radial acceleration
will also cause image distortions such as defocus in
2-8
Like other electromagnetic systems, radars employ ra-

azimuth and range, range walk etc. owever, the most diation with polarization, the alignment (relative to
sensitive effect is azimuth defocus [3]. vertical or horizontal) of the electric vector in the
A real linear array is to be considered as flat as wave. Polarization on both transmit and receive is de-
Aorea lineadearaytistbeonsmasri d al aas termined by the antenna. Scattering objects, such as
long as the deviations from a straight line are terrain features or hard targets such as vehicles,
less than a half wavelength. For a synthetic array have radar cross sections that reflect differently in
this for the two way case must be halved due to the
phase coherency conditions. Therefore, for the line response to the incident polarization.
of sight deviation holds: AR 5 X/4. An acceleration In principle a complete description of a radar target
d2 R/dt2 acting for the whole dwelltime T
2 D corres-
2
ponds to a radial position error AR = d R/dt (if a can be given only if all like polarized and cross-po-
case constant acceleration is assumed). In the two larized amplitudes
radar signal and the
are known. Suchrespective phases radar
a polarimetric of the
gi-
way this coefficient also must be halved. From ra sile knfon u ap argetri rag
theseyes all possible information on a target within the
fromlesonid n rrelative small bandwidth of the modulated radar car-
rier frequency.
At present, technological limitations in the efficien-
d2 R 5 2u2 r. cy of radio frequency power generators are important
= (19) design factors. The "radar equation" shows an increase
dt2 2T2 R2 .X of the average power necessary for higher frequency
This is the maximum tolerable acceleration a target ranges. The dependence is linear: this means that an
may have in line of sight direction of a SAR or the X-band SAR needs about twice as much power as a C-
radar platform may show vice versa. (19) is a basic band SAR, and about 7 times as much as an L-band SAR.
equation for motion compensation. For turbulence di- In addition, other technological difficulties increase
sturbances in Fig. 12 mainly acceleration errors are with frequency. Phased array antennas (desirable for
responsible . beam steering) are much easier to realize in lower
than in higher frequency bands. In L- and C-band a
In Table 3 several sources of phase errors are listed realization with microstrip technology is state of
together with the respective errors caused within SAR the art, whereas in X-band a realization of a large
images. The Figures 16a to 16d show as examples the microstrip antenna with sufficient efficiency for
influence of different phase errors on the impuls re- space applications seems to be extremely difficult.
inlueofofa point
oiffnt ae eState of the art for X-band use
arrays [11, 12]. This is the mainslotted
sponse target, waveguide
reason for the sing-
7. REALIZATION CONSIDERATIONS le polarization of X-SAR against the multipolarization

capability of SIR-C.
Atmosphere and ionosphere produce frequency dependent
distortions.
mit These effects
due to attenuation set an upper
for airborne radarfrequency
(about 90li- from
f h which
iche
a aradar oeratin
radar observation to a etai tage
certain target
mit adutoatenationforaborne radar (about andacan 90 be successfully made and, therefore, the orbit
GHz) and spaceborne radar (about 15 Gfiz)and a lower, atudofscernsyemispwrlie.
limit for spaceborne SAR due to ionospheric granulari-
ties (about 1 GHz). The state of technology sets upper
limits as well. The fsequency bands available for ra-
dar surveillance are also limited due to internatio-
nal agreements.
Error Source Image Image Azimuth Range Range Azimuth Image Image
Misregi- shift Defocus Defocus Walk Mainloge Sidelobe
stration Loss Increase
Range r !
(altitude t
and cross r
track) dr
Azimuth x
(along ••
track) x
dtx
Propagation 1.1
path j 1
If1
'
Electronic ___1____ ___

stability
Table 3 Effects of phase errors,caused by platform or target motions respectively (the relative mo-
tion is from relevance only), as well as caused by phase jitter A9 in the propagation path
and in the SAR electronic.
2-9
0 0,
-40 -20
-5 - 5
-60 -60
-70i -70
-0.6 -0.4 -0.2 0 0.2 0.4 0.6 -0.6 -0.4 -0.2 0 0.2 0.4 0.6
Tim(] Tim.[.]
Fig. 16a Impulse response, shifted by a linear Fig. 16c Impulse response, degraded by a cubic
phase error, phase error.
0 0
-10 -10-
-20 -20 -
S-30 !.-30
40I -40
-50 -30
-60 -60
-70 -70
-0.6 -0.4 -0.2 0 0.2 0.4 0.6 -0.6 -0.4 -0.2 0 0.2 0.4 0.6
Fig. 16b Impulse response, degraded by a quadratic Fig. 16d Impulse response, degraded by a random high
phase error. frequency phase errors.
-
Principally, the power required increases with the capacity. A reasonable goal would be 250 Mbit s 1.
3rd power of the radar distance. (This means a doub- New data transmission systems with splitted data
ling of the distance, i.e. the orbit height, requires links to data relay satellites or ground stations
an 8-fold multiplication of the required power). A respectively are under preparation.
larger antenna beams more power to a required area
(expressed through the so-called antenna gain) than Data storage capability has to be increased also.
a small one, and, therefore, a large antenna would As a first step new recorders with capabilities
seem to be favourable. But the ground resolution of exceeding 100 Mbit s- 1 are qualified for use in
a SAR is improved for smaller real antennas. These space. The present approach is to use more than
considerations lead to a tradeoff, which must be ma- one recorder, i.e. one recorder for each channel
de carefully in order to fulfill a satellite SAR's in multipolarization and multifrequency SAR, as
requirements. In any case the transmitter power is a on the SIR-C mission.
limiting element for the design of a SAR as well as
the antenna. Todays state of the art are a few hun- Image data processing capabilities also are li-
dreds watt (mean power). This implies requirements mited. However, this processing time will de-
for the primary power supply of a satellite. State crease rapidly during the next few years and at
of the art is about 6 kW to 10 kW. These require- the end of the decade real time processing with
ments can be fulfilled with solar power generators excellent image quality will be possible.
and atomic generators as well. The antenna dimen-
sions for spaceborne SAR have at present values of Requirements for high resolution and wide swath
about 2 m x 15 m. in continous strip map mode are in conflict with
data handling capabilities, and with require-
8. FUTURE SAR TECHNIQUES ments for a large antenna to conserve power.
Therefore, new SAR techniques have to be intro-
SAR sytems produce a tremendous amount of data duced and developed which allow electronically
(examples for data-recorder bitrates on the ground: steered beams. For this purpose the spotlight
ERS-1 n1 02 MBPS, X-SAR/SIR-C G 45 MBPS per chan- mode and Scan-SAR modes are under consideration.
nel M,)DLR airborne E-SARZ 28 MBPS) Requirements The length of a synthetic antenna corresponds
2 ") to the section of the flight path from which
for simultaneous high resolution and large swath one target stays within the antenna beam. This
widths make the data rates higher. All requirements fact leads to the requirement for wide beams
for extensions of SAR to multifrequency and multipola- and, therefore, small antennas for conventional
rization capability respectively entail a multiplica- high resolution systems. The same effect, how-
tion of the data rates and this would exceed the pre- ever, can be reached if a small antenna beam
sent limitations of data handling. This seems to be a (from a large antenna) can be continuously
key problem in all high resolution imaging systems. pointed at the target. This allows also a lon-
Therefore, different requirements have to be fullfil- ger synthetic array and, therefore, a finer
led in order to handle or reduce the data stream of azimuth resolution [11]. However, the gain of
future systems either by means of onboard processing azimuth resolution entails a loss of coverage
or with development of advanced SAR systems like spot- due to the fact that during the continuous
light SAR. One solution is to increase the data links spotlight illumination of one small surface
T2-10
area, the sensor passes other parts of the German Patent P 41 24 062.6.
swath which are not illuminated. Therefore,
the spotlight mode can be used for the enlar- [11] Buckreu8, S. Motion Errors in an Airborne
gement of a sector of the observed swath similar Synthetic Aperture Radar Sy-
to the zooming with an optical camera. The Scan stem.
SAR Mode can be used for an extension of the ETT Vol. 2, No. 6, Nov/Dec.
swath in radial direction using more than one 1991, pp. 655-664.
beam generated in a time shared manner.
method increases the swath at the expenseThis
of [12] Keydel, W. Verification Using Spaceborne
azimuth resolution [16 to 20]. A spotlight SAR Microwave Imaging.
in orbit, however, would be a rather expensive IEEE Technology and Society Ma-
enterprise. gazine, Dec. 1990/Jan. 1991,
The combination of all modes allows variable pp. 53-61.
resolution and swath widths as well; the trade- [13] Horn, R. C-Band SAR Results Obtained by
off between resolution, swath width, power etc. an Experimental Airborne SAR
leads to optimised configurations. Sensor.
9. REFERENCES Proc. IGARSS 89, IEEE, pp.
2213-2216.
[1] Skolnik, M.I. Radar Handbook. [14] Moreira, A. Improved Multilook Techniques
McGraw Hill Book Comp., New Applied to SAR and ZCANSAR
York, 1970. Imagery.
[2] Hovanessian, S.A. IEEE Trans. on Geosc. and Re-
Introduction to Synthetic mote Sensing, Vol. 29, No. 4,
Array and Imaging Radars. July 1991.
[15] Moreira, A. A New Subaperture Approach for
[3] Tomiyasu, K. Tutorial Review of Synthetic Real-Time SAR Processing.
Aperture Radar (SAR) with Ap- ETT, Vol. 2, No. 6, Nov.-Dec.
plications to Imaging of the 1991.
Oceans Surface.
Proc. of the IEEE Vol. 66, No. [16] Brunner, A. Concept for a Spaceborne Syn-
5, May 1978. Langer, E. thetic Aperture Radar (SAR)
[4] Kovaly, J.J. Ottl, H. Sensor Based on Active Phased
Synthetic Aperture Radar. The Zeller, K. Array Technology.
Artech Radar Library. AGARD Conf. Proc. No. 459 on
Artech House, Inc., 1976 High Resolution Air- and Spa-
[5] Ulaby, F.T. Microwave Remote Sensing Vol. ceborne Radar, Papers presen-
ted at the Avionics Panel Sym-
Moore, R.K. II, Radar Remote Sensing and posium held in The Hague, The
Fung, A.K. Surface Scattering and Emis- Netherlands, 8-12 Aug. 1989,
sion Theory. pp. 23 Al - 23 A10.
Addison Wesley Publ. Comp.,
1982. [17] Jatsch, W. Concept of an X-Band Synthetic
[6] Elachi, Ch. Spaceborne Synthetic Aperture Langer, E. Aperture Radar for Earth Obser-
Ottl, H. ving Satellites.
Bicknell, T. Imaging Radars: Applications, Journ. of Electromagnetic Waves
Jordan, R.L. Techniques and Technology. and Applications, May 1990.
Wu, Ch. Proc. IEEE Vol. 70, No. 10,
Oct. 1982, pp. 1174-1209. [18] Luscombe, A. Taking a Broader View: Radarsat
[7] Schlude, F. Imaging Radar Systems. adds Scansar to its Operations.
Proc. IGARSS '88 Symp., Eding
Proc. of an ESA-EARSeL Workshop burgh, Scotland, 13-16 Sept.
held at Alpbach, Austria, 16-20 1988, pp. 1027-1032.
March 1981 on "Coherent and In-
coherent Radar Scattering from [19] Raney, K. Canada's RADARSAT.
Rough Surfaces and Vegetated Remote Sensing Yearbook 1990,
Areas", ESA-SP-166. (personal communication).
[8] Moreira, J. Estimating the Residual Error [20] Luscombe, A.P. The Radarsat Synthetic Aperture
of the Reflectivity Displace- Radar: A Flexible Imaging Sy-
ment Method for Aircraft Mo- stem.
tion Error Extraction from SAR Proc. 11. Canadian Symp. on Re-
Raw Data. mote Sensing, Waterloo,
ICCC Intern. Radar Conf., Ar- 22.-25.6.1987.
lington, USA, 1990, pp. 70-75.
[9] Moreira J. A New Method of Aircraft Motion
Error Extraction from Radar Raw
Data for Real Time SAR Motion
Compensation.
12th Canadian Symposium on Re-
mote Sensing, IGARSS '89, Van-
couver, Canada, proc. IGARSS,
IEEE, 89, pp. 2217-2220.
[10] Keydel, . Verfahren zur Extraktion von
Moreira, J. durch die Atmosphire verursach-
ten Phasenfehlern des R~ck-
streusignals eines Abbildungs-
radarsystems aus Radarrohdaten.
3-1
MOTION ERRORS AND COMPENSATION POSSIBILITIES

by
D. Hounam
Deutsche Forschungsanstalt fOr Luft- und Raumfahrt e.V.
Institut fdr Hochfrequenztechnik
Germany
1. SUMMARY along-track position and spatial resolution will

not be influenced, by attitude errors. However, if
The synthetic aperture radar (SAR) technique re- the antenna beam points to a different part of the
lies on knowledge of the relative motion between Doppler spectrum than that which is processed
the sensor and the target. If the flight path of (filtered), the image intensity will be influen-
the sensor is not accurately known or the SAR pro- ced. With serious mismatch, defocusing will occur, j
cessor is limited in its ability to take the due to distortion of the matched filter weighting
flight data into account, the SAR image will be function and errors in the range migration correc-
degraded. Motion errors are particularly critical tion. Also ambiguous responses can occur. Antenna
for SAR sensors on small, low-flying aircraft, due pointing needs to be known to process the SAR da-
to turbulence, and where high spatial resolution ta, or the processor has to derive it by analysing
is required. the Doppler spectrum. The latter can be considered
normal processing practice and, therefore, in the
The lecture discusses the effects of motion errors following, these effects will be neglected.
on image quality and the requirements on the sen-
sor and processor to compensate for motion errors. Whereas, with the above reservations, the focusing
The DLR airborne sensor, E-SAR, and associated of a SAR is insensitive to angular deviations, de-
image processor will be used as examples. Techni- viations in the path of the sensor will lead to
ques using a priori knowledge of the flight path displacement and defocusing of the target in the
from independent sensors, e.g. inertial navigation final image. The path of the sensor does not need
systems (INS), and by extracting the flight data to be a straight line but it must be known with
from the SAR data, e.g autofocus and reflectivity sufficient accuracy so that deviations can be ta-
displacement method (RDM), are treated, ken into account in the processing of the SAR data
to achieve a sharp image. This process is often
The author would particularly like to thank J. Mo- referred to as motion compensation. I
reira and S. Buckreu8 whose work, referenced be-
low, was used extensively for the lecture. When discussing the path of a sensor, the question
arises as to where the reference point within the
2. INTRODUCTION sensor is located. Displacements of the sensor are
clearly only effective on the propagation path of
Every amateur photographer is familiar with the the radar signal. Hence, the reference point wi-
blurring of his photograph if the camera is not thin the sensor is the phase centre of the anten-
held steady while the film is being exposed. The na, i.e. origin of the spherical far-field wave-
action of the lens focusing the scene on photogra- front. The phase centre is the point from which
phic film results in the scene being resolved in the antenna effectively radiates. For many antenna
angular units. Rotation of the camera in any plane types, the phase centre is outside the structure
will cause the scene to move across the film and of the antenna, e.g dish antennas, where it is lo-
blur th resulting image. cated behind the dish.
In contrast to this well known effect, the focu- Hence, by the path of the sensor is understood the
sing of a SAR sensor is in principle insensitive path of the phase centre of the antenna. If the
to angular deviations. This is because the sensor path is not measured directly at the phase centre,
resolves the scene in terms of displacements ra- roll, pitch and yaw angles need to be known to
ther than angles. correct for the offset. Errors in these angles
will then lead to degradation in image quality,
In the plane orthogonal to the flight path, i.e. further contradicting the generalisation above
in the range direction, the position of a target that the focusing of the SAR is insensitive to an-
is determined by measuring the delay of the tar- gular variations. For the sake of simplicity, the-
get's echo (range delay). The spatial resolution se angular effects have been neglected below. If
is determined by the length and bandwidth of the needed their influence can be derived from the
transmitted pulse. If the sensor is rotated in formulas provided.
this plane (roll angle), the influence of the an-
tenna gain pattern may cause the strength of the Clearly, a stable flight path greatly eases the
echo, i.e. the image intensity, to vary but the task of motion compensation. Satellite platforms
delay and, hence the spatial resolution will re- fall into this category even though in some cases,
main unaffected. e.g. space shuttle the orbital parameters as well
as the attitude can be poorly defined. The real
Parallel to the flight path, i.e. in the azimuth challenge for motion compensation is presented
direction, the scene is resolved by matched filte- when SAR sensors are flown on light aircraft due
ring the Doppler spectrum, there being a fixed to the influence of air turbulence and the often
correspondence between Doppler frequency and the frantic effort of the pilot to combat it. The air-
relative position of the target on the ground. The borne SAR sensor of the DLR (ESAR) is such a case
spatial resolution is determined by the spacing of [1] and will be used as an example in the follo-
the lines of constant Doppler and the bandwidth of wing analysis. Fig. 1 shows an image of the DLR
the matched filter. Rotation in azimuth produced centre in Oberpfaffenhofen using this sensor which
by pitch and yaw will have no influence on the li- exhibits typical degradations due to motion er-
nes of constant Doppler nor on the matched filter rors. The aircraft was deliberately flown to pro-
characteristics. In principle, therefore, the duce motion errors which can be seen as blurring
of the image (top right) and geometric distortion
3-2
of the runway and taxiways.
In the following analysis, the influence of motion

errors on the azimuth imaging properties will be
discussed first as this is far greater than the
influence in range. The analysis closely follows
the approach of BuckreuB [2]. These properties are
discussed in terms of the impulse response, which
is the response of the SAR sensor, including pro-
cessing, to a point target.
%
D
P$
Fig. 2 SAR geometry, where P is a target on the

ground, H is the altitude of the sensor,
D is the distance of the target from the
ground track, Ro is the slant range at
closest approach, OA is the azimuth angle
and 9D is the depression angle of the tar-
get.
distance between the target and the ground track.
Alternatively,
OA(t) = L (5)
L
• where Ro is the range at the point of closest ap-
proach of the sensor of the target.
Integrating and defining t = 0 when the sensor is

Fig. 1 SAR image of the DLR centre in Oberpfaf- at the point of closest approach, we obtain the
fenhofen exhibiting strong motion errors. slant range r:
2.1 Impulse Response in the Azimuth Direction
V2 t 2
Let us consider first the azimuth channel of a SAR r(t) = +o R- (6)
system, where the platform flies in a straight li-
ne (Fig. 2). The coherent integration can be ex- Ignoring the constant phase expression, the two
pressed by the convolution of the backscattered
signal S(t), with a reference function H(t), where
way phase change q(t), mentioned in Eq. (2) beco- I
mes:
Z(t) is the impulse response. 9() V2 t2
Z)
(t) " 2-- where X is the (7)
Iz(t)• = Sit) * H(t)I (1) X 2R 0 radar wavelength.
The returned signal S(t), corrupted by a phase er- Or,
ror 0(t) is given by
kt2 4z.V2
j (t ) t(t) - - t , where k = ---- . (8)
S t) = A o e " (2)
Hence, the nominal phase modulation
is a quadratic
function of time, i.e. linear frequency modula-
were A0 is the amplitude of the returned radar si- tion.
gnal and 9(t) is the nominal phase history of a
point target. According to the geometry (see Fig. Let us now consider deviations from the nominal
2) the velocity in line of sight (LS), is: path where the platform is displaced from the no-
minal position A to the position A' (see Fig. 3).
VLos(t) = V-sinOA(t) (3) The range r' becomes:
where V is the forward velocity of the aircraft ') t) 0

and 9A is ther angle between the LOS vector and r' t) - R0 + + 1x(t) A
cross-track plane (azimuth angle). For small va- y ct
lues of A, we can put: AY(t)cosb + az(t)sine,
where ax(t) is the displacement in the flight di-
A(t) = -. , (4) rection, ay(t) and Az(t) are the displacements of
7-7 the platform in y and z direction, respectively,
where H is the height of the sensor and D is the and % is the depression angle of the target.
L___ ii__ mmi

3-3
the phase error 4(t).
window
If the weighting function is a rectangular
W(t)= WR(t), with
WR(t) = 1 for Itl ! T/2 and
(16)
A' A WR(t) = 0 for Itl > T/2
ed and 0(t) = 0, the ideal impulse response [Zo(t) l

becomes:
1Z0 (t)I = Ao' T• ktT/2

sinktT/2) j 17
(17)
It will be seen below that the phase error func-

tion 0(t) caused by motion errors can lead to con-
siderable distortion of the azimuth impulse res-
ponse.
2.2 Impulse Response in the Range Direction
Differentiating, Eq. (10) we can obtain expressi-

ons for the angular frequency error 0(t), due to
velocity components along the three axes.
Fig. 3 Deviation from the nominal flight path, due 4*V r 1
to turbulences. OX(t) =- " • x + t.Ax(t)J , (18)
The phase error 0(t) is, therefore: iy(t) = 4-s cosD •y(t) (
' (19)
0(t) = - ax(t)'@A(t) - Ay(t) 'cose + -X
+ Az(t)sinD]. (10) - sin " (20)
From equation (10) it can be seen that the in- The first term in the brackets of Eq.(17) repre-
fluence of the motion errors in the x direction is sents the frequency error resulting from the di-
independent of the depression angle OD and, there- splacement in flight direction. The second term
fore, the the
in theofother
the position
swath, ofWhereas, two directions (y the
target within and shows aerror
quency relationship
linearand velocity component, it beingfre-
between angular pro-
z), OD needs to be known. Also, it is evident that portional V-t/Ro, i.e. the azimuth angle @A. This
the variations of the displacements with respect angle is the sum of the so-called squint angle,
the i.e the angular offset of the boresight of the an-
to time in all three directions is reproduced in tenna, and half the azimuth antenna beamwidth.
the phase error, a linear variation in the displa- Eq.(19) and Eq. (20) also show a linear relation-
cement
and so manifesting
on. itself as a linear phase error ship b
ship n frequency
between Equency error
eror and v elt compo-
an velocity on-
and s on.nent.
Assuming that the velocities are constant, we can
The normalized reference function H(t) from Eq.(1)
is given by: now calculate the phase error 0. across the trans-
2
ikt mit pulse length %P;
H(t) = W(t)-T f'.e2 , -T2 t +T/1 , (11)
*px = *x'cp , *py, = Sy'p , p2 = z'¢p (21)
and W(t) is a weigh-
where T is the aperture time
ting function to suppress sidelobes. Often a Ham- Let us now calculate some practical values for the
ming weighting function (4] is used: velocity components and the displacement in flight
direction, which would each result in a linear
WM(t) = a + (l-a)-cosIt a = 0.54 (12) phase error of x/2, a value which would need to be
LT.achieved to influence the impulse response. The
results are given in Table 2. We will use the pa-
Using equations (1), (2), and (11), the impulse rameters for a typical satellite sensor (ERS-l)
response of a point target becomes: and airborne sensor (E-SAR) given in Table 1.
Z(t) = S(t) * H(t) = f S(t)'H(t-i) diT-SA

. (13) ERS-l E-SAR
Or, Velocity V 7100 75 m/s

2
+- -(k! Closest approach R, 850 4 km
Z~t)
A0 e2 .eje() N(t-11
Z A l(t-) (14) Wavelength X 0.0566 0.0566 m
2 Depression angle %b 20.55 45 deg

Taking the absolute value of Z(t), we approximate- Max.azimuth angle amax 0.013 0.14 rad
ly obtain:
" Pulse length 37.1 5

IZ(t)l - Ao . e
-I Table I Sensor parameters.
This can also be considered as the Fourier trans-
form of the weighting function W(t) modulated by
3-4
ERS-l E-SAR (t) = (0) + v(0)t + a(0)t2 +
Ax -22830 -75467 m +1
+ 1 a(0)t
a(0)t+3
+ ..3 (24)
x -9339 -6434 m/s
where p(0) is the position error, v(0)
the veloci-
130 1274 m/s ty error, a(O) is the acceleration error and A(0)
is the derivative of the acceleration error at the
beginning of the synthetic aperture.
z -346 -1274 m/s
2IMPULSE RESPONSE
Table 2 Displacement in flight direction

and velocities resulting in a phase0
error across the pulse of x/2. T 1
The values in Table 2 show that the displacement -10
and velocities necessary to influence the range
impulse response are far higher than achievable -15
with both satellite and aircraft platforms. Hence, - 2
-20
~ .- _
motion errors can be considered to only influence
the azimuth impulse response, and this will be -25 - -
concentrated on below.
7 -30 - 4-
3. CLASSIFICATION OF PHASE ERRORS -5-
Let us consider a sinussoidal phase error $(t)= -40
socos(2xfot-yo) with constant frequency fo and
constant phase shift yo. The impulse response of a 45 1
point target from Eq.(15) becomes: -50
IZ(t) - A, 27 (22)I
20 15 10 -s5 10 15 20
2 ) - jkt Azmuthm I
x W(,) - ejobcos( xfo .e rd .
-- Fig. 4a Impulse response with no phase errors and

Equation (22) describes a Fourier transformation the shift due to a linear error of x rad.
of the product of the weighting function W(t) and
a frequency modulated signal. This corresponds to
the convolution of the Fourier transformed weigh- IPL RESPONSE
ting function W(t) with a Bessel function, the R__NS
_____
latter resulting from Fourier transforming the FM
signal. Assuming a small amplitude of the phase
error 00 ( 1, the frequency modulation can be ap- -5 --
proximated by an amplitude modulation and the im- -10
pulse response becomes:[-
jZ(t)I, = Zo(t) + .Zo~t - 2_ ] + -
2t k )(23) -25
*11 +
+ .l0t kJf, ' -30-
where Zo(t) is the ideal impulse response from
Eq.(17). Eq.(23) shows we obtain one main impulse -40 _ -
response and so called 'paired echoes', located at -45

t = 2xfoIk, with their maximum being a factor
2 -50
o/ below the main lobe
For phase errors with frequencies above I/T the

paired echoes are spaced away from the main lobe 60
and result in an increase of the sidelobe level. 20 -15 10 5 0 5 to 15 20
In general, energy from the mainlobe is transfer- Azimuthim)
red to the sidelobes decreasing the Integrated Si- Fig. 4b Impulse response with a quadratic phase
delobe Ratio in(ISLR),
of contrast whichThe
the image. corresponds to a loss
ISLR is defined as: e rspo n w a dt s
Each term causes different effects on the image

Energy of the Mainlobe quality:
ISLR =
Energy of the Sidelobes * The constant term has no effect on the image
If the frequency fo of the sinusoidal phase error quality. In this case, the aircraft can be
is lower than the reciprocal aperture time 1/T, supposed to fly parallel to the nominal track
i.e. low frequency phase errors, the paired echoes and a different area is mapped.
will merge with the mainlobe and cause its defor-
mation. The low frequency phase error can be ex- * The linear term corresponds to a shift of the
panded into a Taylor series: mainlobe in azimuth direction causing geometric
: distortions of the image.
3-5
0 IMPULSE RESPONSE Ge(f)= cos2 eD "

N4-2 GDy(f) (29)
-5 -\and
2
-10- - - G (f) =sin eO .%}) GDZ(f) , (30)
-15~ - (M
-20 where GDx(f), Gm (f) and G02 (f) denote the displa-
-20 cement PSDs of the platform in x, y, z direction.
-30 - ,, The expected image quality can be predicted, if
the power spectral density of the phase error is

-35 - - known. In the following, the effects of the phase
-40 - - __\ errors, classified above, are expressed by means
._ of thePSD.
-450
4.1 Linear Phase Errors
-50 - _
IA At first we consider a deterministic sinusoidal

phase error with a constant frequency fo and a
260 -15
-20 10 -5 0 5 10 15 20 constant phase shift 10 ,
Azimuth(m] 0(t) = Oocos(2xf 0 t-y0 ) . (31)
Expanded into a power series, the linear term

Fig. 4c Impulse response with a sinusoidal phase OL(t) becomes
error of 0.1 rad amplitude and 6 cycles 2
across the interval showing paired echoes. L(t) = *o2xfot for 10=x/ (32)
9 The quadratic term causes a broadening of the The phase history of a point target 9(t), corrup-
mainlobe, respectively a loss of geometric and ted by the linear phase error OL(t) can be denoted
radiometric resolution. as
2
9 The cubic term causes an asymetric distortion kt
9(t) + OL(t) = - 2 + oo21fot . (33)
of the mainlobe and an increase of the side-
lobe level, which also leads to a degradation
of the radiometric resolution and a loss of Thus the impulse response is shifted by the di-
contrast, stance 1, where
Fig. 4 shows examples of the influence of linear, 1 .of (34)
quadratic and sinusoidal phase errors on the im- 2.V
pulse response for an. airborne sensor with a nomi-
nal azimuth resolution of 2.45 m. We can see, that the shift of the impulse response
is proportional to the slope of the linear phase
4. NOTION COhPENSATION REQUIREMENTS error, which is determined by 0o and fo.
The motion error of a platform is regarded as a Considering all frequencies contributing to the
stochastic process, which is assumed to be statio- linear phase error, which are defined by G (f),
nary and ergodic. Thus it can be described most the variance of the shift of the impulse response
2
completely by a power spectral density (PSD) [3]. OL can be evaluated with
2 1/T
The following Eq.(25) denotes the relation between 2= .R) f
the phase error PSD G (f) and the PSD of the dis- OL2 = 2" • f2 G0 (f)df . (35)
placement of the platform from the nominal track 0
GD(f). Within Eq. (26) and Eq. (27) the factor 4.2 Quadratic Phase Errors
(1/2f)2 corresponds to an integration in the time
domain, which enables the conversion from the ac- Concerning the sinussoidal phase error denoted in
' celeration PSD GA(f) to the velocity PSD Gv(f) and Eq. (31), for yo=0, the quadratic term of the
to the displacement PSD G(f). power series becomes
2
f,,t)
GD(f) , (4-X)2.
•
G
=
(f) (25) 0 0 (t) = 0 2x (36)
rl(f) = At the end of the synthetic aperture when t = T/2,

2ff Gvf) (126) the quadratic phase error is
* Gv(f) = .G~f). (27) *,(T/2 ) = i-xfT

o' (37)
The components of the platform motion in x, y and Considering all frequencies contributing to this
z direction are contributing to the total phase phase error, which are described by G (f), the va-
error PSD G.(f). riance o02 of the quadratic phase error, measured
at the end of the synthetic aperture results in
Using Eqs. (10) and (25), we get for the x-axis the expression
Go" )- Gr(f 4 . J .oI)

-212="i 2)
(28) °2 " 4 ff f4 'G,(f) df. (38)
The components in y- and z-direction are denoted

as:
3-6
The relation between the mainlobe broadening and where p, is the mean value of the resolution of
the quadratic phase error at the end of the syn- one look. The looks were evaluated with a Hamming
thetic aperture Q9 (T/2) was determined empirical- weighted reference function and finally root sum
ly, by correlating a simulated backscatter signal squared.
of a point target with a Hamming weighted referen-
ce function (a = 0.54) and is described by 4.5 Determination of the Acceptable Notion Error
Eq. (39).
The acceptable residual motion error, which re-
(39) mains uncorrected, represents the required perfor-
peff 1 + 0.55 2, t *Q(T/2) 5 x mance of the motion compensation system. The fol-
Pideal lowing analysis is based on the E-SAR system para-
meters given in Table 3 and the image quality spe-
The ideal resolution is denoted as pid..a and the cifications in Table 4.
resolution, which is degraded by a phase error is
0elf. For example, a quadratic phase error of x/2
at the end of the synthetic aperture results in a * platform: Dornier Do 228
mainlobe broadening of 6.6%.
e altitude: 1000m ... 3000m
4.3 High Frequency Phase Errors * forward velocity (nominal): V = 70m/s
It has already been pointed out, that phase errors * maximum slant range: Ro= '000m
with periods below the aperture time T will have e depression angle: %= 0- ... 70°
an impact on the integrated sidelobe ratio or the
image contrast. According to Haslam [4], the re- a antenna beamwidth: 6B= 10.
quired expression is * wavelength: XL= 0.2308m
1
r c= 0.0566m
ISLR - G,(f)df (40) Xx= 0.0313m
1/T
* aperture time: TL= l5.0S
Tc= 3.68s
4.4 The Effect of Phase Errors on Multilook SAR
Images
Tx= 2.03s
The multilook technique was developed to reduce * azimuth resolution (l look): p = 0.66m
the speckle of SAR images. This can be achieved by
dividing the synthetic aperture into overlapping
subapertures, which are finally summed up incohe- Table 3 System parameters of the E-SAR.
rently. The impact of motion induced phase errors
on each look can be calculated with Eqs. (35),
(38), (39), (40), where the integration time of
one look TN has to be taken into consideration. a Integrated Sidelobe Ratio: ISL = -20dB
* A linear phase error has the same effect on a & loss of geometric resolution: < 10%
multilook image as pointed out in 4.2.1: Each * pixel shift: < 50% of one re-
look is shifted in the same direction by the sa- solution cell
me distance.
Table 4 Required image quality.
" A quadratic phase error causes a mainlobe
broadening and a displacement of the looks from 4.5.1
eachoth.Spcfication of the P for an Acceptable
Motion Error
The distance dN between the first and last look A power spectral density of the displacement of
[5] is given by the aircraft can be specified, which expresses the
allowable, residual motion error. The specified
d,_ a-rTNV (41) PSD of the displacement is denoted as GD.(f), the
ar 'measured PSD of the Dornier DO 228 aircraft is
denoted as GD(f). The PSD GD.(f) is split up in
where ar is the nominal Doppler rate, subdivisions, which are characterized by exponen-
tial functions, depicted as straight lines, using
2.V2 a double logarithmic scale (Fig. 5). For computa-
-(42) tion, Eqs. (25), (35), (38), (39), (40) were used.
The coefficients and exponents of these subfuncti-
and Ac. is the deviation from the Doppler rate due ons were adjusted numerically to achieve the re-
to the quadratic phase error *0,which is given by quirements in Tab. 4 for L/C/X-band. GD,(f) is li-
=
mited to a frequency range from 0.001 Hz, which
- O • (43) corresponds to a maximum duration of one pass of
TN x'TN2per
2 1000 boundary
seconds of
or 2.2 16 minutes,
Hz. Frequencies
appoximately up-
above to2.2an Hz
From Eqs. (41), (42), (43) and after replacing the would cause a maximum ISLR of approximately -30 dB
look displacement dp by its standard deviation oN, (X-Band) and are not considered to degrade the
we get: image quality markedly. The result is given by
Eqs. (46)-00) and is also shown in Fig. 5. The
QN = 2"kR o o (44) exponent of the subfunction (50) was intentionally

V (T4 set to zero, to reduce the degrees of freedom for
further computation. It is remarkable that in this
The mean value of the resolution with N looks pN, case the requirements for the linear phase error
can be estimated empirically and becomes appro- turned out to be more restrictive than the requi-
ximately rements for the quadratic phase error. Thus only
Eq. (35) was used to specify the low frequency
3 section of GD,(f).
N (45)
3-7
rors were derived from the PSDs of Fig. 5, and im-

pulse responses were computed according to the E-
SAR system parameters, using a Hamming weighted
so6 reference function (a = 0.54). Typical results for
C-band are depicted in Figs. 6-9. The impulse res-
40 ponses of Fig. 7 and Fig. 9 are normalized to the
maximum of an ideal, unweighted impulse response
20 o Z0 (t) from Eq. (16).
0 the aircraft The required transfer function S(f) of the motion
compensation system is shown in Fig. 10 and is gi-
E 0 ven by
.2 -40
-6 for LlClx-band =
S (f) f(52)
S -80 I | !
-,00 .. . . ... .
L ,l
.. i.
Tc Tx
1 t... i ....
1-3 0.01 0.1 to
Frequency [Hz] ' I
E 5
Fig. 5 The measured PSD of the motion of the DO

228 aircraft, compared to the specified 2
PSD of the acceptable motion error. E
0
5
G,.lf) - 6H46-10-5 z (TL'f)-3 ' (46)
0.001Hz !5f s I/T L
-2 0 9 3
GD,(f) - 3.41"10-6 Mz .(Tc'f) . , (47)
I/TL S f 5 I/T, -4 I I I_r _
-S -6 -4 -2 0 2 4 a
- 7 "4 5 6
GD,(f) - 4.11"10-8 z "(Tx'f) , (48) TIme [a]
1/Tc !5 f < I/Tx
2
G, (f) - 4.11"10-8 M i/Tx i f 5 2.2 Hz (49)
Fig. 6 Simulated motion
error, derived from the
GD1(f) = Gom(f) , 2.2Hz 5 f < - . (50) measured PSD of the aircraft.
For easier interpretation of the specified PSD

G05 (f),the standard deviation aDs of the displa-
cement from the nominal track in line of sight of
the antenna can be calculated using: 0
ideal lIpulseo
response
*r : G , (f)df (51) -20
0 degraded impule
In this case the forward velocity of the aircraft .I @on

is assumed to be constant. Thus we get a standard - -40
deviation of oD, - 0.4 m for the low frequency
section of GD.(f) for L-, C- and X-band. This E
means, the RMS deviation from the nominal flight A
path should not exceed 40 cm within 1000 seconds, -60
to keep the RMS mainlobe shift within 33 cm.
For the high frequency section of GD.(f) we get
-g0
L-band: O, -1.84mm , I/TL 5 f S - 1.5 - -0.5 O 0.5 1.5
C-band: a,. - 0.45mm , /Tc 5 f S - ?imd [a)
X-band: at. - 0.25mm , i/Tx 5 f - . .
To obtain an ISLR of -20 dB in L/C/X-band, the RMS
deviation from the flight path should not exeed
0.8 % of the used wavelength within the aperture
time. Fig. 7 Simulated C-band impulse response, degra-
I1p
ded by phase errors due to Fig. 5, compa-
4.5.2 Simulations red to an undistorted impulse response.
In addition to theoretical examinations, the ob-
tained results were verified by simulations, based
on the Monte Carlo method. Therefore, motion er-
(3-8
5. MOTION COMPENSATION METHODS

-0.125 5.1 Flight Path Measurement with a Strapdown
Inertial Measurement Unit (IMU)
Within a strapdown inertial measurement unit, the

0sensor elements are directly mounted to the air-
-o.155 craft frame, whereas linear motions are detected
by accelerometers and for angular rate measure-
- ment, gyros are used. The following information is
* -0.14
usually supplied by a commercial strapdown system:
04 aircraft referenced accelerations,

-o.145 e aircraft referenced attitude,
e inertial velocity,
-0.15 * ground speed,
* heading,
-0.155 I I
0 position.
-8 -6 -4 -2 0 2 4 6 8
Time - 2 To obtain this information, the following evalua-
tions have to be made by a strapdown computer:
* compensation of the rotation of the earth and

the transport rate, both beeing included in the
gyro signal,
F compensation of the gravity, included in the
Fig. 8 Simulated motion error, derived from the accelerometer signal,
specified PSD due to Eqs. 4 - (50). computation of the attitude angles of the air-
craft (pitch, roll and yaw), by integrating the
angular rates,
transformation
0* of the accelerations to the
earth referenced coordinate system,
-to - computation of the velocities by single inte-
gration,
-20 a computation of the actual position by double
integration.
-30 -- 5.1.1 Systematic IMU Sensor Errors
-40 - The inertial sensors may produce an output signal,
E although the IMU system is actually not moved.
-so Farrell [61 already showed, that an accelerometer
bias ab leads to a quadratic phase shift and a cu-
_60 bic phase shift arises from a gyro drift wd.
! 0The
-70 quadratic phase error Oo(t) is denoted as
-0.6 -0.4 -0.2 0 0.2 0.4 0.6
Time [Q(t) - I ab't 2 (53)
and at the end of the synthetic aperture, where

t=T/2 we get
Fig. 9 Simulated C-band impulse response, corres-
ponding to Fig. 7 with an acceptable main- *Q(T/2) = 3 a . (54)
lobe shift, loss of resolution and ISLR. 2 X
_0 Taking the E-SAR parameters for L-band, the most
.............................. critical case, we obtain for a phase error of
o(T/2) = x/2, a total accelerometer bias of ab =
s5o 104.6 pg. The cubic phase shift *c(t) is given by
. 40 Vc(t) : - 9
-gd (55)
0 where g is the gravity of the earth. With t=T/2 we

k get
20 c(T/2) = T . (56)
110 For an acceptable cubic phase error *c(T/ 2 ) = s/8,

Swe obtain a total gyro drift of cd = 2.16*/h for
0 _. X L-band parameters.
lI-$ o.oi 0.1 105.1.2 Errors from Digital Data Processing
FrequuaM [Muj
In addition to the phase errors induced by the
aircraft motion or by the inertial sensors, fur-
ther errors will occur due to the imperfect solu-
tion of the motion compensation algorithms and of
digital data processing artefacts.
Fig. 10 Required transfer function of a motion
compensation system for the E-SAR.
7: 3-9
Using a digital phase shifter, the amplitude of flight-path can be performed online in the sensor
the phase error 0(t) can be adjusted only in dis- or off-line in the SAR processor. Thereafter the
crete steps, due to the number of available bits. data can be processed as if the platform moves in
2
The variance a. , respectively the mean power of a straight line at constant altitude and parallel
the quantisation noise is to the swath.
AS2 From the flight-path the following parameters are

2 (57) derived.
On
- The true forward velocity is used to correct
the range independent phase er.ors by ensuring
o
Quantizing 360 with N bits, we get the along track samples are equidistant. This
also corrects the image geometrically along
As = 2N (58) track.
2 With on-line correction, equidistant sampling
and finally, from Eqs. (40) and (51), the ISLR, can best be achieved by controlling the PRF of
induced by the quantisation noise, becomes the radar, i.e. the PRF is directly slaved to
the forward velocity. Off-line, the same effect
ISLR n
2
= _2 is obtained by interpolating the data in the
3"22N along-track dimension.
5.1.3 Delayed Phase Adjusting The range delay of the samples is needed to en-
sure the range cells are correctly aligned. On-
line this is achieved by adjusting the delay of
Within online motion
an inertial compensation system, uti- lietsisahvdbydjtngheeayo
lising an strapdown system, the computa- the digitised echo window. Off-line the range
sysemthe ompta-
usin aninerialstradow cells are shifted in the computer memory.
tion time will cause a delayed phase correction.
Thus a residual, uncorrected phase error will re-
main, causing an increase of the ISLR. Considering corrected according to the true slant range to
a phase error 0(t) = o0 cos(2Rf 0 t), which will be eliminate phase errors. on-line this can be
corrected after
compensated phasea error
time delay Atd,
0r,(t) a residual,
will remain. un- civdwt
achieved hs shifter.
iia phase
with a digital hfe.OflnOff-line
the data can be corrected by shifting the phase
Ore,(t) = Oocos(2nfot)-oocos[2xfo(t-Atd)] (60) mathematically.
= 2-Oosin(nfOAtd)-sin(2fO(t-Atd/2)] (61) 5.1.6 Example of Motion correction using an IMU

Sensor
The residual signal is sinusoidal too, with the The image in Fig. 1 was taken with the DLR airbor-
phase error frequency fo. Its amplitude depends on ne SAR equipped with a LITEF LTR-81 heading and
trietime delay Atd, the frequency fo and the am-
ptde ife eay ora sirn aaltitude reference system supported by the iner-
plitude of the original signal 00 . tial reference system of the aircraft. The air-
Considering all frequencies and amplitudes, defi- craft was deliberately caused to yaw ± 2 causing
ned by: the motion errors already described.
From the recorded flight path data, the forward
(f), the ISLR can be evaluated using Eq. (40) velocity (Fig. 11) and phase errors (Fig. 12) were
derived.
ISLR - 4" sin 2 (zfAtd) 'G (f)df (62) FomadV*ociy

4 1/T 82.* "
5.1.4 Sampling and Holding the Detected Motion
Within an on-line motion compensation system, the SL2

phase shifter is adjusted in time intervals At,.
This is equivalent to sampling and holding the
phase error. I
Considering a phase error 0(t)= Oocos(2nfot), the 1
power of the residual, uncompensated phase error, .2.o
L,(t) within the interval at, is approximately
At,
L,(t) -1 f(0o2xfot)2dt = (At.'a'0o'fo)2 (63)
0 20 0 so 80
Thus the ISLR, caused by sampling and holding the
detected motion, becomes
(X'At')2" f2"1 Fig. 11 Forward velocity variation for the example

ISLR - ~a,2. 2
G,(f)df .(64) Fi.~ SA image.
l/T The flight data were used to correct the SAR data
and the resulting image is shown in Fig. 13. The

image is now well focused and the geometric di-
Once the flight path has been determined, it has stortion (runway and taxiways) has been corrected.
to be taken into account during the processing of
the SAR data to ensure an undistorted, well focu-
sed image. The correction for the errors in the
IT-
3-10
ciple of the meth.d is based on the misregistra-

tion of the looks if the Doppler rate is incor-
rect. An error in the Doppler represents a quadra-
tic phase error. In Section 4.1 we have seen that
the shift of the impulse response is proportional
to the linear phase error.
Assuming a quadratic phase error (see Eq. 24)
(60)
* (t) = - a(0)
Differentiating we obtain:
- =-a(0)
a(t) t . (61)
From the earlier analysis it can be 3hown that the

relative position of the impulse response becomes:
Ro
s= t a(0) . (62)
For two looks spaced At in time, the misregistra-

tion between the locvs in meters is given by:
Fig. 12 Phase error profile for the example SAP
image. Ro
As = At a(0) . (63)
'0. - --- A-a* * - Hence, the coefficient of the quadratic phase er-
ror a(0) can be del-ermined from the misregistra-
tion of the looks. Note that Ro , V and at are all
I '" ' known.
The misregistration of the looks can be determined

by correlating them to determine Lhe spacing. Kno-
wing the quadratic phase error term, the Doppler
rate can be corrected.
This method works well where quadr, Lic phase er-

rors dominate. This is the case where the platform
flight path is stable but the velocity is not
known, e.g. satellite sensors. Higher order terms
cannot be derived.
The method requires a preliminary processing of
the image before correction can be applied.
5.2.1.2 Contrast Optimisation
This method is based on the re-ationship between

the contrast in the image ai.d the Doppler rate
used for processing. The image (or j.,rts of the
image) is processed and the contrast is measured
by calculating the ratio of the deviatijn/mean in-
tensity of the image. It can be shown [7] that
this ratio is directly related to the spatial re-
solution. This is a trial and error method for
finding the optimum Doppler rate for azimuth pro-
- cessing and, hence, requires considerable computer
effort.
5.2.2 Reflectivity Displacement Method (RDM)

Fig. 13 The scene from Fig. 1 corrected for motion This method [9, 9] of motion compensation using
errors. the SAR data is the most comprehensive and in
5.2 Deriving the flight Path from the Radar Data principle is capable of correcting both high and
low-frequency errors with impressive results. The
5.2.1 Aiutofoue Methods technique analyses the frequency spectrum of the
SAR data after range compression to derive the
Autofocus is a process whereby the SAR processor flight path of the sensor.
attempts to derive the data necessary for accurate Considering the deviation from the optimum flight
focusing of the image from the radar data. The path in terms of a velocity component in
following methods describe typical approaches, line-of-sight VLoS, rather than displacements, we
5.2.1.1 Look Miaregistration Method can write:
With this method the SAR data are processed to ob- 2 V(t)'OA 2"VLos(t,
tain a two-Inok image. The azimuth processing is foper - (64)
performed with an assumed Doppler rate. The prin-
3-11
Assuming the forward velocity V(t) is constant du-

ring the period under consideration, we can derive
the following expression for the frequency shift too
between two adjacent power spectra spaced At in
time :
'E 90
2
Af~oppler - 2 V (t)'At + 2 Vrn (t) -At
-fD +r" (65) 80a
where r is the range of the selected range samp- 'a

les. We see the frequency shift can be separated
into two components, one dependent on the forward 6 so
velocity and one dependent on the acceleration in
line of sight. 501
The spectrum of the SAR data in azimuth is a con- so

t
volution of the antenna pattern and the ground
reflectivity function. If the antenna pattern is 30 "
broad enough not to influence the ground reflecti- 0 100 200 300 400 500
vity function or, if it can be corrected for, the
frequency shift can be determined by correlating SHIrTCz
fREOUENCY
the two adjacent azimuth spectra. If At is much
smaller than the azimuth illumination time, the
frequency shift AfDoppler can be determined very
accurately.
Fig. 15 Correlation of the azimuth power spectra
Fig. 4 shows two such spectra for the E-SAR air- form Fig. 14.
crne sensor operating in L-band with a time offs-
et At of 1.075 s.
HZ
Fgtud VOW0 0. In LOS-drM.
Acc.ler.
{dli F- ou4Vocy
200 " 02. lA10.Il0r0200 In
40
s(fk-1) t I= o -to
o -20
-300 -. 0.-20 032 .003 3.01 2.23 0.1 D.3 I MR.
Fig. 16 Power spectral density of the forward ye-

0i. 5shwsthc-ocity and acceleration in LOS for aDO
-300 -200 -100 0 too 200 300 D10 NE-OF E-SA* SOSTEM .0TI,-lI COIPE l50TIO. PPC 0S$on
FREQUEN~CY
(HZ)
Fig. 14 Two adjacent azimuth power spectra taken

with a time offset at =1.075 s.
Fig. 15 shows the corresponding correlation func-

tion yielding a Afloppier of 12.1 Hz.
The next step in the RDM method is to consider the

power spectral density of the forward velocity and
the acceleration in LOS. Fig. 16 shows the PSD for
the E-SAR aircraft, a twin-engined Dornier 228. It
can be seen that the velocity variations are mai-
nly low-frequency whereas the accelerations in LOS
are mainly high frequency. This is true for most
airborne platforms where the forward momentum is
much higher than in other directions.
This phenomenon enables the forward velocity and

acceleration influences to be separated in Eq.
(65) by filtering the values of Af]oppier. For the
above examples, the filter cut-off frequencies are
at 0.05 and 0.1 Hz.
This procedure enables the forward velocity and

acceleration to be determined. From these parame- Fig. 17 Image taken in C-band with the E-SAR sen-
ters, the necessary corrections of the SAR data sor with 12 % velocity variations. The up-
can be carried out. Fig. 17 shows an example with per image is uncorrected and the lower one
corrected using the RDM method.
3-12
the E-SAR sensor operating in C-band with 12 % ve- [2] S. BuckreuB:

locity variations. The upper image is processed Motion Errors in an Airborne Synthetic
with constant velocity and the lower one with the Aperture Radar-System.
processing parameters corrected using the RDM me- ETT-Journal Special Issue: ETT Focus on
thod. The defocusing visible in the upper image is SAR", Lfd. r. 2, Bd. 6, 1991, pp.
eliminated after motion compensation. 655-664.
6. CONCLUSIONS [3] A. Papoulis:

The influence of motion errors on SAR images has Probability, Random Variables and Stocha-
been
been iflusede and requirements fo
the requres
andmtin aimage hstic Processes.
sationdiscussed
system derived. for a compen- Internat. Student
gakusha, Ltd. Edition,
Tokyo, 1965, McGraw-Hill
p. 336-384. Ko-
Several methods of compensation have been discus-

platforms and the RDM[4 Moio Ssin
inertialshwingthebestresuts.Motion
sed, the use of methd Reir
Sensing Requirements for Synthetic
method showing the best results. Aperture Radar.
As an alternative to using inertial platforms, it Proc. IEEE Conf. Toronto, 1983, Vol.1, pp.
would be attractive to use the satellite naviga- 126-131.
tion system GPS, being a much cheaper solution. [5] F.K. Li, D.N. Held, J. Curlander, and
A GPS receiver is used in the E-SAR airborne SAR C. Wu:
to provide absolute position of the aircraft. Ho- Doppler Parameter Estimation for Spacebor-
wever, due to the deliberate errors introduced ne Synthetic Aperture Radars.
into the GPS data by the GPS operator (selective IEEE Trans. Geosci. Remote Sensing, Vol.
availability), the data are not accurate enough GE-23, 1985, pp. 47-56.
for motion compensation of the SAR data and they
are only used for annotation purposes. Note, that [61 J.L. Farrell:
the selective availability degradation particular- Strapdown INS Requirements Imposed by SAR.
ly influences the velocity measurements. The use the USA.
Conf. Dayton,of OH,
Proceedings IEEE 21-25
1984 May 1984.
National
of an additional stationary GPS receiver to com-
pensate for these errors (differential GPS) could Aerospace and Electronics Conference.
provide an attractive method for measuring the Naecon 1984 (IEEE Cat. No. 84CH2029-7).
flight path and is worth investigation. [7] D. Blacknell, I.A. Ward and A. Freeman:
An additional method of improving image quality motion Compensation and Geometric Distor-
was suggested by Chan,the so-called tuned auto tion in iborn S ery.
compensator [10]. This method analyses strong Progress in Imaging Sensors, ISPRS Symp.,
point targets to identify paired echoes. From the- Stuttgart, 1-5 Sept. 1986.
se the high-frequency phase errors can be derived. [8] J.A. Moreira:
The method requires much trial and error and can, A New Method of Aircraft Motion Error Ex-
be regarded as an augmentation of
therefore, only cte--cnae.traction tActo f r rRaw Dat f or E
from Radar Raw Data for Real Time
SAR Motion compensation.
'Table summarises the capabilities of the various Proc., IGARSS Symposium, Vancouver, Canada,
methods. 1989.
[9] J. Moreira:
Method Forward Low-frequ. High-frequ. Motion Compensation SAR-Processing Facili-
velocity errors in errors in ty at DLR.
LOS LOS EARSeL '90, Toulouse, France, 5-8 June
1990.
inertial
measure- yes yes yes [10] Y. Chan:
ment A Tuned Auto-compensator for Residual An-
tenna Motion in Synthetic Aperture Radar
Look Systems.
misregi- either or no IEEE Trans. Geosci. Remote Sensing, Vol.
stration GE-24, Nov. 1986, pp. 1025-1027.
Contrast [i] J.C. Kirk:

optimi- either or no Motion Compensation For Synthetic Aperture
sation Radar.
IEEE Transactions on Aerospace and Elec-
Tuned tronic Systems, Vo 1. AES-Il, No.3 (May
auto com- no no yes 1975) pp. 338-348.
pensator 1
[12] T.A. Kennedy:
RDM yes yes yes Strapdown Inertial Measurement Units for
Motion Compensation for Synthetic Aperture
Radars.
Table 5 Comparison of the capabilities of the va- Presented at the IEEE 1988 National Radar Con-
rious motion compensation methods. ference 0885-8985/88/1000-0032,IEEE AES Maga-
zine, Oct. 1988.
7. RZFERENCZS
[13] Joos D.K.:
[I] R. Horn: Inertialnavigation in der Strapdown Tech-
C-Band SAR Results Obtained by an Experi- nik.
mental Airborne SAR Sensor. Special Issue. Ortung und Navigation.
IGARSS Symposium, Vancouver, Canada, 1989 Journal 2/1983.
July 10-44.
4-1
THE REAL APERTURE ANTENNA OF SAR, A KEY ELEMENT FOR

PERFORMANCE
H.bttl
Deutsche Forschungsanstalt fUr Luft- und Raumfahrt e.V.
Germany
SUMMARY
For a SAR system flying on an airborne or space- flight
borne platform, the real antenna must be designed
in such a way so as to avoid ambiguities and ach- (azimuth)
ieve the envisaged resolution.
Although a SAR is, with respect to geometric azi-

muth resolution, independent of its distance from
a target, the ground range resolution depends on h
the incidence angle and, of course, on the band-
width dependent slant range resolution. lox
The antenna size and its half power beam width %

(HPBW) in azimuth and elevation define its azimuth
resolution and, for a given off-nadir angle and
chosen altitude, the swath width.
The miniium antenna size, measured in wavelengths,

depends on the altitude, velocity of platform and
chosen off-nadir angle. In real antenna design,
the aperture size will be somewhat larger in order
to allow for amplitude taper (at least in
elevation), for electronic beam steering and pos-
sibly for beam shaping.
This paper explains the interdependence of antenna
parameters with SAR system performance.
resolution
1. INTRODUCTION
cell
Fig. 1 Coordinate system and illumination geometry.
Spaceborne imaging radars and, in many cases, air-
borne imaging radars require enormous antenna di- the azimuth reference function (Doppler function)
mensions (measured in wavelengths) if real apertu- along the synthetic aperture length. This step
re systems are envisaged for ground resolutions of creates the synthetic aperture; accordingly, it
several metres or less. Such large apertures can reduces the azimuth HPBW of the multiple fan beams
neither be launched into space nor installed on in relationship to the synthetic aperture length,
high flying aircraft. Coherent RF-technology al- resulting in a ground resolution of approximately
lows the composition of a synthetic aperture half the real antenna length for a single look
length in flight direction, which defines the azi- image.
muth resolution, while in the orthogonal direction
(in range) resolution is achieved by the short ra- It should be noted that the illustration of res-
dar pulse length or, if chirp modulated, by the olution cells created by multiple beams across
compressed pulse length. range, which are sweeping along the azimuth direc-
tion, is only one possibility of describing SAR.
A schematic view of SAR's illumination geometry is It does not include illumination geometry, that
illustrated in Fig. 1. means changing incidence angle direction and value
during a flight along one synthetic aperture
Each target within the footprint of the real aper- length.
ture antenna will be illuminated during the time
needed to fly along one synthetic aperture length,
which corresponds approximately to the length of 2. GEOMETRIC RESOLUTION, ANTENNA APERTURE AND
the footprint. The data takes are characterized by SWATH WIDTH
time, Doppler frequency and distance between the
flying (or orbiting) real aperture antenna and the It is important to understand the improved along
target (Fig. 2). These data are the input for a track resolution of a synthetic aperture antenna
SAR processor which generates the SAR images. in comparison to a real aperture antenna of the
same length 1 and same azimuth aperture taper fac-
Two steps are generally performed by a SAR proces- tor ahr. To explain this, we assume an antenna
sor. Firstly, the data are range compressed, that carrying platform passing a target. In the case of
means correlated with the range reference functi- a real antenna, the target will be fully illumina-
on. Accordingly, one could imagine that this step ted if the Doppler frequency is zero. The antenna
produces as many fan beams as range bins are fo- beam, originating from the phase center of the an-
reseen. The HPBW of one such fan beam in range is tenna, is orthogonal to the flight direction
dependent on the slant range resolution and the (Fig. 3). Resolution is given by:
incidence angle; the HPBW in azimuth is still ap-
proximately equal to the HPBW of the real aperture H
antenna. Secondly, azimuth compression is perfor- rar = - Phr F (1)
med, which means that the data are correlated with
4-2
Hatched Area:
Antenna Footprint
Constant Doppler Hyperbolae
Fig. 2 Position of resolution cell within a SAR image is determined by slant range, Doppler
frequency and time.
I flight direction
Doppler impact due to

antenna velocity
0
Constant frequency circle
Fig. 3 Very long real aperture antenna illuminates a target The Doppler frequency shift of the
backscattered signal corresponds to the one way case.
4-3
t flight direction
0 constant frequency
/04
0./
4-
4M Doubled Doppler impact due to
two way phenomena of synthetic
O '1aperture antenna
.real aperture antenna flying along the synthetic aperture
Fig. 4 Synthetic aperture antenna illuminates a target. The signal from the real aperture antenna
propagating to the target undergoes already a Doppler shift; the backscattered signal is
"Doppler shifted" again resulting in twice the value of the one way case.
H = altitude of antenna above ground, the HPBW of the synthetic aperture
= off-nadir angle (- incidence angle),
hs = ah, * (4)
=
azimuth HPBW 3hr ahr , (2)
ah, = azimuth aperture taper factor,
X = wavelength. L = length of synthetic aperture.
The HPBW of the real aperture antenna which flies
The signal reflected from the target reaches the along the synthetic aperture (the antenna is much
different parts of the real aperture antenna with smaller than in the above mentioned comparison) is
a phase distribution corresponding to the one way used for the definition of the antenna footprint.
distance. The Doppler frequency occurring at the The length of the synthetic aperture is given by
different parts of the moving "long, real aper- the length of the footprint and therefore,
ture" antenna represent also the one way case.
).H
In the case of a synthetic aperture antenna, a L = rar = ahr 1 COS (5)
small signal source (phase centre) flies along the
synthetic aperture. We have, therefore, a two way Substituting L from (4) and Oh. from (3) in (5)
case for the phase difference along the synthetic leads to
aperture (Fig. 4), which means that phase diffe-
rences (or Doppler frequency) are twice the values ahs 1
achieved in the real aperture case of the same rap a 2 ' (6)
length. This phenomenon causes a phase change of
1800 twice as fast as in the case of the real
aperture antenna, meaning that the nulls of the which is the well known approximation for finest
synthetic aperture antenna main beam have a spa- along track resolution. For ah. = ahr, rap corres-
cing which is half of the real aperture antenna, ponds to half the length 1 of the real aperture
antenna.
Therefore, a synthetic aperture antenna has an
along track resolution (HPBW of SAR) which is ap- However, it should be noted that in many cases the
proximately twice as good as for the case of a amplitude distribution in azimuth of the real
theoretical, real aperture antenna of the same aperture antenna is not tapered, resulting in ahr
length. The along track (azimuth) resolution for a = 0.88, while the sidelobe suppression achieved by
single look image is given by the reference function of the synthetic aperture
azimuth pattern (more than 40 dB below mainlobe)
H leads to aha - 1.3. Therefore, it is a realistic
rP = c Oh. , (3) assumption to expect an rap - 0.8 1.
4-4
The swath width is a parameter which is dependent 2 PR? av H . sinO

RavHIsn
on along track resolution ra , platform velocity uh>2
and off-nadir angle &, as wiYl be shown below. c cos 2 o
Flying along a synthetic aperture length covers,

in the case of SAR, twice the normally encountered av = elevation taper factor.
Doppler frequency shift fD. Since PRF > fD and fD -2 ahb , the physical
For zero offset processing (using in-phase and dimensions of the real aperture antenna are gi-
quadrature channels for positive and negative yen by
Doppler identification) only fD per channel is
used. Ap = h 1 > 4 ahr av H H X sino (12)
cos2o
From geometric considerations (see also Fig. I and
2), fD is defined by In many cases, ahr = 0.88 (constant amplitude)
2hr while av will be approximately 1.3 to allow sup-
fD = 2 sin - (7) pression of the sidelobes in the elevation plane.
Furthermore, range ambiguities should also not oc-
For high flying platforms (satellites), Ph is cur within the main lobe at all (main lobe width
small and is approximately 2 to 3 times larger than ov).
2u 2 u Therefore, empirical estimations of the real an-

fD - hr = ahr 1 (8) tenna dimensions could be based on
The Nyquist sampling theorem requires a pulse re- u sin

petition frequency (PRF) equal to or higher than Ap - 10 H- , (13)
the f0 which is used in SAR processing per chan- cos 21
nel. Generally, a PRF of 1 to 1.3 times fD will be
chosen. Two radar pulse returns must not be recei- f = SAR frequency.
ved simultaneously from the illuminated footprint
(instantaneous field of view, IFOV) of the real It can be concluded that the estimated minimum di-
aperture antenna. Looking at Fig. 5, we notice mensions of the real antenna enforce a trade-off
that the slant range difference between the edges between resolution and swath width. High resoluti-
of the illuminated swath width (corresponding to on can only be achieved with a small swath width;
0,) is given by the slant swath width increasing the swath width reduces resolution
(Fig. 6 and Fig. 7).
3. DIRECTIVITY, GAIN, BANDWIDTH AND DUAL POLARIZA-

TION
H The remarks given below refer to spaceborne anten-
OV nae, which are more difficult to realize due to

0their size and the associated, complicated, fee-
ding network.
These antenna, of several square metres area, are
usually designed as flat (foldable) arrays.
The directivity of such antennae is proportional

to their area but depends of course on the chosen
taper. Compared to constant amplitude distribution
with a sidelobe level of about -13 dB, a good si-
delobe suppression (in elevation) reduces directi-
vity easily by 1 to 1.5 dB, which must be compen-
sated by an increased antenna area.
For a rectangular, flat array antenna of length 1
and height h, the directivity D is given by
Fig. 5 Distance Detween two pulses must be > 2R,
in order to avoid range ambiguities occu- 4z 1 h
ring within the IFOV. R. = slant range D (14)
swath width. x2
if an equal phase and amplitude distribution is

H Ov sint assumed. In this case, we call 1 h the effective
Rs - _ (9) antenna area
The distance between two radar pulses must be > Arf = 1 h . (15)
2R.,
rangetoedge
avoid the reflected
with pulse from thepulse
far
overlapping the reflected For any deviation from constant phase and ampli-
from the near range. Therefore, tude distribution, the effective antenna area will
be smaller than the real aperture. The relation-
i>2H
0 (10) ship between the two areas is called aperture
_.> 2 0826 (0 efficiency I.. It reduces the directivity proper-
tionally and therefore,
c = speed of light.
Substituting av for Ov gives the antenna height h
4-5
\e ~~
,\00..
00rI
\ cell
J~ ~~~~ i / ""
a~~~
~
v*~e~'~~C&
0 -" se tq ' ie to
L
re
Aid~.h (left side

Fig. 6 High resolution corresponds to long synthetic aperture and small swait:•
of igr)while a wide swath width enforces a short synthetic aperture and low resolu-
tion (right side of fiur)
q1a 12 () (18)
B = chirpin vrtial
olaizaton the are
bandwidth. til ona
The antenna gain G is smaller than the pattern di-
rectivity D due to internal antenna losses. This The ground resolution across track is accordingly
thermal efficiency, or radiation efficiency, is gvnb
defined by gvnb
11, = -(17) r = 2B sin S (19)
For a passive microstrip antenna of several square Dual polarization capability has been increasingly
metres or a waveguide ar:" y oi ;everal metres requested in recent years in order to allow for
length (spaceborne SAR an enna , ,l might be 0.5 full polarimetric SAR operation. For instance, a
or even lower (depending on ne frequency) .horizontally polarized signal pulse will be trans-
0 " mited and the back scattered signal will be re-
ceived in the same polarization and simultaneously
o I"
these high
iConsideringwith such arrays, one solution would beassociated
one way losses to use ane ti
active (T with distributed

arraymodules) loss of such sy-o-
transmit/receive
The one way back signal pulse
rized scattered signal be transmitted
willwill be received in the same
andthe
dules
stems is much lower and leads to d - 0.9.
0.8 polarization and simultaneously in horizontal p-
lower microwave
For S-band, frequencies, as L-band
suchconsiderations
and volume larization.
peated. This sequence
Assuming time be
that the will continuously
between re-
2 transmit-
and light weight
lead to microstrip arrays while for higher fro- ted pulses of different polarizations
inlpl
polarize e is
l negligi-
tra s-
oren lor (depenudigoweveefrtquency).bhorotentahlt
e fi
ch cble, we achieve the complete polarimetric matrix
array might be favoured for its lower losses
Thcontains much more information about radar illum-

to easily achieve the
(chirp bandwidth).
desired range
Frequencies resolution
(signals) outside ntdtresignal
scattered lseof onehntecplrzdbc
polarization.
the useful bandwidth should be rejected or, at be noe that SAR
However it shoul polarimetric
lat stogyatnae.operation (also named wquad-pol" operation) requi-

Typically, microstripofand wave guide arrays attain res doubling the PR for a SAR with 2 receiver
sa
relative bandwidth 3% to 5%. That means, for channels. This results in reducing the swath width
a high slant range resolution rw it is advisable to half of the single polarisation case and, accor-
atiue hrrawigh frequencyibad frant e o - dingly, doubling the antenna height h for proper
ample, a slant range resolution of 0,5 m corresponds rang abiuit suppression exceion: ine
approximately to a chirp
is 3% of 1 Go az).
bandwidth of 300 M~z (which SAR usually tonottheaffected
is itdue
problems, low altitude and amiguity
by range smiler
Slant range resolution ra is defined by The polarization ratio between copolarized andi
a tcross-polarized antenna pattern should exceed 30
dB within the mores in azimth and elevation. It
4,6
[km]
1. C
._o
020
--
E
cc
0 __
X100
--- 1-1A-I
0
101in
single look along track resolution

Fig. 7 Azimuth resolution rap influences proportionally the swath with; the parameter is incidence
angle.
is very difficult to reach this performance for ties and therefore reduce the revisit time
the large arrays needed for spacecraft. Screening intervals,
align- - to collect radar data under different incidence
the
ent feeding network
of single radiating elements mechanical
and accurate are a "must" for angles.
a good polarization ratio. The request for 30 dB
is based on the facts that cross-polarization oc- The off-nadir angle range used in various studies ex-
curs in many cases in the range of 5 dB to 20 dB tends from 150 (200) to 500 (606) covering a data take
below the copolarized back scattered signal and opportunity width of about 200 km to 600 km on the
that the performance of a spaceborne SAR sensor ground (depending on the orbital altitude, e.g.
(noise floor) seldom allows a signal to noise ra- shuttle or satellite).
tio S. of more than 20 dB.
For an array antenna with distributed T/R modules, on-
ly phase shifters need to be added in order to produce
4. ACTIVE PHASED ARRAY an active phased array antenna.
For spaceborne remote sensing, a fast off-nadir The number of T/R modules and the spacing of the
angle change capability is needed radiators (or groups of radiators) fed by a single
T/R module depends on the chosen phase controlled
- to allow scan-SAR operation, angle range.
- to increase the nuber of data take opportuni-
4-7
/ e
#.S
%Qt~1
illuminated spots
lesub
Fig. 8 Azimuth look steering mode increases the overall synthetic aperture (for multi look images
with azimuth resolution rap). Only spot-wise data takes are possible in this mode.
m~~~
so , __,
1, (20)
d = distance sfter s centres of single radiators

(or groups of radiators) fed by neighbouring T/R modu-
sow. les,
slm range.
2000- In
u case of At= 20* the spacing must be less than or
Kequal
-AC-B&Mmodes to 0.75 a
require Spotlight
w.beam is electronically
which or control-
azimuth look steering
Ku-Band~ Th requ
A squint
hedeired
anl orhopeainsgis
S du.1
led by phase shifters within a small (squint) angle
These modes allow high resolution observations
rrange.
_Xro nal possibilities before and after each spot (Fig. 8).
SKu-Ba The required squintptangle for such operations is
' i , -- -/ .. gousain azimut
seldom larger fedts
thanil) by
to asciiing th.
The spot
t2a. molehveao
size is fre-
- - quency dependent because of the IFOV, which increases
proportionally with the wavelength (Fig. 9).
The geeato
Equation (20) will not be applied
V atoni for squint angles
bmllaeastng Any sageri atcing fosromteom-
clof l to 2 in order to keep the number of T/R mo-
lag. I a s dules and its phase shifters small. The radiator
.... groups in azimuth fed by a single T/R module havea
ahigh directivity, which is used to suppress the
ta
h grating lobes.
i There are two causes of gain reduction due to elec-
tronic beam steering. Any angle deviating from the me-
....... (dog chanical boresight angle reduces the antenna area pro-
10 20 30 40 50 00 70 4o portionally to cosao or cosag . Furthermore, each sing-
W90 hidda, ewqb le radiator has a pattern which usually has its maxi-
mum at the (mechanical) boresight angle. The reduced
gain at other angles must be multiplied with the array
factor (which represents the far field of point sour-
Fig.9 aea or sot ighmod an/or
Ilumnatd ces arranged in the geometric layout of our antenna).
ook
teeingmodeincease
azimth apro- For instance, the gain loss of a well designed active
azimth
teeingmodeincease
ook apro- phased array at A6 = 20* will be approximately 0.5 dB.
ximately proportionally with the wavelength.
The generation of several "main beams", which are 5 OCUIN

called grating lobes, occurs if the spacing is too
lag.In such a case, several angles exist with Tera prueatnai e lmn o A
equal phase distribution; it is at these angles Teraper e
turan ntenn s ao keyfleent fo |
that the grating lobes occur. Suppression of gra- pefrac.Idfisan/rnlucs
ting lobes is accomplished by satisfying
I.
4-8
- the along track resolution rap by its length 1,

- the swath width by its height h,
- the across track resolution rR by its bandwidth B,
- the range ambiguity suppression by its amplitude
distribution in elevation (taper factor a,),
- the required onboard power by antenna losses
(thermal efficiency ql),'
- the polarimetric performance by its polarization de-
coupling,
- the data take opportunity, scan SAR and spotlight
mode SAR by its electronic beam steering.
It is, therefore, worthwhile investing efforts in a

good antenna design and accompanying technologies.
6. REFERENCES
This paper is a lecture based on principles, which
have been published several times by other authors and
by recent studies, in which the author was involved.
The references below were used as sources, but due to
the general aspect of this lecture, no special refe-
rence is given in the text.
1. Skolnik. M.I., "Radar Handbook", McGraw-Hill Book

Company Inc., 1970.
2. Barton, D.K. and Ward, H.R., "Handbook of Radar
Measurement", Artech House, Inc., 1984.
3. Meinke/Gundlach, "Taschenbuch der Hochfrequenz-

technik", Dritte Auflage, Springer *erlag, 1968,
Kapitel H. Antennen.
4. Jasik, H., "Antenna Engineering Handbook", McGraw-

Hill Book Company Inc., 1961.
'I
5. Ulaby, F.T., Moore, R.K. and Fung, A.K., Micro-
wave Remote Sensing", Vol. I and II, Addison-
Wesley Publishing Company, 1981 and 1982.
6. Ulaby, F.T. and Dobson, M.C., "Handbook of Radar

Scattering Statistics for Terrain", Artech House,
Inc., 1989.
7. Colwell, R.N., Simonett, D.S. and Dlaby, F.T.,

"Manual of Remote Sensing", Second Edition,
Vol. I, The Sheridan Press, 1983, Chapter 9 and
10.
8. Jordan, R.L., Huneycutt, B.L. and Werner, M.,

"The SIR-C/X-SAR Synthetic Aperture Radar System",
Proc. IEEE, Vol. 79,No. 6, June 1991, pp. 827-838.
9. Jatsch, W., Langer, E., Ottl, H. and Zeller, K.H.,

"Concept of an X-Band Synthetic Aperture Radar for
Earth Observing Satellites", JEWA, Vol. 4, No. 4,
1990, pp. 325-340.
10. 6ttl, H. and Wahl, M., "X-EOS, a Multi-Mode X-Band

Synthetic Aperture Radar for EOS SAR", under Pu-
blication in Proc. of ISY Conference, Munich,
March 29 - April 4, 1992.
r.j ...
~5-1
Polarization Effects and Multipolarization SAR
A. Freeman

Pasadena, CA 91109
USA
Introduction Polarization of Electromagnetic Waves
Imaging radar polarimetry has excited much attention The polarization of any electromagnetic wave can be
in the literature over the past few years ([-161), since characterized by the ellipticity angle, X. and the
NASA/JPL first demonstrated a successful SAR
polarimeter system in 1985 [I]. That system was known orientation angle, Vj, of the polarization ellipse, shown
as the CV-990 L-Band radar; NASA/JPL now have a fully in Figure 1. The intensity of the wave is represented by
operational, three-frequency polarimetric SAR flying the parameter Io, where
on a DC-8 platform, which has taken part in many
science data acquisition campaigns since 1988 171. 1,= al+Q (I)
Several other institutions have radar polarimeter
systems which are operational or under development, Some commonly occurring polarization states are:
including the University of Michigan, MIT/Lincoln
laboratories, the Environmental Research Institute of Vertical (linear) - (W - 00. X 00)
Michigan (ERIM) 181, the Canada Center for Remote
Sensing (CCRS). FEL-TNO in the Netherlands and the Horizontal (linear) - (4f - 900. X - 00)
Technical University of Denmark. NASA/JPL are Right-hand circular - (X 450)
currently completing the construction of a two-
frequency, fully polarimetric SAR system which will Left-hand circular- (X- 450)
fly on the Space Shuttle in 1993/1994 191. These systems
generally transmit and receive horizontally (H) and For the two circular polarization states above, the
vertically (V) linearly polarized electromagnetic fields. orientation angle. , is unspecified. Another useful way
Imaging radar polarimeters are usually implemented of characterizing
electromagnetic the
wave is polarization
the Stokes vector, state of an
using a Synthetic Aperturc Radar (SAR) approach to e
give a high resolution image in two dimensions: range
and azimuth. For each pixel in the image a polarimetric
SAR gives sufficient information to characterize the
polarimetric scattering properties of the imaged area
F =
[ I
U
V
Locos2 cos X
sin2V cos2X
1. sin
2
X
2
1o (2)
(or target) as seen by the radar. Using a polarimetric

SAR system as opposed to a single-polarization SAR
system provides significantly more information about where Io, Q, U and V are the four Stokes parameters.
the target scattering mechanisms and allows better which all have the same dimension. The Stokes
discrimination between different types of surfaces.
In these notes a brief overview of SAR polarimetry is

offered. The notes are intended as a text to accompany a
lecture on SAR polarimetry as part of an AGARD-NATO
course. For a more in-depth treatment, the interested
reader is referred to the recent review paper by Zebker
and van Zyl I101 and the textbook on 'Radar Polarimetry aV
for Geoscience Applications', edited by Ulaby and Elachi
IlIl. For a discussion of inverse methods to determine
scattering phenomena for polarimetric radar data the
interested reader is referred to the textbook on 'Inverse
Methods in Electromagnetic Scattering'. edited by
Boerner [121. For brevity, a discussion of the
development history of polarimetric radar has not been
included in these notes, nor has a comprehensive
bibliography. The reader interested in more
informationreferences
examining on these (101-1121.
topics would do well to begin by
Covered in the notes are: the polarization properties of
electromagnetic waves; the concepts of radar scattering

and measuring radar backscatter with a SAR;
polarization synthesis: scattering matrix, Stokes matrix
and covarlance matrix representations of polarimetric
SAR data; polarization signature plots; design and
calibration of polarimetric SAR systems; polarization
filtering for target detection; fitting a simple model to Polarization Ellipse
polarlmetric SAR measurements of naturally occurring
features; and supervised classification of polarimetric Fig. I Polarization ellipse.
SAR data.
5-2
parameters are related via:

Ig . Q2 +U
2 2
+V (3) Measuring Radar Backscatter with a SAR
system
so that only three of them are actually independent. A SAR system is simply a high-resolution active
microwave sensor, capable of measuring the (real or
The polarization state of an electromagnetic wave can complex) radar reflectivity of a surface. SAR's usually
be represented as a (unique) mapping to a point on a operate as monostatic radars, mounted on a moving
sphere of radius 1o. called the Poincare' sphere, where platform, with an antenna looking out and down
Q. U, and V are the Cartesian coordinates of the point, towards one (or both) sides relative to the platform
The angles 2Z and 2i define the latitude and longitude motion vector. Most SAR systems measure the radar
of the point in the spherical coordinate system. The backscatter using linearly polarized antennas,
four common polarization states listed above are shown typically with polarizations given by the & and 1
as points on the Poincare' sphere in Figure 2. coordinates in Figure 3. In its ideal realization, the
resulting SAR image should just represent the
scattering matrix element, Spq. in complex
0
L-circular representation, or the radar cross-section, pq, in
(X 45) intensity, for the receive (q) and transmit (p)
polarizations of the radar. The scattering matrix
Horizontal determines the relationship between the wave incident
(V= 90o, X = 0)) on the scatterer and the scattered wave. After [IIi] the
scattering matrix is defined via:
s
SE Sh, Sw EV
2V 'where E is the electric field vector of the wave
incident on the scatterer, E ) is the electric field
a Vertical vector of the scattered wave. ko is the wave number of

(V =0, X =0) the illuminating wave, and R the (radial) distance
between the scatterer and the radar antenna. A fully
polarimetric SAR system would typically measure all
R-circular four of the scattering matrix elements simultaneously
(X = -450) (or near-simultaneously) as complex numbers for each
pixel within the area being imaged by the SAR.
Fig. 2 Poincare' sphere. Polarization Synthesis
R t sKnowledge calculation ofofthethe scattering matrix allows the

backscatter intensity, or radar cross-
In Figure 3 the general geometry for measurements section, for any possible combination of transmit and
made by a bistatic radar is illustrated. The transmitting receive antenna polarizations (e.g. left-hand circular
antenna transmits an electric field whose components transmit and right-hand circular receive). This
are expressed in terms of a local Cartesian coordinate procedure is called polarization synthesis. For any
system (h, i, i ) with origin at the transmitting given radar receive and transmit polarization, the
antenna. We can define another Cartesian coordinate radar cross-section (RCS) can be calculated [I I via: a
system (i, y, z with origin at the scatterer. The two
coordinate systems are related via: rpq= 4X 1iv S j-P f (6)
sin u Cos i (4a) where S is the scattering matrix defined in (I). and qr
pt are polarization field vectors for the radar receive
v = -cos cos 8i - sin @lcos0 1 y + sin Oi z (4b) and transmit polarizations, respectively. For linear
polarizations (horizontal, h or vertical, v) note that the
- coOs sin Oi sin sin
Oi + os Oi (4c) RCS is given by:
{Yq )is 4x s F (7)
r Athird coordinate system (
origin at the antenna which receives
, n is defined with
the In what follows we will concentrate on linear
electromagnetic wave which is scattered by the polarizations, since most SAR systems just measure
scatterer. This coordinate system is related to the linear polarizations.
scatterer coordinate system in a similar fashion to the
above, substituting subscripts s for subscripts i in (4). Expressions (5) and (6) above represent the quantities
This choice of coordinate systems ensures that the which are directly measurable by a SAR. The units for
primed (h, ;, ) and unprimed ', v', ) coordinate Gpq are in meters squared. Both Opq and Spq are
systems coincide for a monostatic system, i.e. when the functions of spatial position I (, - ) in the scatterer
receiving and transmitting antennas are at the same
location. The radar scattering measurements made by a coordinate frame] . time (t). viewing geometry (Oj, 0
monostatic radar system are referred to as radar and radar wavelength 00) in addition to the
I7~
backwcatter measurements. polarizations of the transmitted 'and received
. ______
5-3
' Receive
Anten r--a
TransmitA
Antenna
I z -
I -/
I - - I
I a .
Scatterer
xY
Fig. 3 Coordinate systems for radar scattering.
electromagnetic waves. If the radar backscatter

5
measurements made by the SAR are calibrated, each M 14 = 0.51m (SI S&) + 0.51m (Sh: w )
image pixel value should correspond to an accurate and
precise measurement of one of these quantities, and be 25 2
repeatable, under the same conditions, by that c:2 = O. (Shm Sh, - Shv Sh* + SW S)
another sensor.
Stokes matrix and Covariance matrix M 23 = 0.5 Re (SI Sh, ) - 0.5Re (Sh, S, )
representations M 24=O0.51m (Si;Sh, )- 0.51m (Sh: S,
In the monostatic scattering case, it can be shown that
the reciprocity principle [101, which dictates that M 33 = 0.5 (Sh Sh,,
) + 0.5Re (S,; S, )
Shy = Svh (8)

M 34 = 0.51m (ShIS,
is generally applicable. This simplifies some of the
analysis of polarimetric SAR data and quantities derived M 44 = 0.5(Sh Sh,*)=0.SRe (SI S, (10)
from it. For example,
representing the polarimetri
another useful way of I
atring information The first element. M I I . in the Stokes matrix is often
covariance matrix between the scattering matrix referred to as the total power. It is related to two of the
elements. Without reciprocity the covariance matrix other Stokes matrix elements via:
would have to be a 4x4 matrix containing all possible
cross-products between the four scattering matrix MII=M22 +M33 +M4 (II)
elements. With reciprocity a 3x3 covariance matrix is
sufficient, i.e., Note that it is possible to recover the covariance matrix
elements from the Stokes matrix elements. It is not
S,Sh, SW possible, however, to recover the original scattering
Shh4Shh
• I matrix from either, since the overall or absolute phase
C = Shy S;u Sh, S,* Sy Sv (9) of the scattering matrix (for example the phase of the
S Sw SShh term) has been lost in forming the cross-prodcts.
S. S* S.,-Sv W S" The relative phase between the scattering matrix
elements is preserved in the covariance matrix and
The covariance matrix is also Hermitian, so that three Stokes matrix formats, in terms such as the Shh S
of its elements (S&. 5 ,Sw Si; and Sw S/r ) contain cross-product, whose argument is the relative phase
redundant information, which is already contained in between the Shk and Sw terms.
three of the other elements.
It is still possible to carry out polarization synthesis on
Another way of representing the cross-products the covariance matrix and Stokes matrix format
derived from the scattering matrix elements is in the polarimetric SAR data to synthesize the radar cross
Stokes matrix format [131. For reciprocal scatterers the section measured by a radar of arbitrary transmit and
Stokes matrix M is a 4x4 symmetric matrix, with the receive polarization [101. A common technique
following elements: employed on SAR measurements of radar cross section
to reduce the variance due to speckle is to perform
M II = 0.25(Shh Shh + 2Sh, Sh + SW S ) incoherent averaging or multi-looking. Using the
scattering matrix data, the synthesized radar cross
section would be calculated for each pixel using (6). The
M 12 = 0.2 5 (SM Sk - SW S") calculated radar cross section values would then be
averaged to reduce the speckle variance. Multi-looking
M 13 = 0.5 Re (S/ S& )+ 0.5Re (SZvS, can also be carried out by averaging the covariance
matrix or Stokes matrix elements for several pixels to
produce a single composite result, then performing
I
5-4
polarization synthesis on the composite result [14). Tr,-

two approaches can be shown to be mathematically
equivalent and give the same answers, but there is a
considerable saving in computation time using the
covariance matrix and Stokes matrix approach.
Example Scattering Matrices
Table I gives examples of some simple scattering

matrices for selected man-made targets. The targets are
the trihedral corner reflector and the sphere (which
have the same scattering matrix), the dihedral corner
reflector, the dipole and the polarimetric active radar
calibrator (PARC) [151. Only the scattering matrices for
the trihedral and the sphere show no dependence on
the rotation of the scatterer about the vector f which
defines the line of sight between target and radar. By
rotating the dihedral, the dipole and the PARC about the
line of sight it is possible to change the scattering
matrix for these targets as seen by the radar. The Fig. 4 L-band total power imageofSanFranciscofrom
scattering matrices corresponding to more complicated
man-made targets, such as vehicles or planes, may be the NASA/JPL system.
very different from the simple ones given in Table I.
In Figure 4, we see a total power image of San

Francisco. California which was derived from L-band
(24cm) polarimetric SAR data collected by the
NASA/JPL DC-8 SAR system [71. The image is interesting
because it contains three distinct types of scatterer: the
man-made buildings and other structures in the urban
areas; the largely vegetation covered areas in the
90
Golden Gate Park and the Presidio (just south of the
olitation 0 or'entatim
Golden Gate Bridge); and large stretches of open water
in the bay and out in the ocean. 180 -45 180 _t ellpticity
CO-POLARIZED SIGNATURE CROSS-GLARIZED SIGNATURE
for visualizing
In Figure 5. an example of a useful tool
the polarimetric scattering properties of a target.
known as the polarization signature plot, is shown. In
this case, the polarization signature plot was calculated
from a composite Stokes matrix formed from pixels in
one of the urban areas in Figure 4. After [131, the Fig. 5 Polarization signature for a predominantly
polarization signature plots represent the (synthesized) urban area in the San Francisco image.
response of the target to all possible like- and cross-
polarized radar transmit/receive combinations. The
Table I: Scattering matrices for man-made targets, normalized so that the largest
matrix element is one
Target type Scattering matrix
Trihedral, Sphere (1
0 0)
1
Dihedral (cos 2a sin2a)

sin 2a -cos 2a
Dipole ( cos % sin a cos a)

sin a cos a sin 2a
PARC ilC a Ok
-sin 2a -sin a cos a)
[a is a rotation angle for each target type, about the line of sight to the radar.
a-valies of 00, 450 and 900 are commonly used, in practise.)
5-5
polarized signature plots have no zero points for any

orientation angle/ ellipticity angle combination. The
like-polarized signature plot is similar in form to that
of the water, except that the polarization signature of
the vegetated area appears to sit on a 'pedestal'. The
presence of such a pedestal indicates that the individual
4 45 scattering, covariance or Stokes matrix measurements
0 oneotatlon used to calculate the polarization signature plot for the
area in question were not identical. In fact, the height
emti¢ity ty
ehiptic of the pedestal can be related to the variance in the
CO-DOLAPIZED SIGNATURE CRCSS-PCLAPIZED SIGNATURE measurements. Note that the polarization signature
plots for the water shown in Figure 6 do not have a
significant pedestal, which suggests that the
Fig. 6 Polarization signature for an ocean area in the measurements averaged to produce those plots were
San Francisco image. fairly similar.
Design of Polarimetric SAR systems
An example of a polarimetric SAR system is NASA/JPL's

DC-8 SAR system, first tested in January 1988. which
operates at three frequencies (L-, C- and P-Band) and
four polarizations, HH, HV, VH and VV (in this notation
HV means Horizontal polarization on transmit, Vertical
on receive). This system is the latest in a series of
45 90 45 NASA/JPL synthetic aperture radar systems that have
yrentrton 0 orentatron 0 been designed, built and tested primarily by the Radar
8 - 180 -45 Science and Engineering Section at the Laboratory.
e A tCAOS
> S Z The radar will serve as a test-bed for SIR-C with which
eShGNATURE
'ECZSIC)AURE COSS- CLAOIZED SIGNATURE it has similar characteristics, as well as acting as a
useful science tool for the remote sensing community
in its own right.
Fig. 7 Polarization signature for a vegetated area in the
San Francisco image. A detailed description of the NASA/JPL's DC-8 SAR can
be found in [71. A block diagram of the basic
transmit/receive circuitry required for just one of the
polarization signature plots are given as functions of frequencies is shown in Figure 8. This basic design
orientation and ellipticity angle, and are normalized contains just one transmitter, two receivers and two
with respect to the total power. Note that the like- antennas (one H. one V). In the DC-8 SAR case, the
polarized polarization signature in Figure 5 shows a radar system generates a pulse at L-Band which is
maximum response when the orientation angle is 900 or frequency shifted using a common reference oscillator
00 (1800). This behavior, and the double-null structure to generate the P- and C-Band pulses. After
,
visible for (W = 450 and 1450 X - 00) is very similar to amplification at each frequency, the transmitted chirp
visiblexp for 4r ar a d a w is alternately polarized by the operation of a switch to
that expected for a dihedral corner reflector with either the H or the V antennas. The antenna patterns
rotation angle, a -00. Thus the polarimetric radar for each of the frequencies cover approximately the
backscatter from the urban area is seen to be similar to same range of incidence angles, between 20" and 70".
that produced by two flat plates; one flat on the ground. On receive, the radar can collect both H and V channel
the other vertically upright, data simultaneously at all three frequencies, the six
receiver channels being entirely separate. The
The scattering matrix model for Bragg scattering from receivers have no Sensitivity Time Control (STC) or
an idealized rough surface, such as wind-blown water, Automatic Gain Control (AGC), but the receiver gains
is: can be adjusted over a 26dB range of settings. A wide
instantaneous dynamic range of "45dB is achieved by
sa O)With a,brealb~a0 using 8-bit analog-to-digital converters (ADC's), which
S _= (12) are clocked together to ensure cross-channel
0 b and <ab*> - ab coherence. The ADC's each operate at 45MHz and
produce real data samples (not I and Q) at a sustained
i,e.,a scattering matrix with zero cross-polarized data rate of between 20 and 60 MBytes/second.
return, HH and VV returns which are completely depending on the data collection mode.
correlated and zero phase difference between tlhe HH
and VV returns. The ratio a/b and the absolute The operation of the DC-8 SAR is such that alternate H
backscatter level are dependent on the surface and V polarized pulses are transmitted. The return
roughness (i.e., the sea-state in this case) but need not echoes are received by both the H and V receivers
be known a priori. Although we specify a rough simultaneously. Thus for one H-polarized transmitted
surface for (12) to hold, it should not be too rough, pulse the H- and V-polarized return echoes are
since second order Bragg scattering can give rise to recorded (HH and VV returns); the next transmitted
significant cross-polarized backscatter. The pulse will be V-polarized and the VH and VV return
polarization signature corresponding to a patch of echoes are recorded. This process is repeated
throughout a data-take. Then the recorded return
water from the San Francisco image is shown in Figure echoes undergo SAR processing (range and azimuth
6. We see that the like-polarized signature shows that compression) to produce SAR images corresponding to
the VV return isgreater than the HH (i.e. b 'a) and the each of the set of HH, HV, VH and VV return echoes. The
linear (X - 00) polarization response in the cross- SAR processing operations applied to each polarization
polarized signature is zero for all orientation angles. are identical. After processing, the HH and HV images
must be resampled in the azimuth dimension so that
In Figure 7 we show the polarization signature plot they are registered with the VH and VV images: prior to
corresponding to a largely vegetation-covered area in this operation the two sets of images will be
the Golden Gate Park from the San Francisco image. In
this case we see that the cross-polarized and like-
misregistered by one pulse repetition interval. After
registration the scattering matrix measurements can be i
5-6
Circulator
H ReciverH Antenna
transmitter }P Switchi~
~Polarization
V Receiver V Antenna
Circulator
Fig. 8 Simple block diagram for a polarimetric SAR

system.
considered to be spatially coincident. i.e. they - i
correspond to the same patch of ground. - S + (R)'N (T) (14)
Polarimetric Calibration
(provided R R and T =T. with denoting an
Calibration of the data produced by polarimetric SAR's estimate). Note that the result still has noise
(e.g. 111. 1161, 117]) is necessary if the data is to be used superimposed on it. At the complex SAR image stage
in a quantitative fashion. Goals for calibrating (which is what polarimetric SAR scattering matrix data
polarimetric SAR data are given in Table 2. The basic is), although the mean noise contribution is zero, the
problem in calibrating polarimetric radar data can be rms uncertainties due to noise can not be removed.
seen from the following expression, after (181. for the rsucranisdet os a o ermvd
meeasuremets e byllowg
axporeric, ar .forwhh After square-law detection, however, the average noise
measurements made by a polarimetric radar for which poe ca bestrtdofth rsuin RC
R ad
ysem Tmatixode isvald:power can be subtracted off the resulting RCS
the 2x2 R and T matrix system model isvalid: measurements (though this still does not remove the
rms uncertainties due to noise). Polarization synthesis.
Ma,'Wei' Shh Svh 16 in which the target RCS, Gpq, in response to a radar
with transmit polarization p and receive polarization q
82 fl Sh Sw 84 f 2 is synthesized from the scattering matrix, involves a
square-law detection step, so the noise bias term can be
rnhh nlh subtracted off the synthesized image pixel powers,
I nhv
+ v ~n( nvv J provided it can be estimated.
We characterize the (complex) noise terms as having
nN
V=K/ e*,Rt ST + v/- (13) two-dimensional, zero-mean, Gaussian distributions.
with the following properties:
where 81 and 82 are the HV and VH cross-talk or

83 and 84 are (nik) - 0
polarization impurity terms on receive,
the HV and VH cross-talk terms on transmit, fl is the =-jni;
channel imbalance (amplitude and phase) between the
H and V channels on receive, and f 2 is the channel (lkfm 0 , for j I or k# m.
imbalance (amplitude and phase) between the H and V
channels on transmit. We shall ignore the radiometric
Ks - I, and kikSlrn)= 0,for any j,k, 1,m. (15)
for the moment, setting
calibration problem
absorbing the - npq .factor into the individual noise
matrix elements, We shall also ignore the absolute where is the noise power (or noise-equivalent
phase calibration problem for now, setting *s

- 0. The sigma-zero) in the polarization channel jk. We assume
polarimetric radar data calibration problem can then that the noise terms are uncorrelated with each other
S and with the scattering matrix (signal) terms.
be stated as follows: to solve for the scattering matrix
from the measured matrix M we need estimates for the An important point to note is that, in equations (13) and
matrices R and T, which represent the rada (14), we have not included any (U. y) dependence in
polarimetric characteristics on receive and transmit. any of the terms. In practice, the elements of R. T and
N may vary significantly within a SAR image, i
Ideally, R and T are identity matrices for a perfect
radar polarimeter. Once we have obtained estimates for especially in the across-track dimension. Depending on
R and T they can then easily be inverted and applied to how quickly the elements of these matrices vary, it may
(13) to recover an estimate for S. i.e., be necessary to calculate them at many different points
across the image. This is a strong factor in deciding
wh..~l~at (se belo).
pietric adroc aibaigdaafo
what type of approach to adopt in calibrating data from
S= (R)M (T) aPoltrimetric radar (see below).
5-7
The validity of the majority of the polarimetric the problem, we need at least another six equations in
calibration algorithms in the literature depends on the the various terms involved. There are three approaches
validity of the system model above for radar to obtaining the required six equations: the first is to
polarimeters, which was first put forward in [I]. This use man-made targets with known scattering matrices;
system model contains just six relative parameters, the second is to make assumptions regarding the
including four cross-talk (or leakage) terms and two general properties of the backscatter being measured:
channel imbalance terms, one for the H,V transmitters and the third is to make assumptions about the radar
and one for the H,V receivers. Determination of these system parameters. Of course, all three of these
six parameters, followed by correction for any approaches may be combined to find an appropriate
deviations from the ideal is then sufficient to calibrate solution.
the radar data polarimetrically, so that the HH, HV, VH
and VV scattering matrix measurements can be The first attempts ([201 - [221) to calibrate polarimetric
meaningfully compared. radar systems used only the man-made target
approach. There are only a limited set of such targets
The generally accepted polarimetric system model available. Table I lists the normalized scattering
relies on the constancy of the system, particularly the matrices for some of the more common ones. Barnes
transmitters, the receivers and the antennas. For 1211 and Woods 1221 both used combinations of three
instance, it assumes that the ratio of the power passive devices to come up with the six additional
transmitted between H and V polarizations never equations required to solve the problem. In his
varies, and that the receiver gains for H and V are kept approach, Woods 1221 used a trihedral and two dihedrals,
.
one at 00 rotation, the other at 450 Barnes 1211 also used
at a constant level relative to each other. This is not this approach, and presented another solution using
always the case for radar polarimeters [191. For two dipoles (at 00 and 900 rotation angles) and one 450
example, in operating the NASA/JPL system the gain of dihedral. The dipole signatures were achieved using
the H (or V) receiver can be switched by up to 6dB, screened trihedral corner reflectors ([231, [241). Three
depending on whether a like-polarized (HH or VV) or targets were found to be necessary to solve the problem
cross-polarized (VH or HV) return is expected. [ As because it was very difficult to obtain more than two of
described above, the NASA/JPL system has two the required equations from each target signature.
receivers, one for H and one for V, and alternately Tueh et al [251 presented a more general solution using
transmits H and V pulses!. The reason for operating the passive targets, in which at least two must either have
radar this way is that the cross-polarized returns from singular or non-singular scattering matrices, and the
natural targets are usually lower than the like- targets must not be the same. Whitt and Ulaby [261
polarized returns, so the gain is increased when showed how the problem may be solved provided just
expecting a cross-polarized return to keep the signal one of the three target scattering matrices is non-
level within the optimum range for the analog-to- singular. Freeman, Werner and Shen [151 showed how
digital converters. This gain differential should be three polarimetric active radar calibrator (PARC)
removed during processing. The ERIM P-3 polarimetric signatures could be used to solve the problem. The
SAR solves the same problem with a different design, sensitivity of all of these approaches to small errors in
havigtansmtte,
oly oe on reeive an an the rotation angles of the devices used was pointed out
automatic gain control (AGC), which varies the gain of in [191.
the radar receiver in some pre-set fashion [8]. The AGC
setting in amplitude and phase is usually different for
HH, HV, VH and VV returns, which are collected in and the backscatter to arrive at a solution to the
sequence. The design for the CCRS polarimetric SAR has
made the following assumptions
Zyl [27! radar
an elaborate switching sequence for it's two receivers, about the van
problem. polarimetric system:
so that HH and VV returns go through the same
receiver, and VH and HV returns go through the other
receiver. In none of these cases is the 2x2 R and T 61 =
matrix system model necessarily valid, unless the gain 62 864
differentials between receivers are properly removed
from the data. fI f2 (17)
Let us now return to the problem of estimating R and T, i.e. R - T. These are three of the required equations.
in order to calibrate the scattering matrix data. van Zyl also assumed that the backscatter was
Expanding (13), we obtain four equations relating the reciprocal, i.e. Shy - Svh, which would at first seem to
scattering matrix measurements with their actual provide a fourth equation, but on further examination
values: of equations (16b) and (16c) this makes one of them
redundant. so the net number of equations does not
Mhh - Shh + 8 2Shv + 84Svh increase. Next, van Zyl assumed that for most natural
++ nhh (16a) backscatter types, azimuthal symmetry holds [3]. for
8284vv hhwhich the like- and cross-polarized backscatter terms
are uncorrelated, i.e.,
+
Mhv- 81Shh fIShv + 8 184Svh (16b)
+ fl84Svv + nhv (Shh Shv* 0
8 + 8 8 +
Mvh - 3Shh 2 3Shv f2Svh
(0hvSIv 0
+ f282Svv + nvh
This gave two more equations, which allow the
8 8
Mvv - 1 3Shh + fl83Shv + f28lSvh (16d) calculation of the two cross-talk terms, 81 and 81. These
+
+ flf2Svv nvv can vary significantly with incidence angle as can be
seen from the plots shown In Figure 9. The solution was
completed by estimating flf2 from the ratio of the HH
These four equations contain ten unknow i quantities and VV measurements from a trihedral. From (16a) and
(ignoring the noise), which are the four (complex) (16d) It Is easy to see that for a trihedral, with
scattering matrix elements and six (complex) radar scattering matrix as given in Table|,
system unknowns (the 8's and the f's). Clearly, to solve
5-8
Table 2. Calibration Goals
Long and Short-Term Relative Calibration (Between ±1 dB

passes and within an image frame)
Absolute Calibration (any channel) ±3 dB
Cross-frequency calibration ±1.5 dB
Polarization Amplitude Imbalance (between ±0.4 dB

polarization channels) (2-way)
Polarization Relative Phase Calibration (between ±100

polarization channels) (2-way)
Polarization Cross-Talk error (isolation) -30 dB
a) Cross-talk amplitudes b) Cross-talk phases

0- 180
140
13 00
-20a 60-
" 20
-30 ' 2
-40 -60-
-loo
; -:4 n~ delta
-6 0 .............-- ---- -------.

... ,. ..--. ,. .. . . , .. ,. .
5 0 15 20 25 30 35 40 45 50 55 60 5 10 15 20 25 30 35 40 45 50 55 60
Local Incidence angle (degrees) Local Incidence angle (degrees)
Fig. 9 Plots of cross-talk amplitude and phase for the

NASA/JPL L-band polarimetric SAR system.
showing variations as a function of elevation (or
incidence) angle.
radar system terms to three. Klein and Freeman (301

M- fl f 2 (19) showed that, if backscatter reciprocity were assumed.
Mhh then only two man-made targets were necessary to
solve the polarimetric calibration problem.
provided second order terms in the 8's and noise terms Sarabandi, Ulaby and Tassoudji [311 simplified the
are ignored. problem by assuming that all four cross-talk terms
were zero, using backscatter reciprocity and an
In an alternative approach, Klein 1281 showed how just estimate of flf2 from a trihedral or sphere to complete
the assumption of backscatter reciprocity could be used the solution. Whitt and Ulaby 1321 proposed that there
to give three equations relating the system parameters, exists a rotation of the H-V coordinate system for which
which could replace those in (30). The remainder of the 84 - 0. Then backscatter reciprocity can be used to give
equations in Klein's approach were similar to van Zyl's. three more equations and a single trihedral signature
This was followed by an analysis of the problem in
Freeman et al 1291, in which it was shown that the can be used to solve for flf2 and 81, completing the
assumptions underlying van Zyl's and Klein's solution. This approach gives a solution for which
approaches were similar and that the backscatter there is an arbitrary (small) rotation of the H-V
reciprocity assumption could be used in a coordinate system, although they are orthogonal.
transformation to reduce the number of unknown Sarabandi and Ulaby 1331 used the simplifying
,m~m
mm lmmlln Il i i
mallmmm
5"
5-9
assumptions that the different pairs of cross-talk terms is short (less than 10 meters), densely packed, and has
were equal (but not that f I - f2). This was followed by a sparse foliage. Yet another wetland forest type is
rotation of the H-V coordinate system such that 81 - 82, swamp forest, consisting of short trees or shrubs
with a trihedral signature then yielding the remaining growing in standing water. Besides forest, there are
three equations. reed and sedge marshes. These are areas of herbaceous
vegetation in standing water. The sedges are
Calibration of the phase difference between considerably taller than reeds (200 cm versus 40 cm).
polarization channels (i.e. determination of arg(fl) and In addition to these naturally occurring vegetation
arg(f2) alone) using corner reflector signatures and types there are agricultural areas consisting of cleared
areas, bare soil and various crops.
signatures of moderately rough surfaces is addressed in
1341. An examination of the assumptions on the system
and backscatter behavior required for this approach to Table 3 contains typical radar backscatter
be successful was presented in (351. measurements for the different vegetation types
identified in the area imaged by the DC-8 SAR. From the
Polarization Filtering table it can be seen that the very low radar cross
sections correspond to open water and bare ground at
We have seen that it is possible to synthesize the all three frequencies. The upland forest spans a fairly
response of any target to arbitrary combinations of narrow range of cross-sections. The range of radar
transmit and receive polarizations from measurements cross sections is given in the table by the two values
of the scattering matrix, the covariance matrix and the shown in the table. Radar cross section tends to
Stokes matrix. We have also seen from the example increase as the angle of incidence decreases. If
polarization signatures that transmit/polarization incidence angle effects are considered, the radar cross
combinations exist for which the synthesized radar section for the upland forest is essentially constant. For
cross-section is maximized or minimized and that the other types of vegetation, at P-Band we see a steady
particular combinations of transmit and receive climb in the HH cross section as the biomass of the
polarizations for which the maximum and minimum canopy increases until we reach a plateau where the
RCS values are reached will depend on the type of upland forest begins. The same feature is visible in the
scattering (e.g. Bragg, dihedral mechanism). Consider L-Band and C-Band data, but at L-Band the plateau is
the problem of trying to detect an isolated target (e.g. a reached for re-growth vegetation (2-3m in height).
ship) against a clutter background (e.g. the ocean) while at C-Band the plateau is reached for farmland
using a polarimetric SAR. In this case we would choose vegetation (<in in height). This type of saturation of
the transmit and receive polarization combination radar backscatter with biomass has been observed
which maximized the ratio of the synthesized target RCS previously. At all three frequencies, especially C-band,
over the average of the synthesized clutter RCS. i.e. the we then see a secondary increase region in the HH
signal-to-clutter ratio: radar cross section data. The sites corresponding to
these data points may have much less biomass than the
lI ) upland forest, but they have a brighter radar return.
t some feature which adds a
SCR . - q(aVr, Xr) ST P(Wt, X) (20) These areas generally have
secondary scattering mechanism in addition to the
q(vr,t
Xr~ Sc p(. At". Ivolume scatter which dominates the upland forest. One
site was a rain forest area where the trees had been cut
SCR is also known as the contrast ratio. Several authors but not cleared (clear-cut in Table 3). The site had an
have offered mathematical solutions to this problem enhanced HH polarization return due to the many tree
([361-1381). The first step is to characterize the trunks laying on the ground. Other sites with
scattering properties of the target and the clutter, i.e. on
to enhanced backscatter were sites with vegetation
growing in standing water (reeds. sedge, high marsh
determine ST and SC. Then the polarizations
transmit and receive which maximize the SCR have to forest, swamp forest, flooded Bajo). In this case the
be determined. Application of those transmit and backscatter probably consists of volume scatter from
receive polarizations to an entire image is known as the vegetation canopy plus double bounce scattering
polarization filtering, due to interaction between the vegetation and standing
water. In other sites, the double-bounce mechanism
Radar mapping of vegetation types may be enhanced because the vegetation understory is
relatively clear (in the areas designated coffee and
Due to the concern in the scientific community over palm forest, for example).
the global warming problem. mapping of vegetation
cover (and changes in vegetation cover), especially in A simple, three-component scattering model
tropical rain forest areas, using remote sensing data
has received considerable attention recently. Synthetic We can begin to understand the phenomena which give
Aperture Radar, with it's ability to collect data during rise to the radar backscatter characteristics seen in
day or night. or in cloudy conditions, is ideally suited Table 3 and in other polarimetric SAR images of natural
for monitoring the earth's remaining tropical rain surfaces by using a very simple conceptual model of
forests. scattering. This model is more fully developed in [391;
here an outline is presented. We assume that scattering
In what follows, we present multi-frequency. from a veRetation laver is a combination of volume
polarimetric SAR results from a tropical rain forest site scattering from vegetation, double bounce scattering
in Belize, Central America, as an example of the ability from the ground/trunk interaction and scattering from
of polarimetric SAR to map different vegetation types. a rough surface. For volume scattering, we assume that
An image of a 10lOkm area is shown in Figure 10. The the return is from randomly oriented, very thin
area under study is dominated by old growth upland cylinder-like scatterers. The double bounce scattering
tropical rain forest. There are also small areas which component can be modeled by scattering from an
have been cleared of upland forest within the last five upright (rotation angle - 00) dihedral corner reflector .
years and allowed to re-grow. In addition, there are In the model, both surfaces in the reflector are made of
areas of almost pure palm forest, consisting of either dielectric material, and the two surfaces may have
cohune or botan palm trees. There are several types of different dielectric constants, corresponding to trunk
wetland forests. High marsh forest has vegetation and ground. The trunk has Fresnel reflection
wateris coefficients and Rtv forLikewise. the and
horizontal vertical
resembling the upland
is present during
rainportion
a large thestanding
forest,ofbut year. Bajo polarization, Rth
respectively. gr6und has
another type of wetland forest which grows in areas Fresnel reflection coefficients Rsh and R3 p. For the
having very poorly drained clay soils. The vegetation
~5-10
Vegetation -and D -and
0
Class OHH HVV P HH/VV HV/HH (JHH IHHVV P HH/VV HV/HH HH 44HVV P HH/VV /U
Open Water -32.6 5.8 033 -6.6 -7.4 -28.7 -3.1 0.63 -7.7 -6.6 -23.7 -18.7 029 -2.4 -5.3
Bare Soil -25.1 -8.8 0.75 -5.4 -9.5 -16.5 -23.7 0.75 -1.8 -10.4 -9.0 -10.0 0.76 -0.3 -8.9
Reeds -23.0 115.3 0.72 -10.9 -10.7 -9.0 98.1 0.54 -3.3 -11.6 -0.3 135.8 0.44 22 -11.9
Farmland -20.3 -12.5 0.53 -2.0 -11.5 -13-3 -18.6 0.75 -1.5 -11.7 -7.6 -12 0.47 0.7 -6.4
Regrowth -15.3 20.7 025 -0.9 -5.8 -10.1 -22.7 0.31 -0.1 -5.5 -5.7 3.2 0.44 0.7 -6.5
Bajo -14.5 70.8 021 -2.8 -5.8 -9.7 15.1 0.17 0.1 -5.6 -6.2 -2.1 0.42 -0.2 -5.6
Swamp Forest -13.8 149.5 0.1 -0.6 -8.4 -6.9 165.4 0.06 0.4 -7.6 -2.1 2.3 0.41 1.2 -8.9
Upland Forest -12.8 32.3 022 -0.8 -5.4 -9.2 7.9 0.25 0.2 -5.1 -7.4 9.0 0.46 0.0 -5.8
Upland Forest -11.5 51.1 0.14 0.4 -6.4 -9.2 7.9 0.25 0.2 -5.1 -5.6 -4.1 0.51 0.6 -6.6
High M. F. 2. -11.0 6.4 0.16 -2.3 -5.3 -8.4 7.8 0.16 0.2 -5.2 -5.3 -1.4 0.43 0.6 -5.8
Palm Forest -11.3 91.4 021 -0.1 -7.2 -8.6 48.5 0.2 0.6 -6.4 -5.0 -2.4 0.56 0.4 -7.0
Sedge -10.3 93.6 0.18 0.5 -7.0 -8.8 -5.7 0.26 -0.2 -5.2 -6.1 -0.9 0.52 0.0 -6.3
Flooded Bald -9.7 75.6 0.16 0.9 -7.0 -8.6 31.4 0.22 0.6 -5.6 -6.4 -2.3 0.42 -0.9 -5.5
Coffee -9.2 137.3 0.4 1.1 -8.8 -8.0 52.1 0.12 1.7 -7.7 -6.6 6.9 0.42 1.4 -7.0
Clear-cut -9.0 31.0 0.32 3.0 -8.5 -7.6 6.9 0.4 2.0 -8.0 -6.0 8.3 0.55 1.4 -8.3
High M. F. 1 -8.2 -32.7 02.9 -1.6 -7.5 -4.0 -2.5 0.32 2.1 -7.8 -3.2 15.0 0.35 1.5 -7.0
" Cross section results and ratios are in dB; phase differences are in degrees
* High M. F. - High Marsh Forest
Table 3: Radar backscatter measurements from typical examples of the different

vegetation classes.
Vegetation - -band L-bsnd C-and

Class Odd Even Vol. Odd Even Vol. Odd Even Vol.
Open Water 0.67 0.09 0.24 0.76 0.01 0.23 024 0.05 0.70
Bare Soil 0.77 0.03 0.20 0.65 0.07 0.27 0.56 0.01 0.44
Reeds 0.02 0.93 0.05 0.16 0.67 0.17 0.11 0.59 0.30
Farmland 0.65 0.16 0.19 0.69 0.08 0.23 020 0.01 0.79
Regrowth 0.14 0.09 0.76 0.08 0.05 0.87 0.19 0.02 0.78
Bajo 0.10 0.26 0.64 0.02 0.11 0.87 0.14 0.01 0.85
Swamp Forest 0.16 0.37 0.47 0.09 029 0.62 036 0.12 0.52
Upland Forest 0.04 0.12 0.84 O.02 0.03 0.95 0.16 0.01 0.83
Upland Forest 0.05 0.18 0.77 0.00 0.05 0.95 023 0.01 0.76
High M. F. 2 0.04 0.23 0.73 0.00 0.03 0.97 0.12 0.00 0.88
Palm Forest 0.02 0.39 0.59 0.08 0.14 0.78 029 0.01 0.70
Sedge 0.07 0.25 0.68 0.02 0.06 0.91 024 0.01 0.75
Flooded Bajo 0.05 0.21 0.73 0.00 0.00 0.94 0.17 0.02 0.81
Coffee 0.00 0.46 0.54 0.07 0.25 0.69 0.23 0.04 0.74
Clear-cut 0.30 0.07 0.63 0.29 0.06 0.65 0.37 0.04 0.59
High M. F. I 0.30 0.19 0.51 0.24 0.07 0.69 0.19 0.05 0.76
Table 4: Backscatter mechanisms from the model fit for typical examples of the
different vegetation classes.
surface scatter, we assume a first-order Bragg model where fs. fd and fv are the surface, double-bounce and
(12) is adequate to describe the backscatter. For all of volume scatter contributions to the VV cross section, P
these three backscatter components, we assume that
like- and cross-polarized returns are uncorrelated. and is a real number and Eis given by:
that the backscatter is reciprocal (HV - VH). Now, if the
volume, double-bounce and surface scatter components Q _Rh R
are uncorrelated. the total second order statistics are a Rg Rt
the sum of the statistics for the individual mechanisms; Rgv Rtv (25)
i.e., the backscattered powers rather than voltages are
added. Thus our model for the total backscatter is: This model gives us four equations in five unknowns.
In general, a solution can be found if one of the
unknowns is fixed. Since neither the surface or double-
bounce mechanisms contribute to the HV term in the
h 2 fs 1 fd Gf (21) model, we can use this to estimate the volume scatter
contribution directly. The volume contribution can
2 then be subtracted off the IShh 12. ISvv 12 and ShhSvv*
IS~Ic
j " 1 +
fd +
fv (22) terms, leaving three equations in four unknowns. After
1401, we then decide whether double-bounce or surface
3
,-. fv/ (23) scatter is the dominant contribution in the residual
based on the sign of the real part of ShJSvv" . If Re(
ShhSvv" ) is positive, we decide that surface scatter is
ShbS,v - P fs " a fd fv/ 3
(24) dominant, and fix a - 1. If Re( SbSvv" ) is negative, we
J
r ~5-l11
Fig. 10 P-band (68cm) NASA/JPL DC-8 SAR image of a

tropical rain forest area in Belize, Central
America.
decide that double-bounce scatter is dominant in the performing supervised classification on polarimetric
remainder and fix 0 - I. Then we calculate fs, fd and 0 radar data. First training sets are selected, which
or Ot from the backscatter measurements. In Table 4 we represent different surface types. Then the elements of
the covariance matrix for each surface type can be
show the fraction of the total power contributed by averaged to find the average polarimetric scattering
each of the three mechanisms (with odd - surface, dbl properties for each class. Kong et al (411 developed a
double-bounce and vol - volume scatter) for the supervised classifier in which the following vector is
vegetation classes given in Table 3. formed for each pixel in the image:
From Table 4, we see that the model predicts that the(

dominant backscatter mechanism for the upland forest -
is volume scatter at all three frequencies. There is a X . (27)
small ('20%) contribution from the double-bounce
mechanism at P-Band, practically none a' L-Band. and Sw
none at all at C-Band. The surface scatter is very low at
P- and L-Band, but contributes "20% of the C-Band then a distance measure
class i. where
d (i) is calculated for each
returns. This may be caused by relatively thick
branches in the upper canopy which may look like
surface backicatter at that wavelength. For the lower
biomass sites and the enhanced backscatter sites, we see
the volume scatter percentage drop for all three
d, ( = * (C, ' X + In I" ln[Pa(i)1 (28)
frequencies, while surface and/or double-bounce Cj is the covariance matrix calculated for the ith class
components become significant. The bare soil results, from the training set and Pa(i) is the a priori
for example, indicate that surface scattering is probability that the pixel belongs in class i. The pixel
dominant at all three frequencies, while for reeds, the under classification is classified as a member of class i
double-bounce term is dominant, if
Supervised Class fication r < d, X for all j
(-' * i (29)
for
Polarimetric SAR data can be a very effective tool
mapping different surface cover types. Several
techniques are available in the literature for
This classification scheme is a maximum likelihood
scheme for polarimetric radar data.
I
5-12
Another type of classifier is known as the minimum 181 Sullivan, R. et al, Polarimetric XIL/C-band SAR,
distance classifier. For polarimetric SAR data, a Proc. IEEE National Radar Conf., Ann Arbor. MI,
minimum distance classifier was implemented by van pp. 9-14, 1988
Zyl and Burnett@. For this type of classifier the distance
measure used is: 191 Carver, K.. et al, Shuttle Imaging Radar-C Science
Plan, JPL Publication '86-29, 1986.
(101 Zebker, H. A. and van Zyl, J. J.. Imaging Radar
Polarimetry: A Review, Proc. IEEE, vol. 79. No. 11.
where November 1991, pp. 1583-1606.
[Il 1 Ulaby, F.T. and Elachi, C. (ed.), Radar Polarimetry
for Geoscience Applications, Artech House. 1990.
S,, SI:
(121 BoernerW. M.. et al (eds.), Inverse Methods in
- S Electromagnetic Scattering, Hingham. MA:
S SZJ (31) Reidel. 1985.
SS[131 van Zyl, J. J., Zebker, H. A. and Elachi. C.. Imaging
S& Sf; radar polarization signatures: theory and
observation, Radio Sci.. vol. 22. pp. 529-543. 1987.
[141 Dubois. P.C. and Norikane, L., Data volume
i.e. a subset of the covariance matrix elements and the reduction for imaging radar polarimetry, Proc.
decision rule is that the scatterers which are IGARSS '87, Ann Arbor. MI. pp.691- 696, 1987.
represented by Y are a member of the class i ir [15] Freeman, A., Shen. Y. and Werner, C.L,,
Polarimetric SAR Calibration Experiment Using
d, nY< d nY-forallj *i (32) Active Radar Calibrators, IEEE Trans. on
Geoscience and Remote Sensing, Vol. GE-28. No. 2.
March 1990.
Acknowledament 1161 Freeman. A., Werner, C. and Shen. Y., Calibration
Part of the work described in these notes was carried of Multipolarization Imaging Radar. Proc.
out by the Jet Propulsion Laboratory. California IGARSS'88, pp. 335-339. 1988.
Institute of Technology, under a contract with the
National Aeronautics and Space Administration. 1171 Freeman, A., Shen. Y., van Zyl. J.J. and Klein. J.D..
Calibration of NASA/JPL DC-8 SAR data. Proc.
IGARSS '9 1, Espoo, Finland, June 199 1.
Bibliography 1181 Freeman. A.. SAR Calibration: An Overview.
submitted to IEEE Trans. on Geoscience and
IlI Zebker, HA.. van Zyl, J.J. and Held, D.N.. Imaging Remote Sensing. October 1991.
Radar Polarimetry From Wave Synthesis. J.
Geophys. Research, Vol. 92. pp 683-701. 1987. 1191 Freeman, A.. A New System Model for Radar
Polarimeters, IEEE Trans. on Geoscience and
[21 van Zyl, J.J., Papas. C.H. and Elachi, C., On the Remote Sensing, Vol. 29, No. 5, pp. 761-767,
optimum polarizations of incoherently reflected September 1991.
waves, IEEE Trans. on Antennas and Propagation,
vol. AP-35, pp. 8 18-825, 1987. [201 Livingston, P.S. and Kaplan, D.J., Radar
Calibration Procedure, Technical Note 1432,
131 Borgeaud, M.. Shin, R.T. and Kong, J.A.. Naval Ocean Systems Center, San Diego, CA,
Theoretical models for polarimetric radar clutter. November 1985.
J. Electomagnetic Waves and Applications, vol. 1.
pp. 67-86, 1987. 1211 Barnes, R.M.. Antenna Polarization Calibration
Using In-Scene Reflectors, Lincoln Laboratory
[41 Boerner, W. M., Foo, B.-Y. and Eom, H. J.. Technical Report TT-65. Sept. 1986.
Interpretation of polarimetric Copolarization
phase term in the radar images obtained by the [221 Woods. M.A., A. Calibration Procedure For A
JPL airborne L-band SAR system, IEEE Trans. on Coherent Scattering Matrix Radar. Royal Signals
Geoscience and Remote Sensing. vol. GE-25, pp. and Radar Establishment. Memorandum 3889.
77-82. 1987. May 1986.
15[ Ulaby. F.T., Held, D.N., Dobson. M.C.. McDonald, K. (231 Kennaugh, E.M., Polarization Properties of
C. and Senior, T. B. A., Relating Polarization phase Corner Reflectors With Modified Walls, Ohio State
Difference of SAR signals to scene properties. University, Report 612-6. January 1957.
IEEE Trans. on Geoscience and Remote Sensing,
vol. GE-25, pp. 83-92. 1987. 1241 Kennaugh. E.M. and Chang, C., Design and
Theoretical Performance of Reactive-Wall
161 Evans, D. L., Farr, T. G., van Zyl, J. J. and Zebker, Corner Reflectors, Ohio State University
H. A., Imaging radar polarimetry: analysis tools Technical Report NOW 64-0215-D January 1965.
and applications. IEEE Trans. on Geoscience and
Remote Sensing. vol. GE-26, pp. 774-789, 1988. (251 Yueh, H.A., Kong, J.A.. Barnes, R.M., and Shin,
R.T., Calibration of Polarimetric Radars Using In-
(71 Held, D.N., at al, The NASA/JPL multifrequency, Scene Reflectors, J. of Electromagnetic Waves &
multipolarizatlon airborne SAR system, Proc. Applications..
IGARSS '88, Edinburgh, Scotland, pp. 345-349.
1988.
I.
5-13
[261 Whitt, M.W., Ulaby. F.T., Polatin. P. and Liepa, (341 Zebker. H.A., and Lou, Y.. Phase Calibration of
V.V.. A General Polarimetric Radar Calibration Imaging Radar Polarimetric Stokes Matrices. IEEE
Technique, IEEE Trans. on Antennas and Trans. on Geoscience and Remote Sensing, Vol.
Propagation, Vol. AP-39, No.1, pp. 62-67, January GE-28, No. 2, pp. 246-252, March 1990.
1991.
[351 Sheen, D.R., Freeman. A. and Kasischke, E.S..
[271 van Zyl, J.J., Calibration of Polarimetric Radar Phase Calibration of Polarimetric Radar Images,
Images Using Only Image Parameters and IEEE Trans. On Geoscience and Remote Sensing,
Trihedral Corner Reflector Responses, IEEE Vol. GE-27, pp. 719-731, Nov. 1989.
Trans. on Geoscience and Remote Sensing. Vol.
GE-28, No.3. pp. 337-348, May 1990. [36) loannidids, G. A. and Hammers. D. E.. Optimum
Polarizations for Target Discrimination in
[281 Klein, J.D.. Calibration of Complex Clutter, IEEE Trans. on Antennas and
Quadpolarization SAR Images Using Backscatter Propagation, Vol. AP-27. pp. 357-363. 1979.
Correlations, submitted to IEEE Trans. on
Aerospace and Electronic Systems, 1989. [371 Dubois, P. C. and van Zyl, J. J., Polarization
filtering of SAR data, Proc. IGARSS '89, Vol. 3. pp.
1291 Freeman, A., van Zyl. J.j., Klein. J.D.. Zebker. H.A. 1816-1819. 1989.
and Shen, Y., Calibration of Stokes and scattering
matrix format polarimetric SAR data, accepted [381 Swartz, A. A., Yueh, A. H., Kong, J. A.. Novak. L. M.
for publication in IEEE Trans. on Geoscience and and Shin. R. T., Optimal polarizations for
Remote Sensing. 1991. achieving maximum contrast in radar images.
Journal of Geophysical Research, Vol. 93, No. B12,
[301 Klein. J.D. and Freeman. A., Quadpolarization SAR pp. 15252-15260. 1988.
Calibration Using Target Reciprocity. Journal of
Electromagnetic Waves & Applications, Vol. 5. No. [391 Durden. S.. et al. Multi-frequency polarimetric
7. pp. 735-751. 1991. radar observations of a tropical rain forest.
submitted to IEEE Trans. On Geoscience and
[311 Sarabandi. K. and Ulaby, F.T.. and Tassoudji, M.A.. Remote Sensing, April 1992.
Calibration of Polarimetric Radar Systems With
Good Polarization Isolation, IEEE Trans. on [401 van Zyl, J. J., Unsupervised classification of
Geoscience and Remote Sensing, Vol. 28, pp. 337- scattering behavior using radar polarimetry
348. May 1990. data. IEEE Trans. On Geoscience and Remote
Sensing. vol. 27. pp. 36-45, 1989.
[321 Whitt, M. and Ulaby. F.T.. A polarimetric
calibration technique with insensivity to target [411 Kong. J. A.. Swartz. A. A.. Yueh, H. A.. Novak. L. M.
orientation, Proc. IGARSS '90, Washington, D.C.. and Shin. R. T., Identification of Terrain Cover
pp. 1089-1092, May 1990. using the Optimum Polarimetric Classifier.
Journal of Electromagnetic Waves and
[331 Sarabandi, K.and Ulaby, F.T., A Convenient Applications. Vol. 2. pp. 171-294. 1988.
Technique For Polarimetric Calibration of
Single-Antenna Radar Systems, IEEE Trans. on
Geoscience and Remote Sensing. Vol. GE-28, No. 6.
pp. 1022-1033. November 1990,
4)
6-I
RADIOMETRIC CALIBRATION OF SAR SYSTEMS

by
H. Ottl
Deutsche Forschungsanstalt fUr Luft- und Raumfahrt e.V.
Institut flr Hochfrequenztechnik
Germany
1. SUMMARY the dynamic range of radar cross-sections (RCS).

Flight passes were also performed with DLR's air-
Most SAR image interpretation performed in recent borne SAR.
years was based on data which were often insuf-
ficiently calibrated. Ground truth data were used Results concerning the transfer function between
for comparison and interpretation, image amplitude and radar backscattering coeffi-
cient a' of DLR's airborne C-band SAR were publi-
The importance of calibration was recognized by shed in [7] and, with NASA/JPL's airborne SAR per-
the need for reproducible data, by the introduc- formance, in [8]. The latter reference describes
tion of multifrequency and multipolarization sy- the cross-calibration between the 2 systems as
stems (interchannel calibration) and the long term well as the polarimetric calibration of the JPL
scope of remote sensing. system.
Hydrologists, especially, requested an absolute Th'nmain emphasis of the lecture will be on the
calibration with tolerances of less than 1 dB. In- spaceborne sensors, because an angle dependence of
ternal calibration schemes, as well as the use of a' has a strong impact on image evaluation due to
external passive and active calibrators, were in- the large incidence angle range coverage of air-
troduced to achieve this goal over the wide dyna- borne systems.
mic range. The in-flight measurement of the anten-
na pattern by means of ground based receivers be- 3. RADIOMETRIC CORRECTION OF NEAR RANGE/FAR RANGE
came increasingly important due to pattern changes DIFFERENCES AND OF ANTENNA PATTERN INFLUENCE
caused by electronic beam steering and the neces-
sity of beam alignment in case of multifrequency Spaceborne SAR sensors usually have a half power
and/or polarimetric operation modes. beamwidth (HPBW) of a few degrees in elevation.
Therefore, the influence of incidence angle chan-
The use of radiometric corrections to compensate ges on 00 can, in many cases, be neglected if the
for near range/far range differences caused by an- instantaneous field of view (IFOV) covers flat
tenna pattern and geometry of illumination will be terrain. The range dependence cannot be neglected,
explained in the lecture. because it affects the signal-to-noise ratio S. by
a power of 3.
The impact of geocoding on radiometric levels will
also be mentioned, including the phenomenon of A similar statement can be made for the antenna
over-/underexposing hilly regions, caused by inci- gain directional pattern G(6), since the signal-
dence angle changes. to-noise ratio increases with the square of G(b),
as can be seen in the simplified SAR-equation be-
2. INTRODUCTION low:
2
For the Radarlab missions, SIR-C/X-SAR Science
Plans [1, 2] were published which described expe- S- '16) K , (1)
riment goals within the various disciplines (such R3 (6) sin6
as geology, hydrology, glaciology, vegetation, Pav 13 ahs rR
technology). At that time, it was not possible to K =
guarantee SAR system performance as well as was 20W kTo F u as
requested by the scientific user community (se- 2 T
lected principal investigators and co-investiga- K is assumed to be constant, because the average
tors). Critical parameters were instrument stabi- p is wvelent zimu
t, te acerage
lity and relative and absolute radiometric cali- power Pay, wavelength , azimuth taper factor a,,
bration. Consequently, a number of selected expe- (of the synthetic aperture), slant range resoluti-
field of technology
in the calibration. are dedicated on Boltzmann's
rR , figure constant k, temperature T.,
riments
to radiometric The engineers des- noise F, platform velocity u and pulse ta-
igning and manufacturing the SAR sensors were re- per factor an do not change with respect to the
quested to provide internal calibration loops. off-nadir angle 0.
Means for X-SAR internal calibration are described
in [3]. Similar loops
• were integrated in the ERS-1 In orderassuming
swath, to keep atS() seeming
the same timeconstant across
that 00(0) is the
[4]. constant across the swath, the following correc-
4
An overview of the efforts and plans of f ve years tion function CF is required:
ago was published in [5], and one year ago a
workshop was dedicated to SAR calibration [6]. CF = R3 () sino (2)
In 1989, the flights of NASA/JPL's airborne SAR

over various European test sites were used in a This correction causes a relatively accurate
calibration campaign. For this purpose, 42 trihe- amplitude distribution within a SAR image (across
dral reflectors, 4 dihedral reflectors, as well as the swath) but, of course, it cannot change the
a receiver prototype and an ARC prototype, were real signal-to-noise ratio.
positioned on grassland and concrete surfaces wi-
thin DLR's Oberpfaffenhofen test site. The corner As an example for deriving CF, the X-SAR elevation
reflectors were located in special configurations, pattern is used to illuminate a "flat Earth* under _____
I
taking into account viewing angles, spacing and 40" off-nadii (Fig. 1).
6-2
Antenna
i4
, Nadir
SI
0I range
to PB. swath width

Fig. 1 Illumination geometry; swath width corresponds
X-SAR RADIOMETRIC CORRECTION

10
6-
0
-3.0 -1.5 0.0 1.5 3.0
off-boresight angle [dog]
Fig. 2 Correction function CF across the swath. Near range is on the left, far range on the
right; boresight angle of antenna is 00.
The computed correction function CF across the 4. IFLUZXM OrU3ORZtIGT MRS DUR TO U
swath within the HPBW is shown in Fig. 2. The ATTITUDE ZRO IN AOLL DIRECTIN
asy metric shape is caused by the illumination
geometry. For known antenna pointing and given (or measured)
pattern shape inelevation, it ispossible to ap-
m ODA
~6-3
I
-10-
c-20
-30
-40
-10 -5 0 5 10
off-boresight angle [deg]
Fig. 3 Predetermined X-SAR elevation pattern.
2 I I I I I I I I I
11
L.%
-2,
-5 4 3 2 1 1 2 3 4 5
off-boresight angle [deg]
Fig. 4 Gain error function for various boresight angle errors; solid line ± 0.3*, and dashed-
dotted line ± 0.40.
ply equation (2)to correct the image amplitude. 5. ANTRNNA PATTERN Z4ZSUENZNT
In case the pointing error is unknown, a radiome- As mentioned in part 3 of this paper, the antenna
tric error will be produced which, in most cases, pattern must be known for precise radiometric cor-
cannot be corrected. rection. In view of strong mechanical stress du-
r For instance, the predetermined X-SAR elevation

pattern (Fig. 3) is used to compute a gain error
ring launch, possible deployment errors and ther-
mnal effects, the actual inf light antenna pattern
should be used to achieve the highest quality res-
function (Fig. 4) for various boresight uncertain- ults. In our institute, we use a large number of
ties (0.1* to 0.4* angle deviations). An HPW of high precision calibration receivers, aligned in
about ± 3 (inFig. 4) is relevant for the swath the cross-track direction (see Fig. 5) to measure
width of X-SAR. the azimuth cuts of the actual antenna pattern.
Every receiver digitizes and stores each received
The gain error function has to be squared to show radar pulse as 16 samples. In parallel UTC is
its impact on image amplitudes (grey levels of SAR co-registered with high precision (10 ), so that
fne image).
the prpecise
the measured azimuth-cuts can later be time-cor-
related and the required elevation main cut of the
antenna pattern can be reconstructed. Of course,
location of the SAR-sensor vs. time
must be known to convert the time of pulse receipt
6-4
4 !
-Elevati-
Schematic m
Fig. 5: Schematic experiment set-up.
160 1 1 V .
120-
00 80- H H
0
40-
0 15 0 15
SAMPLE NUMBER
Fig. 6: Registered pulseshapes from NASA/JPL's C-band RHDC-8 SAR.
into antenna angles. It is also mandatory to pre- The relatively large deviations of the measured
cisely know the attitude data of the platform as points from the fitted curve in Fig. 7 can be ex-
well as the location of each receiver. plained by 3 facts:
Inthe 1991 NASA/JPL airborne SAR-campaign over 1)precise flight and attitude data have not yet
our test site, we had a first opportunity to test been received from,
ourr m2)
measuring system (9].
the positions of the receivers were only known
Fig. 6 shows registered radar pulses from the to about t 100m,
C-band SAt of NASA/JPL's DC-8 (horizontal po-
larization). Slight distortions, i.e. decreasing 3) at the time of the experiment not all of our
power from begin to end of the pulse can be seen, equipment was fully calibrated.
probably due to the wide chirp used and the rela-
tively small antenna bandwidth. These drawbacks will be eliminated in our future
experiments with spaceborne SAR-sensors (all re-
An azimuth cut can be obtained, by integrating the ceivers are fully calibrated now, positions are
pulse energies andofplotting themcuts
against
givestime.
time-correlation all azimuth the The determined using differential
bit and attitude GPS andfrom
data are available precise or-
the sa-
main cut of the antenna elevation pattern (see tellite owners).
Fig. 7).
The co-registration of the time of pulse receipt
.1
6-5
-5 A
C
C
-0 _ ___
.N
ca
M
E -15
0
0 15 30 45 60 75 90
off-nadir angle [deg]
Fig. 7 Reconstructed main cut of the antenna elevation pattern.
.?l7~ -02dB 1 I 1
HPA
'"-d~.-
FGU
Fig. 8 X-SAR calibration block diagram.
can yield two sets of further information: fore and after data takes. It can be used to dedu-
ce an optimum reference function for processing.
Precise measurement of cross-band beam alignment The test signal can either be taken from the low
in azimuth and elevation direction (especially power part of the radar transmitter (HPA is swit-
suitable in the X-SAR/SIR-C mission) and possible ched off) or from the directional coupler between
antenna squint angles. the HPA and antenna. In the latter case, the low
noise amplifier (LNA) will be bypassed. These
Our system will be used in all present and near- loops are indicated in the block diagram in Fig.8.
future SAR missions (ERS-l, JERS-l, X-SAR/SIR-C,
PRIRODA). Internal calibration of active array antennae is
more complicated due to the large number of trans-
6. INTERNAL CALIBRATION mit/receive modules (T/R modules), which can easi-
ly reach several hundred to several thousand.
Spaceborne SAR systems should have life times be-
tween 3 and 6 years without too much degradation These T/R modules f the next generation SAR an-
in performance. Internal calibration loops are ad- tennae will be designed and manufactured in mono-
visable to monitor system stability. For the high lithic microwave integrated circuitry (wIC)ba-
power output, monitoring can be achieved by measu- sed on GaAs technology. Besides control and moni-
ring a small portion; the measuring point could be tor units (power, phase shifter, polarization), a
a directional coupler be..ween the high power am- time sequential check-out of the transmitter and
plifier (HPA) and the antenna. receiver stages is necessary. Different concepts
using separate feeding networks to distribute or
Receiver gain stability and linearity can be mea- collect calibration signals within the active ar-
sured by feeding a small part of the radar signal rays are presently under study at various centres.
via attenuators through the whole receiver chain.
For power leakage problems, the HPX is usually For internal calibration, the balance and orthogo-
switched off during this test phase and the signal nality of in-phase and quadrature channel (after
is taken from a low power section. demodulation) are important. Error analyses for
X-SAR have shown that 50 deviation from orthogona-
Furthermore, the actual chirp can be measured be- lity causes about 0.2 dB amplitude error and 0.2
I 1i .
6-6
40
30 _
20
transfer function
__ 10 - a comer reflectors
TRACK
2
-10 I--grass C - Band / VV
-20 rconcrete col.-factor: (18.6*0.6) dB
-30 1 m 1 , 1
0.0 100 200 300 400 500 600
image amplitude
Fig. 9 E-SAR system transfer function.
40 __________ _______I _____

I'
30
20
10 - transfer function
6 a corner reflectors
b 0-0
-10 TRACK 2
gross C - Band / VV
-20 concrete col.-factor: (6.1 *0.8) dB "
-30 1 1
0.0 25 50 75 100 125 150
image amplitude
Fig. 10 DC-8 SAR system transfer function.
dB imbalance an additional error of up to 0.2 dB. 7. EXTERNAL CALIBRATION
Another source of SAR signal contamination is the External calibration can only be usefully perfor-
total system noise. This noise is mainly composed med after having accomplished the previously des-
of contributions from the sensor itself (ADC, LNA, cribed internal calibration and radiometric cor-
ohmic losses) and the earth surface. In the case rections. The main objective in the scope of abso-
of the X-SAR system, this quantity is estimated lute radiometric calibration is the determination
before mission and measured in a receive-only mode of the overall SAR system transfer function rela-
during mission. On the raw data stage a noise ting image amplitudes Pi to a'-values, as expres-
subtraction takes place by using the resulting sed in the following equation:
noise data.
a' = 10.lg(P i ) - 10-lg(K') - 10.1g(rA rm) . (3)
6-7
40
30^
A
A
E
20
b A E-SAR (28.9±1.1) d8m'
b
10 • DC-8 (29.0±0.4) dBm'

theoretical 29.3 d8m'
0 .0 I I I I I I I a i
1 2 3 4 5 6 7 8 9 0 1 12 13
Number of Cornerreflector
Fig. 11 Amplitude accordance of 13 equally sized corner reflectors.
Pi is the image power in the case of distributed The2 mean values determined were 28.9 dBm2 and 29.0
targets and the impulse response minus an averaged dBm , which are in good mutual agreement and also
clutter (noise) level for point targets; rA deno- in good agreement with the theoretical value of
tes the azimuth resolution. K' is slightly diffe- 29.3 dBm2 .
rent to the factor K of equation (1) because of
the inclusion of the processing gain which is ach- Fully polarimetric SAR sensors, such as the DC-8
ieved by forming the
wit synthetic
refeenc taget aperture. hae
knwn CS-alue External
t be Ful
SAR anda thereSIR-C system,
SAs enr, suchoas require
additionally the t thee
reference
used targets
in order to determine
with knowntheRCS-values
calibration
havefactor
to be apiaino oaiercclbainpoeu
application of polarimetric calibration procedu-
K' nCorner
reflectors and active radar calibrators res. Such procedures imply the elimination of the
. Csensor induced distortions cross-talk and channel
(ARC) are appropriate tools to fulfil this task imbalance. This can be accomplished by utilizing
for the central and upper parts of the system either different types of corner reflectors or
transfer function,
ment whereas fin esecilly
situ RCS measure-
wih sctteometrs eleted eiter
clutterdi ticstogether
statistics of rer trihedral
with rel c corner
or
ments with scatterometers of especially selected reflectors [10).
area-extended targets are helpful to determine the
lower part of the system transfer function. S. GEOCODING
As an example, results of the 1989 Oberpfaffenh- Due to the sidelooking illumination geometry and
ofen airborne SAR calibration campaign are presen- the "azimuth-range-imaging" of SAR, mountains seem
ted in Fig. 9 and 10. These figures show the
C-band W overall system transfer functions of to bow toward the sensor. This phenomenon is known
DLR's experimental SAR (E-SAR) and NASA/JPL's DC-8 as foreshortening and, in the worst case, as lay-
SAR. Both system transfer functions represent best over. The latter happens in cases when mountain
fits of the corner reflector responses as marked peaks are less distant from the sensor than the
by triangles in the two figures. The calibration valley in front of them.
factors
an for better
accuracy both systems
than 1could be determined
dB, showing a fairlywith In Fig. 12 the foreshortening can easily be seen
good calibration performance for airborne SARs.The by comparing the projected slope area in front of
oo correspondence due to the SAR systems involved the mountain Sf, which appears in the SAR image as
has been determined to be within the 1 d error (R2 - Rl), and the projected slope area behind the
hasbee deermnedto
e wthi th I B eror mountain Sb, which appears in the image as (R3-
bounds for two typical earth surface object clas- mUn
ses: grass land and concrete. R2 ).
Cross-calibration becomes more and more important The power density relationship Sr between (R2 -
in order to keep the effort and expenditure on R,) and,(R 3 - R2 ) can be approximated as follows
multichannel and multisensor calibration within
tolerable limits. This is especially true in view Sf R3 -R2 (R1 +R2 )(R3 -R2 ) (D2-bi)
of the upcoming spaceborne missions, such as R2 -R1 b (R3 +R2)(R2 -R1 )(63 - 2 )
X-SAR/SIR-C and the follow-on activities. A measu-
re of cross-calibration quality is, for instance, The second and third factors in the numerator are
the RCS accordance of equally sized corner reflec- larger than in the denominator (see Fig. 12) and,
tors derived from image data from two different therefore, the power density ratio is >> 1.
SAR systems. Such an example of thirteen 0.9 m
corner reflector responses concerning the C-band This can be seen in SAR images in the form of
VV mode of the E-SAR and DC-8 SAR is illustrated bright slopes in front of mountains.
in Fig. 11 together with the theoretical RCS va-
lue. Geocoding means precise rectification. For this
6-8
Antenna S1 = projected slope area In

.01 front of mountain
0, Sb = projected slope area

behind the mountain
2
H
Fig. 12 Illumination geometry for a mountain.
OLR/0-PAF E R5 -1.5A :0GTC
;T!,
~
Fig.~~~ ~~ ~ 13GooedMtehonsee
6-9
purpose, about a dozen control points from maps

(scale 1:50.000) will be used (for instance for
ERS-l images).
No additional radiometric correction will be ap-

plied (see Fig. 13). A very rough correction would
be the inverse application of equation (4), assu-
ming an isotropic scattering behaviour of the slo-
pes.
This assumption is not valid, especially not in

the presence of vegetational layers. A tool to in-
clude radiomeric correction according to the inci-
dence angle could be the incidence angle dependent
o. But again this implies some a priori knowledge
about the ground consistency.
REFERENCES
[1] NASA Shuttle Imaging Radar-C Science

Plan.
JPL Publication 86-29, Sept. 1,
1986.
[2] Ottl, H. The X-SAR Science Plan.

Valdoni, F. DFVLR-Mitt. 85-17.
et al.
[3] Seifert, P. Preparatory Investigations Con-

Bl6tscher, H. cerning the Calibration of
Spaceborne SAR-Systems.
Proc. IGARSS '91, Helsinki,
3-6 June, 1991.
[4] Menard, Y. Design and Performance Assess-

Oudart, P. ment of an Ultra Stable Cali-
bration Subsystem for a SAR and
a Scatterometer.
Proc. IGARSS '87, Ann Arbor,
18-21 May 1987.
[5] Hartl, Ph. Radar Calibration Techniques

Heel, F. Including Propagation Effects.
Keydel, W. Adv. Space Res. Vol. 7, No. 11,
Kietzmann, H. pp. (11)259-(11)268, 1987.
[6] Heel, F. Proc. of the SAR Calibration.

Freeman, A. Workshop, DLR, Oct. 9-11, 1991.
(Organizers)
[7] Heel, F. Radiometric Calibration of an

Ottl, H. Airborne C-band Synthetic Aper-
Zink, M. ture Radar.
Proc. of The 17th Internat.
Symp. on Space Technology and
Science, Tokyo, 1990.
[8] Zink, M. The Oberpfaffenhofen SAR Cali-

Heel, F. bration Experiment of 1989.
Kietzmann, H. Journal of Electromagn. Waves
and Appl., Vol. 5, No. 9, 1991,
pp. 935-951.
[9] Seifert, P. Ground-based Measurements of

Lentz, H. Inflight Antenna Patterns for
Zink, M. Imaging Radar Systems.
Heel, F. To be published in IEEE Trans.
on Geoscience and Remote Sen-
sing, 1992.
[10] Zink, M. Comparative Investigations of

Heel, F. Polarimetric Calibration Me-
Kietzmann, H. thods.
Progress in Electromagnetics
Research Symp., PIERS '91,
Cambridge (MA., USA), July
1991.
7-1
SAR SIMDUATION
by
D. Hounam
Deutsche Forschungsanstalt fdr Luft- und Raumfahrt e.V.
Germany
1. SUMMARY The first challenge for the SAR engineer is the

system design, which requires juggling with the
The use of software tools as an investigative me- system parameters until the required performance
thod is particularly important in the case of syn- is met. This can be described as a 'bottom up' ty-
thetic aperture radar sensors, as the geometry pe of analysis, i.e. starting with a set of per-
cannot be reproduced in the laboratory. Also, the formance requirements and external constraints,
complete SAR system, from the target via the pro- the engineer attempts to find a set of design pa-
pagation path, sensor and image processor to the rameters to meet his gol. This procedure is often
final image, represents a highly complex data called parametric analysis and culminates in a set
chain, which cannot be treated in part. of performance specifications.
The lecture discusses different approaches from The second task arises when the system design has
parametric analysis tools to full-blown simulators firmed up and the engineer needs to monitor whet-
capable of analysing all elements of the SAR sy- her the performance is being met. This is the top
stem. The latter will be illustrated with the aid down approach, deriving the performance characte-
of the SARSIM simulator, which was used for con- ristics from the estimated or measured sensor pa-
firming parameters of the ERS-l Active Microwave rameters.
Instrument (AMI).
The performance of a SAR sensor is described in
Particular emphasis will be placed on the simula- terms of imaging parameters [1]. As a SAR image
tion of target scattering mechanisms, the under- only materializes after considerable data proces-
standing of which is essential if the potential of sing, the SAR processor characteristics have also
SAR systems is to be fully exploited, and on the to be considered in the analysis. Clearly, some-
modelling of sensor characteristics, where the engineer has got to put stakes in the
ground or his design will never converge. It is
The author would like to thank T. Pike and S. Pot- probably fortunate that some constraints exist
ter for the use of material and K.-H.Zeller for such as physical dimensions, platform flight path
useful discussions. (orbit) available power etc., limiting the degrees
of freedom. Nevertheless, it will generally be ne-
2. INTRODUCTION cessary to freeze those parameters which are not
part of the design task. For example, the SAR pro-
The term SAR simulation is used for a wide range cessor characteristics will be fixed when de-
of software tools to help design and evaluate SAR signing the SAR sensor and vice versa.
systems. Clearly, such an approach can be much mo-
re economical than performing experiments, bearing 3.1 Parametric Analysis
in mind the difficulty, and in the case of satel-
lite sensors, impossibility of achieving a reali- Parameric analysis software can be regarded as a
stic geometry. suite of software utilities to aid in the design
of the SAR system. Because of the large numbers of
In this lecture, we will differentiate between choices open to the SAR engineer, frequent inter-
analytic tools and simulators. The former term action with the user is necessary. There is no
encompasses all purely algorithm based programs, hard and fast rule as to where one starts in des-
whereas the simulators attempt to mimic as many igning the system, but the timing scheme, i.e. the
parts of the SAR system as possible. choice of pulse repetition frequencies (PRFs) and
the position and duration of the received echo
In general, an analytic approach to solving a need to be defined at an early stage.
problem is preferable to simulation, because it
leads to a more complete understanding of the 3.1.1 Timing
problem. However, sometimes problems arise which
are not accessible to analysis. Experiment or si- The minimum PRF is determined by the width of
mulation are then the only alternatives. Simula- the received Doppler spectrum, it needing to be
tion, like experimentation can also usefully higher to satisfy the Nyquist sampling theory.
augment analysis and provide valuable confirmation The steps to arrive at the PRF are as follows:
that the understanding of the mechanisms within
the system is correct. * The width D (azimuth dimension) of the antenna
is derived from the required azimuth resolution
It is clearly out of scope of one lecture to dis- Remember the rule of thumb that the single look
cuss the algorithms of such tools in detail and so azimuth resolution is half the antenna width.
the emphasis will be placed on architecture and
performance. * The antenna beamwidth can be calculated from the
physical width:
3. ANALYTIC TOOLS
When confronted with analysing a SAR system, one = D (1)

is faced with a multitude of parameters many of k is the broadening factor due to weighting of
which are strongly interdependent and all of which the antenna aperture and X is the wavelength. k
impact on performance and sensor design, is 0.886 for an unweighted aperture, which can
be shown to be optimum, if the PRF is chosen to
just meet the sampling requirement (see below).
.....
......
7-2
2400
N.Rank: 9
qm. PRF: 1700 0

Geometry: Orbit
2000
1400
0 5 10 15 '20 25 30 is 40 45 so 50
*[eq]
.0 143.0 285.5 427.3 571.3 712 1 16 2 067.9114t'185,0
G o u d ran 9e [ ki ]
Fig. 1 Timing for the ERS-l satellite derived by the Cross-track Geometry Program. The wanted echo
window is identified by the vertical lines.
* Now the width of the received Doppler spectrum The bands are plotted on a graph of PRF versus
BD can be calculated: off-nadir angle (angle between the target and
the vertical) and ground distance (distance be-
2V e5 tween the target and the ground track). The axes
BD = X , (2) can also be swapped around. For the wanted swath,
the user can select the PRF, rank (number of
where V is the velocity of the sensor platform. transmit pulses after which the echo is received)
and the maximum and minimum off-nadir angles, i.e.
a To satisfy Nyquist, the PRF needs to exceed the the limits of the antenna elevation pattern. The
bandwidth: program calculates the position of the unwanted
ambiguous echoes enabling the requirements on the
PR > 1.2 B, (3 elevation pattern to be derived.
The above steps are simple approaches which can 3.1.2 Link Budget
require more sophistication, depending on the case Once the timing has been defined, the geometry is
in hand. In effect, a software utility needs to be known and the link budget can be computed accor-
developed to handle all aspects. For example, the ding to the radar equation. This is a straight
last step, where adequate sampling is considered, forward task and doesn't require much software
ensures that ambiguous responses in azimuth, due
to aliasing unwanted energy into the wanted Dopp- sophistication.
ler spectral band, can also be influenced by the The following form of the radar equation can easi-
characteristics of the azimuth compression routine ly be realised with a spreadsheet. Table 1 gives
in the SAR processor. However, the design of many an example with the values for the ERS-l satelli-
SARs has started down this path. te.
The timing of the received echoes requires consi- The radar equation for a single look in the final
deration of the cross-track geometry and the pulse
characteristics of the radar. The art is to find image can be written:
space between successive transmit pulses for the 2 -r X 2
=
echo with enough margin for the receiver to settle S/NLg. LNoJEc,
CP
[C±tcpPRFJ]
(4M
"
back to full performance. Another constraint is 0
that the echo from immediately beneath the sensor [A] [B] [C] [D]
(nadir) should also not fall close to the echo - r o r 1
window. Having an incidence angle of 900 it can be Llsani2-
strong enough to overload the receiver. Particu- [E] [F] [G] [H]
larly, for satellite geometries, many pulses may where,
be underway before the echo is received (for ERS-I
9) and just as many nadir echoes exist. These am-
biguous echoes interfere with one another and have G = transmit power, g
to be suppressed by the antenna elevation pattern. N. = receiver noise density,
An example of a software utility for analysing the TP = transmit pulse length,
timing is the Cross-track Geometry Program develo- tcp = compressed pulse length,
ped by the DLR. Fig. 1 shows a typical example for Tj = azimuth integration time,
the .tS-1 satellite. The program computes the va- PRF - pulse repetition frequency,
lid bands where the echo is corrupted by neither X = wavelength,
the preceeding or succeeding transmit pulses, nor r = slant range,
the nadir echo. Forbidden areas are shown hatched. Co - velocity of light,
7-3
- IMPULSE RESPONSE AZ I 1l0 Tl
0 Number of Steps t00 000

-5 - r.ot Disp.Res.colls: 22.000
Look flandoidlthHz: 40.000
Target Velocity m/s: 75.000
-1 5 - -Haming Factor- 0.54
-10 -
-AMPLITUDE ERRORS
-20 - - -- - -Linear do: 0.00

Quadratic do. 0.00
* ~-25 - - --- - - -Ripple cycles: t.58
z 3 -- Ripple Amplitude do: 0.00
-35 -- - PHASE ERRORS
- -0 -- Linbar deg- : 0.00

-40 uadrat i c deg. : 0.00
-45 A__ Ripple cycles* 0.00
Ripp:: Ampi. d::. : 5.72
-5( Ripp9e Phase do.
j
0.00
55 ->1SLR too sol1

LL60L Spaooal Res. (m) :2 -45 14
-20 -15 -10 -5 0 5 10 15 20 P S R (dB):5-3562
Azmu h[MJ I S R (dB)21.0206
*Fig. 2 Azimuth impulse response computed with IMPRES.

i = incidence angle,
V, = satellite velocity,
go = backscatter coefficient, 2578 ecBmk E5-4
LATH = atmospheric loss,
Lp= processing loss.w kwsu At.gin B 714
om1h±Ias Pea mr 8MI 36.8
The terms have been grouped to separate the va- 21.ka 4(3 .4
rious influences. Mitel ag C
.5
orRm dkuty -1518.484VfrJ 198.5 337.7( 0S
[A] = RE sensor parameters, called 'nomi- 01 Fe amp Ikpk w 71i
nal gain' for.ERS-1, g ~ PA-
.Mp~ Brm 71.9 Z7.6 cB
rB] = range compression gain, ImkoraA kn tikm flm -9.6
[C] = noise bandwidth, a
0-vApi6m ra -71.9
(D] = free space loss, EF imIt :.1 -0 C
[E] = range resolution,M nsvc Wkrt OM -A9
[F] = azimuth resolution, 10 bibx O.b -MM4.9
[G] = target backscatter coefficient, W-ft a* D- 3
[H) = losses. amtt A1 -36. 1.9
The radiometric resolution y is computed from the ~ 1 po.
jmn*ta
03p-~PA
rclii &,
GM
23.M di
-71.9
4.5
expression: amlz 113140 %.8 14.4 c3
1 F1 Acsth Slat MoP 8363347I 592
r +SIN -±as MU~ffqt .056as -12.5
yO0 I1
log
.e (5I~~t. S~ .
I Stra alfy MM is -38.6
where L is the effective number of looks. 99A 1)F~kM -1 CB -18.0 -2 dB
asmcet 0 awlol -M.0 clo
3.1.3 Impulse Response (3 d
0(3im
The shape of the impulse response of a SAR sensor smo 0... 1

is dependent on the used weighting functions and D-T_ I l 0.13 cS -0.1
the phase and amplitude errors. The latter can be Fwmk 1.5 cR -15 -1.6 dB
due to sensor imperfections and, in the case of
the azimuth response, also due to errors in the ____________I~ 0.7 cE
knowledge of the flight path. The DLR program IMP- Motim Raw o IMN 6 2.45cB
RES (2]derives the range impulse response from
the chirp characteristic, the range weighting
function and amplitude and phase errors. With the Table 1 Spreadsheet of the link budget using the
azimuth response, instead of the chirp, the Dopp- example of the ERS-l satellite.
ler characteristics determine the resolution. The I
weighting function and errors can either be ent- grated sidelobe ratio.
ered as formulae or from look-up tables, Fig. 2
shows an example of the azimuth response showing 3.1.4 Porforuance tatiuatiom
panedechoes. Fig. 3 shows an impressive two-dimen-
sional representation, which however, can be dif- The above tools are designed to be quick and easy
ficult to interpret, to use and run fast enough to be able to iterate
the design. They can, therefore, be called upon to
IMPRES also evaluates the impulse response and perform both the top-down and bottom-up design
computes resolution, peak sidelobe ratio and inte-
7-4
IMPULSE RESPONSE.2 . leO N

*Iu I0l,W'I2 l
0llll llill
A"1#1]6 no
M2s
. A A.F.... 6.4
0 ., *0o2 .o4i.1 69 Ol
-A 00l
ilpell.3
.Pp ob.
l I
0.l 24I A0
'ZIi-20 I
liep221..
l
P... 44
,Id 25 0
I 464
11:11 II1'$
metpon -s20 t u0 l Sep Pabove.
-54
-1 p2
1 ii f .l 4Ioi 300
+3iii
Fig.llIimus 3ll~reposeTw-imnioa
these have the advantage over comprehensive para- a

metric analysis packages that they are easier to PP
master. The DLR Performz-ce Estimator (3] software
is such a package which uses more sophisticated
versions of the above utilities.
The Performance Estimator was originally developed

for the X-SAR sensor to be flown on space shuttle
together with the SIR-C sensor in 1993. It grew
out of the need to provide the operators during
the mission with a tool to ensure that the radar
is optimally adjusted before each data take. Expe- "whop
rience with the preceeding SIR-A and SIR-B missi-
ons had shown that the predicted shuttle position
and attitude was so inaccurate that relying on0""
preset parameters meant that target sites could be CM
missed. This led to an architecture for the Per-I
formance Estimator which has not only fulfilled
the requirements of shuttle missions but also of M 1
conventional SAR satellites like ERS-l and airbor-
ne SARs like the DLR E-SAR.
Fig. 4 shows a schematic of the Performance Esti-

mator. Its basic architecture follows the approach
described above. In the case of satellite plat-
forms, the geometry is calculated from the orbit
parameters and the position of the target site.
For airborne sensors or cases where the orbit pa -
rameters are not known, a fixed geometry can be
entered. The radar parameters are stored in two
$ways, differentiating between those which will Fig. 4 Schematic diagram of the Performance Esti-
eventually be fixed and those which can be changed mator.
by the operator, for instance, by means of tele-
commands from the ground. Examples of the latter whe-e different philosophies can be used, depen-
are timing parameters, gains antenna pointing, ding on the sensor.
etc. The 'fixed' parameters are stored in the so-
called systems file, whereas the parameters under The Estimator first calculates the timing parame-
operator control are calculated and ranked by the ters by running through all the available PFs to
Estimator, providing the user with a choice. see which PFs meet the requirements. Where image
Clearly, the 'fixed' parameters can be changed at quality is affected simple algorithms, as given in
any time. Typically the specifications of the ra- Section 3.1.1 are used for the bottom up analysis.
dar parameters would be used as fixed parameters The PRFs for which adequate solutions are found
in the early phase of sensor design to be replaced are ranked according to criteria, which can be se-
by true measured values later. lected by the operator. The solutions are dis-
played on the timing page (see Table 2). The ti-
The algorithms used in the Estimator are coded in ming page also lists the performance parameters
the program but in some cases a choice of algo- which are dependent on geometry and timing like
rithmns is provided. An example of this is the al- swath width and position, and spatial resolution.
gorithm for choosing the optimum gain setting,
7-5
Azimuth ambiguities occur, due to aliasing of un-

,wanted
,ZKM Doppler frequency bands into the wanted re-
VULE muO 90SO - - 3 , 5 , gion. The latter are suppressed by the antenna
NN = AN 1703.0 70 00 50 10 1 M azimuth pattern. The azimuth ambiguity ratio is
ts I." .
,.s "1.: ::.3 I 59 calculated by integrating the energy in the Dopp-
"M' 2359.3 "94 a359.3
.92. 2.2
4. .2 ler band selected by the SAR processor and the
ml
SO.T WI 11
.0 359.3
,.,I 359.3
,.7 9,.,:35.3
359.3
,, : .7 energy in the amb~iguous bnds and dividing. Apart
2.0 51. 4.5 54.0 5N.0 -.
94 M ,.0 370'.0
,.H 34.5 '1' 2S5
251:0 .5 27.5 5. ,. from azimuth antenna
the bandwidth weightingthefunction
and the pattern, processed
R.
-- Wa
RIM FaORa°29.
sunoorION
no PTRS.
379.0
25.2
274.5
25.1
270.0
23.1 2 2 3 2.1:
5.0
.
391 .5 Doppler used
Y M AT . 1.. o . 0 for azimuth compression have to be taken into ac-
.:,.o
inzAMA 949I9 sc, 494. 491.4 495.9 490.3 474. J:2 count. A large number of weighting functions can
!u -, VA TAR I= In I= be selected. Again the azimuth antenna pattern can
4 UR1 .2b. 3 .4 .9 Zb 2b 3 Z.2 be entered as an analytic function or as a look-up
SS S M , . o o o:
a' 2 .o 2.0 table.
PUA MI "M OW 7 9 9 10 11 12
O M" A 1690.0 ..o 1 1481

4*. M.0
55.
1"4,0
5 .
The distributed target ambiguity ratio is calcula-
M SIN o 591.7. 595.9
9X 909 1o M o541.0 5
0 ,44.0
40: 5,14.0 914 5.15. ted by combining range and azimuth amiguity ra
59,.025,.3 251.3 ,93 tios. Table 4 shows typical values for the ERS-l
5 5...5 5.4 satellite.
52891 99158
S59994 0 9.3 9.0 . 0.
$TM3 G?0 R9030 54.0 53.0 51.0 50.0 4..0 47:1
MMR 19 .3 ... 243. 233.5 23,. 335.0 14.5 21
NM D S 2 2SP . 1 25.3 25.1 99190591Mrt1 02.08t GAIN s
TAR =03SPAT 3U . 1o.8 14.7 19.. 18..7 . .. 19.
On= 9A8,2 910399 0 0.30 . ..0 .90 9:0 .,- .ooz. M
33.02
R9A 99I0 - 463.4 459.0 452.3 449.7 441.1 433.4
98w79.?9 289509.14' M1 Y:0 Y1 15 T VMS NOT5 T
83 1(IN 45.03
3-9 .4 210.9 353.4 M9.9t304.4 320.1 333.3 345.8
MOIR CA9 11,2.b3 Zb b lb 2 b1 b JA 8 31.. 41.9 45.0 48.5 52.1 55. 59.3 1.4 91.5
RX94In 7 0 1 9 .
35.5 35.5 29 9 25. 4.4 .4.9
C M MA RR 0c
3.0 0.3 .
3. 7.3 1 0.0 910
ni to
Table 2 Example Performance Estimator timing page
(E.RS.-25.I).0 ;Om 21.3.23.9 25.9 29.9 23.5
20. :
1.5 5.5325.20.9 2.5
29.1 92.1 2.1 39.1 21.. 29.1 25.1 29.12 5 3.5
29.1
(ERS-l)
. m
Lon
4 19.9
3.4
34.9
33.4
34.9
23. 33.9
S.
23.1
23.9
3
23.
.9
23.9
35
23.
24.9
23.9
Eo n 6 13.9 13.8 13. 8 13.9 13. 1 13.1 13.9 13.9 13.9
The next phase of the Performance Estimator is to MUMSN 21.7 21.7 21.7 21.7 21.' 21.' 21.7
2.7 21.7
calculate the imaging performance for the various TOTAL 21.4 21.' 21.' 21.' 21.'231.7 21.7 31.' 1.7
PRF solutions. Normally, the three highest ranked
PRF solutions will be analysed but the operator
can change the ranking or the number of solutions,
if he desires. Table 4 Example of the Performance Estimator ambi-
guity performance.
on three screens:
Imaging performance is displayed
signal to noise, ambiguity ratios and target de- The target detection characteristics describe the
tection. ability of a SAR sensor to detect point targets
above a distributed target background.
Using built in algorithms for the radar equation
digitisation noise, bit error noise and a variety Man-made targets generally consist of single or
of scenarios for the normalised backscatter coef- combinations of discrete scattering centres and
ficient a., the Estimator computes the signal-to- can be regarded as having deterministic proper-
noise ratios for each selected PRF solution and ties. Such scatterers, often called point targets,
nine positions across the swath. The S/N at the are usually much smaller in extent than the re-
imput to the analogue-to-digital converter, i.e. solution cell size. Although the radiometric re-
the video SIN, the S/N for each processed look and solution can be calculated for such targets, the
the multilook SIN are all calculated. From the user is more interested in the ability to detect
latter the radiometric resolution can be computed. targets above the system noise or distributed tar-
The dynamic range is the range of signal powers get background (clutter).
over which a particular requirement, e.g. radio-
metric resolution is met. The Estimator searches To detect a scattering centre or point target suc-
over a range of a0 values until the radiometric cessfully it must be brighter than the brightest
resolution goes out of specification. Table 3 point in the speckled background.
shows typical results for the ERS-l satellite.
9 M1'02. an "N .doIn principle, the radiometric resolution can be
299
'S
4909
.
M0o4
1r-5
2 1useful
M calculated for point
parameter targets
if the radar and thissection
cross can beofa the
vm 2 o,
0 I1. :. 0: -0.9 -2.0 -3.4 target needs to be measured with known precision.
25. 20. 2. 0 24 0 29.9 23. 25.2 24.9 24.3
-1.1 41.2 41'. 41.. 41.1 ,.5 :o. ., .3 The application considered here, to image man-made
um 109 .92 3 .
1<'
.
.9 . -
s.' -0.3
1. 1. .30 4
-1.3
..
-2.9
-9.3
objects consisting of one or several dominant
L ,w .0 -1. scattering centres, requires that the weakest
1LON 1
I3 l. 19 1.3 0.9 .. -1 .3 -3.79
-1.o -.
L0W 4
59l
-3"1
t,
1.3 1.!
-1.2 -'1 1.1
1.8 -11 1J 5
-3.3 -04
-2,*'-,.1 -. 3 scatterer can still be resolved above the back-
I 0
2,.3 . 2.,3.3 .5 .4 ground. If sufficient scatterers from the target
rim"
.9 04 9.0 9.4 5.5 2. 9.0 4.3
3.7
4.o 3.'
7.9 9.0
1,0 14.3 -
-9 -19.0 -15.3 -151. -4.o-13.9 -1.0 can be detected, the size and shape of the object
W8 2. 21. 2 . 38
20. . 21. 28. 20.3 20.0 can be reconstructed and, hence, the object des-
RWXugW ,I:5 ,9 99 . 19i
Oo5 310M - 2,,947 cribed.
Table 3 Example of the Performance Estimator si- In the approach described here, two parameters
gnal-to-noise performance (ERS-l). from radar detection theory are used as a measure
of the ability to detect a scatterer. The false
The ambiguity performance in range is calculated alarm rate (FAR) is the probability that the in-
from the geometry, timing and antenna elevation tensity of the background in an image pixel is
pattern. Range ambiguities occur when the slant such that it Is falsly interpreted as a point tar-
range differs by half the pulse repetition inter- get. The detection probability is the probability
val (PRI). The ambiguity ratio is computed by in- that a point target can be detected above the
tegrating the energy in the wanted region and am- background.
biguous regions and dividing. The antenna eleva-
tion pattern can be entered aa a look-up table or A scatterer is considered to be detected in an
as a weighted pattern with up to seven weighting image pixel if the intensity of that pixel exceeds
coefficients, the intensity of the surrounding pixels by an
- *... .+ . . . . . .
7-6
amount large enough to meet the required false

alarm rate and detection probability.
Ca I
To analyse the problem quantitatively, assumptions - WbrMF
t
have to be made about the object to be detected Mkd 1W
.1--
and its surroundings. In the following, three sce- d opmbomf=l
narios for the analysis have been chosen.
1) The first scenario consists of a single stab-
le point target in a background consisting of
a distributed target. The assumption is that,
although the point target, by definition, is
CM 2
T"uI"-Pa*pmhtmaghMbMdQMk
\
mA
aqm-
imihk.d,
IM
smaller than the resolution cell, only the
point target contributes to the energy in the pkeibfmee FAt
corresponding pixel. This would apply to the w PD Ms-
case where the point target is surrounded by a bo
a surface of low scattering cross section on.na
large enough to fill the resolution cell and
where the system noise is negligible compared
with the background oo. This scenario, is ty- m
pical for large metallic structures, such as cows
aircraft, where the scatterers are surrounded T =oaowf% om ,t
by large metallic surfaces reflecting the en-
ergy away from the sensor. F OwaU f FAR
2) The second scenario considers a single point ____ty POaa~

(1
target in a distributed target background
noise as
where the background, the system
well as the point target contribute to the
energy in the point target pixel. This is a \FA "we
more general case than the first scenario and
would be typical of buildings and small iso- Fig. 5 Representation of the three detection
lated objects. scenarios.
URTT CM? - 4 M0139 COUnIOLM SI0MO - -21,50
3) In both the above cases it is assumed that ? C?0 ,ZTO
,O,? P-2ELITIS M
the target cross section remains constant. In
reality, it will vary according to the aspect Mm
angle (glint). For such a variation, a Rice SOTf. 247.5 239.9 212.4 204.0 297.2 309.7 322.1 334.5 347.0
distribution can be used. The third scenario r0. 0..
therefore assumes that the energy in the tar- :3. 0 0"0 01 .0 20 0. 0. 0 .0
get cell is a combination of a varying target 1.0. . 0o.o 0 . . .1 0- o* O
0.0
0.0 0.0 0.0
cross section as well as background energy 0.0 0.0 0.0 0.0 0.0 0.0
and system noise. Table 5 Example of the Performance Estimator

detection performance.
The principle of the model is that a threshold is
set according to the background and the required A.-..
false alarm rate. Fig. 5 shows the position of , '.M .. - " ' '.,. ."
the threshold for the three detection scenarios

with the probability density of the image intensi- . W'0"0
ties. The FAR is obtained by integrating the pro- 1

bability density of the background from the thres- . . ..........
/
hold to infinity. The detection probability is the
area under the probability density curve of the L
target from the threshold to infinity. For the ., Ii
first scenario it is always one.
With this model the Estimator can calculate FAR
and detection probability. Typical results for the . ...
ERS-l satellite are shown in Table. 5. Fig. 6
shows the detection probability plotted against "A
swath position with the false alarm rate as a pa-
rameter.
.
'"' ' " "'
. *[
"=
.] : " .
' "
The model elements of the Estimator are sumnarised

once more below: Fig. 6 Detection probability for the ERS-1 satel-
lite as a function of FAR. The object is
* a GEM6 Earth model which includes Earth rota- a large vehicle.
tion, ficient versus incidence
angle,
* a circular orbit (to calculate the platform
velocity), a variety of target models (RCS characteri-
stics embedded in noise and clutter) for
* uncertainties in the height of the platform, False Alarm Rate (FAR) detection modelling,
nadir and target,
the prediction of signal attenuation versus
" antenna and platform pointing errors, rain-rate,
* theoretical and measured azimuth and elevati- 0 a variety of ways of selecting the PRFs, as
on antenna patterns, well as user selected,
* a variety of ground clutter models which re- * the selection of an optimal set of PRFs for
present the expected radar backscatter coef- PRF hopping,
7-7
a variety of receiver gain setting algorithms the system analysis performed to date,
which use either centre swath signal, mean
swath signal and mean or the minimum/maximum * a listing of all system files available and
backscatter coefficient, as well as user se- the switching to other radar systems using
lected, function keys.
a choice of either fine or coarse range res- The Performance Estimator is typical of an algo-
olution, rithm based simulation tool. The software uses DEC
windows and allows the user to play with different
* various system hardware gains and losses, parameter settings and check the performance.
Clearly, the algorithms that are used are tried
0 the possibility of overlapping azimuth looks, and tested, so that few major surprises are to be
expected in the results. Full-blown simulators, as
a characterisation of the thermal and Analogue described below, mimic all the steps in the SAR
to Digital Converter (ADC) noise sources, system and can, hence, be used for scientific in-
vestigations.
0 a variety of amplitude weighting functions,
as well as user selected, 4. SIMULATION
0 the ability to degrade the range ambiguity A simulator attempts to model as many parts of the
ratio due to an expected range of radar back- SAR system without recourse to analytic functions.
scatter coefficients. The advantage of this approach is that genuine da-
ta are generated, which can be examined at each
The software characterising these functions has stage. Fig. 7 shows the elements of a SAR system
the following features: which many be simulated and the evaluations which
can be performed.
partitions which can be displayed on the
screen with descriptions, Although all elements contribute to the imaging
performance, some may be able to be simplified de-
DEC windows designed to allow the user to go pending on the application. For instance, if the
forwards and backwards through all the opti- application is concerned with the investigation of
ons to make and assess system changes quick- SAR data processors, a very primitive raw data
ly, generator with simple targets and ideal sensor may
be adequate for the task. For this reason simula-
the ability to generate antenna patterns from tors tend to grow, starting off being tailored to
a Fortran file, a particular application and later being augmented
to become a more comprehensive tool.
0 a facility to examine graphically the manner
in which, for example, the antenna patterns The first step in simulating a SAR system is to
and the ADC noise are modelled, model the target. The target can most easily be
simulated, if it is described by a two-dimensional
0 fast evaluation of a large number of swath matrix of scatterers each modelled as a complex
points which are split up into their azimuth vector voltage reflection coefficient. Fig. 8
look components, shows such a target model consisting of xy cells.
As will be seen, the simulator only has to form
0 all the timing information that can be paged the sum of all vectors within the instantaneous
for all the PRF and includes for example the field of view IFOV of the radar to generate the
predicted interference cases, time domain signal. The latter process is referred
to as raw-data generation and the final result is
0 tables and graphs of ambiguity, signal to to generate the signal received at the output of
noise and imaging data for all the selected the antenna. The raw data generation can be the
PRFs that are easily copied to a specified most complex part of the simulation chain, even
printer, exceeding the complexity of the SAR processor.
0 tables and graphs of required target size for Generating such a two-dimensional array of vectors
detection or detection probability versus assumes that the scattering process lends itself
false alarm rate for verification modelling, to this representation. Distributed targets with
differing normalised reflector coefficients ao
graphs of the components of the video signal can be synthesised with such a model, as can iso-
to noise ratio and the factors that degrade lated point targets. Mixing distributed and point
this, targets is also possible. The speckle associated
with distributed targets can be generated with an
a the analysis of the effect of changing any appropriate statistical model. Most simulators
system parameter and plotting the change pro- working in this way include utilities to help the
duced in any of the performance measures, user generate target patterns.
run information and a series of warning mes- The following description of the DLR simulator
sages when either algorithms are not functio- SARSIM will be used to illustrate such a simula-
ning or when the user has made a mistake, tor.
0 graphics software that allows the user to ea- 4.1 The System Simulator, SARSIM
sily select graphs, automatically scales and
grids the data, plots requirement curves and, SARSIM was originally developed by Ferranti in the
if applicable, fully labels the output and UK in 1977 taking about 15 man-years of effort.
provides a rescale facility for comparisons
with other graphs. In addition radar sensiti- The original software was extensively modified by
vity data are given under each modelled swath Ferranti in 1978/79, by Logica (UK) in 1981/82 and
point, currently by the DLR since 1978. Two versions of
the program exist, one on the DLR CRAY and more
book-keeping functions for the performance recently on a VAX4000 workstation. SARSIM is des-
plots which allow the user to keep track of cribed in [4 and 5].
7-8
SIMULATED Sensor Plafform

PARAMETERS Transmitter Attitude
Antenna Right path
Receiver
Data handling
Propagation Path
Attenuation
Polansation
Bd-errorsLink
Down
Target
Geometry
Composition
Scattering mechanism
SAR Processing Facility

Raw data analysis
Range compression
I Corner turn
iiIRange migration correction
LN Mult-ooking
Iterpolation
EVALUATION
Image Processing
S Venfication of system design
Influence of system parameters
Investigation of SAR imaging properties
Development of interpretation algorithms
Training of image interpretorsI

Fig. 7 Simulated elements and evaluation tasks of a comprehensive SJhR simulator.
TAGE adr Electronics Processor TARGE

OEL Model Model IMAGE
Fig. 9 Schematic of the SARSIM structure.

For each physical element of the system, there is
r a corresponding software module. These software
Conslat Range modules simulate the physical element characteri-
stics with the important exception of the linear
gain of the element. This means that the overall
model should be seen as a relative, rather than
Xabsolute, assessment tool. A more detailed schema-
tic of SARSIM is given in Fig. 10.
Here one can see that the simulator has been divi-
C ded into two segments. The first segment, signal
generation, deals with simulation of the SAR video
Aateam Fooe rint signal, that is the SAR system up until the point
where the analogue signal (in baseband) comes out
Fig. 8 Representation of the simulator geometry of the receiver system mounted on the platform.
with the target represented by two-dimen- The second segment, data handling and SAR proces-
sional array of complex reflection coeffi- sing, then simulates the remaining on-board elec-
cients. tronics, data-link, and SAR processing to produce
the SAR target image. Auxiliary programs can then
A schematic of the SARSIM model isgiven in Fig. 9. be used to create a hard copy of this image, apply
SARSIM is an "end-to-end" simulator producing as an post processing algorithms or to make image quali-
output the SAR image of the input target. ty measurements.
SARSIM has in-built options to generate several
7-9
processing, where the Doppler history of the tar-

get is split into N segments, each segment is se-
parately processed and then added incoherently af-
ter detertion. This technique improves the radio-
metric resolution at the expense of spatial res-
i olution. Finally the data can be corrected
I (calibrated) if desired.
As an alternative to the built-in processor, SAR-

SIM data can also be processed with the DLR mo-
tion compensation processor.
S4.1.1 The Target
The target is represented by a matrix of complex
• voltage reflection coefficients assumed to lie di-
rectly on the earth's surface. Distortions and ef-
_fects due to local height variations cannot be si-
-- -mulated. Similarly, variations of radar reflecti-
0.3l vity with incidence angle and time cannot be simu-
,. '' lated. The complex voltage reflection coefficient
II~i ~~j IVr + iVj is related to the radar cross-section a
K1 L..--.. .. __-_of the corresponding ground element by the follo-
wing equation,
mV+ Vj = ko
where k is a constant. The target size is restric-
. .ted to 192,000 elements.
... 7* l The target generation segment of SARSIM consists

of a two stage algorithm. In the first stage, a
matrix of cells is generated according to some
pattern corresponding to the ideal radar cross-
d section of the target. In the second stage, each
subcells valuea of
and the into
cell is divided the reflection
number coeffi-
(at least one) of
cient for each sub-cell is derived from the value
for the whole cell. This second stage permits the
..... introduction of speckle into the model, by assu-
,----, ming that the amplitude of each sub-cell value is
I ."
drawn from a Rayleigh distribution. The mean of
the Rayleigh distribution is specified by the cell
value generated in the first stage algorithm. The
,__________-- ____ "phase of each sub-cell value can either be fixed
or drawn from a uniform distribution (-x, x) Each
Fsub-cell value can be thought of as representing
the reflectivity from a point target located at
standard types of targets. Slope effects, such as the centre of the cell. Therefore, the target,
layover and shadowing, are not simulated. The ba- even though representing an extended target, con-
lyvp l sists of a matrix of point targets. This is a
sic principle of simulation is a "pulse by pulse standard method for modelling extended targets
simulation, that is the relative positions of c
platform and target, together with platform atti- Superimposed on this, the user can specify up
tude are calculated for each radar pulse and then to 16 deterministic point scatterers.
this is used to calculate the returned signal for Fig. 11 illustrates a possible target example. He-
each transmitted pulse. The returned signal is re- Fig ec illtas a pib tar exampl.
presented by a sequence of floating-point (com- re, each cell has been split into four sub-cells.
plex) numbers obtained by convolving the transmit- The amplitude, Ai of each sub-cell can be determi-
ted pulse form with the target, weighted by the ned in one of three ways:
antenna pattern and taking into account any spe-
cial receiver chain features. . Stochastic Background
The sample spacing of the returned signal is cho-

sen to match with the Analogue to Digital Conver- pit)
ter (ADC) sampling frequency. As the simulation is
performed directly in baseband, RF system element . .....
effects must be carefully accounted for. After the o 2n
ADC, the signal may be prefiltered or resampled if ,ub-cef (Ai . 61 A -u"
such an option is present in the real radar sy-
stem. Specified Scatterer.
(Determilstic)
In Fig. 10, alternative options of on-board analo- I
gue range processing or on-ground digital range I Stochastic (low reflectivity) background with Deterministic
processing are shown, these simply reflect alter- (high ret ectivity) superimposed point scatterers
native positions of performing the range proces-
sing in the SAR system. The range processing (and
azimuth processing) can be performed in the fre-
quency or time domain. After the range processing Fig. 11 Target examples.
the data is reordered (corner turn followed by 1) Ai is fixed and proportional to rJ.
range migration correction) before the azimuth
prccessing. Illustrated in Fig. 10 is multi-look 2) One value per cell is drawn from a Rayleigh
7-10
distribution (mean = Fo) and all Ai within In

in options B - E,
the pattern are values
set to not explicitlyvalue.
a background specified
the cell take this value.
4.1.2 Geomatry
3) An Ai value for each sub-cell in turn drawn
from a Rayleigh distribution. The same di- Two models exist for speoifying the geometry of
stribution is used for all sub-cells within the SAR system. The first model uses a flat earth
the cell. model together with a platform model assuming eit-
i of each sub-cell can be de- her constant linear velocity or stepwise constant
Similarly the phase, oacceleration components. This is suitable for si-
termined in one of three ways: mulating an airborne SAR. The second model uses an
1) 01 takes a constant value for every sub-cell ellipsoidal earth model, together with an ellip-
in the matrix. soidal orbit model. The second model is suitable
for spaceborne SAR simulations. In both models the
2) One value per cell is drawn from a uniform platform coordinate origin (x,y) is defined as
distribution and all 0i within the cell take the platform position when the beam centre and
this value, target centre are in alignment. The platform posi-
tion is expressed in Cartesian coordinates. In the
3) A 4-1value for each sub-cell in turn is drawn case of the ellipsoidal earth model, a local sphe-
from a uniform distribution. rical approximation is made to the earth's surfa-
ce,
roid.andThis
theassumption
target is assumed
is validtoso lie
longonasthis
onlysphe-
a
In the case where the sub-cell and cell are iden- r oid s assumpt i ai somlngda onlyh
tical, then options 2 and 3 in both of the above fewaseonds ofkflitt are simated th
are similarly identical. A cut-off point can also rotation is taken into account by combining the
be specified where option 1 is used for all azi- earth's surface velocity with the platform veloci-
muth bins greater tnan the cut-off point, and p- ty to produce
difying a relative
the platform yaw velocity
angle. vector, and mo-
tions 2/3 are used for the remaining azimuth bins.
This is referred to as a semi-diffuse target. 4.1.3 Platform Position and Attitude
The following
of the (first options are provided
stage) matrix for generation
of cells: The platform position and
Tepafrpsto n attitude are calculated
tiueaecluae
for every pulse in the time interval [-T/2, T/2],
A) User specified pattern: a two dimensional ar- that is the simulation is constructed time symme-
ray of amplitude data is read in from a file trically about time zero, time zero being the time
when beam and target are in alignment. In the case
provided by the user. of the elliptical orbit simulation, the orbit and
B) Cyclic pattern: a set of cycles is generated geometry parameters are sufficient to determine
in either the azimuth or range direction, the theoretical position of the platform for every
Each cycles consists of an alternating se- pulse. However, an additional perturbation may be
quence of high reflectivity and low reflecti- introduced in the
leration term. The form
flat ofearth
a constant linearrequires
acce-
vity amplitudes. The length of the sequences instead input data relating tosimulation,
the linear velocity
within each cycles is fixed, but increases by
one cell between successive cycle-series. A components at time zero and linear acceleration
example is given in Fig. 12. components before the platform position per pulse
can be calculated.
,st SNo 3The platform attitude per pulse is expressed in
CYCLE-SERIES CYCLE-SERIES CYCLE-SERIES yaw, pitch and roll angles. These can either be
AMPLITUDE -'f considered as constant or a.:!onforming to some
2 CELLS LS specified angular motion.
HIGH- I -C!
4AMPLITUDE 4.1.4 Transmitter

In SARSIM, the transmitter is ideally modelled,
with the only feature characterized being the
ideal form of the transmitted pulse. It is not
Low-.
AMP~UDE possible to simulate any system.
form ofMismatch
noise orbetween
distor-
!,ion in the transmitter
the up- and down-converters is assumed to result
T Oin a residual carrier frequency and this ismo-
-.4 I.--LENGR OF ,CELL AZLONG
SINSANC delled in the receiver simulation.
BININ NUMBEROF CELLS
The following forms of transmitted pulse can be
Fig. 12 Cyclic target pattern, simulated:
C) Linear pattern: a set of point targets of li-
nearly increasing amplitudes in one (or more) 1) A pulse with constant frequency.
, range bin or azimuth bin. 2) A linear r' pulse; the frequency of the
D) Two point targets: two point targets of stransmitttaly during thepulse
pulseincreases/decreases
trafismPsion. linear-
cified amplitudes are placed in either one
range bin or one azimuth bin. 3) A quadratic FM pulse; the frequency variation
E) Single point target: a point target of speci- of the pulse during transmission is a quadra-
F)
flied amplitude at a specified position.
Checkerboard pattern: a pattern of alterna-
tic function of time.
4) a binary phase encoded pulse; the frequency
of the pulse is constant, but the phase can
I
ting squares/ rectangles of high amplitude take two values: 0 and x.
and low amplitude.
The phase structure of the pulse is specified by
G) Square of
square wellzero
pattern: a uniform
re~lectivtwy target with a
inset. binary string.
~jk -~ -
7-11
pulse, the
In all four forms of the transmitted
amplitude of the pulse is constant. -, ,
4.1.5 Antenna -, -
range gate
- m
The antenna is defined by a two-way antenna gain
pattern, specified in azimuth and elevation. .... i....
The antenna gain function is assumed separable, 1 . . m
that is the antenna gain in any given direction is
obtained by multiplying the appropriate azimuth 1 t T t. r 'targe
and eaNd elements
for gain variations during the simulation period.
The following gain patterns can be used in the si-
mulation: Fig. 13 Target element contributions to the mth
range bin return using signal generation
primary option (time domain).
1) Uniform pattern G = 1.
range bins. It should be noted here that the tar-
2) Modified cosine pattern, get cell range dimensions should be chosen equal
cos(x) wxd to, or smaller than, the range bin width. This is
G = 2I where x sine , to minimise problems occuring when few target
(S)2 - X2 cells happen to fall in a given range bin, a re-
sult of the fact that the target is in fact model-
where: D is the effective antenna aperture led by a number of discrete point scatterers,
(in azimuth or elevation), rather than by a true extended target.
X is the radar wavelength,
a is the azimuth/elevation offset The primary method of signal generation inherently
angle, includes all range migration effects due to earth
3) Sinc pattern, rotation, earth curvature and radar wave front
sin(x) curvature. The simulation of the range modulation
G (x) where x is defined above, can either be performed in the time domain (by
W 'convolving the returned signal directly with the
transmitted waveform), or in the frequency domain
4) User specified pattern. The user must specify (ymliligteaporaeFuirtasom
a table of gain against azimuth/elevation (by multiplying
and thenthetaking
appropriate Fourier transforms
offset angle.offst anle.together the inverse transform).
The beamwidths of the antenna pattern (in azimuth Two secondary options for received signal genera-
and elevation) must be specified by the user. The- tion exist, both of which attempt to reduce the
se values define the extent of the generated an- amount of computer time required. The first of
tenna pattern, these options generates the expected return (phase
history) from a point target located at the centre
4.1.6 Propagation of the target matrix assuming an ellipsoidal, lo-
cally spherical, earth's surface and an unpertur-
The radar equation is not included in the SARSIM bed elliptical orbit. Appropriate account is taken
model. Most of the terms in the radar equation are of the azimuth antenna pattern. The azimuth modu-
constants (for most system simulator applica- lation is performed in the frequency domain and so
tions), apart from the range gain dependence (r4 ). the Fourier transform of this phase history is
For satellite simulations the lack of range depen- used to calculate the return range gate values.
dent attenuation is thought to be of negligible The same phase history is used for all range gate
significance (though of more significance for air- values. Range migration effects are not accounted
borne applications). for in this model. Note that the target element
spacing is assumed to be defined by the range bin
Atmospheric distortions are only included in as width in range and the pulse repetition interval
much as the optional addition of a phase noise in azimuth.
term. The noise follows either a uniform, Ray-
leigh, exponential, or normal distribution; in the The second option extends the above idea to partly
case of the normal distribution an auto-correla- include range migration effects by taking into ac-
tion function can also be specified. The auto- count that the return from a point target appears
correlation function can either be linear, quadra- in several range gates, or alternatively expres-
tic, exponential, Gaussian, or user supplied. sed, that neighbouring target gates contribute to
the return from one range gate (Fig. 14). The
4.1.7 Signal Generation transition points between neighbouring range/tar-
get gates are calculated at the centre of the tar-
The primary option for received signal generation get (mid-swath) and assumed to apply to all ran-
generates the returned signal for each pulse in ge/target gates. The phase history segments are
turn in the time domain. The distance to each tar- convolved with the target in the time domain. In
get element is calculated and then the returns both of the secondary options the range modulation
(after weighting by the appropriate antenna gain) can either be performed in the frequency or time
from all target elements within a range gate are domain, as in the primary option signal genera-
summed, taking into account the appropriate pha- tion.
ses, to give the final return range gate value.
This operation is performed for all range gates 4.1.8 Receiver
within the sampled return pulse and for dll pul-
ses simulated and is, hence, not an insignifi- The receiver system typically consists of a low
cant task. Fig. 13 illustrates the target returns noise amplifier a down-converter, an IF amplifier,
which contribute to the return range gate value, and an IQ detector. The low noise amplifier is as-
sumed to be perfect apart from the addition of
The exact form of the curved radar range bin de- thermal noise. The down-converter is similarly as-
pends on the orbit and earth geometry simulated. sumed perfect apart from the possibility of a re-
The width of the bin is determined by the sampling sidual carrier frequency resulting from a mismatch
frequency of the ADC, which, together with the re- bet.een the up- and down-converters. This residual
ceiver gate times, also determines the number of frequency produces an additional term in the un-
7-12
range
' ' ' - ' '
gate
'
! ! '------"7-----
t- ----
I ' I . "-'-,targetI
elements , .. -t
Fig. 14 Target element contribution to the mth

range bin return using signal generation ,
secondary option partially accounting for ..-----------
range migration.
compressed radar pulse. The IF amplifier is cha- ..--

racterised by a filter response and its non-linea-
rities. Fig. 15 Optimum equally spaced level quantiser.
The filter characteristics can either be specified quantisation threshold.

by the user in the form of a gain against input
signal amplitude table or by an analytical expres- 2) Quantise the data.
sion.
3) Rescale the data to compensate for scaling
After the IF amplifier comes the IQ detector. This introduced in 1), if required.
is modelled with the following error sources:
4.1.10 Prefilter/Presummer
" phase and amplitude errors,
" orthogonality error, The purpose of the digital prefilter is to reduce
* DC offset, the quantity of data to be transmitted or stored :
" gain imbalance. The filter can operate either in the range or azi-
]
muth directions, or on the transmitted chirp re-
Receiver system phase noise can be introduced im- plica.
mediately prior to the IF amplifier. The receiver
noise form (both phase noise and thermal noise The filter can either operate on both channels
from the low noise amplifier) follows either a (full quadrature) or just on one (real) channel
uniform, Rayleigh, exponential, or normal distri- (non-quadrature) by setting the remaining (imagi-
bution; in the case of the normal distribution an nary) channel to zero. The prefilter is assumed
auto-correlation function can be specified. The perfect apart from errors introduced by the finite
auto-correlation function can either be linear, word length within the digital filter. Word-length
quadratic, exponential, Gaussian, or user sup- effects (both before and during the filter) may be
plied, simulated by specifying the number of bits avai-
lable in the digital filter registers.
4.1.9 Analogue to Digital Conversion
4.1.11 Datalink
The Analogue to Digital Converter (ADC) can be
thought of as consisting of sampling and quanti- The datalink between the on-board system and
sing operations. The sampling operation is inhe- ground based processor can either take the form of
rent in the SARSIM model, as the analogue signal a telemetry link (satellite SAR) or a High Density
is represented by a set of discrete samples, Digital Tape (aircraft SAR). The model restricts
each sample consisting of a (complex) pair of the datalink simulation to allowing the user to
floating point numbers. The quantisation process specify a link transmission word length and then
then converts these floating-point numbers to a simulating bit errors over the link. The bit er-
finite set of quantisation levels consisting of rors can either be stochastic, where the spacing
both positive and negative values. Values excee- between bit errors is a random variable drawn from
ding the largest qunatisation level (ignoring an exponential distribution, or deterministic,
sign) are set to this level. The number of over- where a user-specified cyclic table (of up to 50
flows is counted. Quantisation of both the radar values) is used to determine the distance between
signal and transmitted chirp replica can be simu- one bit error and the next. The setting of the bit
lated. The quantiser itself can be specified in in error is switched. The datalink is simulated by
two forms: the following algorithm:
1. The user specifies up to 14 threshold values 1. Scale and quantise the data to the relevant
and corresponding output levels, link transmission word length, if necessary.
The scaling factor must be defined by the
2. The user specifies the number of bits n, the- user.
reby defining the number of quantisation le-
vp , (2n). The optimum equally spaced symme- 2. Insert bit errors.
i-ic quantiser is used as shown in Fig. 15.
3. Rescale the data to compensate for scaling
The quantisation process consists of the follov ng introduced in (1), if required.
steps:
4.1.12 Processing
1) Scale fhe data and add a DC offset, if requi-
red. The scaling can either be user-specified For a frequency modulated or "chirped" pulse, the
or automatic. For automatic scaling, the ma- first step in the SAR processing is to remove the
ximum, mean or mean squared, of the sar.ples frequency modulation. This is known as range pro-
in the first pulse, is adjusted (scaled) to cessing and can either be performed on-board in
be a user-specified fracticn of the maximum the analogue system using a direct replica of the
7-13
transmitted pulse, or, on-ground using either a

replica or theoretical values for the reference
function generaion. As far as the simulation model
is concerned, the difference in these two approa- "age range
ches to range processing is apparent only in the point target -.
fact that the range processing simulation has two response RM. C.
possible positions in the radar data chain (Fig.
10). Range processing can either be performed im- azim,,otme azntu,/tme
mediately prior to the ADC (on-board analogue ran-
ge compression), or immediately after the datalink Fig. 16 Range Migration Correction.
(on-ground digital range compression). It would
also be possible to simulate on-board digital ran- migration effects in the time domain.
ge compression if required. For each range bin in turn, the range walk for a
The first step in the range processing is to ge- target cell is calculated and applied. The range
nerate a range reference function or copy of the walk is calculated by considering the range histo-
transmitted pulse form. There are three ways of ry used in the azimuth reference function genera-
doing this: tion, and so the range wdlk function is updated
whenever the azimuth reference is updated.
1. Ideal replica. A copy of the transmitted pul-
se form used in the transmitter segment is The feature that distinguishes SAR from other ima-
generated. ging radar systems is its ability to achieve high
azimuth resolution through coherent integration of
2. Standard reference function. The possible the returned radar signal. As a target cell passes
forms of the reference function correspond to through the beam the phase of the returned signal
those possible for the transmitted pulse, changes, and this can be picked out by convolving 4
i.e. linear FM, quadratic FM and binary phase the (azimuth) returned signal with the appropriate
encoded. The user must specify the appropria- reference function. The convolution can be perfor-
te reference function parameters. med in one, or split into segments (subapertures
or looks), each segment being convolved separate-
3. User-specified reference function. ly. The resulting looks are detected (amplitude
taken) and summed to give, what is commonly cal-
Once the range reference function has been genera- led, a multi-look image. The simulator can perform
ted it is conjugated, and then used in the com- all these operations, starting with the azimuth
pression. It is also possible to specify a refe- reference function generation.
rence function weighting. Compression can either
be performed in the frequency or time domain. For The following reference functions can be genera-
frequency domain compression, the reference func- ted, either as one reference function (single look
tion and signal must first both be transformed to processing) or as several subaperture reference
the frequency domain, multiplied together, and functions (multi-look processing):
then inverse transformed back to the time domain.
1. Curved Earth Time Invariant: the same phase
For time domain compfession, it is necessary to history is assumed to apply to all target
directly correlate the reference function with the cells in a given range bin. The user must
signal. Two methods of time domain correlation are specify the bin number before the reference
possible in the simulation model: the overlap add function can be calculated.
technique and overlap save technique. These tech-
niques are specially designed for the case where 2. Flat Earth Time Varying: the phase history
one function (reference function) is considerably for an individual target cell is calculated.
shorter than the other (range line), and are fully The user must specify the coordinates of the
described in [7]. target cell. This option allows several dif-
ferent azimuth reference functions to be con-
After range processing, the radar data on disk volved with each range bin, by generating re-
must normally be re-ordered before the range mi- ference functions for target cells with dif-
gration correction and azimuth processing can be ferent azimuth coordinates within the bin.
performed. Up until this point the radar data file
has consisted of successive range lines or azimuth 3. Curved Earth Time Varying: as for option 2,
bins. The azimuth processing requires the data are but for curved earth geometry.
ordered so that the data file consists of succes-
sive azimuth lines or range bins. This process is In all three of the above options the azimuth re-
called corner turning. ference functions are generated for a particular
range bin. The reference functions generated
Range migration is caused by the fact that the di- though, may be applied to all range bins, or al-
stance (hence time delay) to a target cell changes ternatively updated (new reference function
as the target cell moves through the beam. This is generated) every n range bins, as required, where
due to earth rotation, curvature of the earth, and n is specified by the user.
the ellipticity of the orbit. The end result is
that the response from one target cell is not in The platform geometry used in the azimuth referen-
one range bin, but split into bands over several ce function generation is, in the basic case, the
range bins. Therefore, before the radar can be same as that used in the returned radar signal
processed in the azimuth direction, it is necessa- generation, i.e. it is assumed that the platform
ry to re-order the data so that the response from position and attitude are exactly known. Errors
one target cell is observed only in one range bin can be simulated by generating a new platform data
(Fig. 16). This must be repeated for each pulse file, specifying different geometry and/or error
and across th swath, and can involve some form of vdlues, this is then used for the reference func-
interpolation. Range migration itself can be split tion generation, or, more simply, by specifying
into two components: a linear component (range geometrical or Doppler errors. In the case of ge-
walk) representing the difference in range to a ometric errors, the user specifies slant range and
target cell between when it enters the beam and yaw angle errors, which are directly used in the
when it leaves the beam, and a higher-oder terms reference function generation. For Doppler errors,
component (range curvature). The simulation model it is necessary to convert the Doppler errors to
has an algorithm to correct for only the linear equivalent range and yaw errors. Doppler errors
7-14
are specified in terms of a Doppler slope error

(slant range error) and a mean Doppler error (yaw 0.0o P0.E SLOPE-4.0000E1 8A0055I
angle error). The azimuth reference function can P0S5 ,R4WI0ih.Z.36 -1

be weighted.-oo
Once the appropriate azimuth reference functions

have been generated, it is necessary to convolve 0-4o.0
them with the range migration corrected signal da-
ta. It is possible to perform the azimuth proces-
sing in either the frequency or time domaain, using 0.
exactly the same techniques and algorithms as in
the range processing. After azimuth processing,
the signal data are detected, that is the moduli
(or moduli squared) of the complex values are ta-
ken, and, in the case of multi-look processing,
the individual looks are summed. Word length ef- 8. 4.5 4.6..
fects in the summation can be simulated, and the SARSIKH E2
summation can be weighted (for example to accommo-
date antenna gain pattern effects). Fig. 19 Signal after azimuth compression.
4.1.13 SARSIM Application Example series of small plates or facets enabling the tar-
get to be entered as a complex shape. The model is
In this section an :xample of a satellite SAR si- capable of handling hidden surfaces (radar sha-
mulation is given. The configuration of the system dows) and can generate polarimetric signatures, a
is based on that for the ESA Remote-Sensing satel- feature of increasing interest for the next gene-
lite (ERS-l). Errors and noise sources within the ration of SAR sensors.
system are not simulated in this example.
5. CONCLUSION
Fig. 17 shows the radar target used in the example
simulation: a single point target on a non-reflec- Parametric analysis and simulation software packa-
ting background. Fig. 18 is a plot of the range ges for investigating SAR systems have been des-
compressed SAR data. Fig. 19 shows the final image cribed and their merits discussed. The usefulness
plot of the single point target. of simulators is often called into question as the
cost and manpower invested can be large. It is,
therefore, worth considering two applications of
0.0, the SARSIM simulator.
In 1984, the ERS-l project was faced with the pro-

- blem that the coverage of the synthetic aperture
radar mode of the Active Microwave Instrument
known as the 'wave' mode was found to be marginal
for the application it was designed for. The wave
mode images small areas of the ocean surface, the
data being stored on-board the satellite to achie-
.ve global coverage. From the images, the ocean wa-
ve spectrum was to be derived by generating the
two-dimensional Fourier power spectrum. The limit-
ed storage capacity of the recorder meant that the
imaged area was limited to approx 2 x 5 km. This
O0.0 0 .. area was found to be too small to detect long wa-
5ARSH l yes.
Fig. 17 Simulated point target The only practical solution was to reduce the
amount of recorded data by halving the number of
bits used for quantising each SAR data sample from
P01,0fL-0oCHh401515 4 bits 1/4 bits Q to 2 bits 1/2 bits Q. Analysis
5E ,..0-2.36
-1 showed that quantisation noise would be acceptab-
S ,___ le, but would sufficient information L2 retained
to derive the wave spectrum?
__--- ____-- -- _---_ The problem was analysed using SARSIM [9]. An
ocean scene was simulated (Fig. 20) and the image
S n =' (Fig. 21) used to generate the power spectrum
(Fig. 22). Analysis of the latter showed the wave
-- spectrum was still reproducible. The new quantisa-
tion scheme was introduced and today the ERS-1 sa-
I .tellite has a coverage of 5 x 5 km in the wave mo-
S% 6 -- de.
SARSIA
I42 041801, ,0054, The above example shows how simulation can remove
final doubts even if analysis of the problem is
Fig. 18 Signal after range compression. successfull.
4.2 Simulator with Facet Backscattering Model Another example also concerns quantisation noise.
When plotted against input signal power, the quan-
The two-dimensional reflection coefficient target tisation noise shows a distinct minimum increasing
model used in SARSIM is adequate for many applica- at low powers, due to the discrete quantisation
tions, for example sensor investigations but it is steps and at high powers, due to saturation (see
very limited if scattering mechanisms of complex Fig. 23 and Fig. 24). These curves were derived
targets are to be examined. The facet backscatte- theoretically (10] (dotted) and SARSIM was used to
ring model used in the SARAS simulator, developed confirm the results. The simulation results are
by Naples University (8] models the target as a joined by dashed lines.
7-15
50.0 # ERS-1
Imp 05.A0 N
S IO EOO-IO..i~. Dy.AW ft'U
-II l~i*Sed D0W 00'

40.0
UNFYEAED S.
0.0
-00SCALE IAEFig. 24 5-bit quantisation noise curve.
AZIMTH PIXES1 atchd flter inthe SAR processor. This is an

example of simulation throwing up surprises and
Fig. 20 Simulated ocean wave scene. leading to further investigations.
4SIT RO . MISMATCH 0 0O 00 M CEA4 WW
A-ALDUI IN OARDMANG
G1191511"6. REFERENCES
Hounam, D. Derivation of the Techni-

05[1]
Pierschel, D. cal Specification of the
ERS-1 Active Microwave
3.0 Instrument to Meet the SAR
Image Quality Require-
'1 ments.
Proc. of IGARSS '87 Symp.,
Ann Arbor, 18-21 May 1987.
[2] Chorherr, G. IMPRES, A Program to Gen-

SRRS!M a. G.S 1.0 :.5 2.0 2
..E2 Hounam, D. erate the Impulse Response
.4..,f)IX5 VOLAOA:uTTP!0L5~of a Synthetic Aperture
Radar.
Fig. 21 SAR image of the scene in Fig. 23. DLR-IB 551-3/92.
f'
Z'
11597&008
RNLGEONBADPotter,
000000TSIDN
MANGE
2TINE 5.5DLR-IB
INI
31 Hounazo, D.
S.
Schmid, R.
Algorithms and Software
User Guide for the Perf or-
mance Estimation of Syn-
thetic Aperture Radars.
551-4/92.
[4] Pike, T.K. SARSIM: A Synthetic Aper-

ture Radar System Simula-
tion Model.
DFVLR-Mit. 85-11.
PP[51 Pike, T.K. Analysis of ERS-1 SAR Per-
ERS ~.formance through Simula-
1 'EE2
Proc. IEEE '86 National
--A- ALIRT"
March 12-13, 1986, pp.

12-18.
Fig. 22 Fourier spectrum of the image in Fig. 24.
AW-? i W
IWV[6] Krul, L. Principles of Radar Measu-
*,. rement. Proc. EARSeL Radar
Calibration Workshop, Alp-
bach, Austria, December
S.~TN0NT I 1982, ESA SP-193, pp. 11
.... 0Oow" 20.
(7] Rabiner, L.R. Theory and Application of

B. Digital Signal Processing.
T\Gold,
Englewood Cliffs, NJ:
:2: II
i~g.0I..1.0Riccio,
/
I I 81 Franceschetti, G.
Migliaccio, M.
D.
Prentice Hall 1975.
SARAS: A Synthetic Apertu-
re Radar (SAR) Raw Signal
Simulator.
Schirinzi, G. IEEE Transaction on Geo-
Fig. 23 2-bit quantisation noise curve, science and Remote Sensing
Vol. 30, No. 1, Jan. 1992.
For both the 2 bit case (Fig. 23) and the 5 bit
case (Fig. 24), the simulated values are much less
than the theoretical results. An at least partial
explanation of this is the non-white nature of
quantisation noise [11] and the influence of the
7-16
[9] Wolfram, A.P. Quantization Study for ERS-1

Pike, T.K. Wave Mode.
DFVLR Research Report,
DFVLR-FB 84-39, October
1984.
[10] Sappl, E. An Optimal Quantisation of
Coherent Radar Echoes from
Terrain or Sea Surface.
DFVLR, Institut f~r Hoch-
frequenztechnik, Report,
Sept. 1984.
[11] Li, F. Simulation and Studies of
Held, D. Spaceborne Synthetic Aper-
Honeycutt, B. ture Radar Image Quality
Zebker, H. with Reduced Bit Rate.
15th International Sympo-
sium on Remote Sensing of
the Environment, Ann Ar-
bor, May 1981.
it
__ _ _ _ __ ___ _ _ i
8-1
Multi-Frequency Multi-Polarization Processing for Spaceborne SAR
J.C. Curlander and C.Y. Chang

California Institute of Technology
Pasadena, Ca 91109
United States of America
Abstract
The Shuttle Imaging Radar-C (SIR-C) is the third in a series of space shuttle based synthetic
aperture radars (SAR) sponsored by the National Aeronautics and Space Administration
(NASA). The SIR-C ground data processing system is to process the playbacked SAR signal
data into a variety of data products for distribution to the science community. This paper
presents an overview of the end-to-end ground data processing system with emphasis on
the unique characteristics involved in the system design. Included in the discussion are
science requirements, radar system specifications, input data format specifications, system
operations design, data products design, processing algorithm design, hardware architecture
design and software design.
1. INTRODUCTION ucts. Table 2 shows the orbit characteristics and the radar
specifications.
will for the first time
The Shuttle Imaging Radar-C (SIR-C)
provide simultaneous data acquisition of eight radar chan- This paper presents the system design of the SIR-C ground
nels (two radar frequencies: L and C, each comprised of data processor, which is being developed at the Jet Propul-
two like-polarized and two cross-polarized channels) from a Table 2: SIR-C orbit characteristics and radar
spaceborne synthetic aperture radar (SAR) (Curlander, 91a], specifications.
[Curlander, 91b], [Jordan, 91]. This instrument is accompa-
nied by a VV-polarized, X-band SAR (X-SAR) which will SAR Orbit
operate simultaneously with SIR-C. The SIR-C/X-SAR is -Nominal Atitude 215 ± 25o Km
-Eccentricity s 0.002
currently scheduled for three space shuttle flights, the first - Inclination 57*
being in late 1993. Each flight is planned for a six to eight Attitude Measurement Error(3o)
day data acquisition period. Table 1 summarizes the SIR-C - Roll ±1.24"
science requirements on the image quality of output prod-
Yaw ±1.430
- Pitch ±1.73
Table 1: SIR-C science requirements on the image Attude Drift Rate Eror(3o)
quality of output data products. Roll +0.03-1soc
- Yaw ±0.03°lsec
- Pitch ±0.031saec
Reslution broadening s 20% Transmitter frequency
Integrated Sidelobe Ratio 1 -14 dB L-Band 1.25 GHz
- C-Sand 5.30 GHz
s -17 d8
Peak Sidelobe Rato Polarization HH, HV, VH. VV
Ambiguity to Signal Ratio s -20 dB Pularidt 2H, 10.MHz

wahWIdth Swt kt 15
5KKm to
o990 Km3.8
mPulse duration 33.8, 16.9, B.44 usec
Radlometric Accuracy (Sdence Go Sempng ate 45. 22.5 Ihz

- Relative Crose-Swath (to) t 1.0 dB
- Relative Band-to-Band (I) ± 1.5 O1 Data quanhizaton lormat 4-bit, 8-btt, (8,4) BFPQ
- Relative Channel-to-Channel (lo) t 1.0 dB
-Abeolute Each Channel (3o) ± 3.0dB P.lge~epatit Frequency 344, 1395, 1440 148
- Phase Error s 10 1512. 1620, 1674. 1736 Hz
Gom Accuracy (o) Antenna Olmalon

-A.eolaeo Locatin S 100 m - L-Band 12.1 m x 2.8rm
- Chennel-toChannel Registration Error S 1/1 pixel - C-fand 12.1 m x 0.74 m
- a er 0.1%
- Skew error 5 0.1% hIddaene Angle 17" to 63"
II
8-2
sion Laboratory (JPL) under contract with NASA [Curlan- multi-band, polarimetric SAR image [Freeman, 89], [Klein,
der, 91b]. There are two European space agencies working on 92]. For radiometric calibration, the built-in-test-equipment
the X-SAR processor development: the German Aerospace (BITE) data are designed for probing the health of the an-
Research Establishment (DLR) and the Italian Space Agency tenna, receive only noise data for estimating the noise power,
(ASI) [Runge, 90]. calibration tone signal for monitoring the receiver gain and
temperature measurements and T/R module failure informa-
The major challenge to the SIR-C processor design is to cope tion transmitted via the downlink telemetry. These ancillary
with a large number of radar modes. Nominally the SIR-C data and calibration site data are essential to derive radio-
science team has selected seventeen data acquisition modes metric calibration parameters which are applied during the
from all the possible combinations with eight radar chan- data processing to produce calibrated image product. For
nels (See Table 3), two pulse bandwidths and three data geometric registration, special consideration is required in
quantization formats. Additionally, data will be collected in the processor design to ensure that the output images are
two nominal attitudes over incidence angles from 170 to 630 registered in both cross-track and along-track dimensions.
with a variety of antenna elevation patterns controllable to
provide beam spoiling at the steep incidence angles. The The remaining paper presents an overview of the SIR-C end-
large number of radar modes complicates the logics in han- to-end ground data processing system design, which includes
dling different types of data format and increases the scope input data format specifications, system operations design,
of testing the integrated processor software, data products design, processing algorithm design, hardware
architecture design and software design. At the end of the
The second challenge to the SIR-C processor design is to paper, we give a brief summary of the status anld plan for
attitude
cope with the large attitude uncertainties and high
drift rates of the space shuttle platform as shown in Table the processor development.
2. The large attitude uncertainties create PRF ambiguity
problem in Doppler centroid estimation and large Doppler er-
rors for data acquired over high terrain relief areas. Special 2. INPUT DATA FORMAT
techniques (multiple PRF technique and attitude steering
technique, respectively) are required to resolve the problems The signal data is recorded across four recorder channels on
[Chang, 92a), [Chang, 92b]. The high attitude drift rates in- the High Density Digital Cassette (HDDC). The data rate
duce fast Doppler drifts in both cross-track and along-track for each recorder channel is 45 Mbps for a total of 180 Mbps.
dimensions. Frequent Doppler update is required to main- The signal data is quantized into one of three types of for-
tain the image quality, which complicates the geometric rec- mat: 4-bit, 8-bit or (8, 4) block floating point quantization
tification procedure to produce a seamless image. (BFPQ). Nominally, the data is collected over a period called
data take using the same set of commanded radar parame-
ters. The length of the data take varies from 3 minutes to as
The third challenge to the SIR-C processor design is to pro-
duce radiometrically calibrated and geometrically registered long as 15 minutes for ocean site data. The average length
is estimated to be between
4 and 5 minutes.
I,
Table 3: SIR-C radar data aquisition modes. The nominal SIR-C data take consists of a turn-on sequence,
followed by the science data collection and a turn-off se-
WaE CHANNEL-1 CHANNEL-2 CHANNEL-3 CHANEL4 quence as shown in Figure 1. The first four seconds of the
I tHH IHH LHV IHV turn-on sequence consist of (one second each): receive only
2
3 LVH LVH LVV LVV
044 0l C4V CVV noise data, caltone scan data, low noise amplifier (LNA)
4 CVH CVH CVV CVV BITE data and high power amplifier (HPA) BITE data.
5 UH UHH LVV LVV These four second data are used for radiometric calibration.
6 044 01H CVV CVV The remainder of the turn-on sequence consists of one sec-
7 LHH LHH CVV CVV
a tHH w
C"1 044 ond of PRFA data and one second of PRFB data which
9 LVV LVV CVV CVV together with the first second of PRFc data are used for
10 1 LV LVH GVH I CVH resolving PRF ambiguity in estimating the Doppler centroid
II UH LM 044 COf
12 LVH LW CV4 CVV frequency [Chang, 92a]. The system remains on PRFc for
13 U4H LVV 04 CVV collection of the science data. The turn-off sequence
S '14 UNH LYN LHV LVV sqec is simi-
14 . LM 2HV WL. lar to the turn-on sequence in that the science data collection
Is LVH, LH LVV, UHV CV, CHN CW, CHVi is followed by one second each of PRFV and PRFA. The
T last four seconds of the turn-off sequence are receive only
T=~~ CUM, I CVV noise data.
8-3
Null-Unes at every one second time tick time Rn
4Turn-On Sequence Target Turn-Off Sequence
FN CAL. LNA 4PA PRF PRF PRF C FCN IIN I:N

Scan BITE BITE A BA
or or
FC3N FECN
4 6 seconds ------ Average 4.5 minutes - 4 - 6 seconds
Figure 1: SIR-C input data run format. Each segment in turn-on and turn-off sequence is 1 second duration.
RON: Receive Only Noise, LNA: Low Noise Amplifier, HPA: High Power Amplifier, CAL scan: Caltone Scan.
At every one second time tick, a null-line is inserted. The 4. DATA PRODUCTS DESIGN 4
null-line is obtained by setting a half of the phase array el-
ements with a 1800 phase difference to create a null around The SIR-C output data products include three image prod-
the center of the antenna elevation pattern. The null-line is ucts: survey image, standard multi-look image and standard
used to estimate the shuttle roll angle drift, single-look image; and one reformatted signal data product.
The throughput requirements are
to produce 24 survey im-
A sinusoid waveform, called the calibration tone (caltone) age products per week during the phase 1 operations and
signal, is injected in the receiver electronics and recorded to produce 9 standard multi-look image, 1 single-look image
together with the return echo data. The caltone is used to and 1 reformatted signal data products per week during the
estimate the receiver gain change as the temperature varies, phase 2 operations. The expected processor throughput far
exceeds the requirements.
3. SYSTEM OPERATIONS DESIGN
The survey image is a 4-look, single-polarization strip image,
Operations of the SIR-C processor is comprised of two main stored in the byte amplitude format. The image is deskewed
phases: phase 1 survey processing and phase 2 standard pro- to zero-Doppler and resampled to the ground range domain
cessing, which last for a total of one year. During the phase 1 with a 50 meter pixel spacing. The resolution is approxi-
operations, a quick-look survey processor is employed to pro- mately 100 meters. The length of the survey image is equal
cess single-frequency band, single-polarization channel data to the length of the data take. The average length is approx-
into low resolution strip images. These survey images will imately 4.5 minutes or 2000 Km. The survey image will be
cover all the SIR-C ground sites albiet with a single radar recorded on Alden thermal prints and CD-ROMs. The CD-
channel. By-products of the survey processor include un- ROMs will be distributed to all the principal investigators
ambiguous Doppler centroid estimates history and roll an- (PIs).
gle estimates history. Additionally, during the phase 1 op-
erations, some selected data segments (covering calibration The standard multi-look image is a multiple look, polarimet-
sites) will be processed into single-look, full-resolution com ric (single-, dual- or qnad-pnlari7Atinn) frame image. The
plex imagery. These data will be analyzed to derive the image is deskewed to zero-Doppler and resampled to the
parameters used for antenna pattern generation and polari- ground range domain with a 12.5 meter pixel spacing. The
metric calibration. These parameters will be applied during azimuth resolution is chosen to be 25 meters. The range res-
the phase 2 standard processing to produce phase and am- olution is chosen to be 25 meters or the natural resolution if
plitude calibrated data products [Freeman, 891, [Klein, 92]. greater than 25 meters. The image data is stored in a com-
pressed cross-product format [Dubois, 89]. The basic frame
The system operations schedule is planned as follows. Six size is chosen to be 100 Km. The image will be recorded on
weeks are allocated for processor check-out upon receipt of Kodak prints and CEOS formatted tapes.
the first signal data tape. Phase 1 operations will begin
following the completion of the processor system check-out The standard single-look complex image is a single-look, po-
and last for a period of twelve weeks. This is followed by larimetric (single-, dual- or quad-polarization) frame image.
phase 2 operations for a period of forty weeks. The image is processed to full-resolution, deskewed to zero-
I
s-S.
8-4
Doppler and presented in the slant range domain in natural is performed using the spectral analysis (SPECAN) algo-
pixel spacing. The image data is stored in a compressed rithm which requires fewer azimuth FFT's than the tradi-
scattering matrix format. The basic frame size is chosen to tional matched filtering algorithm. Following azimuth com-
be 50 Km. The image will be recorded on CEOS formatted pression, radiometric correction is applied to compensate for
tapes and a reduced, detected image will be printed by the the along-track radiometric modulation. This is followed by
Kodak printer, a geometric rectification step that resamples the slant range-
The reformatted signal data contains the signal data refor- Doppler image into the ground range cross-track and along-
matted in the range line byte format. The signal data to track domain. The rectified burst images are then overlaid
pouetefnlmlt-oksrpiae
to produce the final multi-look strip image.
gether with the decoded radar parameters will be stored on
CEOS formatted tapes. For the survey processor, the initial Doppler centroid fre-
quency is determined using a clutterlock routine and a ambi-
5. PROCESSING ALGORITHM DESIGN guity resolution technique that requires a multiple PRF data
collection at the start of each data take [Chang, 92a]. The
5.1 Survey Processing Algorithm unambiguous Doppler centroid frequency is then tracked by
a burst mode clutterlock algorithm during the data process-
The SIR-C survey processor utilizes a burst mode processing. The Doppler frequency rate is solely derived from the
ing algorithm [Sack, 85], [Curlander, 91b]. The algorithm ephemeris parameters. Analysis results show that the accu-
flow chart is shown in Figure 2. The survey processor is de- racy of the ephemeris is sufficient for generation of survey
signed to process an entire data take into a strip image in products without employing the autofocus routine.
approximately one-seventh the real time data collection rate.
To attain high throughput rate, the data is bursted in az-
imuth (slow time) with a one-quarter duty cycle factor. The
data volume is further reduced by a factor of four in range Prior to standard processing, preprocessing is employed to
(fast time) by processing the data using only one-quarter iiteratively refine the Doppler centroid frequency and the
of the range chirp bandwidth. The azimuth compression Doppler frequency rate estimates using clutterlock and aut-
ofocus techniques [Li, 851. Doppler centroid frequency is
Selecled Null Lines Bursted estimated from the azimuth spectrum by locating the en-
segment Adjacent Lines Range Lines
of data ergy centroid. Doppler frequency rate is estimated from the
look registration error by azimuth cross-correlating the look-
Caltoe Null-Line RneWak 1 and look-4 images obtained by spectral division. Identical
P Ing Poel Doppler parameters are used for processing all polarimetric
gain angle Range ange data channels to ensure the phase coherency required for the
estimate stimate Reference Comprsslon polarimetric data analysis. This approach will result in some
Cross-Track Generation
Radiometric oincrease in azimuth ambiguities if the antenna beams are not
i Turn exactly aligned.
ramp Deramp The range-Doppler processing algorithm (i.e., the rectangu-
Generation R lar algorithm) with secondary range compression and fre-

Azimuth quency domain range cell migration compensation was se-
Forward lected by SIR-C for standard processing [Vu, 82], [Jin, 84],
Cluttorlock/L [Curlander, 91a]. The algorithm flowchart is shown in Figure
Doppler r3. The range compression and azimuth compression matched
Radlometric filtering operations are performed using the frequency do-

main fast convolution technique. All the signal data, in-
Geometric dependent of the final products, are initially processed to

Rectificatio single-look, complex imagery using the full azimuth process-
ing bandwidth. This is followed by an azimuth deskew op-
eration where the resulting deskewed, single-look complex
corrected.
is then radiometrically
Simage
Image Following standard processing, postprocessing is employed
Figure 2: Survey processing algorithm flowchart, to generate-the final image product (Curlander, 91b). Data
Tlj
8-5
raw data reduction is the only postprocessing function for generation

of single-look image products, where the data reduction func-
Range FFT tion is applied to the scattering matrix. Major postpro-
cassing functions for generation of multi-look image prod-
ucts include cross-product generation, multi-look filtering
Generatne and data reduction where the multi-look filtering combines

Fange
i multi-looking as well as geometric rectification functions. For
Inverse FFT SIR-C, all the multi-look images will be filtered to a 25 m
resolution in azimuth and a 25 m or natural resolution in
Cornor range. The pixel spacing is selected to be 12.5 m in both
1 rsndaj fd, fr__ Turnrange and azimuth. The filtering is applied to the cross-
Az t products. The data reduction function is then applied to
the multi-look filtered cross-products data.
Migration|
6. HARDWARE ARCHITECTURE DESIGN
Single-look
Full-aperture
Azimuth I Reference
Iutniaid I Figure 4 shows the hardware architecture design of the SIR-C
Reference Iground data processor. The entire processor system is com-
Generation Azimuth posed of seven subsystems. The Data Transfer Subsystem
nvers FFT I (DTS) performs raw data reformatting and line synchroniza-
Azimuh I tion. The SAR Correlator Subsystem (SCS) processes the

Deskew SAR signal data into survey and standard image data. The
Output Products Subsystem (OPS) performs image data re-
azimuth deskewed, formatting, recording and display. The Control Processor
single-look Executive (CPX) controls the processing sequence of the
complex image
above three subsystems. The Catalog Subsystem (CAS)
Figure 3a: Standard processing algorithm flowchart, stores the information concerning the processing request and
processor status into database. The Calibration Subsystem
(CAL) is used for generation of calibration narameters and
analysis of calibration site image quality. The Radar Data
azimuth deskewed. Center (RDC) archives all the output data products.
single-look
complex image The SCS consists of a STAR array processor with three coin-
putational modules, an Alliant FX/8 mini-supercomputer
Single-Lool Multi-Lool with eight compute elements and an Alliant FX/2800 mini-
Complex Complex supercomputer with twelve i860-based CPU's. The STAR
Multi-Lool array processor is the main compute engine for survey proces-
Detected sor. Its FFT performance is measured at 120 MFLOPS using
Oversampling Oversampling three computational modules. The Alliant FX/8 is primar-
and Slant-to- and Slant-to-
Ground Range Ground Range ily used for standard poetprocessing functions. Its aggregate
Conversion Conversion FFT performance nears 20 MFLOPS. Two SKYBOLT accel-
SLo - erator boards are installed to speed up the FX/8 computer,
Reduction Filtering
Isatoi
(intensity
II Filtering
(cros'productl
which provide additional 100 MFLOPS compute power. The
Jdt )atrin r FX/2800 is the main compute engine for standard processor
matrix dat) data) data)
Tta
and standard preprocessor. Its aggregate FFT performance
is measured at over 300 MFLOPS. Computational tasks are
Reduction
(cross-product distributed over computers for concurrency processing in or-
data) der to provide maximum processor throughput.
The DTS consists of a high density digital recorder, a DE-

B
C id MUX and two data quality analyzers (DQA). The DEMUX
is used for selection of recorder channel for data processing.
Figure 3b: SWndwd poslprocessf algorithm flowchart. The DQA is used for line synchronization and verifying the
0-
8-6
Dat
D Transfer Subsystem (OTS)
__ IF24]I .6'
m Cutalog Subystm AS WN
LAWa iA 401_411
eam
"I
Figure 4: SIR-C ground data processing system hardware architecture.
data quality and integrity. The OPS consists of Exabyte rically registered. Another challenge to development of this
tape drives, Alden thermal printers and Kodak color print- large software based system is its complex interfaces among
ers. Three subsystems, OPS, GAS and CAL, run on three the many software programs. Clear interface definitions are
separate SUN Sparc workstations. The image display and essential to successfully deliver the operational system on
~operator interface display are handled via X-ternminals. schedule.
7.SFT AEDEINCurrently, we are in the middle of developing all the proces-

~sor software. Major computer hardware will be installed by
There are a variety of software packages used for develop- summer 1992. The end-to-end system integration will take
ing the SIR-C processor due to the need of specific applica- place in early 1993. The entire system is scheduled to begin
tions. The major part of the signal processing software is operations in late 1993.
written in FORTRAN while the input and output format-
ting software is written in C. The image display software is o t
developed using X-library routines. The operator interface experience from the previous and existing spaceborne and
software is developed using a graphics user interface software airborne SAR processors, such as SEASAT, SIR-B, and JPL
calldrns
Tleusn whch
tp ofMOTF. Te iage AIRSAR. Experience accumulated from the SIR-C proces-
~notation is created using a commercially available software sor will certainly benefit future processor design and devel-
i package called PV-WAVE. The 7.
catalog SOFRWFRERDECIS
subsystem software opment, such as EOS SAR and RADARSAT.
1'
~uses both FORTRAN and INGRES.
8. SUMMARY
nmjosotrparwaeoftis
tion.pocesiTe opeatins nlte 993
he ignl
SACKNOWLEDGEMENTS
The research described in this paper was performed at the

writen
andoutpt
i FORRAN
fomat hiletheinpu
! Jet Propulsion Laboratory, California Institute of Technol-
Design and implementation of the SIR-C ground datapro- og, under the cntract with the National Aeronautics and
casing system is quite a challenge due to the large number Space Administration.
of radar modes and the large attitude errors/high attitudemn
drift rates. In addition to the correlation software, there are
ny sfrware. prograM required for deriving parameters i d b
from the ancillary data in order to ensure that the output [Chang, 92a C.Y. Chang and .C.Curlander, "Applications
inmge products are radiometrically calibrated and geomet- of thlleMule PRF Technique trvete is Doppler Cetrod
8-7
Estimation Ambiguity for Spaceborne SAR, IEEE Trans- Transactions on Geoscience and Remote Sensing, Vol. GE-
actions on Geoscience and Remote Sensing, September 1992. 22, No. 6, November 1984, pp. 592-597.
[Chang, 92b] C.Y. Chang and J.C. Curlander, "Attitude [Jordan, 91] R. Jordan, B. Huneycutt, M. Werner, "SIR-
Steering for Space Shuttle Based Synthetic Aperture Radar, CX-SAR Synthetic Aperture Radar Systems, Proceeding
Proceedings of 1992 International Geoscience and Remote of the IEEE, Vol 79, No. 6, pp. 827-838, June 1991.
Sensing Symposium, Houston, May 1992. [Klein, 92] J.D. Klein, "Calibration of Complex Polarimet-
ricSAR Imagery Using Backscatter Correlations, " IEEE
[Curlander, 91a] J.C. Curlander and R.N. McDonough, Syn- Transactions on Aerospace and Electronic System, Vol. 28,
thetic Aperture Radar: Systems and Signal Processing, John No. 1, pp. 183-194, January 1992.
Wiley and Sons, 1991. WLi,85] F. Li, D. Held, J.C. Curlander, and C. Wu, "Doppler
[Curlander, 91b] .1C. Curlander and C.Y. Chang, "Tech- Parameter Estimation for Spaceborne Synthetic Aperture
niques in Processing Multi-Frequency Multi-Pol. Spaceborne Radars, " IEEE Trans. on Geoscience and Remote Sensing,
SAR Data, ' European Transactions on Telecommunica- Vol. GE-23, No. 1, January 1985, pp. 47-56.
tions, Vol. 2, No. 6, pp. 605-617, November 1991. [Runge, 90 ] H. Runge and R. Bamler, "X-SAR Precision
[Dubois, 89] P.C. Dubois, et al, "Data Volume Reduction for Processing, " Proceedings of the International Geoscience
Imaging Radar Polarimetry, " IEEE International Sympo- and Remote Sensing Symposium, College Park, Maryland,
sium on Antennas and Propagation, 1989, Vol III, pp.1354- May 1990.
1357. (Sack, 85] M. Sack, M.R. Ito, and I.G. Cumming, "Appli-
[Fe n 8cation of Efficient Linear FM Matched Filtering Algorithms
[Freeman, 89] A. Freeman and J.C. Curlander, "Radiometric to Synthetic Aperture Radar Processing, " lEE Proceedings,
Correction and Calibration of SAR Images, " Photogram- Vol. 132, Pt. F, No. 1, February 1985, pp. 45-57.
metric Engineering and Remote Sensing, Vol. 55, No. 9, W
September 1989, pp. 1293-1301. [Wu, 82] C. Wu, K.Y. Liu, and M. Jin, "Modeling and a
Correlation Algorithm for Spaceborne SAR Signals, " IEEE
[Jin, 84] M. Jin and C. Wu, "A SAR Correlation Algo- Trans. on Aerospace and Electronic Systems, Vol. AES-18,
rithm which Accommodates Large Range Migration," IEEE No. 5, September 1982, pp. 563-575.
I:
isi
1'1
I 1
9-1
INVERSE SYNTRETIC APERTURE RADAR
J.P. Hardange
Thomson-CSF
178, Bd Gabriel P6ri
92242 MALAKOFF Cedex
FRANCE
1. OBJECTIVES AND APPLICATIONS well known that the resolution in range is

inversely proportional to the bandwidth of
ISAR is a technique, based on time and the received waveform. The coherent
Doppler frequency analysis, which is used processing is to short in time to allow any
for imaging of targets having rotational resolution in Doppler frequency (Ref. 2 p.
motions with regard to the radar. In the 3-18).
sixties, observation of the Moon and If several targets are located in the beam
planets by a radar located on the Earth was in the same range cell, they cannot be
one of the first applications (Ref. 1, 2 resolved. Furthermore, there is no
chapter 33). Closely derived from these information at all about their range rate,
first trials, imaging of objects in acceleration, rotational motions, etc...
terrestrian orbit by ISAR techniques is The conventional radar is basically a
performed with interesting results (Ref. mono-dimensional range-only sensor.
3). Specific means have to be added to give to
However, the most current domains of the radar the ability to provide the
application of ISAR are now measurements of position in azimuth and heicht of the
targets signatures and target recognition. target, and all the parameters listed
above.
The objective of the first one is to
measure the complex reflectivity of each To try to get all these informations, the
reflecting point of a target. Although the principle is to realize a filter which is
processing has to compensate for various matched to a long time observation of the
degrading effects of the image quality, it target. In these conditions, the Doppler
is in general the most simple case, as far frequency can be analysed, providing an
as the conditions of the experiment can be additional dimension to the image.
perfectly mastered. This happens if the
target can be put on a turntable for the One question is here to recognize the axis
analysis (Ref. 5). The result of the which corresponds to this Doppler analysis:
analysis is a two dimensional (2-D) image Is it the azimuth? Is it the height? Is it
of the distribution of the reflecting something other? This requires some knowing
points. The two dimensions are range and about the conditions of observation. If the
cross-range in a fixed plan. axis of rotation of the target is known,
like in the case of an object put on a
The second application has been subject to turntable, it is shown in paragraph 4 that
numerous studies and experiments in the the dimension which is measured by the
field of aircraft classification from a
ground based radar (Ref. 4, 7, 8, 13, 14). Doppler analysis is a cross-range axis, in
1-D (cross-range only) and 2-D principles a plane orthogonal to the axis of rotation
have been tested. ISAR has been envisaged of the target.
to equip air defence radars and eventually Generalizing this remark, one can realize a
airborne fire control radars with a
non-cooperative target recognition mode. reception filter which is matched to a
longer and longer analysis. Several terms
of the conventional ambiguity fonction,
In the same domain, the most demanding which are negligible for a short or medium
configuration is obtained when the radar analysis time, become preponderant. The
and the target are moving simultaneously, output of the filter is not only a function
and when their relative position and of time and Doppler frequency, but also of
velocity can be described only with the derivates of the Doppler frequency: Doppler
help of an analysis of the returned radar rate, Doppler acceleration, etc... It is a
signal itself. This is the case of ship generalized ambiguity function (Ref. 2 pp.
imaging with an airborne radar (Ref. 11, 3-14 and 3-15).
16). In this case, the particular behaviour
of ships at sea allows the production of The output of this long time matched filter
3-D images: range, azimuth and height, is a function depending of 2, 3 or more
parameters. Through an a priori knowing of
the behaviour of the target, it can be
2. iON To GT %:2 InAGi ? possible, in some particular cases, to
establish a relation between each parameter
Let us consider a Lonventional radar, in ,f the ambiguity function and an axis of
which the receiver is matched to a single analysis of the target: Range, azimuth,
pulse return, that is to say to a very height. As an example, a relation between
short observation time. In this case, it is the parameters of the generalized ambiguity
= ,
... ~ ...~~~....
. ....... ..
....
.... . . . .. ... ...... . ....
..
.......
2..
9-2
function and the two cross-range axis of a Equation (6) becomes:

target is established in paragraph 6 in the
case of ship classification. 4 n R
#,(t) - 2 n fc t ----
- + #0 (8)
X
2.1. Doppler Frequency
R is in fact a function of time. R(t) can
The expression of the Doppler frequency is be developped, using its successive
well-known, but is established again in derivatives (up to the third order, in the
this paragraph, as an introduction to above example), around its value at an
further developments. These developments initial reference time (t-0):
leed to the estimation of the resolution
that can be obtained for the different t2 t3
parameters of the generalized ambiguity R(t) - R0 - u0 t - Y0 -- - Y0 -- (9)
function. The computation of the resolution 2 6
in range, using a Taylor development of the
correlation function of the received dR I initial velocity of the
signal, is not performed. But the u0 ....-- target, relative to the
computation of the resolution in Doppler is dt It-0 radar (positive when
done, and then the computation of the closing) (10)
resolution in Doppler rate, by the same
method. d 2R I initial acceleration of the
oI 2 target, relative to the
Let us consider a radar, transmitting a dt It-0 radar (11)
constant frequency continuous wave. The
.hase of the transmitted signal is linked d3R I derivative of the
to the time and the carrier frequency by yO- ---3 I acceleration of the target
the expression: dt It-0 relative to the radar (12)
#.(t) - 2 n f t + o () Substituting (9) in (8), we obtain:
to: Initial phase 4 R R
The signal is reflected by a target, -2 a f t-+--0-- 0(t)
located at a range R. The signal which is
transmitted at instant t, is received at: x
2 R + -u t
t, - t + --- (2)
c 2
4 n t
Approximation: We consider that the + --- Yo --
velocity of the target is far lower that X 2
the velocity of light. As a consequence, we
admit that the range of the target doesn't 4 n . t3
vary between the time of transmission of + --- Y0 -- (13)
the signal and the time of reception. X 6
At this is
signal time, the which
the one phase of
had the
beenreceived The three last terms of equation (13) are
transmitted at instant t: characteristic of the motion of the target.
We can define a "Doppler phase", which is
,(t) (the phase that remains after demodulation
of the received signal by the transmitted
Taking the instant of reception as signal:
reference time, equation (3) becomes: 4 4 i t 4 n .
2 R #(t) uo t + ---- O -- + --- 0 --

.
-,(t) #.(t - . .) (4) X X 2 X 6
c (14)
Using equation (1), we get: The Doppler frequency is deduced, being a
a 1derivative of the Doppler phase:
#r(t) 2 n fc (t ---2 R) + #0 (5) 1 do,
c f 0 (t) - (15)
2 n dt
4 aR fc
,(t)- 2 n f t - --------- # c2 (6) u, 2.... yo YO t2
c fD(t) - - u. t + -_ (16)
Introducing the wavelength: X X

c We have obtained an expression of the
(7) Doppler frequency and phase, depending of
fc the parameters which are describing the
motions of the target. It can be usefull to
settle the expression of the Doppler-phase
as a function of the Doppler frequency and
its derivatives:
9-3
fD(t) ff + D t f t (17) (31)

2 '
2 u,- 2u0 SIPf~ 2

'' ... "tt 2 ..t3 t
exp 32n[(f+f.x.. -- )t (f+ f.c)--+ f --jdt

fa . Average Doppler frequency 2 2 6
(18)
2 yo 2.4. Ambiguity function

fD - ---- Doppler slope (19)
2
IX(t,f,f,f,f.,f.)1 is the generalized
2 y. ambiguity function of the waveform.
- Doppler acceleration (20)
3- RESOLUTION
3
0D(t) - 2 n f + n. D 2
+ f4" --t
n • (21) 3.1. Resolution in Doppler frequency
3 To evaluate the resolution in Doppler

frequency, we suppose that the derivatives
of the Doppler frequency are null. That is
2.2. Received siqnal to say, we use the range-Doppler ambiguity
function.
which is
We consider now that the signal
transmitted, s(t), is no more a continuous The ambiguity function is a correlation
wave at constant frequency, but a narrow function. Its maximum is at the origin.
bandwidth waveform on a carrier frequency.
Z'osimplify the notations, the amplitude is To estimate the width of the correlation
normalized to 1. Expression of the received peak on the frequency axis, we perform a
signal: taylor development around the position of
the peak (0,0):
s(t) - s(t) exp [ j #D(t) ] (22) dx(0,f) I
2
t2 .. t3 X(Of)P1 - lx(0,0)1 + f ----------- I
sr(t) - s(t) exp [j2z(f~t + fD -- + fD -- )] df f.0
2 6
(23) V2 d21x(0,f)11 I
+- --- ----------- 1 (32)
2
2.3. Matched Filter 21 df If.0
The filter, which is matched to a signal of The first derivative of the ambiguity
Doppler parameters fa, fa, fa, has the function at (0,0) is equal to zero, because
following impulse response: at this point, the function is reaching its
t
2
.. t
3 maximum:
h(t)-s*(-t) exp [-j2n(f. t + f. -- + f. -- I dIX(0,f)1 2 I
2 6
(24) df I
if-o
- 0 (33)
Notations:
The second derivative can be computed, as a
f "f, - f. (25) function of the value of the ambiguity
function at (0,0):
- fo - f. (26) d2Ix(O,f)j1 I_
-2(2n) 2
- - - - - - -- ----- f t Is(t)['dt[x(0,0)j
df' Ifo (34)
*- (27) f0 is the half width of the peak on
the
Output of the matched filter: Doppler axis, at 3 dB:
2
Ix(0,f0 )1 1
X( ,f,f,f f. (T-t)s,(t)dt (28) ---O -
2
2(35)
IX(0,0)1 2
X( ,f,f,f ,f.,f.) - s(t) s*(t-r) eJ# dt The width of the peak is then:
* I I..(29)
with: 2 f .....-- Is(t)- dt)/2 -- (36)
t 2 t
3 21 (ftls(t)2 dt)1/2 T.
- 2 it fo t + fo -- + fD -- (30) T. is the "equivalent duration" of the
2 6 signal. For the signals of finite duration,
which are considered in our application, T.
(t- )2 . (t-r) 3 is very near from the duration of the
- f. (t--) - f signal.
2 6
-_----.-
-
I~Eape
9-4
duin
T.,thereslutonof equation (36) gives:
6' 0,78
an aprue h f
x~ it4ftepaki
t4 Is(tfl 2 dt
2 I(t)1
16
=
2
hn
X(0,0,0)1
dt)/ 2
2
(42)
(3
2
R T. T. itui dt)1/
Example:
3.2. Resolution in Doppler acceleration For an observation during an aperture time
We make now the hypothesis that the Doppler Ta, the resolution of equation (43) gives:
frequency is null, and we use the same 85 5.7
principle to evaluate the resolution in 2 0 - ------ (4 i
f~ IT, T2
The Taylor development of the ambiguity

function around (0,0,0) is done hereafter: 4- BASIC ISAR
IX(f)P
,O - (T ~ 2 jift 2
4.1. Description
(38) The radar is fixed (point A of figure 1).
The target is rotating at a constant rate
w. The center of rotation is C. The range
dIX(0,0,f)121 between the radar and the center of
-Ix(0,0,0)P1+f ------------V. rotation is R.
dt jf-0 We consider one reflecting point of the
target (point M of figure 1), which is
located at coordinates (x,y,0), at initial
f2 d2 jX(0,0,f)jIj time (t=0). x and y are small compared to
-+ ---------- 1. (39) R.
IX(0,0,0)1 2 - --
2
f
if dt2
"t 4 IS(t)
If-0
2
dt (40)
The coordinates of M at each time are
(u,v,O): o t- in t(5
f. is the half width of the peak on the v - x sin w t + y cos w t (46)

Dopper axs,
eriatie
t 3dB:The range of the reflecting point is:
Ix(0,0,fO01 2 I RM2 - (R + U) 2 + V2 (47)

----- ----- (41)
2
IX(0,0,011 2 R'- (R + x cos w t - y sin &)t)
+ (x sin ut +ycos Wt)2 (48)
y,
V
T U
M (X,
, 0)
1< AC X
Figure 1: basic ISAR geompetr

9-5
R,2 - R02 + 2 R x cos ot is a function of only one parameter +a-

- 2 R y sin ot (49)
whith a0 - 1 at t - 04.3. Projection plan
R x cos w t It comes from previous developments that

R- ( - -1-+ the resolution of ISAR processing is
R02 obtained on a cross-range axis, which is
orthogonal to the axis of rotation of the
R y sin w t target.
- RC-- -
--- (50
(50 ) The image is a projection in a plan
R°
0 containing the radar-to-target axis (range
The Doppler frequency of N is then: axis) and the cross-range axis which is
orthogonal to the axis of rotation.
2
fn- - w (y cos o)t + x sin w t) (51) Figure 2 gives several examples of basic
projections.
For a short observation time T. around t-0, - In figure 2a, the target is oriented to
the Doppler frequency is depending only on the radar and has a pitch motion. The
the y coordinate: range resolution is on the lenght axis of
the target, and the cross-range
2 yu resolution is the height of the target.
- (52) The projection plan is a range-height
X vertical plan.
We have established a direct relation In figure 2b, the target is oriented
between the Doppler frequency and the perpendicular to the radar range axis and
position of the reflecting point on a has a roll motion. The range resolution
cross-range axis, relative to the center of is on the width axis of the target, and
rotation. the cross-range resolution is still the
height. The projection plan is a
We can note that the reception filter has width-height vertical plan.
to be matched to signals of constant
frequency and finite duration. As a In figure 2c, the target is oriented to
consequence, the ISAR processing is reduced the radar and has a yaw motion. The range
to a simple Fourier transform, resolution is on the length axis of the
target, and the cross-range resolution is
the width. This case is comparable to
4.2. Resolution SAR. The projection plane is a
length-width horizontal plan.
As a result of equation (36), the
resolution in Doppler frequency, for an
observation time T., is:
In several applications, the configuration
k is not so simple. The target principal axis
rfD - -- (53) are any possible orientation compared to
T. the range axis. There can be simultaneouly
pitch, roll and yaw motions. The
where the value of k is near to 1. orientation of the instantaneous axis of
This result is well-known for every rotation is not known. Consequently, the
conventional spectral analysis. It could be position of the projection plan is not
known either. I
resolution, by
possible to achieve a better
the mean of estimation methods.
Unfortunatly, these methods require a high
signal-to-noise ratio and a large amount of 5- RELATIONS WITH SAR, REAL APERTURE, AND
computation. They are not currently used in TOMOGRAPHY
this field of application.
5.1. Relation with sAR
The resolution on the cross-range axis is
then: Figure 3 shows a typical SAR configuration.
The radar is flying at constant speed, u,
X rfo X and height on a straight line. It is
ry - - (54) observing a first reflecting point designed
2 w 2 w Ta by N The azimuth of the target is #1. The
Doppler frequency is:
r . (55) 2u
2 ** fD .... cos # (57)
with
There is a second reflecting point, 32.
- c Ta (56) which is at the same range, but with a
*, is the angle of rotation of the target slightly different azimuth:
during the observation time (figure l).It 2 * + 64 (58)
is interesting to note that the resolution
.9-6
a Height
Pitch C~p
Range= LMht
R~oll N D =Height J
Flange Wdh
=Wkdth
Dopper
Yaw AxQ
Range= L-ht
Figure 2: Projection plan of the image
The Doppler frequency of the second

reflecting point is:
2u
f,2 Cos (+J + 64) (59) __
x U
At the first order:
2u
f'1 - fD2 . ...-sin *, 64 (60)
For a short observation time T., , can be

considered as a constant. In this case, the
signals received from the targets are pure
constant frequencies, and the matched
processing is done by Fourier transform.
The resolution in Doppler frequency is
l/T.. The associated angular resolution in
azimuth is derived:
2u 1
--- sin #, r# - -- (61)
.T.
r# - (62)
2 u T. sin +,
The projection of this angular resolution
at the range of the target is:
S-- (63)
r 2 u T3 sin #1 (
During the observation time, the

displacement of the radar is the length of Figure 3: SAR qeoet
the synthetic aperture: _______ 3 ___________
L - u T. (64)
9-7
The variation of attitude angle of the X

target relative to the radar is: 1 - -- (69)
L u Ta
(65) 1 - 2 L (70)
R R
The reouinr ca
The resolution
function of :r can be written as a 5.3. Relations with tomography
It can be shown that some kind of ISAR can

Xbe interpreted as a tomographic
ry (66) reconstruction problem (Ref. 10).
2 ** sin *1
Computer-aided tomography
(CAT) is a
technique for providing a two- or
The best resolution is achieved for 4* - three-dimensional image of an object
900 (side-looking radar). In this case: through digital processing of many 1-D
projectional views taken from different
look angles. It is used is the medical
r (67) field for imaging with X-ray scanners.
2 4
The projection-slice theorem, that is used
This equation is identical to equation 55. in this technique, can be applied to ISAR
processing.
SAR and ISAR are identical. In both cases,
the parameter, which fixes the resolution,
is the variation of attitude angle of the
target, relative to the radar. To change 5.3.1. Tomography principle,
from SAR to ISAR is only a change of projection-slice theorem and backprojection
reference: reconstruction method
- Rotation angle refered to the radar: SAR Let a(x,y) be the function to map.
- Rotation angle refered to the target: The measurement are made by the mean
ISAR of several 1-D projections on
different directions. The direction
We have established this result, of the line-of-sight for one
considering a short observation time. We projection is #. The value of the
have
order limited the computations to first
developments: projection at each range u is:
p*tu) - alx(u,v,#),y(u,v,+)] dv
- In SAR technics, it is the hypothesis of . (71)
the unfocused SAR. The processing is the
same (Fourier transform), whatever the with notations of figure 1:
range of the target is: it is focused at
infinite range. x(u,v,0) - u cos + - v sin + (72)
- In ISAR technics, the equivalent name y(u,v,#) - u sin 4 + v cos 4 (73)
"unfocused ISAR" is not usual. But the
principle is the same: second and higher The projection function has a
order terms in the variation of the range Fourier transform:
* are neglected.
For a longer apeture time, there is a P#(U) - p(u) e du (74)
difference between SAR and ISAR:
The Fourier transform of the
- In a pure ISAR configuration, the range projection function is one slice of
between the radar and the center of the 2-D Fourier transform of the
rotation of the target is constant. image a(x,y) (projection-slice
theorem). The orientation of the
- In a SAR configuration, this range is slice is given by the angle *:
varying following a quadratic law.
However, the effect of this range P#(U) - Z(U cos #,U sin #) (75)
ij variation can be compensated through an
appropriate demodulation of the received Z(X,Y) is linked to a(x,y) by the
signal. relation:
E(MY) a (x,y) • dx dy
5.2. Relation with the real aperture (6f- y (76)
#(x,y) can be reconstructed by the

For SAR and
synthetic aperture is: length of the
ISAR, the use of a 2-D inverse Fourier
transform:
L-4, R (68) / J21t(xX + yY)
The real aperture providing the sam e(x,y) - XY) • dX dY

resolution is: . (77)
9-8
Generaly, the reconstruction is not 2f

2 f
made by the mean of an inverse
Iourier transform, ''cause the value s(f,#) P+(U), with U-----
ok(X,Y), in recta qular c (83)
coordinates is not known. It has to The back-projection algorithm can be
be computed, using interpolations, used, following equations (80) and
Another method is used, based on (81):
filtering and backprojection in
polar coordinates. Multiplication by lUI
The measurements are obtained in . Inverse Fourier transform. At this
polar coordinate. Let us write step the signal is compressed on
equation (77) as a function of these the range axis.
polar coordinates:
Spreading of each point (amplitude
X - U cos #, Y - U sin * (78) and phase) of each profile on a
(t(.- line which is orthogonal to the
d(x,y) - IriZ(U cos CU sin *) line of sight of the profile, and
frf. sum of the resulting signal at
j2nU(x cos + + y sin *) each point (x,y).
e UI dU d#
(79)
j2nU(x cos , + y sin ,) 6- EXAMPLE OF APPLICATION: IMAGING OF SHIPS

*(x,y)- P(U)IU~e dU d# AT SEA
) ) (80)
6.1. Modelization
The integral over U is identified as
an inverse Fourier transform of Imaging of ships at sea is a complex
variable (x cos + + y sin #): application of ISAR. It is interesting,
because it is a case where the
O(x,y)- [p+ * k](x cos # + y sin #) d+ discriminating parameters are not only the
ir
where k
(81)
is the inverse Fourier
time and the Doppler frequency, but also
the derivative of the Doppler frequency.
transform of IUI. Figure 4 is a geometrical representation.
The radar is flying on-board an aircraft,
The processing is made of the at velocity u.. Its initial position is 0
following successive steps: and its average course is along the X-axis.
a high-pass filter, H(f) - IUJ, The antenna is oriented toward the ship,
which is producing a constrast defining an x-axis. The angle between X and
enhancement, x is the azimuth of the target, #.
a sum for all the values of # The center of rotation of the target is
located initialy at point 0'. Its velocity
Other reconstruction methods are is u, and its course makes an angle a
existing and are listed in Ref. 12. relative to the x-axis. Two cross-range
axis are defined: y-axis, in the horizontal
plan, orthogonal to x, and z-axis, which is
5.3.2. ISAR as a tomographic reconstruction the height.
problem
We consider a reflecting
point M on the
We consider now a radar which is ship. Its position, relative to 0", is
transmitting several continuous (0, y, z).
waves at constant frequencies. The
different frequencies are regularly The ship is supposed to have sinusoidal
spaced and can be addressed pitch and roll motions. The angular
simultaneously or sequencialy. This velocity in pitch and roll are%
type of observation is repeated for
several values of the angle of view QP - o sin (2 n fp t + +P) (84)
The signal received for a given 2, - Wr sin (2 x f, t + #,) (84)

frequency and a given angle of view, The Doppler frequency of the signal
after demodulation, is proportional reflected by point M, and its derivative,
to: are:
s(f,) p(u) j2n2f/cu du f, -
J- (82)
(ussin # -u, sin a +Qtzsmin a +Qrxcos ct)'
where 2 L is the dimension of the _-- i ---
target along the line of sight, and - R - - -
p#(u) is defined by equation (71).
s(f,#) is the Fourier transform of ussin * - u.sin a
the projection p#(u): y (86.2)
ft I
9-9
0 0'YZ
R
Figure 4: Ship imaging by an airborne radar
+ (Q,(t)sin a + Qt(t)cos a) z (86.3) 86.5: The spurious Doppler frequency due to

aircraft motion is supposed to be
+ ucos + u,cos a
+ (86.4) sufficiently small to have no
coupling effect with other terms. It
+ fD,(t) } (86.5) is evaluated by inertial means, to
allow a compensation of Doppler
The terms in Qt and Qr of equation (86.1) frequency before any further
are negligible, processing.
fD. is the effect of spurious aircraft Equation (86) could permit to resolve the
motion. ambiguity between y and z, if the aperture
time is sufficiently large. But this also
2 (u sin 4 - u sin ) can be done, considering equation (87), in
- - -------------------- (87.1) which only z is mentioned:
X R
87.1: It is a constant Doppler acceleration
term, which is identical for every
+ (Q,(t)sin a + 2t(t)cos a) y ) (87.2) reflecting points. It is eliminated
by the same way than term (86.1).
+ f'.(t) (87.3) 87.2: This term is proportional to z. It
allows the measurement of z.
Let us give an interpretation of each term
of equations (86) and (87) and explain 87.3: Effect of aircraft spurious motion,
their interest: which is corrected by the same way
than term (86.5).
does not
86.1: It is a linear term, which
depend on y and z. It is identical To simplify the problem, we make an
for each reflecting point. It brings additional hypothesis: The ship has only a
no discriminating information. The roll motion, and #, - 0. After correction
parameters u., u., # and a can be of non discriminating terms, equations (86)
estimated, by the use of tracking or and (87) become:
other method (Ref. 16). This term is
eliminated by Doppler compensation. 2 u sin + - u-sin
fe - n-n - Doppler c.- Y
86.2: This term is proportional to y. It R
establishes the link between the
Doppler frequency and the position of + a, sin(2mfrt) sin a z (88)
the reflecting point in azimuth. It
is depending only on the displacement t 2
of the aircraft. It is in fact a SAR f - - u cos(2af,t) sin a z (89)
effect. X
86.3: This term is proportional to s. It We come here to a fundamental property of

establishes the link between the ISAR applied to ship imaging (figure 5):
Doppler frequency and the position of
the reflecting point in height. It is - The Doppler frequency is a sinusoidal
to of
function. the
ship. It isonantheISAR
depending rotation of the
effect. function isTheproportional
average value the azimuth!
of the reflecting point. The amplitude of
86.4: This term is proportional to the variation is proportional to the height
velocity of the ship relative to the of the reflecting point. The use of a
aircraft. It brings no discriminating time-frequency transform is very
information. It is eliminated by the interesting to visualise the evolution of
same way than (86.1). the Doppler frequency of each scatterer
during the observation time. It can be
j_
tI
9-10
fo A
azimuth
shift
t
1/ fr
Roll motion period
Figre 5: Eolution an interpretation of

U e Dopler frequen
either a wavelet transform (Ref. 17.) or Example:

a Wigner-Ville transform (Ref. 12). X - 3 cm (X-band)
Doppler frequency
- The derivative of the
is a sinusoidal function, centered on u. - 150 m/s
zero. Its amplitude is proportional to
the height of the reflecting point. R - 100 km
- 900
W " 3*/s
6.2. Characteristics of the image
If we realize the filter described by T, - 1,5 s
equation (31), the resolutions in azimuth
and height can be estimated using equations - 450
(37) and (44), (88) and (89). Resolution in azimuth: r. - 5,2 m
The best result will be obtained around
t-0. At this time, the Doppler frequency in ResolutiOn in height: r5 - 1 m
is null).(88)Considering
equation to .y (velocity
is due onlyU>>U in
(the term of Taking into account all the approximtions
l othat have been done to come to this result,
the aircraft greater than the velocity of these values are to be considered only as
the ship) the resolution in azimuth is: orders of magnitude.
r r- -------R The resolution in azimuth is not so good

than the resolution in height, because it
sin
2u, is proportional to the range. At far range,
it is useless to try to have an exploitable
- (91) top-view of a ship.
2n T. u, sin *
7- SPURIOUS EFFECTS
Considering the height, starting from
equation (89), at first order we can write: The spurious effects are identical to
spurious effect in SAR, but interpretation
2 9can be different. The most important are
fD " - W1 sin a z (92) listed hereafter.
2 7.1. Range-cell Nigratio

w
7.1.1. Description
(94) Due to its rotation relative to the

---- (94) radar, the aspect of the target is
R T.2 changing during the aperture time.
44T'xThe range of each reflector is
r. - (95) changing, providing the Doppler
IT(o, sin a effect. If the change of range
becomes greater than the range
9-11
resolution (or the ran3e-cell), the If s(t) is supposed to be a signal

reflector does not appear any more of constant amplitude and frequency
as a point scatterer on the image. during the aperture time, then
There is a degradation of the range equation (97) can be considered as
resolution. the Fourier transform of:
Furthermore, the processing is most j*(t)

of the time organized in two e (98)
successive steps: It is usual, in the
study of
• First step: range processing, spurious aircraft motion for SAR
performing a matched filtering on processing, to distinguish low
a short time. frequency motion and high frequency
motion.
. Second step: Cross-range
processing, performing a further
matching on a long time. This 7.2.2. Low frequency motions
processing is usualy made on
signals at constant range of the Two main situation can happen:
radar, that is to say,
independantly for each range cell. . Error on the rotation rate
It is evident, from equation (52),
If there is a migration of the echo, and without making any computation,
greater than the range resolution that this error will modify the
cell, the processing in cross-range scale factor on the cross-range
is mismatched. The aperture time is axis.
reduced to the time of presence of
the echo in one range cell. The If the value of the rotation rate is
resolution in cross-range is not known, the scale factor is
degraded. anyhow unknown to.
Slow variation of the rotation
7.1.2. Compensation methods rate. This will cause a change of
the value of the Doppler frequency
To compensate for this effect, the during the aperture time. We can
principle is to perform the make the hypothesis of a linear
cross-range processing at variable variation. It is equivalent to
ranges. There are two possibilities: introduce a quadratic phase shift.
to shift, in time, the received The result is a degradation of the
signal, cross-range resolution.
to shift, in time, the impulse o(t) - w0 + kt (99)
response of the cross-range
matched filter.
i kw - -(100)
The main problem is that the k(
range-cell migration is not the same
for every scatterer. It is The corresponding variation of
proportional to the distance between Doppler frequency during the
4 the scattering point and the center aperture time is:
of rotation, that is to say, to its
Doppler frequency. 2 6u y
6f . ......- (101) V
7.2. Unknown radar and target motions
A good approximation is to consider
7.2.1. Modelization that the linear variation of the
frequency, 6f, provides a widening
The modelization is made for the of the peak, at the output of the
basic case of ISAR: Short aperture Fourier transform, according to the
time, developments limited to first expression:
~order.
rf, - rf,0' + Wf )P/2 (102)
i The signal received from a point
reflector is: 4 6ua y'
J+(t J2nfvt rtD " ( rf 0 2 +--------- ) (103)

A
sr(t) - Sat) e e (96)
2
2 Sol y2
where 4(t) is the phase generated by r, - ( r7 0 + 2)/2 (104)
unknown spurious motion. 4%
The output of the matched filtering

is deduced from equation (31): 7.2.3. High frequency motions
+ j+(t) j2nft It is the domain of periodic motion,

X(T,f) " 3s(t) &*(t-) 0 e dt having several periods during the
f (97) aperture time: Vibrations.
9-12
r -2 A reflector can have a vibration

action on the target. For example,
on a ship, objects located on the
mast are vibrating with frequencies
7.2.4. Random motions
Random motions is a generalization
of high frequency motions. Instead
depending on the natural frequencies of one vibration frequency, there is
of the mast and the engines, a complete spectrum, creating a
continuous level of side-lobes.
The spurious motion on the range x(t) is the spurious displacement,
axis is:
X2 is the variance of x.
x(t) - x0 sin (2 x f. t) (105)
The spurious phase is:
The spurious phase is:
4 n 4(t) . . .X(t) (111)

4(t) .... x, sin (2 n f. t) (106) X
The variance of the phase of the
Writing (106) in equation (97), it spurious signal is:
comes: 4in
4nX o (..___.--)2
X2 (112)
X(0,f)" -xp(j--- sin(2nfst)) e dt X
f- X0 (107)
The output of the processing filter
2nX is:
X(O,f)- 6(f) + ---- [6(f-f.) + 6(f+f.)]
X (108) X(O,f) -. ej[2nft + *(t)Jdt
dt(113)
The usefull echo is bordered with u
two side-lobes, located at Doppler j[2nfu + #(u)]
frequencies f. and -f. (figure 6). Ix(0,f)1 2 - du
The position of the false echos is: -[2fv + *(v)]
X f. (114)
y - / 2 ..
w (109)
(109) /.
e *(u)-#(v)] -j2xf(v-u)
2IX(0f) - e du dv
The level of these echos ist (115)
2 x xVariable change:
Ix(o,f)l X...... (11o)
ix0f)I(1)w - v-u (116)
(with Ix(O,0)l - 1, level of the ie[(u)-+(w+u)] -j2nfw

peak) IX(0'f)1 2 e due dw
(117)
4
IX(o,0 1
IX (O,0) I
IX (0,0) I =
IX (0, fm) I 2 xo
- fM fm Doppler Frequency
y=' )AM Xfm Cross - Range axis

2 jo 2o
rigure 6: Side-lobes due to vibrations of

theftrget or the radar
Y 9-13
f e ti(u-w+ufdu -
fr I+j[#(u1-#(w+ul]
WAEFR
GENERATION
UP-CONVERTER
2
I [+(u-#(w+ul
- du
2 (118)LO
T. - T. #2 + R++(w)
(119) RECEIVER
2 2
Ix(0,f)1 - T6(f) -T.# 6(f) + S (f)
1to) [JDOWN CNERTER
with: " e 7: Transission/Reception

-j2nfw bloc-diagran
S+(f) - )R#(w) e dw (121)
The average power of signal at the
ouput of the processing is: 7.3.2. Effect of local oscillator phase
noise
2 2
E(IX(0,f)1 ) - T.6(f) -T.# 6(f) + E(S+(f)]
(121) The same calculation than in
paragraph 7.2.4. provides the
E(IX(0,f)1 2 ) - T. (1-#2) 6(f) + -- average peak-to-sidelobe
&5 ratio:
BD (122)| 4 sin 2 (KfTt ) £(f) df
S/L - 2 -1h ----- -- ----- --
The peak-to-sidelobe ratio is: Bn" T (127)
4x 1
S/L X 2 ------
P()2 (123) 8- REFERENCES (unclassified publications
BA, T. onlyl
1. Brown W.M., Fredricks R.J.,

7.3. Spectral purity of the radar "Range-Doppler Imaging with Motion
through Resolution Cells", IEEE Trans.
7.3.1. Modelization Aerospace and Electronics Systems,
VOL. AES-5, N01, Jan. 1969, pp.
The effect of local oscillator 98-102.
(L.O.) instability is identical to
the effect of spurious radar and 2. Skolnik M., "Radar Handbook", Mc
target motions: Graw-Hill Book Company, 1970.
Low rate frequency drifts: 3. Gniss H., Krucker K., Magura X.,
Degradation of the resolution Perkuhn D., "Problems of Signal
Processing in a high Resolution Radar
Spurious lines: Side-lobes, - Synthetic Aperture Imaging of
isolated rotating Targets with narrowband and
broadband Signals", SEE 1978
Phase noise: Average level of International Conference on Radar,
side-lobes Paris, Dec. 1978, pp. 243-250.
The local oscillator frequency is 4. Chen C.C., Andrews H.C., "Target
Motion Induced Radar Imaging", IEEE
used for up-conversion at Trans. Aerospace and Electronics
transmission and down-conversion at Systems, VOL. AES-16, N"l, Jan. 1980,
reception (figure 7). pp. 1-14.
This operation is equivalent to a 5. Chen C.C., Andrews H.C.,
single delay filtering of the local "Multifrequency Imaging of Radar
oscillator signal. It is performing Turntable Data", IEEE Trans. Aerospace
a weighting of the phase noise and Electronics Systems, VOL. AES-16,
spectrum: N01, Jan. 1980, pp. 15-22.
E,(f) - 4 sin 2 (n f Tt) £(f) (125) 6. Walker J.L., "Range Doppler imaging of
Rotating objects", IEEE Trans.
- 2 /c is the time corresponding Aerospace and Electronics Systems,
to the range of the target. VOL. AES-16, Nol, Jan. 1980, pp.
23-53.
The effect of this weighting is an
enhancement of the high frequency
noise power, by a factor of 2, and 7. Dike G., Wallenberg R., "Inverse SAR
an attenuation of the low frequency and its Application to Aircraft
noise power, by a factor: Classification", IEEE 1980
International Radar Conference Record,
e(f) - 4 V f2 T,' (126) Arlington, April 1980, pp. 161-167.
II
9-14
8. Prickett M.J., Chen C.C., "Principles

of Inverse Synthetic Aperture Radar
(ISAR) Imaging", IEEE EASCON'80
Record, Sep. 1980, pp. 340-345.
9. Mensa D.L., Dean L., "High Resolution
Target Imaging", Artech House, 1981.
10. Munson D.C.Jr., O'Brien J.D., Jenkins

W.K., "A Tomographic Formulation of
Spotlight-Mode Synthetic Aperture
Radar", Proceedings of the IEEE, Vol.
71, n"8, August 1983, pp. 917-925.
11. Wehner D.R., "High Resolution Radar",
Artech House, 1987, pp. 273-339.
12. D'Addio E., Farina A., Morabito C.,

"The applications of multidimensional
processing to radar systems",
International Conference on Radar
1989, Paris, April 1989, pp. 62-78.
13. Bethke K.H., Rode B., "A fast

ISAR-imaging Process and its inherent
degrading Effect on Image Quality",
AGARD CP-459, The Hague, 8-12 May
1989, pp. 31-1 to 31-12.
14. Ender J., "lD-ISAR imaging of

manoeuvering Aircraft",
AGARD CP-459, The Hague, 8-12 May
1989, pp. 33-1 to 33-9.
15. Marini S., Pardini S., Prodi F.,
"Radar Target Image by ISAR Case
Study",AGARD CP-459, The Hague, 8-12
May 1989, pp. 35-1 to 35-14.
16. Fenou M., "L'Imagerie de Cibles
marines S la Fronti~re entre le SAR et
l'ISAR", AGARD CP-459, The Hague, 8-12
May 1989, pp. 28-1 to 28-10.
17. Daubechies I., "The Wavelet Transform,

Time-Frequency Localization and Signal
Analysis", IEEE Transactions on
Information Theory, Vol. 36, n05,
Septembre 1990.
ji
I
10-1
Special SAR Techniques and Applications

R. Keith Raney
Canada Centre for Remote Sensing
tion leads to the possibility of terrain

1. SUMMARY height estimation, with a precision on the
order of the system resolution. Differential
SAR systems as considered in these lectures interferometric techniques allow obser-
are fully coherent, and are characterized by vation of changes in the scene on the order
large time/bandwidth signal structure in of the radar wavelength. For time separa-
both range and azimuth. These properties tion of the signal pair, the full potential of
allow additional and specialized perfor- SAR MTI may be realized. Implementa-
mance to be achieved through innovative +ion in the time domain and in the
system variations, frequency domain is described. Airborne
versus spaceborne constraints are com-
Using one signal sequence, resolution may pared. The principle of conservation of
be improved in azimuth through increased energy of moving targets is presented, and
target Doppler bandwidth. Broad beam used to estimate target velocity.
and Spotlight SAR are techniques used.
For the Spotlight mode, high resolution in
azimuth is achieved at the expense of image 2. INTRODUCTION
size, so that the Stretch range bandwidth
reduction technique is useful to allow The special techniques considered in this
increased resolution in both dimensions. A lecture depend upon the coherent charac-
subset of these methods is the Squint mode, teristics of SAR signals, and often of
whereby the side-looking antenna is pointed greater importance, on coherence properties
forward (or aft) of the zero-Doppler direc- encountered in the complex image domain.
tion, leading to increased complexity for This section introduces the main concepts.
image processing. Wave domain or polar
format techniques are required. 2.1. Coherent Signal Structure
The requirements, capabilities, and limi- Signals encountered in SAR analysis have
tations of single beam moving target phase as their most important attribute, and
indicztion (MTI) for a SAR are reviewed, thus must be coherent over each sequence
Special processing for multi-look SARs to of received samples. Coherence in the
enhance SAR ocean wave imagery contrast azimuth dimension may be exploited to
and directional spectral estimation is achieve remarkably good resolution, and
highlighted. may be used for special techniques as well.
The classic treatment of SAR analysis is
Since the SAR signal is coherent, signal that of Harger [1], who makes the usual
phase comparison is possible between pairs assumption for most of his treatment that
of signals. For spatial separation of the the two SAR dimensions, range and azi-
signal pair, interferometric signal combina- muth, are uncoupled, and that the antenna
©CanadianCrown Copyright
L - -
10-2
beam is so narrow that range curvature is of the signal is given by

not an important consideration.
" Y) C 2k 'y '
For most advanced techniques, detailed
treatment of two-dimensional coupling must
be included in the model. One such analy- exp{-jb'[y-R,(wx)] 2 )
sis may be found in [2], which is based on
the generic signal received from a point 2 12
scatterer as observed in a wide beam geo- .exp j2kR1
metry. In studying the azimuth resolution e( 2
capability, looking at the Doppler proper- (4)
ties, or analyzing methods of preserving
phase in the derived (complex) image
space, an expression for the range signal,
azimuth image domain is essential. Such an where C is a complex "constant",
expression opens our review of special R
topics. R.(to ) "
( 2 '1/2 (5)
Consider a large time-bandwidth signal (1--
ss(xy) from a scatterer at (x=O, y=R,) 4k 2
observed in signal space (x azimuth, and y
range) having wavenumber k=21r/A, and and
two-dimensional pulse envelope b
a[xy - R,,(x)] according to b 1 (
ss(x,y) - a[x, y-R.(x)] 4k3(1 - WXI4k2) 3 /2

exp{ -j2kRn(x) } exp{ -jb[y - Rn(x)] 2 }
The key result is the generalized Doppler
domain expression of Eq. 4, derived in [2].
where Range curvature refers primarily to the

range/azimuth coupling found in the range
R2(x) - + x2 (2) delay term of the envelope of Eq. l and the
locus of Eq. 5 in the Doppler domain of
Eq. 4. This coupling is itself range depen-
Then the azimuth Fourier transform dent. (This little subtlety is at the heart of
processor design for this class of imaging
systems.) Doppler properties are set by the
Ss(caxY) (3) width and angular position of the envelope,
- fss(x,y) exp{-jw,tx} dx both of which are range dependent.
Whether or not significant range curvature

is present, the signal expression may be
transformed and processed to derive an
image. In its most fundamental form the
I
10-3
image is complex, having a phase structure platform, these are sufficient to establish
that may be exploited for certain applica- the Doppler bandwidth of the signal. The
tions. Ideally, the complex image is simply class of sensors includes SAR, certain
a linear transformation of the input signal, SONAR devices, Spotlight SAR, tomogra-
and has no phase errors introduced by oper- phy, inverse SAR as used for imaging Earth
ations in the processor. (Note that very few satellite and nearby planets, etc. The
SAR processors in use today satisfy this azimuth resolution is given by the inverse
objective.) Selected advantages of the (spatial) bandwidth, which leads to
complex image domain for calibration _.
purposes are explored in [3]. Consider- Paz- (7)

ations essential to arriving at a complex 2 (sin 0, +sinP2)
image with robust phase is considered
further in section 2.3 below, where the angles of observation are defined
in Figure 1.
2.2. Comments on Resolution
Senr path
A nice discussion of resolution restraints for
a SAR may be found in [1]. For the pre-
sent purposes, we highlight the central
result, presented in a form rather more
general than that usually found in the
literature. Resolution is generally (but
loosely) defined as the width of the "point
spread function", the "Green's function", or
Object
the "impulse response function", depending
on whether one has an optics, a physics, or
an electronic systems background. More
properly, "resolution" refers to the ability of Figure 1. Geometry of viewing angles,
a system to differentiate two image features general case.
corresponding to two closely spaced small
objects in the illuminated scene when the
brightnesses of the two objects in question
are comparable (according to Lord Ray- Extremum: The minimum (limiting best or
leigh [1879]). finest) resolution is obtained for an effec-
tive field of view of 1800. The limiting
A SAR is distinguished due to its ability to value is readily seen to be
achieve "high resolution" (by which is
meant, small response function width, and lim P - (8)
therefore "good") in the azimuth direction, i,2_904
which is parallel to the sensor platform
velocity vector.
which might seem unrealistic. However, it
General case: It follows from Eq. 4 that should be noted that results approaching
azimuth resolution is determined only by this value have been achieved in certain
the wavelength and the span of angles over applications such as active seismics.
which the object is viewed. For a moving
am.-
10-4
SAR Application: For a SAR, the subten- in which there is no dependence on range
ded angles of view are generally very small, or wavelength! This is a fundamental
and, furthermore, are oriented nominally at characteristic of SAR systems, and makes
right angles to the velocity vector of the them well suited to spacecraft platforms.
sensor. In this case the small angle sine NB: The magnitude of expression of Eq. 12
approximation applies, and one obtains is an approximation only, as it violates the
X correct diffraction limited result that follows
PSARZ 2 P,
- (9) from Eq. of
benefits 7,and it does
orbital not account
geometry. It for the
is this
subtlety that justifies revisiting SAR reso-

where the angle expresses the total effective lution in this lecture. Although the best
viewing angle. It should be noted for known of the "SAR facts" to be found in the
satellite radar systems, where one should open literature, Eq. 12 remains an unaccep-
account for the sphericity of the viewing table approximation for many space based
geometry [4], that SAR considerations.
S-,2.3. Comments on Processing

Peff - S/C p(10)
Vm Means of handling the coupling in the con-
text of range curvature correction in a SAR
which expresses the fact that the angle over processor are well reviewed by Bamler [5].
which the object is viewed in a satellite The several cases discussed in that refer-
setting is increased over the azimuth (free ence differ primarily by the method and
space) beamwidth of the antenna pattern scope of approximations employed. The
due to the rotation of the spacecraft as it consequence of most limitations used to
progresses along its orbit. (The parameters date is to restrict applicability of any given
Vs/C and V,,m refer to the velocity of the method to narrow beamwidth around zero
spacecraft along its orbit, and the velocity Doppler offset, or to narrow beamwidth
of the antenna beam footprint over the about a given Doppler offset. Implemen-
surface of the Earth, respectively.) Note tation of several traditional techniques is
that in virtually all available literature this covered more extensively in [6].
factor is not included.
For phase sensitive applications, particularly I:
A common approximation, familiar to SAR as encountered in interferometry (see sec-
people, is that tion 4 below), requirements on phase
X. precision in the derived complex image
,- (11) surpasses the ability of most traditional
processing algorithms. In response, Caf-
forio, Prati, and Rocca [7] introduced
where D,, is the (azimuth) aperture size of seismic wave domain techniques to the field
the antenna. Thus, from Eq. 3, we reach of radar. In the context of seismic inversion
the famous representation problems, typically characterized by rather
Dh large angular fields of view, and hence
PSAR, (12) severe range/azimuth signal coupling, Stolt
2 [8] had developed an effective change of
variables in the two-dimensional frequency
10-5
domain. A practical difficulty is that the resolution may be maintained to within

Stolt change of variables requires an inter- about a factor of two over "perfect" side-
polator for its implementation, which leads looking SAR resolution out to squint angles
to increased cost and decreased precision on the order of 450, and certain systems
impacting particularly phase fidelity in the support squint up to 80" and more.
complex image. Improvements have been
demonstrated (e.g., [91), although until In squint mode geometry, since the range
recently interpolators are still required. and azimuth coordinate systems are no
The main difficulty arises from the range longer orthogonal, targets appear to move
dependence of the range curvature para- through many range resolution cells during
meter, a quantity not available in the two- their time of radar illumination. For
dimensional frequency or "wave-number" narrow beam systems, this "range walk" may
domain. usually be approximated by a linear phase
term. Processing adapted to squint mode
For phase precision, it is desirable to be radars is covered in [6] usually implemented
able to cope with large angular field and digitally using interpolation. A more recent
range depth without the need for an inter- wave domain contribution described in [91
polator. There have now been processors has been demonstrated at squint angles in
demonstrated that satisfy this objective excess of 45' .
([10], [11], and [121). In future, the
coherent structure inherent to the complex A second complexity arises in squint mode
image of SAR and related system data sets geometry. Recall that the radar is a
should not be limited by processor fidelity, sampled system in the azimuth dimension.
The remainder of this lecture assumes that According to the Nyquist sampling criterion
complex image data is available with suf- (see, for example, l11), the Doppler
ficient phase accuracy to satisfy processing spectrum of the signal sequence must be
requirements. sampled with sufficiently high radar pulse
repetition frequency (PRF) so that spectral
aliases, or ambiguities, are avoided. The
3. ONE SIGNAL SEQUENCE mean Doppler frequency is set by the squint
angle of the antenna according to the terms
Specialized systems have been demonstrat- of Eq. 7. It is in general not desirable to
ed that depend on coherent properties of a increase the PRF in proportion to squint,
single sequence of signals in order to im- since that would reduce the unambiguous
prove resolution, or to derive information range interval available to the system.
about moving targets.
Fortunately, the bandpass version of the
3.1. Squint Mode Nyquist sampling theorem applies, so that
the PRF must be greater than only the
In many applications, particularly for tact- Doppler bandwidth of the signal sequence.
ical systems, it is desirable for the radar to This leaves open the question of estimating
look forward (or aft) of the zero Doppler the Doppler centroid which may be many
plane by a significant amount. The at- times larger than the PRF, a common
tainable azimuth resolution for such a challenge in most SAR systems. (Doppler
"squint mode" SAR follows from Eq. 7 and centroid estimation is covered in another
Figure 1. With suitable processing, azimuth lecture in this series.)
I
10-6
3.2. Spotlight Mode proportionally more Doppler bandwidth

must be obtained.
If one attempts to sharpen azimuth resolu-
tion in a conventional SAR, either in a Processing for Spotlight SARs tends to be
squint or side looking mode, a smaller specifically designed for the task. A so-
azimuth antenna aperture must be used. called "polar format" is often employed [13].
This is the "ground rule" for any strip This had its roots in the days of optical
mapping mode SAR. Unfortunately, the processing, and creates a data format ana-
effective gain of the antenna is reduced by logous to the polar coordinate system in
the square of its aperture, so a severe and which the Spotlight SAR naturally observes
usually unacceptable penalty must be paid its target. The first unclassified work on the
to push the antenna aperture to very small subject was [13], and an interesting inter-
dimensions. pretation may be found in [14]. Treatment
of Spotlight SAR in the general context of
The fundamental resolution expression of SAR processing is included in [6). Modern
Eq. 7 may be exploited in another way, systems use on-board real time processing.
however. Resolution for a given illumi-
nated region is determined primarily by the 3.3. The Stretch Technique
set of angles over which the data sequence
is gathered. A Spotlight SAR is designed to Having achieved very fine azimuth resolu-
observe a selected patch of terrain with tion through the Spotlight technique, it is
dynamic angular pointing of the antenna natural to seek comparable resolution in
pattern steered so as to maintain illumi- the range direction. This is not easily done
nation of the desired area as the sensor by direct means. For example, for 15 cm
passes by. Azimuth resolution much small- range resolution (symmetrical with the azi-
er than the antenna half-aperture may be muth example above), the range bandwidth
achieved for one area, with the trade-off required is on the order of 1 GHz. It is
being that other adjacent areas are not desirable to maintain fine range resolution
imaged at all. without paying such a penalty.
Antenna size for a Spotlight SAR is gov- In the early 1970's the "stretch" technique
erned by the desired size of illuminated was demonstrated [15]. For a linear fre-
area, and by the required signal-to-noise quency modulated (linear fin) signal, de-
ratio of the received signal sequence. Both modulation of the received signal by a
of these objectives are range dependant, delayed replica of the original results in a
leading to larger antenna size for larger difference signal of constant frequency.
ranges, all else equal. Phased array anten- The frequency is proportional to the
na technology is highly desirable for such relative delay of the received and reference
systems, although impressive performance signals. Useful results occur only when the
may be obtained with rather modest scan signals overlap substantially.
angles. For example, from Eq. 7, one may
find that resolutions on the order of only 5 Stretch is perfect for the Spotlight appli-
times the wavelength (15 cm at X-band) cation. It reduces the bandwidth require-
may be achieved with an illumination angle ments needed for all portions of the system
of about 3° (in the nominally side-looking following demodulation. It does this by a
Spotlight geometry). For an N-look image, trade-off between range and bandwidth, a
I
J2
10-7
convenient consequence of the linear fm The solid lines in the figure represent the
waveform. It is thus restricted to scenes (two-way) power profile of stationary scat-
with relatively small range extent, which fits terer return as limited by the azimuth pat-
the Spotlight requirement. Stretch modula- tern of the radar antenna. The zero order
tion applied to the Spotlight SAR case is Doppler response is shown together with
described by Walker in [13]. two of the ambiguous spectra each centered
on a multiple of the PRF. The width of the
3.4. MTI clutter spectrum is B. = 2PVIX, where p is
the two-way half-power angular width of the
An ability for moving target indication antenna pattern, and V is the radar vehicle
(MTI) has been for many years an objective speed.
of both strategic and tactical radar systems.
It is of interest to explore the ability of a The dotted lines represent the envelope
SAR in this application. Early work in this returned from a single target having radial
area may be found in [1], and especially velocity v with respect to the radar line of
[161, both of which are restricted to rates sight, leading to a Doppler frequency shift
rather small comparedto that of the radar 2v/A which appears as an additional phase
platform. For larger motions, the problem term (normalized to azimuth spatial fre-
rapidly gets more complicated [17]. quency by V1) in Eq. 4. (This may be
derived from from Eq. 1 by using
Attempting to simultaneously achieve both R. = > R.+ vt.)
fine spatial resolution and spectral reso-
lution flies in the face of physical principals We are interested in describing the way a
(e.g. [181). The general formulation of this slowly moving target appears in a SAR
principle is through the ambiguity function. image, and in the possibility of detection of
These fundamental limits apply to the con- moving targets.
ventional SAR configuration, suggested in
Figure 2. For a SAR, the small (coherent) radial
velocity component of a scatterer leads to a
shift & in the mapped azimuth position of
IT its image. This is a natural consequence of
the fact that the azimuth coordinate system
is derived from Doppler information, and
*,* the processor has no way of knowing th.,
the target itself might have a Doppler com-
ponent. For the aircraft case [16], this
argument leads to Ax = R, v/V. Since the
• .1 range to a scatterer is relatively large, there
is often a shift in scatterer position many
0
PRF 2 times the azimuth
less there resolution.
are tell-tale signs However,
of where un-
the
Figure 2. SAR Doppler space in the object should be located (such as the wake
presence of a moving target. (Two of a ship), it is impossible to identify the
ambiguities explicitly shown.) return as being associated with a moving
target. This is an example of the spatial-
Doppler ambiguity inherent in the problem.
10-8
In order for a signal to be detected as A qualitative prediction of SAR wave imag-

4 moving, its Doppler shift (see Figure 2) ing appeared in 1978 [191. Most quanti-
must be greater than the Doppler band- tative analysis was concerned with the effect
width of the return from stationary on a SAR image by the advection of each
scatterers, which, for satellite SARs, may be scattering cell on the ocean's surface by the
many km/h. Furthermore, in order to passage of longer waves such as swell [20],
detect a small moving target against a [21]. Since motion coherently sensed by the
larger clutter background, and to guard SAR causes image shift, systematic motions
against false alarms, usually the Doppler from the waves' orbital velocities leads to
shift must be much larger than the system "velocity bunching", actually helpful (within
Doppler bandwidth. limits) for forming a wave-like contrast
pattern in a SAR ocean image. However,
In order to achieve such a result, the the detailed structure of a wave changes
Doppler MTI passband, defined as the with time, sometimes rapidly, so that there
space between ambiguous Doppler spectra are coherence time limitations on image
free from clutter energy, must be made formation that lead to inherent azimuth
large, requiring a larger than normal PRF. reolution constraints, the so-called "azimuth
Normal motions from vehicles and other cut-off'. The state of the art of ocean SAR
moving targets of potential interest are imaging at that time is summarized in [221.
usually more complicated than these simple
results consider. In general, defocusing Independent investigations, both theoretical
arises when all velocity components are [23] and experimental [24], raised the
included and the possibility of range importance of non-coherent aspects of SAR
acceleration is allowed [161. More complex wave imaging. Since waves move, and since
motions and vehicle vibrations may lead to they are imaged from a platform itself in
loss of signal coherence. These consid- motion, both phase sensitive and position
erations argue against effective adaptation sensitive motions should be of importance.
of a conventional SAR as an MTI device.
(More may be done, however, as noted For the time scales encountered in airborne
below.) SAR systems, the two types of wave motion
may be exploited. The separate looks
3.5. Imaging Ocean Waves normally created from a SAR data set are
spectrally separable in the Doppler domain.
The most subtle of "conventional" SAR Since the azimuth signal is of large time-
imaging applications is that of ocean waves, bandwidth product, the Doppler spectrum,
A SAR observes essentially only the surface for any reference time, is proportional to
layer of salt water which under typical the time of actual data collection with
circumstances is constantly in motion. Ships respect to the reference time. It follows
moving on the surface may reasonably be that looks may be separately processed,
expected to follow the "rules" outlined in relative image shift between looks compen-
the preceeding section, but what about the sated, and then the shifted looks combined
water itself? The question has value in that [23]. Properly done, this leads to an
SARs are promoted as potential sources of "optimum" SAR image of ocean waves, and,
global wave climatology data. Today this of more importance, to an optimized direc-
I
promise has yet to be fulfilled. tional spectrum derived from the SAR data
[25]. Furthermore, the technique naturally
_I
,J "
10-9
leads to removal of the 1800 ambiguity of d

wave propagation direction so typical of -
wave imaging systems.
Imaging ocean waves from satellite altitudes R1 R2

and velocities is in general less than
optimum. The time variable aspect of the
geometry is scaled to first order by h/V
which assume less favourable values for
most orbits than for most aircraft SARs.
Ocean wave imaging is an active area of N-mhu ,Vpow X
research.
Figure 3. Basic geometry for topographic
4. SEQUENCE PAIRS interferometric imaging radars.
One could visualize a SAR as simply one

example of a classical coherent imaging
system. It is well known that such systems
reach their pinnacle of precise performance geometry are shown in Figure 3. The
when used in an interferometric mode, i.e., essential feature is that, for known antenna
one in which the scene observed is made to separation dh, and known slant ranges R1
yield its secrets at scales of the wavelength and R 2 to a scattering centre, then its
of illumination. This is the objective of relative height may be estimated from the
SAR configurations designed to use a pair phase information in the interference
of signal sequences, either spatially or pattern.
temporally.
In the mid-1980's, a brilliant innovation was
4.1. Spatial Interferometry introduced by Goldstein of JPL [27]. He
suggested that an interferometric pair could
A SAR image of terrain is a two-dimension- be created by two relatively closely spaced
al mapping of a three-dimensional surface. but separate passes of a (single antenna)
* Unless more information is available, there SAR satellite. Two pass interferometry was
is no way to quantitatively estimate terrain proven using selected passes of existing
height. Of course, as with aerial photo- Seasat data, from which interference fringes
graphs, a pair of SAR iamges each taken were generated, and relative terrain height
from a different point of view could be used estimated.
as a stereo pair, but as a coherent system,
SAR has much more to offer than that. Large area image interferometry placed a
great strain on the SAR processing capa-
In the early 1970's, Goodyear introduced a bilities then available. In response, the
SAR system [26] that carried two antennas 'wave-domain" algorithm and its derivatives
designed to function simultaneously. were introduced: collaboration between
Rocca and Goldstein led to better phase
The two received signals were coherently performance of SAR processors, and
combined analogous to an optical inter- proliferation of SAR interferometric work
ferometer. The main elements of the to Europe and elsewhere.
10-10
The underlying assumption of two pass of slope continuity through which progres-
interferometry is that the terrain being sive phase shift may be integrated across
observed is essentially unchanged (with changes in excess of 21r. Use of low
respect to phase characteristics) in the time resolution methods such as "shape from
interval between observations. The shading" is also helpful.
condition may be inverted, since the
absence of interference fringes is an Within the last decade, interferometry for
indication of (randomly) changed phase in SAR is one of the two most significant
the affected area. developments. (Quadrature polarimetry is
the other one.) The topic remains an area
The basic SAR interferometric principle of very active research, and may reasonably
leads to terrain height estimation with a be expected to offer valuable quantitative
precision on the order of the resolution of results to the user community in the years
the SAR [28]. Its accuracy, however, is very to come.
sensitive to the value of the baseline dh.
For applications in which a known "level" 4.2. Temporal Interferometry
terrain feature is present, such as a
shoreline, then the elevation contours may There is an alternative way in which to
be calibrated. In principle, the technique build an interferometer. For a SAR on a
may be used with any satellite SAR. moving platform, a pair of antennas could
be arranged to lie along the flight vector,
The measurement capability of interfero- thus providing essentially identical views of
metric SARs has been extended through the the illuminated field, but at slightly
use of differential techniques [29], [30], [31]. different times. The basic geometry is
Through this approach, one looks for illustrated in Figure 4.
differences in the fringe pattern between
two pairs of observation opportunities. In
effect, in each case, one of the signal pair
establishes a reference phase for each
element in the scene, and the other signal
provides an estimate of phase change with [
respect to the reference. In principle, the md
method is sensitive to physical changes in D
position of a reflecting element to less than
)/4, which has been demonstrated [29]. It l
has been proposed as a method for estima-
ting crustal movement either before or after
an earthquake [301. Figure 4. Dual antenna time-sequential
Interferometry is not without its problems. interferometry.

One of the most challenging is "phase
unwrapping" [31], [321. Interferometry rests
on phase estimation, which, unless other
information is implicitly or explicitly
available, is multiple valued. Current
techniques in effect rely on an assumption
p. I
..
10-11
The time delay 6t between the pair of where, neglecting noise, one may show that
received signals is DIV if the antennas are
operated separately as transmitter and MA ( )
receiver. The time delay is D/2V if, as is 1 5
usually the case, only one antenna is used (15)
for transmission and both are used for
reception.
as the indication of the presence of a
Scene elements having radial motion v,ad moving target. The detected difference
cause a differential phase shift image will contain only moving targets (plus
differential clutter and noise), each
A0 - 4 n vra 8t (13) weighted by the squared magnitude of the
X radially dependent phase term of Eq. 15.
which may be detected by interferometric Implementation of this algorithm using

combination of the signals from the two digital techniques in general requires
channels. For area extensive motion, such interpolation since there must be
as ocean surface currents, the detection continuous adaptation to variations in
technique normally employed is correlation aircraft velocity V. There is a neat way
[33], from which excellent results have been around this objection, however, by taking
obtained. The technique has been extended the (azimuth) Fourier transform, yielding a
to wave spectral estimation through a domain characterized by a pair of signals
variation on the basic theme [34]. such as given by Eq. 4. The delay operator
becomes simply a phase multiply
4.3.
! MTi and Velocity Estimationex(-D (2whnwitnnspia
exp(-jDw.,,2) when written in spatial
When most of the imaged field is expected frequency, and under the assumptions that
to be unchanged, the more direct processing only one antenna is used for transmission,
strategy of "delay and subtract" applies, and that the PRF is maintained
This is the case for MTI. It has been proportional to ground speed.
known for many years that the dual antenna
configuration offers a way around the The companion operation on the signal pair
Doppler lower limit imposed by the clutter of Eq. 14 is to form their sum. Thus
spectrum [35]. Performance estimates for
the SAR case are available in [16]. mE (t) - 2 ss (t)
If ssl(t) is the signal from the front antenna, + mm(t) + exp( -j4v -a 8t)]
and ss 2(t) is the signal from the rear
antenna, and mm(t) is the signal from a (16)
slowly moving target with radial velocity
Sv,,,d,then in which the effective common mode

mA (t) - ss2(t) - ss,(t - 8t) (14) (complex) signal is doubled, and the phase
modulated term is present as the comple-
ment to the differential channel. One may
show [36] that an effective doubling of the
i' image SNR follows from the first effect.
* -a..-
-.- .-
10-12
The two components of the moving target Improvement of image quality, both through
are shared proportionately between the two summation of extra non-coherent looks, and
through coherent integration in special
channels, with sin2 (kv,.5t) in the circumstances, has been demonstrated using
separate data sets, available, for example,
difference channel, and cos2 (kv,.6t) in from nearly spatially coincident orbits.
the sum channel, where k = 2w/A. These By taking the two-pass topographic inter-
expressions show that there is a "blind ferometric geometry to the limit, in princi-
speed" vr,, B,at which the target motion is ple one could generate multiple range offset
through one (round-trip) wavelength data sets sufficient to create full holo-
between observations, and thus not graphic (three dimensional) images of the
detectable by phase shift. Blind speeds are illuminated surface. In fact, this was
characteristic of MTI radar systems. demonstrated nearly 20 years ago using
aircraft based radars, but has certain
Energy partitioning of the moving target physical limitations when presented as an
signal between the sum and the difference
channel may be used for target velocity optical analog.
modern It could
era of digital have value
imaging, in the
however.
estimation, analogous to the same technique
used for calibration of dual channel Both spatial- and time-delay interferometry
polarimetric radars [36]. have been implemented using a SAR. The
Extrapolation of this principle to orbital classic (partially coherent) field analysis

velocities shows that along track antenna tool, however, is the mutual coherence
sepaatios shofwseal aontrk mtesa function which utilizes simultaneous
separations of several hundred metres application of a spatial and a temporal
would be required for detection of targets interferometric baseline. Such a tool in the
with small components of velocity towards SAR context should have value, for
the radar. the radar.oudhaevau, example, in deeper analysis of SAR ocean o
5. ADDITIONAL COMMENTS wave imaging.
This lecture has attempted to suggest ways Mostthe

and of the techniques
underlying described
capability above,a
to form
in which the basic capabilities of a SAR
basic the
SARurig
image, all ality tor "bistatic"
have their a
system may be stretched to new and some- counterparts. A bistatic system is one for
times almost unbelievable domains. The which the transmitter and the receiver are
discussion has been far from complete. in different locations, sometimes widely
Some closing comments are offered here. separated. We have restricted the discus-
sion to the case for which the radar trans-
One of the richest areas for SAR system mitter and receiver have the same physical
advanced technology is in quadrature polar- location, the so-called mono-static case.
imetry. This topic is covered by other lec- The fundamental requirement for bistatic
tures in this series. One should note in configurations, just as for the conventional
passing, however, that the underlying coher- csi htteeb sse oeec"
ecofteSAR channels is rqieto cssthtthere be "system coherence".
achieve full pne o tmei
polarim ynyelsis euitr
etric synthyesis either in
in In
p each
a e r situation
b s n s asrmeansq i e offoromaintaining
tesignal doanor the image domain, phase robustness
to be utilized. is required alll signals
i nl
i
10-13
Under the constraint that phase of the [7] C. Cafforio, C. Prati, and F. Rocca, Full
signal ensemble is available, then there are Resolution Focusing of Seasat SAR Images
many varied measurement possibilities open in the Frequency-Wavenumber Domain,
to the clever inventor. One may reasonably IntemationalJournalof Remote Sensing, Vol.
expect that in the forthcoming years, more 12, 1991, pp. 491-510.
special SAR techniques will be developed,
and be adopted as needed for specific [8] R. Stolt, Migration by Fourier Trans-
applications of this exciting device, form Techniques, Geophysics, Vol. 43, 1978,
pp. 49-76.
[9] R. K. Raney and P. W. Vachon, A

Phase Preserving SAR Processor,
6. REFERENCES Proceedings of the InternationalGeoscience
andRemote Sensing Symposium, Vancouver,
[1] R. 0. Harger, Synthetic Aperture Radar Canada, July, 1989, pp. 2588-2591.
Systems: Theory and Design, Academic
Press, New York, 1970. [10] R. K. Raney, An Exact Wide Field
Digital Imaging Algorithm, International
[2] R. K. Raney, A New and Fundamental Journal of Remote Sensing, Vol. 13, No. 5,
Fourier Transform Pair, Proceedingsof the 1992, pp. 991-998.
InternationalGeoscienceand Remote Sensing
Symposium, 26-29 May 1992, Houston, Tx. [11] I. Cumming, F. Wong, and R. K.
Raney, A SAR Processing Algorithm with
[3] R. Touzi, K. Raney, and A. Lopes, On No Interpolation, Proceedings of the
the Use of Complex SAR Data for Calibra- InternationalGeoscience andRemote Sensing
tion, Proceedings of the International Symposium, Houston, Tx., 26-29 May 1992.
Geoscience and Remote Sensing Symposium,
26-29 May 1992, Houston, Texas. [12] H. Runge and R. Bamler, A Novel
High Precision SAR Processing Algorithm
[4] R. K. Raney, Conceptual Design of Based on Chirp Scaling, Proceedings of the
Satellite SAR, Proceedings of the InternationalGeoscience andRemote Sensing
InternationalGeoscience andRemote Sensing Symposium, Houston, Tx., 26-29 May 1992.
Symposium, Strasbourg, ESA Publication
SP-215, 27-30 Aug, 1984, pp. 801-807. [13] J. L. Walker, Range-Doppler Imaging
of Rotating Objects, IEEE Transactions on
[51 R. Bamler, A Systematic Comparison of Aerospace and Electronic Systems, AES-16,
SAR Focusing Algorithms, Proceedings of No. 1, Jan 1980, pp 23-52.
the International Geoscience and Remote
Sensing Symposium, Espoo, Finland, June [14] D. C. Munson, J. D. O'Brien, and W.
1991, pp. 1005-1009. K. Jenkins, A Tomographic Formulation of
Spotlight-mode Synthetic Aperture Radar,
[6] D. A. Ausherman, A. Kozma, J. L Proceedings of the IEEE, No. 7, Aug 1983,
Walker, H. M. Jones, and E. C. Poggio, pp 917-925.
Developments in Radar Imaging, IEEE
Trans. on Aerospace and Electronic Systems,
Vol. AES-20, No. 4, July 1984, pp 363-400.
1 -
1I-14
[15] W. J. Caputi, Jr, Stretch: A Time- [23] R. K. Raney and P. W. Vachon, Syn-
Transformation Technique, IEEE Transac- thetic Aperture Radar Imaging of Ocean
tions on Aerospace and Electronic Systems, Waves from an Airborne Platform: Focus
Vol. AES-7, No.2, Mar 1971, pp. 269-278. and Tracking Issues, Journalof Geophysical
Research, Vol. 93, No. C10, 1988, pp.
[16] R. K. Raney, Synthetic Aperture Radar 12,475-12,486.
and Moving Targets, IEEE Transactioinson
Aerospace andElectronic Systems. Vol. AES- [24] P. W. Vachon and J. C. West, Spectral
7, No. 3, May 1971, pp. 499-50f. Estimation Techniques for Multilook SAR
Images of Ocean Waves, IEEE Transactions
[171 E. J. Kelly and R. P. Wish"-r, Matched on Geoscience and Remote Sensing, to
Filter Theory for Hign-velocity Accelerating appear, 1992.
Targets, IEEE Transactions on Military
Electronics,Vol. MIL-9, Jan 1965, pp. 56-59. [25] P. W. Vachon and R. K. Raney, Ocean
Waves and Optimal SAR Processing: Don't
[18] J. R. Klauder, The Design of Radars Adjust the Focus!, Transactions of the IEEE
Having Both High Range Resolution and Geoscience and Remote Sensing Society, to
High Velocity Resolution, Bell System appear, 1992.
Technical Journal,July 1960, pp. 745-808.
[26] L. C. Graham, Synthetic Interfero-
[19] R. K. Raney and R. A. Shuchman, meters for Topographic Mapping, Proceed-
SAR Mechanisms for Imaging Ocean ings of the IEEE, Vol. 62, No. 6, 1974, pp.
Waves, Proceedings 5th Canadian Sympo- 763-768.
sium on Remote Sensing, Victoria, B.C.,
1978, pp. 495-505. [27] H. A. Zebker and R. M. Goldstein,
Topographic Mapping from Interferometric
[20] C. T. Swift and L. R. Wilson, Synthetic Synthetic Aperture Radar Observations,
Aperture Radar Imaging of Ocean Waves, Journal of Geophysical Research, Vol. 91,
IEEE Transactions on Antennas and Propa- No. B5, 1986, pp. 4993-4999.
gation, Vol. 27, No. 6, 1979, pp. 725-729.
[28] C. Prati, F. Rocca, and A. M.
[21] W. R. Alpers and C. L. Rufenach, The Guarnieri, Effects of Speckle and Additive
Effect of Orbital Motions on Synthetic Noise on the Altimetric Resolution of a
Aperture Radar Imagery of Ocean Waves, Interferometric SAR (ISAR) Surveys,
IEEE Transactions on Antennas and Propa- Proceedings of the International Geoscience
gation, Vol. 27, No. 5, 1979, pp. 685-690. andRemote Sensing Symposium, Vancouver,
Canada, July 1989, pp. 2469-2472.
Raney, W. J.
[221 K. Hasselmann, R. K.
Plant, W. Alpers, R. A. Shuchman, D. R. [29] A. L. Gray and P. Farris-Manning,
Lyzenga, C. L. Rufenach, and M. J. Tucker, Two-Pass Interferometry with Airborne
Theory of Synthetic Aperture Radar Ocean Synthetic Aperture Radar, IEEE
Imaging: A MARSEN View, Journal of Transactions on Geoscience and Remote
Geophysical Research, Vol. 90, No. C3, May Sensing, to appear, 1992.
1985, pp. 4659-4686.
I
10-15
[30] A. K. Gabriel, R. M. Goldstein, and H.

A. Zebker, Mapping Small Elevation
Changes over Large Areas: Differential
Radar, Interferometry, Journal of
Geophysical Research, Vol. 94, No. B7, July
1989, pp. 9183-9191.
[31] F. K. Li and R. M. Goldstein, Studies

of Multibaseline Spaceborne Interfero-
metric Synthetic Aperture Radars, IEEE
Transactions on Geoscience and Remote
Sensing, Vol. 28, No. 1, Jan 1990, pp. 88-97.
[321 C. Prati, M. Giani, and N. Leuratti,

SAR Interferometry, A 2D Phase Unwrap-
ping Technique based on Phase and
Absolute Value Information, Proceedingsof
the International Geoscience and Remote
Sensing Symposium, Washington, D.C., May
1990, pp. 2043-2046.
[33] R. M. Goldstein and H. A. Zebker,

Interferometric Radar Measurements of
Ocean Surface Currents, Nature, Vol. 328,
1987, pp. 707-709.
[34] D. R. Lyzenga and J. R. Bennet,
Estimation of Ocean Wave Spectra using
Two-Antenna SAR Systems, IEEE Trans-
actions on Geoscience and Remote Sensing,
Vol. 29, No. 3, May 1991, pp. 463-465.
[35] H. Urkowitz, The Effect of Antenna

Pattern on the Performance of Dual-
antenna Radar Airborne Moving Target
Indicators, IEEE Transactionson Aerospace
and Naval Electronics, Vol. ANE-11,
December 1964, pp. 218-223.
[36] R. K. Raney, A "Free" 3-dB in Cross-

Polarized SAR Data, IEEE Transactionson
Geoscience and Remote Sensing, Vol. 26,
No. 5, Sept 1988, pp700-702.
Review of Spaceborne and Airborne SAR Systems
R. Keith Raney
Canada Centre for Remote Sensing
1. SUMMARY idea. This lecture describes the current

state of system development. All civilian
This Lecture provides a concise summary Earth observing synthetic aperture radar
of the state of the art in SAR systems, (SAR) satellites are reviewed: past,
both spaceborne and airborne. The first present, and future. The current status of
civilian SAR mission in space was the representative airborne SAR systems is
United States' SEASAT (L-band), outlined, and several other airborne
operating July-September 1978. It was systems mentioned. The discussion
followed by two Shuttle missions of one includes more than twenty SARs, hardly
week duration each, SIR-A (L-band, an exclusive field as of this writing!
November 1981) and SIR-B (L-band, Taken together, the implied financial
October 1984). The 1990's is witnessing a support is on the order of 5 billion dollars.
flurry of orbital SAR activity, with Almaz
(USSR, S-band, March 1991), ERS-1 There are several published sources that
(ESA, C-band, July 1991), J-ERS-1 are more complete than this brief lecture
(Japan, L-band, February 1992), SIR-C/X- is able to be. The reader is referred to a
SAR (USA/Germany and Italy, L-, C-, special issue of the Proceedings of the
and X-bands, three launches planned after IEEE [1] for articles that describe
1993), ERS-2 (ESA, repeat mission), and currently operating and approved future
RADARSAT (Canada, C-band, 1995) satellite SAR systems in some detail.
taking place in the decade. System Those papers include planetary missions
parameters are included in Tables, and as well as the Earth orbital radars of
general characteristics of these radars are interest here. In the context of pre-
compared and contrasted. The lecture paration for the EOS SAR mission of
also provides an overview of airborne NASA, an excellent review article [2] has
SAR systems, including those of NASA been prepared that covers several air-
(USA), CCRS (Canada), Intera/MDA borne SAR systems as well as orbital ones.
(Canada), DLR (Germany), and TUD
(Denmark), among others. Acronyms are For satellite SARs, this lecture is
defined in appropriate sections of the text. organized according to epoch: previous
missions; present missions; planned
missions; and proposed missions. The
2. INTRODUCTION airborne systems, typically in a constant
state of change, are described in terms of
After decades of development since the present known capabilities.
concept of "synthetic aperture radar" was
published in 1951, the 1990's is witnessing As is the case for any area in which many
an unprecedented level of technical (and sources might be found, there is not
financial) activity in imaging radar always agreement in those sources on the
techniques based on Carl Wiley's original detailed specifications for any given
Canadian Crown Copyright

I1-2
system. This is understandible, particu- to a massive short circuit in the slip ring
larly when technical change as well as assembly of the solar panel primary power
language and terminology confound the system. (There are system evaluation
issue. In this review, primary sources are reports available from the committees
used whenever available. Furthermore, charged with investigating this event,
when there is an important conflict in should any reader be interested in
published data, comments are included in hardware and programmatic details.)
the text to help clarify the matter for the Data provided by Seasat proved to be of
interested reader. high quality and immense interest to the
science and applications communities, and
still appears as prime material for recent
3. PREVIOUS ORBITAL SAR SYSTEMS papers in the professional literature.
For some individuals, history hinges on the In the Table, the antenna is described as
wisdom of Lao Tzu (1st century, China), "corporate". This is a shorthand notation
and for others on the inspiration of Jesus for a passive antenna using power dividing
Christ (1st century, Judea). For those techniques to distribute the signal to be
interested in synthetic aperture radar, transmitted to a network of radiating
however, history hinges on Seasat (20th elements. In the case of Seasat, the
century, North America). Virtually all antenna was a flat microstrip array, built
Earth observing SAR satellite systems in on eight panels. The size of the antenna
this lecture owe both their inspiration and for Seasat is of interest. Its long length,
baseline performance standards to Seasat, and large aspect ratio, is "typical" of
and by implication its design and science "standard" (satellite) SAR antennas. The
teams. Of course, there are solid logical reasons for this may be found in [3], and
reasons for the technical similarities, but in other lectures in this series.
the fact remains, Seasat led the way.
Seasat was initiated by an inter-agency Typical parameter values important for
study team, and its implementation was image quality for satellite SAR systems, in
managed by the Jet Propulsion Laboratory particular resolution and number of looks,
(JPL) (of the California Institute of were firmly established by Seasat. Unless
Technology in Pasadena, California) one is willing to give up other image
funded by NASA. Seasat was a magnifi- aspects such as swath width, there is
cent achievement [1]. rather little flexibility available.
Principle parameters of Seasat are The NASA objective for Seasat was digital
sketched in Table 1, abstracted from [3], processing, but NASA ran out of money
also available from a widely distributed (after the potential processor contractor
literature. As in the remainder of this had been selected, but before the contract
lecture, most comments in the discusssion had been awarded!). The project was
on this SAR are keyed to entries in launched, literally, with optical processing
pertinent columns of tables. as the baseline. In parallel, Charles Wu at
JPL (and MDA, Canada) developed digi-
The design lifetime of Seasat was two tal processing techniques for Seasat.
years. Unfortunately, the spacecraft failed Seasat was a resounding success largely
after three months of SAR operation due because of digital processing, as the
I
1~1
11-3
TABLE 1. Previous Orbital SAR Systems
Seasat SIR-A SIR-B (Kosmos 1870)

General
Country USA USA USA USSR
Agency NASA NASA NASA Glavkosmos
Spacecraft Seasat Shuttle Shuttle Salyut
Launch date 26 Jun 78 12 Nov 81 5 Oct 84 25 Jul 87
Lifetime 3 mos 2.5 days 8 days 2 years
Radar
Band [wavelength (cm)] L [23.5] L [23.5] L [23.5] S [10]
Frequency (GHz) 1.275 1.278 1.282 3.0
Antenna corporate corporate corporate waveguide
Size (m), length x height 10.7x2.1 6 9.4x2.16 10.7x2.16 1.5x15 (2)
Polarization HH HH HH HH
Incidence angle (0) 23 50 15 - 64 30 - 60
Range resolution (m) 25 40 25 (-30)
Azimuth resolution (m) 25 40 58- 17 (-30)
Looks 4 6 4 (4)
Swath width (km) 100 50 10 - 60 20 - 45
Recorder on board? N Y Y Y
Processing (Optical, Digital) O,D 0 0,D 0
Noise equivalent 0o (dB) -24 -32 (-28)
Mission
Nominasl altitude (km) 800 260 350, 225 250-280
Inclination (0) 108 38 57 71.5
Sun synchronous? N N N N
Down-link data rate (MB/sec) 110 (none) 30 analog
Repeat cycle (days) 17, 3 nil nil variable
Operation time per orbit (min) 10 - 3 min/tape
11-4
remainder of the Tables will support. After the conclusion of the Kosmos 1870
mission, Glavkosmos started to advertize
Seasat used an analog downlink with a its SAR products on the international
bandwidth of 20 MHz. The data was market, using agreements with such
recorded, thence a/d converted (5 bits, I organizations as Space Commerce Corp-
and Q, fixed point) to an effective data oration (USA). Kosmos 1870 was a radar
rate of 110 MB/sec. limited in several regards by its "analog"
foundations. Image products from that
The Shuttle SAR missions extended the radar appeared to be far inferior to
foundations established by Seasat in the expectations based on paper specifications.
dimension of incidence angle. Both SIR- Optical processing in series with analog
A and SIR-B [2] provided valuable results, data links seem to have significantly
but with technical limitations. SIR-A used reduced the effective bandwidth of the
only optical processing, depending on data, leading to fewer looks and more
airborne film recorders adapted to the coarse resolution than the radar itself
Shuttle mission. (SIR-A was an should have provided. (N.B. "Digital"
achievement in its own right, however, as image products available from Kosmos
it was the first payload to fly aboard a 1870 are digitized optically processed SAR
space Shuttle platform.) For SIR-B, two data, not to be confused with digitally
problems interfered with the SAR data processed SAR data.)
quality, including a partial failure of the
signal line connector to the antenna (with
severe increase in noise level), and failure 4. PRESENT SAR SATELLITES
of the data downlink system (TDRS
antenna tracking system) so that the data In contrast to the almost total absence of
recovery originally planned was curtailed. (civilian) space radar capability in the ten
Never-the-less, much valuable data was years following Seasat, the present decade
collected within the constraints of the is witnessing substantial activity. All
flight mission. known space based SARs presently in
service are listed in Table 2.
For two years at the close of the decade,
the USSR operated the Kosmos 1870 4.1. Lacrosse
radar [4] [5], at the time a classified
system. Kosmos 1870 was one of three Although a classified mission, there have
satellite SARs built in the late 1970's built been several open literature publications
by NPO Machinostroenye. (The other two (e.g. [6] and [71) from which the entries in
are known at present by the name of the table for Lacrosse have been gleaned.
Almaz, discussed below.) The Soviets had The key items include resolution, down to
a series of well known real aperture radars one metre [7], and the antenna size, which
in space (the Kosmos 1500 series), but may be estimated from an illustration
Kosmos 1870, in addition to Venera 16 published in [6] and Paris Match (date
and 17 to Venus, was its first publicized unknown). These two facts lead to the
orbital SAR. Kosmos 1870 and its Almaz conclusion that Lacrosse uses the Spotlight
relatives are the only space radars with SAR technique in its high resolution
two antenna assemblies, directed to mode. This requires a steerable antenna
opposite sides of the orbital plane. pattern, so it follows that the rather large
I
TABLE 2. Present SAR Satellites
(Lacrosse) Almaz ERS-1 J-ERS-1

General
Country USA USSR Europe Japan
Agency USAF Glavkosmos ESA MITI/NASDA
Spacecraft (Atlantis) Salyut ERS-1 J-ERS-1
Launch date 13 Dec 88 31 Mar 91 16 Jul 91 11 Feb 92
Lifetime (design) 2 years 2-3 years 2 years
Radar
Band [wavelength (cm)] X [3.0] S [10] C [5.7] L [23.5]
Frequency (GHz) 9.5 3.0 5.25 1.275
Antenna array waveguide waveguide corporate
Size (m), length x height (-8x2) 15xl.5 (2) 10xi 11.9x2.4
Polarization HH HH VV HH
Incidence angle (0) steerable 30 - 60 23 38
Range resolution (m) > 1 15 - 30 26 18
Azimuth resolution (m) > 1 15 28 18
Looks variable > 4 6 3
Swath width (kin) variable 20 - 45 100 75
Recorder on board? Y N Y
Processing (Optical, Digital) D D D D
Noise equivalent a0 (dB) - -24 -20.5
Mission
Nominal altitude (kin) -275 300 -780 568
Inclination (0) 57 73 98.5 97.7
Sun synchronous? N N Y Y
Down-link data rate (MB/sec) - 105 30 (x2)
Repeat cycle (days) nil 3, 35, 176 44
Operation time per orbit (min) 3 min/tape 10 20
I
11-6
antenna must be a phased array. Orbital the heart of which is a SAR whose
parameters are based on the capability of parameters are listed in the table. After
the Shuttle Atlantis launched from Cape several months in the three day repeat
Kennedy, for which 570 is the limiting orbit used for system verification and data
inclination. (Lacrosse is manoeuverable validation, the orbit was changed. The
[7], however, so that current values of first two weeks of April 1992 used the "roll
inclination and altitude may differ from tilt" mode in a 35 day repeat pattern. The
those in the Table.) The press reports at "roll tilt" accomplishes a change in
least two other Lacrosse systems awaiting incidence angle to about 35'. After mid-
operations, one of which may already have April, ERS-1 resumed normal operations,
been launched. Imagery from Lacrosse is still in the 35 day repeat orbit. The 176
not available. day repeat orbit will be used for an
extended period later in the mission to
4.2. Almaz support altimetric experiments. Data is
downlinked when within range of a
Almaz (11, [41, [5]), or "diamond in the receiving station on X-band at 105
rough", isvery similar to Kosmos 1870, but MB/sec using 5 bit quantization (fixed
up-graded in several regards. Data is point) on the I and Q channels. Image
stored onboard in four parallel video quality from ERS-1 is excellent.
recorders whose capacity limit the length
of each data take. The data downlink is With two operational SARs in orbit, it is
analog for both realtime data and interesting to compare their respective
recorded data. The most significant views of the Earth. Figure 1 shows an
improvements in the Almaz signal chain area near Whitecourt, Alberta as seen by
followed from conversion to digital data both Almaz and ERS-1. The contrast
handling and processing. The standard between these two images is striking.
processor is patterned after a VAX 780, ERS-1 appears to be more sensitive to
and uses the conventional range/Doppler topographic effects, and less sensitive to
algorithm. Image quality is variable, but differences in vegetation. Almaz, on the
some excellent examples are available, other hand, easily displays the clear-cuts in
the forest, but is less sensitive to the
From an applications point of view, Almaz terrain relief.
is a very interesting system. Its S-band
wavelength and selectable incidence angle 4.4. J-ERS-1
make it a good choice for certain Earth
resource issues. An example of Almaz The Japanese launched J-ERS-1 ("Earth
imagery may be found in Figure 1. Resources Satellite") early in 1992. After
early difficulty with the mechanism, the
4.3. ERS-I SAR antenna was successfully deployed on
9 April. At this writing, the SAR system
After extensive studies and preparations is still undergoing inflight checkout; the
from about 1975, the European Space initial imagery is excellent. J-ERS-1 is a
Agency launched ERS-1 in 1991. The joint development of the Ministry of
primary payload [8], [9] onboard ERS-1 International Trade (MITI) and the
("European Remote Sensing" satellite) is Science and Technology Agency of the
the Active Microwave Instrument (AMI), National Space Development Agency
11-7
,
p
oI
i0
Figure 1. Satellite Imagery of the Forest Test Site in Whitecourt, Alberta.

Top: ERS-1 image (C-band, 230 incidence angle)
Bottom: ALMAZ image (S-band, 44' incidence angle)
, (Processing courtesy Canada Centre for Remote Sensing)
11-8
(NASDA) of Japan [10].

5. PLANNED ORBITAL SAR MISSIONS
J-ERS-1 is like Seasat SAR in many ways.
In fact, itswereoriginal
identical performance In the sense
specifications to those of missions areused here, for
those planned
whichorbital
the
* Seasat. It turned out that two things sponsoring agences have committed the
combined leading to the present para- necessary resources, and the flight systems
maters (Table 2). First, the test results for are under construction. The "approved"
* the solid state power amplifiers showed systems are outlined in Table 3. They
much better output power than had been include: SIR-C/X-SAR, a series of
thought possible in early design stages. Shuttle-based missions each of relatively
Thus, greater range could be accepted. short duration; ERS-2, identical in most
Second, the driving application for J-ERS- regards to ERS-1 (see above); and
1 is global economic geology, for which RADARSAT, the first remote sensing
larger incidence angles are desirable in satellite mission of Canada.
order to reduce image distortions from
terrain relief. The nominal incidence 5.1. SIR-C/X-SAR
angle for J-ERS-1 was changed accord-
ingly, now at 380, rather than the nominal The SIR-C/X-SAR mission represents a
230 of Seasat. (Note that the literature major milestone in space-based radar.
almost always quotes a "look angle" of 350 Principal features of interest are suggested
for J-ERS-1, which refers to the angle of by the entries in Table 3 which guide the
the antenna beam at the spacecraft, and comments in this section. More details
does not include Earth curvature. Inci- may be found in [2] and [11].
dence angle is the preferred specification
for the antenna effective viewing angle.) SIR-C/X-SAR is a joint venture between
the United States (NASA-/JPL), and a
As a consequence of the larger incidence European consortium of Deutsche
angle, the ground range resolution is Forschungsanstalt ffr Luft- und Raumfahrt
reduced from 25 metres to 18 metres, with e.V. (the German Aerospace Research
no change in system range bandwidth. Estab-lishment known as DLR) and the
Having a range resolution of 18 metres, Agenzia Spaziale Italiana (ASI, Italy).
the azimuth resolution was chosen to be JPL is responsible for the C- and L-band
the same, achieved by a simple adjustment systems, and DLR/ASI is responsible for
in the processor. Finally, the number of the X-band system. Through extensive
looks available is reduced by about 25%, collaborative efforts, these radars have
leading to 3 looks, or (18/25)x4. Con- been harmonized so that they may be
clusion? Although the specifications for J- operated simultaneously, clearly a "first" at
ERS-1 look different from Seasat, the both organizational and technical levels.
imagery with respect to image quality Since the Shuttle has limited orbital
should be about the same. The increased duration, each of the three planned
incidence angle should be of value, missions are expected to last for about 10
days, although an extended (16 day)
Shuttle capability remains a possibility for
the later launches.
I
-- .. ...
..
--
r11-9
TABLE 3. Planned Orbital SAR Systems
SIR-C/X-SAR ERS-2 RADARSAT

General
Country (See text) Europe Canada
Agency NASA/DLR/DARA ESA CSA/USA
Spacecraft Shuttle ERS-2 (dedicated)
Launch date '93, '94, '96 1994 1995
Design lifetime 10 days 2-3 years 5 years
Radar
Band [wavelength (cm)] L,C; X C [5.25] C [5.7]
Frequency (GHz) 1.25, 5.3, 9.6 5.3 5.3
Antenna arrays + WG(X) WaveGuide WG array
Size (m), length x height 12x(3, 0.75, 0.4) 10xi 15xl.5
Polarization quad-pol L+C; Xv VV HH
Incidence angle (0) 15 - 55 23 < 20 - > 50
Range resolution tjn) 10 - 30 26 10 - 100
Azimuth resolution (m) 30 28 9- 100
Looks -4 6 1-8
Swath width (kin) 15 - 60 100 10 - 500
Recorder on board? Y (+ D/L) N Y
Processing (Optical, Digital) D D D
Noise equivalent o ° (dB) -40 < -28 -24 -23
Mission
Altitude (km) 225 -780 -800
Inclination (0) 57 98.5 98.6
Sun synchronous? N Y Y
Down-link data rate (MB/sec) 45 (TDRS) 105 74 - 105
Repeat cycle (days) nil TBD 24
Operation time per orbit (min) (60 h, total) 10 20
11-10
The radar bands selected represent orbital from this radar system. These, coupled
SAR heritage (L), current operational with a degree of freedom in bandwidth
preference (C), and extension to a new and incidence angle selection, lead to very
(civilian) orbital wavelength (X). The complex planning for the operations of
antennas required have been matched in SIR-C/X-SAR. In addition, there are
two regards. First, in order to allow the options in the number of bits to be
same PRF for all three bands, dictated by included in the signal data path, and
the requirement to have simultaneous allowance for experimental passes in the
multi-frequency operation, each of the ScanSAR format (see RADARSAT
antennas should have the same length. below), and for squinted beam data
Second, in order to have nominally the collection. (At this time, however, there is
same elevation beamwidth, the height of no plan to support data processing for
the antennas must be scaled in proportion these special imaging geometries.)
to their wavelength. The size of the total
array is about 12m by 4.1m. The three Processing for SIR-C, and particularly for
antennas will be mounted on one common X-SAR, poses its own challenges. Normal
structure in the cargo bay of the Shuttle, Shuttle angular motions, in combination
vertically stacked. with Earth rotation, lead to relatively large
and time varying Doppler centroid
Antenna technology for the X-band system variations in the received SAR signals.
is slotted waveguide, horizontally For the Shuttle, pitch, roll, and yaw are
polarized. Elevation steering is by allowed to vary within 1.5'. (The attitude
physical rotation of the antenna radiating control system for the Shuttle is not
surface about its longitudinal axis. "proportional", but depends on angular
momentum impulses from gas-jet thrusters
Antennas for both the C- and L-band to correct angular position when the
radars use active phased array technology, design "dead-zone" is exceeded.)
There are 18 panels for each antenna, Tolerance by the coherent radar signal to
a each one consisting of transmit/receive
(T/R) modules for H and V polarizations,
(yaw) attitude is determined primarily by
azimuth antenna beamwidth. For a given
By using phase control of the individual aperture size, the beamwidth of the
T/R modules, the antenna patterns can be antenna pattern decreases in proportion to
steered in both elevation and azimuth. wavelength. For example, the azimuth
beamwidth for X-SAR is about 0.150,
The H and V phased arrays, backed up by which is ten times smaller than the Shuttle
parallel receiver and data recording attitude uncertainty. It follows that
chains, may be cycled to achieve reception Doppler centroid estimation and tracking,
of the complex scattering matrix of the and its compensation in time and range, is
scene, the so-called quadrature polar- the most challenging SAR processing task
ization technique (described in another among all others represented by the
lecture in this series). SIR-C/X-SAR will mission. This circumstance has led DLR
be the first time that this capability is to new frontiers in SAR data processing
available from space. techniques, Le., [12] and [13].
Clearly, with choice of polarizations and

frequencies, nearly 20 modes are available
~IlIl
5.2. RADARSAT ERS-1 have no ability to look to the

south, or, in the case of Almaz, have
On 13 September 1989, the Government orbital inclinations such that Antarctic
of Canada announced full commitment to coverage is incomplete.
build and operate RADARSAT, Canada's
first Earth resources remote sensing NASA is a major partner in the
satellite [14]. It is being prepared for a RADARSAT mission contributing services
launch in 1995, and is designed for five for the planned 1995 launch from
years of service in orbit. The only imaging Vandenburg AFB using a medium-class
instrument is a SAR. A variety of expendable vehicle (McDonnell Douglas
resolution, image swath width, and Delta II 7920-10). The satellite payload
incidence angle parameters are available will consist of the SAR and its associated
that may be selected through ground downlink transmitters, tape recorders, and
command. The designated agency command and control computer. The
responsible for RADARSAT is the spacecraft is being procured from Ball
Canadian Space Agency (CSA). The Aerospace (United States), while SPAR
mission is the result of more than a Aerospace (Canada) has prime system
decade of work and initiative by the responsibility under contract with the
Canada Centre for Remote Sensing Canadian Space Agency.
(CCRS).
Data downlink for RADARSAT is at X-
The nominal configuration of the band, with the maximum data rate chosen
spacecraft has the SAR pointing to the to be compatible with the ERS-1 data
north side of the orbital plane. This rate. RADARSAT will use 4 bit floating
makes possible regular coverage of the point quantization, I and Q. There are
Arctic up to the pole, but coverage of the two downlink channels, required to
central Antarctic region is not possible support simultaneous readout of the
with this geometry. However, for two onboard tape recorder together with "live"
periods during the first two years of the data. There are three downlink
mission, the satellite will be rotated 1800 transmitters, a "3 for 2" redundancy
about its yaw axis so as to direct the radar approach.
antenna beam to the south side of the
orbital plane. Each such yaw manoeuvre For a satellite using a radar sensor, good
is expected to be maintained for two solar illumination of the spacecraft is
weeks. The purpose of this manoeuvre is more important than sunlight on the
to obtain a complete SAR map of Ant- Earth's surface. For this reason,
arctica at the times of maximum and RADARSAT will use a sun synchronous
minimum ice cover. (To the extent that dawn-dusk orbit. In this orbit, the
spacecraft thermal budgets allow, this will spacecraft is illuminated by the sun
also allow opposite side viewing for throughout each orbit. (There are short
selected regions north of the Equator periods at southern latitudes during the
during these periods.) Optical sensors Austral winter during which the spacecraft
such as Landsat have no access to regions is in eclipse.) Perhaps the greatest
closer to the poles than 810, and are operational advantage of this orbit is that
further compromised by clouds and the SAR can be fully dependent on solar
darkness. Other space radars such as derived rather than stored battery power,
11-12
which means that there is no limiting imaging modes at a variety of incidence

distinction between ascending and angles, several additional imaging modes
descending passes from an applications become available at rather small marginal
point of view. Thus, nearly twice as many cost. The design philosophy for these
viewing opportunities are available to the extra modes has been to base the system
mission than would otherwise be available, specifications on the stand rd imaging
Another operational advantage is that the modes, and to optimize the p.-rformance
ground station data downlink periods for of the additional modes within the design
RADARSAT will not conflict with other envelope determined by the standard
remote sensing satellites, most of which modes. Image quality in the additional
use near mid-day orbit timing. modes is predicted to be comparable to
that of the standard modes.
RADARSAT will carry tape recorders
with sufficient capacity for more than ten Imaging modes for RADARSAT include
minutes of full quality SAR data. As a Standard, Wide Swath, Fine Resolution,
consequence of the orbit yaw manoeuvre Extended, and ScanSAR (see Figure 2 and
and the onboard recorders, RADARSAT Table 4). In each mode, data are
will be the first fine resolution satellite collected continuously along a swath
system capable of complete global parallel to the sub-satellite path. Swath
coverage. This feature of RADARSAT is length is limited only by the duration of
relevant to global environmental continuous radar operation, and may be
monitoring as well as being of scientific thousands of kilometres long. Swath
value, widths and positions are determined by
the antenna elevation beam patterns (and
5.2.1. Modes the radar range gate control), and have
been chosen for the standard modes so
RADARSAT has been designed in res- that there is at least 10% overlap between
ponse to user requirements that demand a adjacent swaths. Range resolution when
variety of incidence angles (from about projected onto the Earth's surface varies
200 to 500) in the standard imaging modes. with incidence angle and hence ground
An antenna with electronic elevation range. Three range bandwidths are
beam steering is part of ihe baseline available (11.6, 17.3, and 30.0 MHz) to
RADARSAT design. Although this allow choice in ground range resolution
enables user requirements to be met, it achieved at each incidence angle. Nom-
does add further complexity to the entire inal range resolution for the standard
system. In order to provide a (nominally) beams has beCn specified at ground ranges
constant ground range resolution over the of 400 km and 675 km from the sub-
range of incidence angles, three different satellite locus.
pulse bandwidths are needed. It also
follows that very fine control of the The additional modes are generated by
transmitter pulse repetition frequency appropriate choices of antenna beam and
(PRF) is required. range pulse bandwidth. The fine
resolution mode, for example, is achieved
Having moved to build in an antenna and by selecting the widest available
control system with the flexibility (and bandwidth, and using a narrow beam in
complexity) necessary to support standard elevation at incidence angles larger than
A
11-13
TABLE 4. RADARSAT Imaging Modes
Mode Resolution Looks 2 Width Incidence

(R1 x A, m) (kin) (degrees)
Standard 25 x 28 4 100 20 - 49
Wide (1) 48-30 x 28 4 165 20-31
Wide (2) 32-25 x 28 4 150 31 - 39
Fine resolution 11-9 x 9 1 45 37 - 48
ScanSAR (N) 50 x 50 2-4 305 20 - 40
ScanSAR (W) 100 x 100 4-8 ° 510 20 - 49
Extended (H) 22-19 x 28 4 75 50 - 60
Extended (L) 63-28 x 28 4 170 10 - 23
Nominal; ground range resolution varies with range.

2 Nominal;range and processor dependent.
Fiurs. AARAte Track

/ - Extended Beams
(low incidence)
9!
SCANSAR
/ Fine Extended Beams
SResolution (high incidence)
Wide Swath Beams
Beams
2'St andard
Beams
Figure 2. RADARSAT Imaging Modes

11-14
(nominally) 45' . A narrow swath results to image a section of one sub-swath must
from the requirement to minimize be from consecutive pulses in order to
beamwidth in order to maintain good provide adequate sampling, and must be
signal to noise ratio, and also from the of sufficient length to allow formation of
necessity to maintain data rates consistent the synthetic aperture needed for the sub-
with downlink channel capacity. Wide swath at the required resolution. The
swath modes are supported by wider imaging operations are therefore split up
antenna beam widths than normal at into a series of blocks of pulses, each
steeper incidence angles, and use of the block providing returns from one of the
smallest available range pulse bandwidth sub-swaths. Each block is processed to
leading to more coarse ground range provide an image of a section of the
resolution. Signal to noise ratio and data corresponding sub-swath. The imaging
bandwidth arguments apply in these operations cycle around the full set of sub-
modes that are counterparts to those for swaths sufficiently rapidly that the imaged
the fine resolution modes, but with the sections in any one sub-swath are
result of broadening the usable swath adjoining or over-lapping.
width. Extended modes result from
selection of beams outside of the nominal RADARSAT will be the first operational
500 km accessibility region, either closer satellite radar system to implement the
to nadir (steeper incidence), or further ScanSAR technique.
away (more shallow or grazing incidence
angle).
6. AIRBORNE SAR SYSTEMS
5.2.2. ScanSAR
This section provides a glimpse at civilian
In order to allow imaging of a swath much airborne SAR systems now in service.
wider than ambiguity limits would Since most of these radars are meant for
normally allow, the RADARSAT system technology development as well as
has been designed to incorporate an applications experiments, the hardware is
alternative and less conventional mode frequently changed. The parameters listed
([15] and [16]) known as ScanSAR. In in Table 5 are thought to be an accurate
this mode, for which rapid steering of the representation of the basic performance of
elevation beam pattern of the antenna is those systems. Additional comments, and
essential, extended range coverage can be a resume of other airborne SARs are
obtained using a set of contiguous beams, included in the following sub-sections.
enabling images of total swath width up to
about 500 km to be produced. This is 6.1. Comparison of Selected Systems
accomplished at no increase in mean data
rate from the radar, but at the cost of The most widely deployed airborne SAR,
degraded resolution of the resulting image. and the one having the most modes, is
that of NASA which carries the radar of
The principle of ScanSAR is to share JPL (the Jet Propulsion Laboratory). This
radar operational time between two or radar, known as AirSAR ([2] and [19]), is
more separate sub-swaths in such a way as relatively new, having been designed and
to obtain full image coverage over their built to replace its predecessor which was
combined swath. The set of returns used destroyed in an aircraft fire in 1985. The
11-15
TABLE 5. Current Status of Selected Airborne SAR Systems
AirSAR C/X SAR E-SAR KRAS STAR-1

General
Country USA Canada Germany Denmark Canada
Agency JPL/NASA CCRS DLR TUD Intera
Aircraft DC-8 CV-580 DO-228 Gulfstream Cessna
Nominal altitude (km) 8 6 3.5 12.5 10
Nominal airspeed (m/sec) 130 70 300 175
Purpose exp'l exp'l exp'l exp'l operat'l
Radar
Band C, L, P X, C X, C, L C X
Frequency (GHz) 5.3, 1.25, 0.44 9.3, 5.3 9.6, 5.3, 1.3 5.3 9.6
Antenna length (m) 1.3, 1.6, 1.8 -1.2 0.15, 0.24, 0.85 1.2 -0.8
Antenna motion controller ? N Y N Y Y
Polarization diversity Y Y HH, VV VV HH
Quadrature polarization C, L, P C No No No
Incidence angle (0) 20-60 0-85 15-60 20-80 45 -80
Range resolution (m) (Slant) 7.5 (S) 6 - 20 2 (S) 2,4,8 6, 12
Range cells 4096 2048 8192 4096
Azimuth resolution (m) 2 < 1 - 10 2 2,4,8 6
Looks 4 1-7 1-8 2-16 7
Swath width (km) (Slant) 7-13 18 - 63 3 (S) 9 - 48 (S) 40, 60 (S)
STC ? Y Y Y N
Processor on board ? Y (1 ch) Y Y (QL) (Y) Y
Noise equivalent a' (dB) -30, -40 -40, -30, -35 -42 -30
Special modes InSAR, At InSAR
Ii
11-16
quadrature polarimetric capability at the do the previous two systems. Its recent
three frequencies of this radar offers a extension to X-band is in support of
unique and very rich data source that is DLR's vested interest in X-band for the
made available by NASA to investigators X-SAR radar. The system includes a
around the world. AirSAR also offers two quick-look (QL) onboard processor having
other special modes that merit attention, 50 m x 50 m resolution.
as noted in the Table. InSAR is an inter-
ferometric mode created by using data The radar of the Technical University of
from two antennas, one mounted above Denmark (TUD) has been designed to
the other on the side of the aircraft. The offer a variety of incidence and image
interference pattern between them may be parameter values within the constraint of
used to deduce terrain height information, being a single polarization C-band system
For AirSAR, this capability is at C-band [24], [251. Within the limits set by their
only. The other special mode also respective imaging geometries, the radar is
requires two separate antennas, this time matched to the ERS-1 SAR, aided by the
spaced along the line of flight, use of an aircraft capable of high speed
Interferometric measurements with these and high altitude.
two antennas may be used to observe
phenomena in the scene that change over The leading civilian SAR that is used for
the short interval At between observations, mapping surveys is STAR-i, owned and
such as currents on the ocean's surface. operated by Intera, of Calgary, Alberta
[261. This X-band system has performed
The SAR flown by CCRS (the Canada more than 75% of all of the airborne
Centre for Remote Sensing) is on a radar mapping done for commercial
Convair-580. Both X- and C-bands are clients world-wide since 1986. Data
fully supported by onboard real time products from this system are digitally
digital processors [20], and have a variety rectified to map accuracy standards, and,
of modes and data combinations available, with the recent use of GPS, allows
The standard image products are pro- accurate mapping with no need for
duced at seven looks. The system has surveyed control points. Under satellite
recently been modified to incorporate full navigation control, the radar map itself is
quadrature polarimetry on C-band, and the most accurate source.
may be operated in an InSAR mode.
Signal data is recorded so that ground 6.1. Overview of Other Systems
processing may be used for specific
experimental purposes, such as inves- There are several other airborne SARs
tigations requiring access to separate looks that deserve mention. The Netherlands
for optimized oceanographic SAR wave for many years has been supporting the
imagery. development of an advanced system
known as PHARS [27]. Currently,
The airborne radar of DLR (defined in PHARS is a C-band SAR mounted in a
section 5.1) continues to be upgraded with Swaeringen Metor II, a twin engined
new modes and capabilities [211, [22], and business plane. Early results meet or
[231. It is designed primarily for high exceed specifications. Nominal resolution
resolution and technology development, is about 5 metres with about 6 looks. The
hence it has a narrower swath width than antenna is W polarized to support
11-17
experiments with ERS-1. The program is II has already been built, and is virtually a
committed to implementing a full twin to Almaz. There is an intention to
quadrature polarimetric version within a launch and operate this radar, but the
few years. changing infra-structure of the former
USSR space segment places the future of
One of the pioneers in the field of SAR is Almaz II in doubt. Progammatic responsi-
the Environmental Research Institute of bility has been assumed by Russia for
Michigan (ERIM), whose civilian PRIRODA ("nature"), a complex payload
experimental SAR is managed by the that includes a SAR as one of several
United States Navy Air Development instruments [17], [18]. Much of the
Center (NADC), and is mounted on a P-3 hardware for the PRIRODA payload has
aircraft. The radar operates at X-, C-, and already been built.
L-bands, and is fully polarimetric [28].
Nominal resolution is in the 3 metre The EOS-SAR (Earth Observing System
range. SAR) [21 is at a different stage. There is
an active design and science team at work
The French have been actively involved in at JPL on EOS-SAR, but the program has
SAR for many years. Their SAR (a yet to receive funding approval from
civilian version of a former Thomson-CSF NASA. The Proposal by the EOS-SAR
classified X-band system) is known as team to NASA is to be in 1993.
VARAN-S [29], and has supported several
remote sensing experiments in Europe. The EOS-SAR as presently conceived
Polarizations are HH and VV. Nominal would be a most ambitious system. It
resolution is on the order of 5 meters, would carry the multi-frequency,
with four looks, nominal, quadrature polarization capability of SIR-
C/X-SAR into a long design life satellite
In addition to these relative well known implementation.
systems, others might be mentioned for
completeness. The Chinese Academy of In addition to these SARs, there are other
Sciences developed their own X-band initiatives being developed. France for
SAR several years agb. The system is several years has been promoting SAR-
called CASSAR, and has seen limited de- 2000, an X-band space radar intended to
ployment in the Far East. The USSR complement the SPOT program. Recent-
maintained an airborne SAR capability as ly, France and Canada have entered a
a test-bed for satellite programs, both bilateral agreement to do joint studies of
Earth oriented and planetary. India has a combined satellite radar system building
developed its own airborne SAR as well on the SAR-2000 and the RADARSAT III
as an indigineous digital SAR data planning. RADARSAT II, essentially a
processing capability, replacement for RADARSAT, is under
consideration by the Canadian govern-
ment. The European Space Agency has
7. PROPOSED SAR SATELLITES sponsored several studies of SAR concepts
that look beyond ERS-2. Undoubtedly,
There are three SAR satellites likely to be other space agencies have exploratory
approved for completion and launch in the space radar studies under way.
next decade, as noted in Table 6. Almaz
I
1I1-18
TABLE 6. Proposed SAR Satellites
ALMAZ Il EOS-SAR PRIRODA

General
Country (TBD) USA Russia
Agency TBD NASA TBD
Spacecraft Salyut (dedicated) MIR
Launch date (1994?) 2000 + (1994?)
Lifetime 2 years 15 years
Radar
Band S L, C, X L,S
Antenna WaveGuide arrays WG, array
Polarization HH quad-pol VV or HH
Incidence angle (0) 30 - 50 15 - 45 35
Range resolution (m) 15 - 30 20 250
- 100
Azimuth resolution (m) 15 8 250
- 50, 150
Looks > 4 1-10
Swath width (kin) 20-45 30-360 80
Recorder on board? Y TBD Y
ii
11-19
8. REFERENCES [10] Y. Nemoto, H. Nishino, M. Ono, H.

Mizutamari, K. Nishikawa, and K. Tanaka,
[1] F. L. Li and R. K. Raney, Prolog to Japanese Earth Resources Satellite-1
Special Section on Spaceborne Radars for Synthetic Aperture Radar, Proceedings of
Earth and Planetary Observations, the IEEE, Vol. 79, No. 6, June 1991, pp.
Proceedings of the IEEE, Vol. 79, No. 6, 800-809.
June 1991, pp. 773-776.
[11] R. L Jordan, B. L Huneycutt, and M.
[2] J. Way and E. A. Smith, The Evolution Werner, The SIR-C/X-SAR Synthetic
of Synthetic Aperture Radar Systems and Aperture Radar System, Proc.of the IEEE,
their Progression to the EOS SAR, IEEE Vol. 79, No. 6, June 1991, pp. 827-838.
Transactions on Geoscience and Remote
Sensing, Vol. 29, No. 6, November 1991, [12] H. Runge, Benefits of Antenna Yaw
pp. 962-985. Steering for SAR, Proceedings of the
International Geoscience and Remote
[3] R. L. Jordan, The Seasat-A Synthetic Sensing Symposium, IGARSS '91, Espoo,
Aperture Radar System, IEEE Journal of Finland, 3-6 June 1991, pp. 257-261.
Oceanic Engineering, Vol. OE-5, No. 2,
April 1980, pp. 154-163. [13] R. Bamler and H. Runge, PRF-
Ambiguity Resolving by Wavelength
[4] W. B. Wirin and R. A. Williamson, Diversity, IEEE Transactions on
Satellite Remote Sensing in the USSR: Geoscience and Remote Sensing, Vol. 29,
Past, Present, and Future, Remote Sensing No. 6, Nov 1991, pp. 997-1003.
Yearbook 1990, Chapter 3, Burgess Scien-
tific Press, Basingstoke, UK, pp. 49-64. [141 R. K Raney, A. P. Luscombe, E. J.
Langham, and S. Ahmed, RADAR-SAT,
[5] S. Chenard, Soviet Earth Obser-vation Proceedings of the IEEE, Vol. 79, No. 6,
gets Less Remote, InteraviaSpace Markets, June 1991, pp. 839-849.
Vol. 6, No. 1, January/February 1990, pp.
17-22. [15] R. K. Moore, J. P. Claasen, and Y. H.
Lin, Scanning Spaceborne Synthetic
[6] USA Today, 14 Dece.nber 1988. Aperture Radar with Integrated
Radiometer, IEEE Transactions on
[7] J. T. Richelson, The Future of Space Aerospace and Electronic Systems, Vol.
Reconnaisance, Scientific American, Vol. AES-17, No. 3, 1981, pp. 410-421.
264, No. 1, January 1991, pp. 38-45.
[16] K. Tomiyasu, Conceptual
[8] ERS-1 Special Issue, ESA Bulletin, Performance of a Satellite Borne, Wide
European Space Agency, 8-10 rue Mario- Swath Synthetic Aperture Radar, IEEE
Nikis, 75738 Paris Cedex 15, No. 65, Transactions on Geoscience and Remote
February 1991. Sensing, Vol. GE-19, No. 2, April 1981, pp.
108-116.
[9] E. P. W. Attema, The Active
Microwave Instrument on-Board the ERS- [17] Glavkosmos, Orbital Station "MIR":
I Satellite, Proceedings of the IEEE, Vol. Complex of Remote Sensing of the Earth
79, No. 6, June 1991, pp. 791-799. "PRIRODA, Moscow, 1990.
I
11-20
[18] N. A. Armand, A. A. Kalinkevich, B. System: Design and Initial Tests, IEEE

G. Kutuza, and S. M. Popov, SAR Transactions on Geoscience and Remote
Facilities for the "PRIRODA" Mission, Sensing, Vol. 29, No. 3, May 1991, pp. 417-
Proceedingsof the Intematir -1Geoscience 426.
and Remote Sensing S'ymposium,
IGARSS91, Espoo, Finland, 3-6 June [25] J. Dall, J. H. Jorgensen, E. L
1991, pp. 1790-1792. Christensen, and S. N. Madsen, A Real-
time Processor for the Danish C-Band
[19] J. J. van Zyl et al., AIRSAR Reference SAR, Proceedings of the International
Manual, Jet Propulsion Laboratory, Geoscience and Remote Sensing
California Institute of Technology, 1991. Symposium, IGARSS'91, Espoo, Finland,
3-6 June 1991, pp. 279-282.
[20] C. E. Livingstone, A. L. Gray, R. K.
Hawkins, and R. B. Olsen, CCRS C/X- [26] A. D. Nichols, J. W. Wilhelm, T. W.
Airborne Synthetic Aperture Radar: an R Gaffield, D. R. Inkster, and S. K. Leung,
and D Tool for the ERS-1 Time Frame, A SAR for Real-time Ice Reconnaissance,
IEEE AES Magazine, October, 1988, pp. IEEE Transactions on Geoscience and
11-16. Remote Sensing, Vol. GE-24, No. 3, May
[21] R. Horn, M. Werner, and B. Mayr, 1986, pp. 383-389.
Extension of the DLR Airborne Synthetic [27] P. Hoogeboom, P. J. Koomen, H. i
Aperture Radar, E-SAR, to X-band, Pouwels, and P. Snoeij, First Results from
Proceedingsofthe Inter-nationalGeoscience The Netherlands SAR Testbed "PHARS",
and Remote Sensing Symposium, Proceedingsof the InternationalGeoscience
IGARSS'O, Washington, DC, 20-24 May and Remote Sensing Symposium,
1990, p. 2047. IGARSS'91, Espoo, Finland, 3-6 June
1991, pp.241-244.
[221 R. Horn, C-Band SAR Results
Obtained by an Experimental Airborne [28] A. D. Kozma, A. D. Nichols, R. F.
SAR System, Proceedings of the Rawson, S. J. Shackman, C. W. Haney,
International Geoscience and Remote and J. J. Shanne Jr., Multi-frequency, -
Sensing Symposium, IGARSS'89, polarization SAR for Remote Sensing,
Vancouver, Canada, 10-14 July 1989, pp. Proceedingsof the InternationalGeoscience
2213-2216. and Remote Sensing Symposium,
IGARSS'86, Zurich, Switzerland, 8-11
[23] C. Dahme, R. Horn, D. Hounam, H. September 1986, ESA Publication SP-254,
Otti, and R. Schmid, Recent pp. 715-719.
Achievements of DLR's Airborne
Experimental SAR System and Image [29] R. Albrizio, P. Blonda. A. Mazzone,
Processing Equipment, Proceedings of the F. Pasquali, G. Pasquariello, F. Posa, and
International Geoscience and Remote N. Veneziana, Digital Processing of X-
Sensing Symposium, IGARSS'91, Espoo, Band VARAN-S Airborne SAR Images,
Finland, 3-6 June 1991, pp. 245-246. Proceedingsof the InternationalGeoscience
and Remote Sensing Symposium,
[24] S.N. Madsen, E. L Christensen, N. IGARSS'89, Vancouver, Canada, 10-14
Skou, and J. Dall, The Danish SAR July 1989, pp. 2203-2208.
B-I
BIBLIOGRAPHY
This Bibliography was compiled to support Lecture Series 182 by the

Fachinformationszentrum, Karlsruhe, D-7514 Eggenstein-Leopoldshafen 2 (Fiz
Karlsruhe), Germany, in association with the Lecture Series Director, Dr W. Keydel.
St
I
B-2
64. 1g0. t- *0-*40-600 1 04!

-4wMw 02-4 OA .406 -40-
10 F 4k O-e 6.0 OA * 0 4 -4 6
4N. 41360 >.
60 4 4 4j 4
--
UC .- 4 CO 0 6- 4j
N-I~ .0 N 0..0l6 6
02..-I~~~ 02 6 106V6 0.4 U
'-~ 0~6SO '- 666-4 . 4

4.M %-40
rVC
02I 69
6
4JM
~0, 4~
ma as
06 4j1-
4
Ci
a0 vi
00. .
or- ~~~ ~Q 34C10o6-
0 0qw 041
0 = 04M (
4k 640
C 4 -- V lo
0. V6 0C '.> 0 ±, 41IC1
0~~~> 0.'4 . 6 ooo o
0.34 V 06 0 V *60k00 .4
Ak S 3"a 41664,
.0 0 .4 -4 4j -4 *-K CO .0.02 IO0 0 4
E4O kV 00 '4*M 0 'o

0.4 00
Vr i > *4 4 CWVA> 0* - U K 0C 6 a r C I
6 C M-
10~~~~0 H*00-.a
00. m0 .- 4V6 0 a c 0 0
v j : 40 2. 0-M C.-141 6 4 4
k > 4; 4020 6. i 0 6 4.n0 4-440 ~ K C > e.C
Qu
m
410400
C.
026
-4
64
C.1.
622
Or4--4I
M.-6
-kC
6 -4 V
000 04160-a
6 a---
'-D0020-I
H4-4 V . 4 41; CON4-I .C H g.
W6-~
020202 0 0' 6 .6 0 W1- 4J2 0 H m a--4
2 .6
-104I -43.
04664 06416k
0.-
0
4.
~ w
44 a0-4
k40 054
6 410 k.-41
1..
- IV NCI r 6k C0026,
-46Eoc m o V6
4.woo .6- r
4) tCrc W4 -- 0 4 -460 g-IC *
.0~~ kW a. a. "046 V k 0 4
V 1" C4 A-
41<64- D4% 0 - % -6.606 0 2k
%-- x
O6CV4 0:wI a 6,02
r..36~ v02X0.$ 0 66666
04 0 .2 -0 6026kg Z . I0 %
.- 41 U kOA02s
v4 - 41-4 4
0 0 6) 0 OkH 6 0 ;6 N
C 064-
4402 a-60 4A 1 41 -4 464..100" 0 z 0. 1464
NU40 4W8 '-0 X021 IDZ).
4 0. 604to0 600
N4
' -0 -0;W - 00 6 0 .
41 0- C4.s.2-0
02 .0 X "-40
60.64>0A o -4D a 66C k&4 600
U3 60 to r-.~6-06 93-0 4 ma 00'

, 0
k~ ~~~~~
04Z 0
0206
r4.N
0 6O
00
C k - . 6
.0
640 0Co
VO 0 g0o6
0.-Is, r.2-2 *N-k.-1.4
0 06
.004A0805002$ 1 02 '-za.0 6.-4 - - 04 j 6- 0-4 r k C-
100 64166. 4 00 M0 .0 - -A6 4 6
i4 0 . X$4t VU-$4 J 0.. r 0 1 J 4W
HO
4 6 C 66 41640 .0 0 C1
0 O 410 .02.-
C-
En t' 0 U -4 .002.0 ~L0 "a6 O~
*-0.4.6.I~k
)m0"'
41- -0 41 .4C. 0 m 46 jp 614a
H412-4m60 660.
066 H W V20C-0
4-s 6 64 064) ca 0 .4640
k0 - 44C6 0 E- S 'M6 000r 6
6a-4 0-4604
4 '-V C 4.k
0 6V 6--.M
4 0 0Mw -OC .5>
foe..I .0.
46066 '4 V 41 0 > 6V

0 660 6-Ilo~
U6-C1060 v.
40 V404'
Q2 V U 0
0-
60 v
4) C* -0a)
6-I
C 6-4440
-4 00- 02
0 4 0.-6EVC416U-w4
. CI v >6 0 .4 0-
02
C*
s- +1 0 M. C 410 0 6 V-0.0V6-1
V
w
6 0 .06K.4
r- wN O416
N
41U.
~~~ 02 0 VJ( 4 -C0 a
0N. 44
.0
64 04440 N .I-02K
-N WOk
41I
r-
. 0
a-6 04141C0 -41
o' 4 66 0 CU 6- C
6k 4 0 06 --46
*
w'a s 0 0 0k 6V 4 k4 1 0k 0 I 02
0060.g 0 0 61 60 w2 r2 .I a- 0 -41.6al 1
6
00"-~~~
0LE.
3
6) V OI I :20 k'-" C.-'
20 ~ 0
6C.0 6
KX
0
0 66
V4200
5 6k6
4 .
6
WV ~1
cc m- 6 U4- V6U 6>024
43.0
6. 414
we60 0
IV
04~-
6 02 02 4-4-. 101 41 6-43 0 k > P346. 40 06 k C0 6
4100 "
0426
.1. 6 6 6 -1 2 00
0z~ Q0 000V.0-i 0- k4 E6- r.W-0a03U C
C 1 0 C00A re6-HU!44 40W0 *4 .0
41 3 0 na - W>> 0208 o
C) )0 0 C.- 411v a :-II
-4-4 04141c - . >.C
026k0- 0 0 - C-IkVW O6
02
-4 ~ 66 0
0 02
fe
60.-
0
~ 10
06 64
04110
C6-4) 6P- 4 - .04 0. 0.a -4.
_ 6 C 0&0. C j -- 6 v26 0 41 0 0At 1C6V. 0
r,to>awc 0 j4I 4c20 r . NtC 404 V 664
k- 020m0.0 60 2 06.0
416 0U *-04 0 6-44- C 01
V--I $4 0 V 61 .4
0
z6
a,'
01~ 6 4
a-4 a .4 k6r
6a vkEIW
~~ 1
12
- to00
0 M
6
V
-r
>1
k
V
V6O
6
6
6
.0
II.C1-
J 4) C
4
6
Q 02 6. 60er
dF0
m
Q0.
U4
2 I
j 14
160
.0 A441-
0100
4,
1
o
40
a4
cl
Pr0
wk 0- 6- 02 r 6...46
4 0 -- 6 C
444 0. -. ".~ 1- -.
0 .4 o F4 1., 0.041 Z 0.. 004 0 J4J
cikO
V0 a2 9 , .A A 04" .
,t 14 0 00 ar 5 V
U)C4
R-3
0.4j 0-S
M>.-
-.0 6 4 rS.
41 0414414w 4
0 4 0 04.5
0 4 -M5U
000
.0 A
4 8-o-s
4.4
00 -4 0 C0.S -t a4.4 06 a500 00 .0

In -04t
- 660 -5 .0 0 C*- 61204 44 0. 0 C Iwo .S>
~ 4 -X 0 W1 -.40 6 0 ,.44 0 06.4 60
04 * 40
_4~ 0- C1 In-*430V4
- 6100 wsn. 44 -4 660 4 D.01:
C - 040 66 > U)
4J. V ja M * 0 mP a 0-44
.41
00
C
m0. CZ>
140 w14 a
6
8-0 u0014-go4-00140
4
60 - I4, 00
050.Aa 0
0 44 4 j
a I
-4
a 60 0 r I o a - 4 04 4 .00441 4 C 0 COOm
0190 a ME4 0 0 0410 140 0
0 -0 .0 04-O 00 6 0 46 00 C
0046 -6 '6 So -40.064
u4 0C .44lso 06006 0 6
- 11 CC-00
Q a44A0 C -0M4i 0-400Wa >
0 Cz> .64
NJS 140141 Um C 6 66 06 00v 1 U 0
0.00
m 00-14 C 66 6-6 4 440 6.0 S. %.0.Cso@
aooo0 v M 0 ) 0 0 C4 .46 r. 0 4 0001z44 .
r41w S0
V40 r lo
0000'aM C.a>
000464
4
046 01 > 0
446M% U
rH
1
.I:
0 9-4. 6
6..014
44 04--4&. . 66 6 M.406464 0 4 601 160. 0I C

U4 to- 614
444.440040 41 13 0.44a4 - 0
0--.S-
BO O 140
Z"4
44164044-
.6
W.
43
ow>
~C Z 104M
61
4 >u 06 -4
14 300;-s-64
0)0 0
-41.4k -- =1460 60.0a1=05v ,-4C .1 10 N0 441 444 0 01 0
60 ,-414to6M46 w4 44 CI 0,..460 r--S 0 40- 1O 00 4.
m 0Gv 4
0 1 E66r- M 0 0- . 0 4J . 0>0. 4J 66c's C 14 06
MC
4J~0O
00U044-
AU 64A.4 6 0 . 0 *' 4 b. 0 .C
6 4 00a -46
0- 4 0-0U-46 40a
00->40 1 >.440 =CJ04)D-
0 k 01 -44 60 0 -04- %00. Cr9
14 04 004 to-0k -S 0 660 14 C 4O
re4 -. 4.46>4 oe ' 4
p41 60. 10
10 90> 0 0 0 4 W04 a 0C-4 0 1 44014-1 a64 06
06.6- 0 M414
460644 i 404- > -a
9 6 0.
-- 0146 0 044
u 64 6' a50 14146000 4.C - kO N0 C 44661 0--w 14 66-4
0 V -. 0 6
4h 4).4we06 0 .0-4. 0 614 W3 6 to-6 A 400C
-
0 0C. a0.-
-414u00 0146vc4Jk 40 U-- ,40
4 --4 0 000
04C <
*0-'4 0Q
0>. as0V 14 01000 41 -6
% o 0 0 0 ,4, C6 64
N 4 - C O 1
4141 6C o0 0 1 a 0.5610 v 65 6C 0.41 040 A 0
-4Z440 . 066U- :
C,0 4J40S 1 6 4 060C
1 166M 0 0-
N-4 W004 I 0soS ) -.4 0* 04uOa--
- 6
v Ik r.0 V
044
%-146SX4 N
,-
00.040
02-5
k4
446 060 0
0. OKCOw.4044
046I..4E-
0
14
>
6
4)
V0
6.6-A0 61.C
DA
U06-- 440 41
0-44 v51-40
v---v 0001 04CC 01
4 4.3 0w e'1 0-4600 M 0u 10601 4m4r.0K1 44C-V-41 400 u.-S
.4 IC 6C .5 060. 0VA -'.64 10 605040 $66, q3 C 00. 41 6
0. Kk -'0 C-0 4v 0 606to 4440>064 4 MK. 0C-4 'CV0 V.4 >MCA
4 0 1141 o 000 000.00 00 0 4-4 O41 1 040.0 06.-4
0- 00 a4 0
.404444.CA3-4440 gos~ aI- 0C640 46000 1
Q
v0- C u14 0C
44 0- V 4.) a 10 W-4
4 W 0 40-.- C
60. Q-. 44 06 -C 0S-1 0 14 06 0;00
C r4-. ; CC-4
0w 04 0w 6 0 41M 66X 44006 3 C
I6U4
U)00
13
5
4 0KC
0 C
>-14 $-.41.HOA 10
60kv0
, 40I0-01CM
C 0r 40-
>
004
004w4c4W014)
406.0.00
z 041 h r 0.4 -4 a$ 0 C ' -H 14c10
M04 4 00 > 44 44a
06 064 6 00 4 6- 6 0-444Co 04-
06 34 1100 a
600.60006 04 ,4 0. w6a6v 6 4 000 to 0-.A
0.0. 14C 14> -S-4 -- 0.C0 0, 6 1
44C - U44 u C 14
0 . %A 0; 6 0 00 0 maiCs
44- 0 N V41 >.6 0 14,- 0 4404
xw
140 N
444 MDAI
I0 10 -4 -Q -. %
6--It 004 0 006014
>- w ->0434 V-4 0a-564 00 10 440 1- C V 04J C C C04
000446 0 m54 r- 6-'~0 J0C4 oo 41 C vC00144M
-46.
m 00 -4:P
0 0. -M 4 a 40>6
U404 41V-- %w zI 00a14 6.4
6
o040C1 r ~ 1450
0>0
616
6
r4060'
Ca
C 0>.
0 64K0
'~I!
0-
r644
S 10 q00
.4004
-4
a) 6Cr4 06..- 0 >,- r o 145r. 0 006CC r 040 .

14 100. 010044
P0 01040 m a OrNO0 0 416vV
40 4011 o aQ
C4 0 cS.. 0 44
004 6 00 0- C 6446 0 0 60.3-A
,
0 o
66 .0 011 8 1 4 44 0.m4 10 41C6. 03 &44
0 040 ,4
410 06 1* 604
0 v 0 0- 0 4l4-6 v040 H. 00010- 0>
6-4 0 k010 WA0- M4 - ) 014-1 .1
aa 4 6 U4444
0100 a k to 1444
C 1 0a>0 014) 1 sO4 4 .4)i 0.
c' 04
6 05- N q40a00a3 0 U C 5 N 0 1m
M' S w 1'- C 10a44 8 o,0 .4 r4 N 4 116U C 0,4 0
o600 0 a) ~-0 6 tL40, -W 0 04
14 0-H.1'--
4 0 r 0- a44 14 04 44 I2- k ) - 4
a rC - bl -P O I C 4 6
u0 041, 0.4410
41O~01 04 -- sM00-0
rem in o tH0-C 0 0 94 . >.. .4>
.wQa .C U
c'46-4.' we6 14 041
w Acm
MCIOO 1 H L 0.M 0 G4 UCE 4 4
B-4
H9 044 4114-4J14~ .4 a 1.UO14

4 414. 0 4-
OD 4JnN
k~~ 0 j.~ 0 0. 44 44
Z A M00.44. 4.1M 0r 0.
4 ON 410 Q -
4 F,- 80. %4 111 '0 01 0 ." O
1 ar~10100.-IV
I O>
14 0 w0 0
>."4 MM
g -IIA
oY>-4
4j 2 . 41 a 00.
Mi-,:,
4 -1100
LHON 04me
4 .4.0
GC0-4.C 41 0 0 M 06
141
4 4 041.4
00- t* 01D 0. -Z..e~ I 00 0 N k

)- 0
>.O10
a0
~
0
-14
MM
90;
4
00 0. .4
40
r. V 4
4A
ei03.4100>
-00400 00 146 0 No
41 10.041 0.0V 00 >00 0 -0104
C
4140 0 -4 00 V.0 3 041 .

11441 0,-4 00 coo-0oo
0 4
&0$
000
-1. 0-
0is
1409 C6;4 0Vq
0
000
A'jo
0920 M
0 1 ~ 0
30
$14
0
~4.4.9
W
it%0o-4
.0-4
top0
4400
: In1
>.- I 04
r-4 .--..00 0.0 0. 1444.0
=I O
" At0 .4 0w 9-1 00 D 0.- 014. 0V 0 400 0 e 0
HM.1 -MV1 -42-- 04.0 Ile, .-40.O
t .4c,14.0I 0
m 40J C 0.
-42A 0 0 04J43~0 1-I 0Go M: M w' 0 0 w0 14
004 0 RICo -40 -4 04, 4&0 40 4.4 . 0 14.4J4-0
-I.41
44 -M O 0 0 ) 400 12.4 g 11gge:0 -c c ww
-. V 1.00 0 t0 -0 III,
I.011g- In4 -A0n
0.01
20
It 01004 014001go~ 46H0 0.10 14. V4 0H0. ?
g 000 0 '-IM1 g.'00.. - 4~ .I 0
41
4.
0-U .4. 0.0
C0 4 -4-4
V4 1 0.00&-4
0
TO 0 . 4J
0
~ 04
H@ V
4j ~ 4O0.0
-~- @>
. 0Q4a
0
0-414 -14a0 0 .H0-4411
4 0 040.> MO .41
0 04
10< ~ 44.-
-1404 V44 4 X.0
04r 041:0.I.012
4g.- A.A.- 1 -4M *0
.
14
00
0
-4
04 0
41
0.0 0*. 0>.
..- 0 .0 00.- 1,4 0 N-0 4 0'4
4000 14. 4J0 4 0 () 111410.0
CI -00 -404 41.0401
to 1 'A -.'-400. o-4
r0.00-4 Son .0q0P4-441x--00000 0 "4
%- ur-3C V V A ' n05 4 1
do.4 $0000.0 00C6
c 00
4 a,)
0rl
4 t
I44 - 0.
4M
QW 0.0.
V$45
0.0
0%4i-I MA.
0
3410
0--0
0j 0A01
0140 HO
w 0C04 4 4 0 .4 so
0 >M.
890.0
a%
0 0
VV.0
0 I
.- 0-.-441 0M 00 0) -0r .0% 01 '--C0 41C Q-401"
27 . 4
0a 2
4l 0 01 ,
- a100
. * 0 40 cc410 01 4 0
1 000 > 0 01 0 V r0 - 00.O >0 M00 t- 41 0 0m-
a u2-02 v 1444
.- 1 0.0 0 4 02-00 41-4- 0V 1>, 4 0
.- $41 Q 14004
41 v 414M0 0 10 Va -W A is0040 00410
4040 Ai04 44 0 *14-.000 0 0 440240 0 000.14u
'-4~X.---0 NM $4V-A4)41 >.4 a4 0 '-.44004-0W --- 40001
>.4 4O 0 400 P
3000 0>14 'A
1 >0 > rI0 014J.1
0ZYU 0>0.04110 O4J0 '14il 0 0- oV0 0

- - lN x U4 >.0 V no>.-
0 M 04 0..0.MOD0 0
l~
-. .0 0.I.C
0 0 1A 0. 10- 000
?A CZO
4 00-14C41444 0001 0- 4 >0
-V 010 14
41N01 00 0N -' 000 C-- N2. .41 0I
0q
P.0 0104V-40000 0 0-M '011 C "0-4~0
J 40
.3
-4 0 00040 w
-I I 00.14 00
4 0 44>1-w0000i;0) > 014 1o4)a
0 o o VOe
K000.0
21 0-'V0 OVO
r) w-.CC 04 -. 2, 000 041 -. 14
00
Xfl 40 0 4000000
5-10 00 44.0
0-.U)0 11-I
IVO 4)0 C) 1
4)0V01c041iu vvo -CO
141
4. 1CH . 00 4 >044V14
$M0Q
W t~ -. . -I 00140 a14 0 V 0 rA-'-4 000 414 $01M00
4) 00140 0 6400) k. 00 . W00 9L0-400
000.0 1430)1 asx 0-.0-.0 >.-44 w0 00. -400410
0~~>0444.>041.
004 44 0. 001$4 1410 M 0 14
41 0 0Cf- -V 4-441-0 4410 0 04000001

W-. 4014W
00 0 4J 4) 0
144J 00J 0J
1 -)Nt 04 00 04-414
X40 14
% 004)4 00I- 0A.4100
4114000j~ 4)11- l0 0 a-I0 Mr w0 -41414114 [-0.0w4
41 a, -4
(A 0 0..4 4.au 013 1 g
Q.4 14 S404.
:11 0 S -141-0 (14r
0u.4 -0000u 0- 41 C4 09 4 = 00-4 0 -1.0410
-I
4)001
0 041
40)0)1-I
00.-; 0
0
'a 41a .0 1M401
0 00 4 o 000e
I--M MN 001 0. 04, 0 410 r-ok .1 0 M~
0 - 0 V 0 41V 1 'ago- OZ0 0)> 4 0.1*
0 ~ ~ ~
0V.,0 $0-1A
004Z4 0 0 >. 0 0 i0
1 Z 040-40 0 > 00$0.44
00 .0 4u-0 0 0.=0 0 a
00-004 010411404 14 400140 0 .1 00 0 0 C 100 4 0M-H$
0. 0. OOX 44 0 0100 4 0-
".V 0 004
0 -4 in0_ 0 4 U , 34VM V0 0j04
00- 4t $4vM 000
V0U 0 0 4 CIC 0
4 0 r go 0CV -4. 0 k%0-4am0 w0. .0
.4~40
041
00-4 000V0014014
.40 >V 10..0
0 g 41W 04)- 140
-.
00
.4$
>1
0014 ~0 0
30041 0 4 M
1,4
41V C 04 0 -.44 0. 141 c.1 SO 0>0~ 0w
11 14 41O
Z4O 44 1 0 '44 W.v
0414' 14
in M-.44 14a>1 04 440u-44

-4014 00U-4-M4I
V4 0 0. 0 -4V 41009 : 00*
4 0- V C c4
5
A 0 . 0- 0 >,a C 1 0 0 4 U.I0
m V .000
0 $4-4J 0 0-1 0 k00 0%- 0- *00W000-414-40-
00 00 0.0 0K 14 -'-- .00 a-I64 V0 0 S -0 16'10
r
4V)u0>.
A1u4 m041 010, .Ba~44 A 604 -40- .a
0 00-4 va a1 A w-.40 $ C4 W44>.0 >~ 0 04.4 $
O 0 V w.1 V>J000041 1440 041-C 0 - -0O
.1 U 0 1 0 0 - . IN 00W01 C 9 0
'A.r
:2 V44
0
A41 C.
uM4
j .Q
4 0a co
1A1 G 01,41 0M .0 0 0--i0-0
10 V. C 40 0 4..0
C 100V 4 0 HU
1114
no0.
0009 14 a
4 COX$A
140
.
00 0 0 1 0
I0 V1-01 4 >0nI
C,4~ 414 A2
8,,w.0 0 44a Ig w 1 041 01 Coo $4140 X-.4 A -4J-
w* 0 a. - I -0I 0
-A 5v
0. M
41 0 1,. -4045-
0 ow0.00. ND >
$41N r II 40 4
- I-C 0 J0Ga T %- O1 C
B-5
0 4 -$&4t
N W '"40 .$4wMN to0 %4 t
0 NO 0 0 -.4. 0 c
$
-0 400 0. 04
0 1 .44 .1 $40 0y
m0 I .. 0S$.
OO -.4 4
.- -44)0 c 4)
14 %4 A 0-A 4J A
go. 4.454 M V- V
M0 -4.0
c 80-1
1.
r -a
.. 0
0 -0.. )co
0, -c
~ 0NOO
-10-40
CO 04) 0 Ix 8t
44 4)0 .0 C$ CL
04
'4.'q0 $.F03 -4 -4 'L0 C -4
041.4 0
0.0~~
00
.00
.1a4
04
0 00 ~ 0 0' 0
0P cc40k 0
$4 0. 6-0
00 $4 10.441 .4 4rn C a4 ,
C.0 44
$4C 0 4 )00 * .. 04).00 .0 40V 04w 4J.1-0
04 w OC'
W4 a -. 40 -4 00 0 $4 .0 0 .I0
c-
1- 0 -4 010 w C -4 0O $4 >.4
0 0 00''- 000t0%0: '0 a0CIO
V .0 0 OC0 w A a 00' %
I 0-t0o
C-'4Or..r
41 ., .4 :1 4 $4 0 $4 0143a
-1ZO
A00 0C 0-4-- 44) 0 .("4C IX 0 -'4-0
400 0 -4400 -. 0 u to0

00U 44 O0$4 0-
) 4400~4
OV0I.MV0 $400 00 '-' Z -410 3-'
NN4'cC 4 0C 04 CC w~ko~
x-00- 0q
w H $4
00
40 n0 00q~ (SM-
00 0. -W4 k M04) 4~ $4 0' 41
41 4 cw0 ) ) a-A 0 0 04) 03 )-
to M 0
w-4 M4J -44' 00 (00 *0 -4C4
tr 'A o 4 01 0 y
> w . A 41 t U
4) 21 = 10 - O 01.D M'-40
Clr 0w -A $0 C-to 0 0 E-$4A 0 > 00400

NO -'.0
>1 O0'Dc AO0 -'4
0,4 0) .r0 U4 0 -r .- $E : 0
0c 0%
'A4'4 -1 0 V 14 M) . 0 'a400I
to-A4-' 0 0
wto4
Is $4 Ck 4-4
V~>'4$ 0 V)0$4 4
44V a
$4 k 0 q4 $4
0 c c 0.
A
M .-4v 00 0
0 4) 0 M 0, 4-
NN "% 9 m 0 4.0
N 400
u- Ac)
C COO
CA M4 J o 4 -4c OD 0 N 44 )0>
FA4j0 0 -.0 0 E 40. .
40 ---40 0 0w (a s4 r.0 - m .140
)
U $4J C CO Q 0 0 U 0 VC4 0 3 C4O 0 r
to 0 0 0 w V)4C
- 0 . 00 000
X V E - >, ,
-1 0 00 00041 . 4 01
:34 m r$ 14 0 CA 0. $4 N $ 4))04 ''.0
0' C 0>4 0C00IA400 ' or
000' -4. -A0
0 .0 . 4 4
0.
-. 4 . r 0 . 0c40
a,14
r. 0 44 0 4 ' 0 .4
0c 00 0 $4 w44 0>
00 0
0).4.4 M..fa.4-
0ua 04) m 4))4 F>'.
0 V' >
0~!
W 0 0~C -'0
Z Z .
144 0 (3 M.-0 00
0
0 g 00'
0Al4
Cl. 44 -001 4Jt0 0. ) 0 W3-M1 s '4 0 41 0
I &,a -I r. " 0 o. 4 0. 4 0 4 )$0 0
000 0.140
00 0 C 42O0 $4I- to 00 0
'
-4r.4
C OU 0 ZC0 0...
0 4
44,
U 0 .1H 0 OA 0 m 014 V
0~4
N ' H0-4 A X ) 0 0 to )
$40 0>.C-4C-C-D
4.1IA 0 4H
a.
0.
0
~M0004)1H
-Cl
02Z0 $4
00%f
CLQ 1C
0 14
000
4140
CAz 14
c
c9 y
0.0 P 4
O
k
N
40
'4
00r .0)4 to) 0.4V0
W 4)14.000
t
E0 > V( Vq 0.)
-
004 0,$ 4 - OO'4

E. aix 40 . -- 4Jor-r1
rC34-
' 'Dto M 10C4-E . 0-
'4- O0. 0014 4 0 0C
4 V ~ -. m. W 4
V1 U 4 000 1 04 0 0 0
H r
w14 '436
r 8 U cr0
0 0
~ C0 $
.0C -0
0 a
0 to
* v" v0 > A)1u 0
440 0 ' A'C 44 0w 14 4)-4 o4)04
-ii
-4 0 1 040 40 V40-4 A0$
244 r~ IN) ~~4
', n -0 41
0o 4 w0
t 4-
--1034 3$
'4 0.>A
tr>1 -. u 0 $4 0-- C
c CL.- 14 0 a OU . . r
B-6
41 -- C.,0 1044 0 '06 54- '0 4

4 t'o
0.6.44.4 0 6'0; 0Ca
H 6-4 M >. . .45
>' r. Q N
4
C04e
04 5010
6
'4C
.~
~ 0 +
"a A .0
00 I I I
0, 01 10 50 0~
0
N 0O2
ow@
0.201C
4544 ~ 00. 0
'0 .. 4
-4 0a 4 5,40 '0
A ~ u 6 4
4. 00 640.
m, 54 110 Cie6..
641 04 44 0,444J4. 'A
f -.4 4j9 1.i, a.4 0 A
.. ~4 H
04
0
w
64.
co5
0 R. -a =1-p
A
U
-456. 64
0
10'
6 6N
0
0l
4
0.
6MM"
a
f
00
040m
A
> 64 06
&44 56 ~ '
0 0 00- W4-
.A a4 -0
-444~4...4 4 ~~~
-I
. 60r ~ ~ 14,1 4J'..4.
>.
~ '
A
46
a
V0.4
6 - yx a164
H4-..--
0N 0.-fm k0 '
2M 4-41J440 C'i55 a t 0 >It SV 066
I6 4-150 04..0' 0. 06 A >-I a s
CL 6 6050W
6 ~ ~ ~
006
00
~64 ~1 .4 0
0 C
W4 > 0
5 0 6-0
2054)6-0 00 4)00 63
009.a 56(. C 540 M '-40. 4'*
at
6 0 5 -40
00- -0, -t 0.4i40.4 0 00f k'- 06 04
-'6W604 .- 444 %4
40 C 00.
0l
0-100k6606 VJ r 2 1 0440a041
6 6 44 0 '0 6.
04-0 6.Oo
("40 0 Q 5k - . - k .-.. 4 4 0 0.0 4J 4
04'54*-*.o 5 14 ('4a
eq40 6
r, 00 0~'4~ 00- 66.45 00Si 0og4 1 14546
0 014 aS 4542 0 00 44 64 0'C
M0 Q00 N- 14 M 5064446
0 4 ark60.1
".4 04'5 ~ ~ ~ 045 0 ~ '
WH
0 6 0.
t-540 40-1.2
0.0
a%200;- a 0~4ev . .' 1~ 4)0#-M
*'1'4-'-.4 0 $.54 6,4 4180 -5 0~ q4 X-
60'- 04 .04 $4
CC. 54040.C c4 $4 :3
544 00 1.400 .00. 10~41 V a,0 -'w'M'c4
6c ..4 64
.c.-O ) - 5--4 -6.0 'A,0
-4250'.C. k", goA 3,2 .4 H> r. CQ $4 . 0
,a
04
0 6 e
C
0 U H4).--4 Och04 g1g 4 41
cm
404.
11.4 01 4.v04
2646260
-f =3
5406 4 6
0Q
2 Ic 0 0
4
1-I
60
C3l-4'0
00 6-5
~
64J44400 0$4604'6 6 '
*0 U0 H 4
'- : .0 1 V'5
4 4446 a. 02 0 54 4.4 -400
=1
06 -01 co to 4k -'
U-4-6 560 4 . 04 -46 0)
a -"
-- 4) ,0 v H a60 0 --0
-0440201004M442 644 -*2000!481 41 2 0x 0.>0o-4 0
UC4 4--4---
u' r 40 44*0o w -4a'
0144 1040 060 o"'05404v C104
C C4O
0 00 0
M
N0 0 C 34'A
4- CM0-.0..6.
' 54a0 ClN %A
9m-,0( V'
600 0kCa.4.-4 C 4)$ 0 N W-. - 0 )3
11.4)4
0 2 4) 0 " 4 . 0 46660r
>- : 4 .4, 0 0t
0M E4050
0 . C~ C
.. 0146
0400. > 0 0 W
6C0 00GA
E0 w4-('40 0 0 M 0C 10 05H
M L Zv 0-0. 60
m'0 6540 14. w.4 03F

C)
0 C 6-4 >
014 0004 MM 0 O.
$40V0a4- 0 0 SIO 'Ia 04 0
V. H4 0> 6445 5460
06 , -0 w6C0 440
0604
00+C0 0
.0054 -0 1436-i +)
C ' '3 w6
.C 5 E-
0) 6- go .4 4 W14
05 1
0' 60 0- r60
to 0 041 M .
440 44 60 041 0qk0 0 a0 0.
446 5 540 0 A4 - 6 % 04 va,)0..44
10 01 a 0.
0+)
6.-46
u
U4
0 .0 ~ 4'. j- 1. C r.
-C 1,0 -'404t44
F46. 54 446 '0m0a00
'.44 . a
%4 90 CA'
Ik
C.40u9
'0.C4
-4 0-060
~~0. . I-. 6006
014'0
Z 0 OC a6 6 046 54 0
655w 1. 66 *6 0 400- 635014'-0
'4 6 6 CO 0
4- V 4W -4 0 ' 0 A '-404 50 - 0 - -4 0 054P4.
4 14 C 10 0~~~~64
V4 C
.004 4 6 4 M. a 5 a46 4
Co%
54C V N6.4A3.44$ -4C0- p6 6 a 0
03 60 +). 664
44. 6M 0a E40 6--0P442 1. r
9. 0 0- , _
069 54 0
A 6., 446 +)5496M'.
'A'I440i'40 09
0.6 t645 "Is a
6~
'4F00is 0M - '01- 4. 0 6W.4
S1" 0545 60J aC 3I -
44.0 '6 0. 4 5 6~ -0 CA
04CJi 0m0. w-A 414

so~
00 544. 6'4 a650
"05I. 0- -4 0- %04
4 066 6-4' 64
D2~
5.4~ - 94 ~ ~
a4564 0.4 6 A
146 0
-1 No a 5ISO
a~ jiP-n- 66
C CM CV1.0 1,-
-r Or.re 0 106 0 0--0 o40.N
0 *-AC00.Ca-11
aC a .... 6 14 - . 0 wo C0-4 4
>k4a0 -0 0 Cr
04 0o w-. ~6414644.40 k 0 > 0
C-0 0 4C 'a 0 44
_C2 Q .0 4J a0W
0 06a V W 044 6.- 4 wkC
4 4 1 ca4s
a4 i0
4&4. $-4 -. 0 =1 6 z2- .4
t46 k 3
0~
4-.
00 6z0.0444 a~~
640k*44kCO)-
0. ~
10OCOMOOO ~ ~ .0 ~
4
.~ 0 4- AO -4-
>"O 0000
t .000044.k0
0-10
-
0
44 c
!'-44 4 W0- 01 44 06.10014 k C-4 >1V~ 4)
4 M4U4VW40,r044
0. 0 XE W to.oA434 0 0 S. %60 MOr-14 C6.
4 ,0)-
6006C
V'k614w
~ 40 0 010 $40 tGo
A433k04. ,
0J
HC
CH'.4C0 -4444J
to044 0 Cf*4*4(
Id04.
0.~0.4 ma0Vto
4~Ok-4
~
4 0
14 0 040a a
06.-k.40 ZO4jI.
.43414 U0. .40 60.4 4 4U) 4 4ONO
V.4-- 6 - 0 6- 4j $4 04X~
qu k4 .4$1
0.4-
A3V~ al .owea
r W-4 0 4 1 C 6 6 1
w-44. a1030
C,44.-6- *0
li t.o r. 4o .. o-i- V V00 '-I44-0- 0~0 I 0, Cz 00- 0 1 0a40.4 4
00
so 44
a 4 61"o 0 A) c.-. 140 4Q 4 -Cn0 -4 C 00 0 6a
6~ k00 0 0 V14 a40W,4 w041 0 .0 c a .0 0 0a V C00 0
NW
k 1
00V*
C
60
63-4X.4Is6--4
-
0-00rfi
0 4J$
644.644-04-u
~U 0
-4.o0
-. 0' M
uCVON
60.C
. 0A
43 w -0k -P
11" w
00
0 6
0V
C0
C 06 0 6 M 0 C 8 0)-44044 0k
2
0.30 0044.CC 6
40 V 004-C
6C0 0-4 .
1" 00.
044to 00CO6 40 40M0A04 0-
k4
0tw 0
0 00 00 A6 00 ,2 .4
-4 r,4 0-40-
44141 ' $4C 4 V--40 0 60U .- 0§4)C403
4 -Ck 0 0 4C
14 Id.-- W, 4C V W
14 c 0 44~0) 4. 004-M
04 04 0 0I44 0
0 4 3 0 041 00h11 D,00 - >10 4 " 4C0k 0-40 0.M
to 20 W 0x 4m -4 kN-4v A 4~ W a03 VV U'.400. 0
w 6 >* X434J M-.A
do4.0
t0
1)0 ' - 4 i 0 4W4 60 41 0k 0 0 k (A ONV)0 44i) 43 4 00O0-k4 4-4 4 4- 4 O4 0 -. CO 40
C 6 60
) ak.44 0 Ch '4 0'K 4 C"0 O 0 A c
10.-(a 0W -6V2 3 0 WJO00a 4-.k0 0w 4-00 0 004 -4 44C r. -4 V0k. WO I0
, 00 3 14r. 0 3 .1 C W 0 0.40C 0t 4. 0
V.. k64..I0 0 >0 U V44 334-4 00 W'K4 0.4 6 043 w
4j 4' N .-44C1 4U)WC4 44 04-4 44 0 0 -. 4kA4d 40r 6 (n0 00.-m W
-40C ON 00 W-43.V00-0 >-~' V- 010'
kOV 0 3 400000 . .
4j% 4 , . 2-4a0r0 H V401 0 Ok'.0. 0 a 4 ' W44. 0 0.00.0
W
0 0 404ul 4 No 0 > 0)a14 0 w4-404 0
V).0 61 0r'tV4 VM 1.a. 0 0 ) R-4 0k00C4.4 V C 0: 404 *-46 ,
F3 1Ca4'-1
C40 kOC 6 0 C404.C46,00 *.04 IA4 6 H0V444 44.43 0 04 044.8 * 0o0iA
('0444 r-.-4 C -40"4 - a. N 04 0 104n a-$440IO C
W.E4 00 4 r0 00o 0 14 >4- 04 14 V 0 0 .10 4V 60 V 0.a.
1%r ) $0U941 0..-4~~ C-.4 0VC43H.4, 0.0. 4)O"
44 $~4 00 0 V4k r. 0
Ln142
w4 ~0 V
-. *0~
0
40l3
- 000 *4.4
0
z4400.
04.
to
0,
toO
-W0a
WVS
.-
k O
14
N 0034kO
4).-44 0 k 9C
4Oo W-11
0--
4
CA
X
0 44 a
0 C-4 3 O 060-44J 0
43. 0 k"4j
-144 00
44430C4 .0-
6 04 -.
0) 4444 a. c 0- 0 *6
w4 -4-0 Ck)0 0 4 C0 .
-' 014 .4 0U W w 6
021 0 0-404' -4a C 0I4-4 *.42m
00 614 10
n04 .O 0 14C - 00j40 CL- a- 0 Q 0 %4
.00
N10*- 10-l D .4 0
k LO. to 0aE-.. 4 o"-. 0 o 0 -6 30
gor w44k~ 0 V 140 W'4. V 41 .00 V W
4Jfl~0% 40 IOU we 4043 M WA4 . NO 4 0 ,
0 C4'-Q' 4 04"a a VJV
z014430 Qa;6)6r4 100' . W' 0 s-I 004 *0rCA U00U)4-4.H0
.4
C3?.0 4
~~ ~ ~
00
W-.- 6
% ..
10$~k..k.4.O0
>IN--40-4 0 00-0
rk
l 4j c~ k-401-H40-4
4 (3~-
440
fk4) 04 -401
0 01444 3-
t)oO a.C - CV.
0-.0k4
00
4-
-- 4-4 0 64
00-.44AV44I-.VA3000
04 100Z a04 00
M-4 0 0 1
-00
E. M (D c0 04-.
U)W.-414~~~~~
4 644
4 r J 444.
0~4.
V
0 tjN0>
400 ~ )-.C
0 44
4 6)
41 kOHC6444.~~ ~ ~
40400;-
*4
0k 00400-0104V
0-'-.4 0 4C 00.4C0
0.040OOCI 14 440.44 006-4 w10- 6Oq > -M-440Q6 Vk14 C C 00043 463
w - V U 43 .U0k0 4j0 0 0 k-0 4.- 10)-10004 - 04C4 00 C $4 VC
*- $ 04 10O~C
CO0 - 4 -A , r60
-14 .C60 0 -106 .4 M 0 i
sm .C O0 V OR Nr-- 40C0$44-a) 910 Z-0 -40 014-.to -460c 0 W V- 4VaV
43 04. 10CQ 4 r A 6 MW &14 H rA C-414 CD414X '.4 a0 C:Vk4 C
14 A '.H -4 0 U)-4 . A U- 06 4 aV C D U04) f 0JX$
Z 6C9 NO k-4 o A06k 430 V4z9 i >4 0U0 A 4 00.
4
r. C 10 CL 04 - 40-.' .4 60W41 3 a- 44-3q-'V. 4 4-.. 4VCK06 M
z<0 00 140I 000 a
04330- 00 CV14~ 4 4 A 0-I 4 C
v4' *4 40r.-i 43 V A404'. aC'.-. 44 !V- q a &OVr
04*- -M
=33o~'-i~ 44
W C 0MCCto 0
w A 01 4jM 0 -M 0 'A 0'A a 2; 0 n 4 30 0 4 aC'4) 44
44 .r: l0 C4JV) 40 0U 4.00-w400 60 -9. 04 4 > 0 60 M-
0 'n
6) 0 CDVCN 0-. 1 , 4)0 , a.44 c v-4 4 .00440k 4 E
C,'o > 00&U- U4- 04j Z)W0V0.CV4>30.'4.C1S04M0"3
C
43 14U)C- , 0 3 0 C V06
0 0 00 000 0cc4 C >
C H U -4.M4r COW 0.00. C16-.-4C . d an H 4H CO -4 -01
0
144 $ H a- 11'G
C,'aN . toW 0(4 3 o 0 0 r.o&104<44 WY - 04J 0 toM..04 M0046 -4C ) %
M10 j 1 4 1 0. toCA- 11 014 .0I 00a q cQ 6 -a 0 > 0300-.V -1 0W 0
40-1 4000 0 04 1w66)w H4 40 m2 04 as 4 4W>O 0
140 -0 0 . 6K0'000-604)-.-E0
,0860-AC a$ 0 -44x0 3-4 CD Z 0A 3u = c a 40 >r4
: C- 0 0 44~C 0 C41X >'-.0 0N 0p 0N00 -40
E.0 a01
6-4O-4 VWL 4o Aa J E t 4 a 4 040 ""C x Z 0A
04 66 440 .4 4) 44 0a6ko 0- 6001 40 IV0 to 03 01400

6~~~
V.1 6041 ~
4 4j 0 V 0U~O~ 0)1
00 44
w.4
04OO 00
-6V 000$
-1044
to1441
No V C--4 60 1-IX. 3 44 02 -. 404 9
-j 0.0 CV 00 4j4 -. -4. .IC -t 142.10C. v0
w co.-
-3 444- 0 VA 60 0 06 C. -,14(A0 44 .060
6C 4V0 -.-.-.0440 m~ 0kN 404 06 C1

- C)O . 0 4-.
. 60OI 44.000 -W - ZI 4N -o V a r. C
In- 0
0j
**.4
.43 k- ~ 0
r:,
-.4
4
w'
c4 RI
1.4.
3 4 0
0'4 4jz
-
40
14
66
4
O 0VC 40 4
a0
44
14Id
014
.4.
-4
- l 4 QU r.1 4004. 140 640 144 0.
4 I Cfn~r4 .V~M 0 0 4 C.i4 CO14 441
4 r.
140e1 weI.an
C 0 me. (00 104
24 00 c
004 0 V0~' O '00 43 6 4
00 44 Q co -A'
t0 C0 -4 -
>C1 0 04 0V Ut
0 01V.4j N(4 A0 0~ $4 A A p4c t
-41404'-04- -.-- 4 6 0)4 to a444 04I J-.4g 3.
0 N 0- V,~ 0 a4 z4 0 4. t0 0044 00C
0 QNc ) 4 0 1 J 0 0 41- I440 4149 0
N 0,4~. 02* K 14 31v v, 0 .4 0 0
-'414X N4
C. .404 14000 -M 004,34 --4 0k -4. 004
4j3-.014m4-0-4-4a 14 0 -kr .'~ a- sl
00 . t-. 14 r.4 0
00 0 V-04 4 $4 z
0 )
.4 V14q 0-.>444 401i~ 40 0 -010 06 063
CY4 -044
62 -4 0 k104 4 0 1460 444.. 0A30
C0 *0 4330 60E 14
O c~0
-44
0 140 430 14 C 4
0 4~ 304404 r is 0 0 o3
61 0
110
006W0V
I 61
0 4 ~ 04J
4)0~M4 -
0 04A4'43-V
-
W 0).
0 w 0 4J 0 0 wW-
to6.40 c 10 O00 4.) w41 too 90 0 0 1400H2UZV
0
NON
4410 A4O c0 a0.W0 -4 a -4 0 $4.
to~-4 Ins~> 4 .1 24 Sam 0V W$.
0 4-4.4 V44440 f4 0 4 X 03r 0 0f0-I
00 14 4 643 4 1 04 1
1'.I603E
44
-4
4- 7 4- W 044.-4 - a.4 ,t
.4-
4 tor IV 0
0.>"044 0 '1 0 ) 0 4 c w m00
4 14 4- 0 044 0 -40-
W004 -- u s42 4 4 .k44 60440 W 06m -~ %) 0O- 014- 60n1
)'- 0 0 V I0 v4-V4 0 to-
-. 031 4 4-44444V414 44-44-4 0e r. 340 g k
B-8
04.4
.4C4 0 44 40410 90 04 0V "A 0
Cc a4 4
41
c "40~~040Mo0 -4 At
04 .4
0. S. 0 144
.r 0 0.444
'At - c.- .. 00
44~~~000. g444 4 *
g
5 0 -.
44
O.
C-SOC
IV"
m
90
.0
0-45Z01
00.V
11.0i
0
O0
I4Sf
0~
002,0
o 410
''UI
0.40
0
.0
140 WS
c
0 a'44 0 . U
C-.
C5 k4. 0
0
m
'.0
%0 I
:zN
k;j; IC.0r.4
V>.
C1
01
00
0 14
44 1,
4400
: 4m- tpfl A 80
4j049
0 r0.4.4.
1
0u~
~0
0444.g44
-.>00"40
0A
4.CW
k4-4A00-
.14
0 %4 '0i
x 101k
410J4C
00
r4 8930
040I
4404.
0k %4 1
-A0 a.044a.
00000004 -A
c 0 00 -4011 14 -41 C
0C: 0 4041 0 144~00
04
%44Cm,
04)4.O ~ 0- 11;
4>4
4 *U 0 o0 k 04J rS4C -
10 N O- N0 4 _ CI" 44
0
00 ; 002
4 al~
.4J 5>
N
84
HO 0
.04J
0.
V1
mW
-A0 0 14
-
0 -M>0
0
0
V.04.1
4 4a
01
-41 0 C
C'.
0C
a
44 0 C4400,400
.. 4 0 3
1"4 aa.qk
.4 0 0a 4 goo
4 >M0 10 0 0. 4. A t 40 .0C 0 g4C -00c
000~~04 0.'4
084 - 0w0kQ
444
m>1m 0 vtoV
4
I = H0-
to 0 44)& CO 0.00 ,00Ho 0 %440. - 2- 04 4) a
Co 00 4
40
V-4-
H 1-40. V0C004444V),- 0.
-A C 6 0.04
cc
-4.4
4,0 k m 0
-. 0 C6 1 VCA,0
f -.4
*4*004
0 a.,
Q1f4-.q-o
CC040
40
004so
4 0C
440
P
M40 .C 0 01- 1% -M"a P 1
>
4)0 404
040 40
0 0
~
Co -.4HkO-
4 NO-.-0 -M0OC W .14 4V43t0-i444 000.O01 04444 0. 4 *-'0>
a.4f0:-- 100 0 $ 4 0444-4N440
~4) 14 4 r4 $ C 4a 0A 4 , '- 4 41~ 4 jq
r.0 % 004 oo 1 04
V V 0, - 4 C.
at CV4 44-40 04.
m a0 0 'N - 3,toM 4 r 0 0 V440 4020 =A 0
C".4 m 41 0100.0-.'
eq00P 0 4 0- 40 4 0 >.0N444 $44 = 0c r4ON 009 .o0k
4 r r 1 OC-40.0 .404444
k _ )I 0 0 -00 %4 44 C3'4m0
4
-4$ NON0r 41 410O 440A 044o 0 k4 40C E-444-0 "qq k r
C--0... 4
0--4)1NV' Aa
- 0 ONCw0
a VCO44r. .. '. V '4 -41 00. -GO' a
a0 0 0 0 M00 .1
P 44N 0" H V 044k C 4 04 No
44
)..O-0
44030
C om0 0.41 0 *. g C On
i*. V *40
v)0.4.- 1D o44000. 0 04J4 CC1a U'40 4 Ac
0 0
.G 0 10 4.4
00.000_ -.j 4J
44
4J0N-.4
04
H
C0. 0 000
)0
r-4 a 8 '>
-OCU
H 4 MO o 14 0 0$ ._
C
0~~~~ ~ ~~
CN%4
a- ocC
CI W1
rk 10001 3 4
00301 0 0 4
H~ 43440. 414404
F 4 49 c
MM
4 0 9
X40-444-.4
14 w v
Q04 40 0 ME0 - 1) C00r
0 c 2c 9O0a 4 ,a00030 04 0 C A000.4402a 044-2 -00 0
4 Q4 C-0 00 Hv0 0
..0 z I"4Z
-s r, C 0O a 44..41- 0 4 040 440044000 0
- $0) k 4UM 0 4) 8- 0 V 004 . 0 414 0 0Q.H04 k1
mt4O
"m34 M 4 0 $4 - 0 0 40k 4.3 4 .4I
4 90 C00(a >H. 0 04MA4
4) 06E- C m4 0 r 1 - 0 '
In -P41 - 0 44 0 00 -4 *to
U
a.A-H0 4 140
%0 84
Cq 0 '04-i WO4 0 -4 4 44
4 C -S 0 ri 044-s 40 0- 0 " 040k a4.00 14 4 0i r
Ic- 0. "40 4 0 so
m W~~ go444 0' z 30 r .t"C
>. 0 x~
~
0, o4) E4a, 0Ic -0 0- M v mk j0=
14i1440M
4440
44ON 0 4 tsO v 0
AWA- a0
00
V_
c q 1
E-(Y0 0 0% 4 4a 00 4 4 0 >C0.
40A440v W40 >
fC
K 04 f0 0._00'XQ3 r0 jU
44U 0 0. 00 4 0 04
0
44
-A N .44
- 00
4 00.
a
44 0
0
0 04
"4
044 C
"U 04.
S4
*
0
4.
r.4 Q
0
0
UN
04 4 C00 0>
0 , 44
004.a4 0-t 40 'aN44 0 4 44 .. 041
040
a 4-k 0 v 406 v00 0 044v 0 - 4 r
0 40 4j 0)00 00 r. 1 040O Va 4 0 01-44
H-5 'at
0. -4v0440
00
00
.-
1 ,4 0.a 'L
0 x
00
.
",a-4-
0-V4
44
0
r
11 .
Sr 0 m4 t's44 r.-$
14 ,j I p to Q 0'm.44
M0
-'4060M0
t
04-
04
V491, A444
Il N 00
5
a
4(A
'540 to
41C o O0 o0.
0,4 -.
01
1
44r4
440
>14 00
440
W44 ~
- MN
4 0
M 4' q0
30.8
a043 u a k4Cc
"40
to'.
0
00
0 44 V 04
4u 0
0 to
.>-
010a0
Z 0 L.
01
- C -4 V44'C 0 .0 ft k 00
456 ka
0 M. 0 14 0 k 4,a444V0 -4040 "4 as5'0
41.-SC .0 C 9
4C0 00 001444toQ0 C 15)4C0
%4 0 40 Mo. 4 %'0 004
NO-v
4Cq.
0-fS W
-SOON
or O
044
soa 0
C
44
4
4 N
40
44 41 J
ty0-
~ 4
04
020 0 4 9 4.
040 M 04k. 0
4r.5-
00J 0. Co-s 041 9 40a
4..4C
v-. a 0
k*M.4
00 4.4 4 4 4 43 au * C3
31041@ 0 >
I'll, I Me
I~4' 0 (A~44
!' ., m ,. . j " -0 AlO O400 C V 00N.
gn.' goo 0 0U44--
4- 040
A 9M - 44 .1 OM 0 0 1 1
0Mg0 4 00 0
0. 1. 09-' 0- ".041 C O
4 0 au 440 04 0
44 A4 000" 0 0 - j a4
810 m o A a 1.0 0 0 4 4 k C u3 , -
U .1 .)0 u aIj j &~~V4~4
0 a4 4 1
B-9
M-0 0 -I.V0'.4J
40f -Or 4X44-4-lIEU1
~~~
El-- SO
ON~
.4M-C
O
x
*.-.0
~~~
$40)I
0S
$4 -- ri
1440M
40 W
0 a.CM 0O4114 C
X0j1044.4
.4
. 0 0
$40 00I
044 j 41 a a)S.0.. aI 44 Ue 04 - N 0 41 *$4
0.40
'N a4U~ I
in03$ ouE$44)
o C,99aI- t 041 4) -0 a A uUCS04
01
0C4.-44 ' 41) 0- 0 -4 Mo 4)
0, ~ 0 C.0 4 aECa
04 14CZ
M 0C
5 0 a4) t 4 00
41%-* 5
00 r
0'4
C041
0 V -1kU
V a 0N
o
4JI54
SC to
'4j 0 L
V
gE0
04.
~0v
-A -A a
-00.45 4 4 $4 I 0 O0
00 04 0-4C4 w 4 0 650 0 .0 ' >..CC
O a0.4 a a k 0% 46.
00 0 w
4) 044 410 0W 4) 0.4x0 W
41 C 0 ' -A445
0
0-
0'- I 4
~ ~ a000 4C4$C
0 1040
do*
E 0 0-4
0
a
u
a-00-
VO0 gr- 0 40.
4
C
co'300: -I40 04> 0 04- $44V04 0$
"4$4~ 0
$4.4 .44, ) ..U -0C.00 04H 06 .0 04 2"40
40 004 U 0 -44 .4-k4041 ot
S. 44 > 0U 4) E.-I .U I IU.C.1
1it .100404.0 14
.0000 0U$400'5
r.I 1 0'- C
V.- 0'.
, . v .0C-Hog-0' $41E CO
-. 10 V0 a 0 a.1. S4a $40 0, 0.4UOO CO 0 J)4 w
O 01 4 , 0E. 4jU,EU
C''.0 rQ 00 Ar. fMo $41 a A
a
050.4.00'a o4 40 0- 0 0. W -MU4 4 )V4
00 p4. A5 0C I0 E. 0 0 4
o0 - CE1"
w' .~ 044 0 0 0 44H 44
0 -)- U0. M. 4) .0 0 0 4pEA :04 0.
X E : -4J
14M>..0
V US00N$
4 0 u &0 $4,01 5.
.0 0 (. 04C4 040CS 0 % -4 v
.4 04,4 54
*4.10044k n. a$ 4 4 0>$4 ( $
$4MO VOUq~~> - 0C
4A0 -'D5 $44 a .4 $~-00I. a00M40 W. $4
$NA *-c44 -H $4. 0) go0 . C 004 01 0$V .
44.-I *0 rE $4 4 0 A-4 0 10 0 0. moCS C '0o; 4 Ac 0
CM O1 .0, -4 M 40 4010 400 Cr$. $40 1 - [ 0w Go CU8
,nC4 a j- 0 $004~ 0 G.400 0 0 C1CS
0 to0 3C . ic C' aI~4400O to00 4 h41 44 0.0 , 0, 00
U C
r w
NCrI$4 o1104, 4.1)4) r44 0 0 rI 0.4 $4 J
I 4% N 0-.4.0 0 000 0 0 4 40 W 44J0 40 *. 4 1 505
40$4fl-4
als-. oW0 -
004)CI MO4- VC V4%40-d
UV 0 .4) CS
L
-A)0 U24 W44 1 5.0.0
00a..
4
(A0 VQ 0 H ON0k444 0 f 0 -- - 00C $04o.1 W 0
t -4 V
.AM 0 A 401W $0U 0. 0U0 C CE r
00 50 C.
]
04 a- 0am 01 0 .04tE a :0 00k
Un >-w.- CC.4C.-IC440It0EC C 44V 4 .4Hw.-S CC 40 -
1445 -0 DN w to.04 0EA0.44 0C u240O 301 U04.1V000
C40'b . 0 to1 0 A0 C. CA'a0
0 00 0 $404 00 a
0 V as . 0 5>. 40. 0 uy04 SE o $4 02 0.4 H0 9C
J 4.)0I
z k-403-.4 4 $0 . CC1
E$ 44 $4W0. C4 44 E44 A 3. 40
NUN'.400 N$ 0C04 0V. Vp
Pk00 0C4 %U%0 00.0 V 50.0
0 zN)
ty.4-I--0
.
N
4U00.2~.-I0O'P)
4c
4 a
04
E
00 ' Mo.44-
4. 1 - 4).C
; 14$40 -
r a
4$'a 44
4 $-4 104
4 $4$4
>.a4 0
N4Il. 4G-0
0-z P a
$O40)
041-4
'00
00.0or
4OZUUV)0
.0.4O
C-4.
4
aE .
4)
.C)00
0004-4 0
40 .0 V 0.
a4-4 a$...IZ 9 .0 0.01 to 0 .4
c0 4.4$000_J4'.
01 0 w 111 1 . 4E)04) 0.o0--MC0 0
0,
I.4MC r4 0EA0 MOEES4EU4
04.>~rf- -10
0.0C040 000.M0a) $4C4
E-.0 '-4 ~~~
$4$001444.444 04-0 04000
41l 4 M 4C-4C C4 V 0.0*

a 0 4 C
$00 .C4
w' to4il)0 ,0 T .0 wF $4V"&0A0 0 $4 0 V 0'
k E4$ 0
a' >01, 00 c 0 . j CA)
Q 1 .400 oo U $4.0 0.
0- 1 0.4 0 $0.)$ 0 .40
H 01O. owM 0 0AM.4 .10 C-04 $44 $400E-0
0~~ 0.00 .4 N .$ 4)4C4. $40 4

C~~
44-U$404 U $4C .0 UN $4 0
S 0.4 5 .4 I00' 0I)(4. -40 4.
4) r. I1;4 04)% 4 )0010 -M 0 ON 0 14C$ 0
CC ,
4 0 M40.04 VM0 0' 1,. a#40 : F
U0 .V404
03
$44t
a' 404
U.4-4 01 0'w0
.I'If4.
4.p4
4004)r.W
U0
0
a'
Q4 r00M 404JA4
4) 40
0
5405N 0V0.I4
M4
4) $
4_0
4
A
4 . 0 04400404
A* 00 C 00C .00-I> n q00400C 0 4 0
0 ) 0 400I C-)
4) ;:.0 9-4))0 U0414 0%4Z S.0
04 0 M 4A '0 UO.I0)-I0
0 00 -)4 *A
.44J.0f4OU 0 0444)0
$4 $4 0E %.E 0 0 -4 .-40 4 0
00 M .00?SO
-
S-IMCS- M'0 IQ a SP0O44h 4 0 v0 .4.a
04.
U .40 4 O4 4 ) 0 . 1 2 444 w 44C.W4 0 00CO.
.4 *~4 .4-C $.U 4 -4 4)0 104V0 v UO0
V) .0 0 Q.* 4) %04.1C40r 0 00 U001 4J. 0 0000 .4t
0 *,4
en0-$4 0).--A ) 0 - 4 )0 4441 00. .0 C44I
V.4aso~ ~0
a00v .M~ 0.4 4J 4 * 41-440
0 J.C
4I '. 413.8 0. 001 ) r In t 0 4$4 4) IV S4
3 4 0 -. 40 4).0 A W $4 40 000 4C . $
00 00 4 r E v 0 $4.a.Cw -0 .4 04 11
Ok$4
& M
0
~0"E
.
0...
04
44)0
0-
s x 0
0$4*. &
00 4
)0 5
14 0 .
40
'00
00444
0 'j
00- V 004 -. 00 %444 500 1 0.4 .0 0
0 00 U V0IV O
04 0'0-I
m a C 09 0 P3UV5 401 V JV4$ A 444)
40 k.C.f 0.0 14 4-100 k 04 vow Co 00 0 4) 04)UI
4001 401At4 a a V0 4 4454 C. w4
-.- 0 a 0 q 4
N$00 W0V0 A040 In 10 0 1 0A %
400 000"4 %4.1; a4 -S .
00 0 4).04.1 0 0 o0 -40 c. ou'
M, - 0 . 0%q C)4 RAN.4 C .
$419
",0
aori 31In$ A r 4 4
K 0 404 0 ,40 r 4 ( $4 -M 44 9) k4 a04

a. 2 04V r UI L 044 44 ) S x
.00.00rIo0. 0woo4
04 C a,1 &0 a
SO
U 000 p
9.51"54050. r-44 4 0 $4
!,
.4 0 10k. MS4 I) -P 0%04 H
a v , 10 0
.Wa fn.~
rfl
is SI0 0 4 , o
n044-
-4 IM4. lo.A0 w ~ 0 14
4 i %4
1.04$
4)0
--. -1 ; 'p. .i
-: ~~ ~
4M ~ ~ 0r~. 4(
.-
1 a U 1 10~ '. 1 :1 "1~.
1)30 40 4 0 ov.
44'l4 1 q 14) IS44a4 14 04 W 0
2.- 0
44 N
01 .- 0
1.40 .0041 2'.0'4

B-10
.. 1~~~
Jj 3C0 .
6 41
14 N
O 6.-44 04.e-
V USC
t3 10
0 C O
4
t; N6 **4 0 t;C~' 0 IIa I6014a

a 4 04 t$)0 V.
0r410. n oo m 6 06 r0
za -o ,$
6*
1 8 t4'e 606V Oa o1 -a0a.".
-
4 C 6% % .44- 1* 4 04 0 V -4 0 4'4
V O01 " 40 I C
el 01: .a:. 611 .1 4' J ' C
I-1 r. 0 S O 4' 16 Q
1-.0 . 444o-'v0C
0,00, 0 4'. 4' k a ,- -01VC
at ;C66IS 0 -4 .-I
8A.
0 ol' IF'0 4t4 010a$ 4 2r C 0 p4' 0 r
C~~~~1 -00 OR'1 kn44 - .. V
00 2 OO
aC 0x . 016> mC 0 04 O $4 0 C AO'1 '4 U %
414 W
6A6A -4 6-4 C 0 114 0 - -OC.00
14
O' CO 0C -4.0 ~ 4V 1
31 0 W A~
0.v - MA
o10tI
V-4
C
-0 -, N2 C
~
0, 0 .IN -14CC 'C . C0
OMCH 14 .4 0. I0 Ad 4,46 .01j
0
~00 81CWC-I0
0OO 4 C 4C6 0 64 0414
$,C 16-.4 00m COo-
0 k 140 m 6 0 .0 k
mM CC >OC
40 1
Cr.U'W 0 -4
' 0,6 00 01 WV V U>. -, 6-
-4r4 O ' . I- - 0 4 4 4 C g o;V-
O C -I O4 140 I
~M% - 6 6 0 0 .0 45 M l V . . C- If6
400~ CWco4' C N.0
Lo,1 624
(f.*.1(-at -4. 0A >OQ-- COO
i10 v 14- 10 r. 1 11 .00-g--A' 1-4 Is' 0
9. 9L 4 . ->0U4 0 14 4 C4 'U CIO r 0 Co. W.6 C
0 011
>
m
1- 0.0Q61
CCbl 4o04 0
g 010.0.
0 .0
~'44~4
00
-t
t 'W"
C0 cO P 0
-400
&-'
4.
01414-S4a0' 14V
wo-.46. 4
6
6.0r
414
A - 04'
040 - 4'6 .04 0 A 0 %4 ' VMC 0.b > 64 6
C 4Os-06446 4 a.r 0 >,c 0 0a1 0 C
toC
$-414 C4 0 C 040 2 6~ -04 0-4 so.06
C
.420 0 S4 t.I 0- W4'1 vW04 C6 C -
'40~. 14000 4 VV A646 v4.I 1 24
:1 0 C140
>0"0 6O1.) 4 COVO 00.. 0.-C v%0 '' so40H
V 14 slo60
4 - 6 ,O% ,- ag
000 . 0% ' 0 N f 1 Z0 60 . 9 1 1-40 MOC 0a4Z60
'00 O
0 4UI. W I4' 0 -4 60 0 0 0 . 0 - 0 O - V0C
' -A--- 0 40 .4O0 O 4' N
0V H 0V 000CC 0 4 V 4 CVC
4.'6 0
u v 0% >.-
oW1 A a MW .4I. 0 -4 04", - 44 0 .C 04
01 1oC 01W
. _841 4. 4 . 0 C40C444V 0 64
U-
1 a C a 0. 0 40 .. 0.C0 o 6- T'.
u o S>V06 00 ..4 0 a 8 --
AVOO 1- 0 0 S
-4N00 a04401 '~ 44I
*M 6OW
co.- 00,40.0 4 9.0 0 a a W4.O V26
1%c INO1V 0 0 6M 0 41 4' 1,1 1
14 0 IA4 0m0 . 4 V-4;41V14 0 -4 4 9n 4O tP .
0~~~~~~0
-40C.1600 C OCM *S00' 14 . A V~f0
4 40400- 40 -4
R4C 0-W u m 06HUM01 A 00V .4 0 $4 $64 0
14gEC
06.44
-4 60. OO 0O061 C 44
01O~ all 00 C 1
0
.
~N
.06 4
0 .
0% O 01 0 61.640 k4
E CYOa.
' H 0 v >O IV10
me-
I NO4 0 0. 0 .V -V~ -14~

g. 04J W- V -004M4 0 . 10>
V 14t CIO4A
0- V U c.14 0%
0 4 0i j0V 4
C0 0 -.c
06 00 C 04
*0 C 4 14 061r0 0a A1 p$
LE
0 C.1
0140 0 cc.-a C
C Ct 04
A1 00 DoC1 6OCV
0-. 04 > 14M 6V'AV
4-4 V. ZM114 C% 0 4 -4
-.- '0 0A' 4"
0 4O 10CZ%~ j V O 01-k 04'. 1"a0 - 14V 140,14
o.'A"
000
4'M W- U IM CA 4 0 0>. n006a C4% MgOO1 4 1 C 06
Wi a ,14 a-04
00 .~
MU
0..
0 +0 0 4'-awn
o C 0r 0-66 , 0 6 0 0V6
144 09 1c 6142M..0$.- .6>
r; Cc4
6- C 4 0 71U 40. 6 6 k>.44 O 4 6
a( lw- 00.0
to 20 OO C 0143
V4
C4'~ a. 4'A '00
0 60" X
14 4 ~ 1,
A
4 -10 -m
D41-C
0
C6 A
4606
~
*i~-* . M4
6H4460'V.0
.
' 0 4 -M
44 - UOU .. o
4
, cc,4
0~~~~~~~~~~
0M1 O -
10001*
4 0 P.I 0V. A
6 0"%W.-4
iC 0 A ..4
w8W.W.41.-
.C. 4 v4 0
1.4464 1 -4010V
,
.0.U4 .1 0
mw 0 6 6 0 6 044 V01 14. -0 - 0>0 1 014 4 0
46344 0MO 00.4'I 0-M. 440 4 64 M-4 6C60%
4M %
CC 44
r-4 $4-M 0 a00
VOW P-004-0 MOO'14 1 t o C0 0s c " 0
0061 040 1.$4

IS 10 V 2>V4
1 0 0 - 11T 4
10 %4-00 044aI a Vto -
m W IOU >.a A.. m o

4~~~B -I -14A.
1600 1, U) V a0V6ca 0a44 .
04) 4 -OCO M404

4.40. ba0 0n 0
m0 004.4..6
.0 C
It8-
.0
0
0
04
w43 00
UC
co I.4v
re-04
44r04
44 .
) 4
. ~
4.06 C 004
0 r-
11 z4 C0 .8 r 0 4 0 0 l .
in 0
0 0 '4M0 4n. 4
OR- C.4.X43n
014r4-0 .0 0 2 1 0 8 ar
.004C)C 03g0 "
141 10
.044 60 *~ 0) 04 4A4 400 0 0.44
+
6-0.-4 '-4 .4 M0
0. 4Ov D 4 0m4 0I4440
0N
C, k 1 -3* 06
0~410.0 C-OkJ 0 OC 4 0 00-4.0-
0
0.0040v 4 4 34 8 0 0 %04 C OO 8 43
0
0 04
6J 0.-OC
,0w6
a~0V.0)40
v 04 0 044.0 140
009 .0 0-4v40C 'o .k

o o to4J ,* I0-04n4*-
go to MoC 40 .4
0 0 0 0 C.0 v, 4 0O4~0
o f C 0 .- 0 4 4 ). 0 0 -444 4 .0 . 04
0 J4 044 0 4 -4 1
00
0. 43
mC r0
t4
0 .
0021.4
4 6 CROD4)
4%30
~w 4
0-M 0 11 44-
- 0
0
14 16 -46 3C
00%400U .C044
0.43004.0 04 0 4a t( 0, 00 I 'A
04400M0 4444440 44 -g400
0. C0040-.4 j-4 00 4 40 0CV
44 -000 * O 4.0
k0O D.-010 to 0 1
N O4 0 0 'n4 1.5 4. 0 4
0 4 i 4 .400 CC. 0 0 0 0 4 03 60 ON0 0

0 o~
A404-40~ 0->4.- 4CCZ-
0
4.04.
."0 06 w> -4.4 C-4.4 -3$C0r30 4
0 -4.04
00 t'.-4- 1-0 1.
to6 4 40J o-WI > 4 0
toC 01404
0-40 O 0 440. .0 4mIl01 1 00- 0jAC S0. 4C0
400 6 0.Cr0 >0~44 .0. 0C4
. kc44.404
-4 0 $04~ 04U.4 44 C~n m .. -4 IL C0
N at0004.
014 0C 4 0 0104
--A00-rC4o4.0
0-a- I, Q4 *~-II043. 0 00 4
k ) . 144.a
0C4 .0.C to04
0 0 000 U)JU
40. X4.0 V.04. 0 V,4
41C4. 060
03
C
00
0U C V44r-4
44 3Ca at
0
43
.04-
C 44430
CU).
04- C0044
0044 -V4. C.r-
09.M C C-0-b1>. 0o
C 443 0 014, 04 4 V40-C04.0 0 z 040, >0
0 e 0(.4 3 -4 00 04C 0 .0. 04I 0U).0C-4
N 0 00 4 IM
*4 4$4v 0 40 40 4 3 04. a C 44*
.C6
40o4.0 )4O44 vC44 40 40 443D -.. 0. kW . .0 a'4
0400 000-4.000.4.0 *4 >0 0 x4Ca C0.a -4 04

0.>.4 0 0 3 0
4 0104J --1 4-4M 04 0 0 40P 1-
-V
-4-1w C x0004.0
0
C6
000 3( t0
40
cc 0C 4.
5. . 0
4so4
40
4 . 0 0 C>0 4. P,33
0. 4-j.13N4
0.4 Ja 00
C. * a 044
g 0
- (00
34.- a 0
0 1 .
c c.
*4 C4
a
.4 0 4
z .c >.4 .0 0 0 44 4 0
9 06Io000 4 04 C04.0 60 - 0e014 V04 k 0u v C 0. 044
-~~404( 44.) CO U) >. C..0
0 0 C. 0
V.00- 0
3CZO 0 00z 4 4 4 00C 304 0M% 0. C 4 . . 0 . 4 4.,.
M4
.4
0D0 ,
v0
.4- 4V-
0
~00.4
-4 . 0>.a.44440 0 * to
-
-00
k
0-
0C-- 1
0.M
>0
*.40ACm6 0
.
.40- 0
40
44
6j 0- 0a0. 4 04 C .40
01CCV 0-4C4 C U040 C
w4 - 04 0 4 4>U -4400 4 0 4j g 34.44
00 04 644 0 C4I
A. 4 4 0- Ci 0' 0 0 k I a 41 0 w 00 00
0 04V 1 4 U09%4 09 00 -1 . 0 014S4 g5) 4 0 43 C .4 0
"J4 4 -4 004 0-0-.04 4.,0 04 504 4 1344
4I4450.00
C- -44
.- fl0.4.
a. M SC
-a,-440
4 00044
43 0
4 0.-44
0F 4 z0 00 .Os )- O.004 00
4 M
044C 00.00000k0
>0 ON
0 411-400
0 434443C000
0-0- .043
9 2 440
004-
-460i0 C VC - r3130414.40.>4
C -k 040 C coot C -4-4g I C
-4
CO 0 00 .-4 00-1O 4w40 V0 a0
. 40 4 0 4.0 b>, 434
0 0 0x 0U b404
OC
4 >-46 0 4 1 0 F w C aC0 - .4 -4a
0 4 .0 IC-b 44 0 4.00 4
V0044, .0 04 440
001C. ; '0 4 k u4 AS.
. 0 0 .0> 0 . 0.4440
Or--4 M00 0 0 1
00 0.CDQ
.40-.4004. .401o 44 0o 01400 041 0 0 0 a 0
.6004 a
.l C000 C6 ., 4 t)
4044 :0I * ,,!---4. 0 C0043 43 40 0 0 .0
0C 3 0to 430 0 4 0 6M0 4 04 .40 40 W.0C-
644-4 C 0C 0 a 0 C-0 W4 04 00k 0> . p. >. 0.0 a asol
. 0 4 %.0 OC.4 04 . a A0.r- - o .080.
414 I00 .
"43 g o -u 40
-. e . 4 4 0.6 4 00 Q-A 44 U.000 04 -
444J4 CU4I 444-4 04 6. 0U)4D. 0 000 - 4 0 0
4 0 4 0 6 14-4 0 ..
10
U C 404
4 -k -A 4A~ V0 Wg0 k .40
0 a-4 a w34 %~04- 40M4343044
140 V0.0 0 W.4 0. 04 4141E 0o 0 044 444 . 4 . g 0.40 0
- >4 0 00 m
40-. 0 41 a C
~kq
00 400in
a4. 041 -A 19 4a C- %>0 c.0 43440
, *o
e 36"4--
4 1
. 0 a~ 4.000~40
k
43 -443
000
r02
9
lax4
1,0 01 0 0. a840 -44 - . 44
144440
0.4
0 44
4 41 0
004
44A 04 . I v444
Z0 ; ~mC .!1 0A0
40
A~)
r 1"~. C>, 440
0~ -0000 0 4 4
I.4 -II4 0 . .4 0q 1 V) so-4a0
.40 440
0 1'm R C 0d Z-0. 4 0 N 4 0>I .0.0zw00 4 1
aUA 0 > 0 0 0> 0J-
4
ko ,0 0 .3 C 04 -U 0.4 I 444 0 I444&I>goI 0
044 0 0044 0-1 44 In- 60w 6 .'a0
1 r
4.43
3 0 0 0 1 %.I.444to C *.
a . A,0 0$ 0 $4 w. j 4 4 44- 0 O .
B- 12
.w
11
e $ In ~ 1 4 1 4C)
r4~~ 4 O.14130
1400 m 0m0
.4. ~rC48 W.d1d
4v 0
4041.
048-in.o F8 444)0 84. rv 1 0 v6
iv- 6
.4 4
e..404~r-a G-.0 *~~~0 6 6 0 . O N..~~
.~
-a 14 a46O 04~
604' Sal
*6 *0 00 0 14 .46
wop k IeU C) 0 1 600 4
w a. .A.60 t" 6 k~6 14
0 oe A*,oeva. -4~ ~ 1 'D $
348es
0.403.4
0C)
%0o
4 :4 4 14-414u fm 06.It
6440.
r- -. 44 00.
0 6'C.. 14
0,.4!g
v 0.4 1. 400ko 0 .4MA 11
'08 4 64$41
,8 A1 r.
I s -4 % Is-z4;6
% z.04 f
6. I: -X.00 .4 6. 0.01 n6*4
4 4 0 4 01 4414 Ua4844
140. .. 0-48 .. W
o~Z.. 0 414 0 0. 0 . 6--4 60
.4 4 r4 04 In '4 4
c' 0 4 0a '
- 0 .0 11 4o$0Q 4
00 0 0..4 .40 061.4V .004
$0,80 >,
:.41 0.-0 006Q4.00OC) - 14
ag 9 0ac a
a4.
H 8 c~4
4 0M <
0 rA Mo
.GC 48
00 ao 4 9G goo a ~ c
. 1mGOC - . U N $
6 0 O t* .
*..4 r8
G400 4 0 'I6 $J6~ .4 4. 06 01404' .c1 0 0 I : 4 9"14
0 0 a-~ .684 0 0
16.
Or a**
400 1
Vt0
*046041.8-
6
0 44.e-4104 w 6146of soC)0
&V
.e 4 0 *466600c1'..
.40. 'aV0 '-4 CEO -G.
0,16 10 rO 11 10
. 4 00W me_ G6 o401r4 0 .61 .
0 4a .G. 04 ow10
1, 6.o 0 $1 00;.3 >. U. 146 - 0 - .0
1 0 040 466
-4G-48M mcoG 0 =a6. 6.M4 C840
I 4 .6 IC a. 4 c4 0. 40 *ft.0tpp4 0 6 48
400 0 C00-0 41 G 4' 6 %4C 0I0
04144 4,0 1 Il4. 0",
01 o 61 m 060 i.4
Z..o
Xi.. 1
A~.6 or- . 0' 086 l 1144, 0>7
306 N, C) -481 c.. 40 . 0 u 14
64-
.q) -4.40% > 001.4 X4 U C -. 6,0' 0 -4 0 060
O 4
6-.'~G48
RIO-
0M
.41400001
0 14-
.0404
~g 0
0-
0 W N
a c .0486)4.4
-.4
000486eG C-G4
4
:.-
- o00 x
*6
4NONO.-4.~
6 . 0 0 4W'*
0 00.$_0
.4 &1
G
-M
61 a 40
*640
000 ~0 ;~.0414G2U41"-
'I.
NON
1 40G 4
0 4 h6
(A016. 80
6
6...
g
1H
-4. 6
H ol 14 '0- ON" 0' -
08-
E -3
oom0~ls
-AP4C
- 4 0-.4104 4 00 4 6 80, I
q4 4 W~ 00.' "I40414148.0
0 - 0
40.4 "
))~
a~. 113O 0040G~ ~ 4
$
6x 0'004. o c0 -3.41 0 4 0 600 0064
%4caM. OU .G2480 c 0~ 61.444
q04 8').
0 041~--9. 0 to 060 a
04-.4 4-040 6
1.a4oo000. 0 .16W6 0 jO ... r- r.46666V0
4 4-4
4 ) 6 m s0 1404 0060041G4
46c14o,"U *I'-G F or$ .- *- A6C1 66C 04JI 041 z
U4 1 iP4 00
-- 4
430U 1,
61400
6 00.-i)4 148 6- 6u 43 4 a V Sot o..4 A068 *-A0.4. a
m48.4).8 GA&0
.46
* 4 06 0i06 U
1444'.G 46
0I.
e0 OV3N4 $4G48.44
k4 S80-06 V
iG'448 4 f m 0
('Z04 -O 0 4 01M
6 W.mG
c 0 A A0 1 014C-
0do 0 4r;G2 t -4 8
a.- .4 '60 > 0. V T C 00
-A)..3 -1 I t ' r661.
4.-AG
V4 m -14t*
8
14 0.06 -06 0.1
r. "
a;0
.00
1-46 1
0v*>,q
)3..6 M
.
I 14 40,
1
IV 0%
04- 0 46
0i 0001$C4
c%
0-40
04'480 - .% 48 040-4 mm .A.41)0
644 40 4 al r). 0 c~ go q
le64814 also~
C..4 6.60 006 V06040 C 6 64K 0614
a-% -A 4
*00
0 40
10811
00
0404 6-4m'C0) 0 ~ 0- 0 , '
$.484 M. 4k48G44 C 5io 1 ",
" _a 48 0at G0-4 0- .4 0 0
60
0 m 61"401 0 44
00 484 '4
4"M
44 ..4 6 60400 8
ow~6. 06.6'4
$4 r g0
0-a 14 0 0 4 a s
In~6.
0486.-44 %
0-6 -4 n 30 to4G40 - 6 6 .
m46 0 1-4 08 6
m141 - 140- 0 0
114 9 1 :0-4
66 0 C 4-1 ; 3 o
00- k - 40 -, 40 4 4 . 0- 04go aZ l &4
.4 V
$61 *.0 r-00.-
Fc ; 83r .04 .44 a 46Ag
.4 to
04hIj~t 00
06.-~~~e
144 - 04I.L
B- 13
-40. 30 W949 Od
2,4 000 A6
VIco4 0> 0.4-A0 aJ.a4v
- 4 9 ,O N - 6: 0 e k ..
c-a A~ . -
*.to 0 63 N04 IOcM 04
a.00 0
'461UAtr40
4
;-s~c9 -00
$40.4 % H
a1.
U C
6 f
64
0
.4
448
c 1 0 -4 0
0 1.4 14 n 0 z Ove v a 0 w
E40.4 vea
00o .0 9440fZ4"44"J
0. tl 4 1 k 0 4J: 4
0-40' *64~ .- 096 O49...
a40
~
Ij,9 94
C 104
014 49
-46bL.4
... ,
.
*0
sum 4 -6094
LO t)4 14 ' 00,W C44 0 0w14 ~ kc.60
0 0 .06 j
-1494 -
M
6
rQ 04
'44. 0-14 6C.6.040Om
94~~~.0-4 111 0-
goof.....
40 >_
010 c 0 94- 99 4 ---
6 0. .04 4
I~
0 .. g.0
a..~' 0 .64 490,a. a4..
,4~
r 10 -41 1 O v3 a
00. 0k00CA 0
w4 006146 0 0 14000490-40w
(
'oZ-96-449 4 40. 0 0 k4 w 0 .4.-40,..406-o
4 W a'4 O0>1.00 0Z9U. j a 26
-.414 4 449 c
0~ V 0 E0.M .*44', 0
,q .0' -4 O6S6-.C -00 w a00 * .4 00
111 M0 0 N ' 4 104 0. 0..
-6 0 26
1-4 _1 0 1;A -4;.- 0 0 Ma 6. VS
C44a 0 0 -4 -40 1- V=
494 00414 49--0,40
;4 10 .0 1 496
a44- 090
r 4 - 01 0 0
0 ;q 0, W 414 a A414
M.1 066c 0 061
940 04.m>0 _ 0-1 0V6a
4 14W 0
*; 0. -
94'0606400694'
0--V0.0.00E-0C#C WI'6I940a-0 2~ - g.
'460 0 .
94 0- 00 6 %4
.0 c$:1
.NN 1i0 49.4)49~~I
049 v9 P4S CL -4.9 soV404
66. 006 0
464
to . 00 W.f-.41
6,, g40.0 00944 494. - 6
1000 a
%40461400
q.*00 0.-4-l
009
066~~~~~~
6CV00wU o4. a -. 0'460
x j-4M--~a
'4 f6 9 )O.4 at* 0-.64
0 4 0-49 0A
~
P A - C (A0 60 C a- -9
c ' 0 14
U .0.-
' 4C-
E .2 94 116 Go 9 0 V.4 0- %4
E- 01 1'0. 060-449v -4 00a4Fa
. ~ . 401 6. $, 49..l..0
'4oC1 C61 Z 90 M
'A490o
C4~ C4 014af0
& 0
C
,.
t 49
.. a.
MO Vaa
* 03 .
-
V1-
4a
4
~4C000006V
-4
.4N 449 .
m4f230
6A
-40 I
84
100 66 14N
CK)4 040-0
- k
04
A k
C o 0 90 64! 10 94
40 4l 00
4
4 .409C4 Cc A, 31m
00 0a
Me 6 00-
I .06 94 W 149
49
04 000 6 U40
0.404900 a C001 H.01
14u. ca
63-49
4 ..4 14. 0490 - 94 90 -4 .44 0'44 4r- I.kx.4
C 14 140 no 049'9 lowr
Cc0 EM - 49 c a6k.4 9 4N0.4445 14 0 Nag
004- me Z'4.a4l-c0 6 A0). v 9 U4010
cc
I0 $4 s 4
0 N 3
0 w0% 00 $ '4n C 0
-409 00 0W 09
>000 Z j
~ %. 4 C, 4 004V*~04 0 0 3-1 4-00
0 tru 0
.8~~
94 p 0
4'4
4
v8 0Ik .. 3,
K~ oq 4 o -.
4 4'00 JoA 01-O
810 16696
~~41 Il 0
r 4
I~A4
P A 41 W0.0~:''
I 4~
43 0k% %4. 0
.4 AZ A se s >4 1 1 InIC
94c rI. 0w0 .In a44 0C410 .4
ZVS40 VZ4 1
01 00 0 3,- -4. . ai4
-4
FA a~
~-144 ~A tP~-4~
oa-o
4.4w1
~~~~.JQI48A 4 00 -
v111no 0a_4
0 -. 4 . 04I
B-14
1"~r--*114" C lo- 4; weZ"14'"a .4 66-o46

6 ONj 0 r4 4 ON.Ca6.- 664
14~~ 6 ""C
1 0 4 V of f6
SC
90 0 tp &C"*A 0060
0112
0 ..
101,
0al 1401
a.;;S
-4
0V
~ In
I-C
0
m.6-4
4
C"
Isla 6 .401 46 0 6 a16 6 -4,0 V64,0

440' quaT . 140 a'
3 -4 A$ (I C41
01410%.0 a~ -044
C .
'4 -6 0 4'64V
N 6
-- ON
0 4 rC
14 No1-0 0-I
04 In 4I6M.61
C115
M.1 43 04)
10. k OR00..p
t9.00 11, &6 I
66 0.4
El a,1040 . 1 H e 40.0! 9 O N
s :4 M.. 14 A-61
"14 6 C
6m 641 04040 54 6 6 1 .
a~~~4 -4 , I~u1, .
.~t.1 .~).c
C10 6 1a 14
61V 6 14 00
60 44
14 4-0 6C 10046 -4 5 0A411 a'
6 OC
a-oo 96* a 4J r-061
"4C~ -441 56 -.
UA '.- 6 2 a6 C~~~ 144.444 4
5$C)4
u 0.
6 no4-
00
QCU)I ~
a0.0a n
04 0 11046
4040
.1 1644a .4-.1
V6-
k 0' 0! 0 44 I 10 -11.4 a.IS%04449 car0 -4,0 ;i48
.4VC In 06 416C CS-k"4
604 In~~
0
.4 0
m N
0.~c 6
V
660-4
404Y0 0
C
-0 146.2
04604
~ In466
. 9SR401 a-A
a.6~4 0 0
q64141
04
40
0.V 06 00 -0. 06.41 0 0 4-I
0 I'l 6..6N
140s4 e
06 W.44~
.- 0 0 0.
44 )0 v 01 01& -4 c 614 06 qu441-
14 .% 4 *0 0 1
44 r > C 1 4 C 6 r006 - 014-4 E464 -,- 06 41 1f
6-.$-4 C.VV460 64 aC 0I4C I'0
6--.- :6C-'---614
rNj0 40 6C 06 . No 4 "-r CD0 >'. 4 0 >.140.6-
v g1. 4 &, ,4 *6 141,A%4m0o %C
Ma4> 0 0 400 - 60
aa 4 £6Iw 41 V inC V464 0- 106-4
61O-3 * 4 10O 441re c S 604 -00a 0.6 > &, 0 4.
k
.qU) .
0614fl'5-0
3 1h4 ii
C9t;.414
-.
-. ~.0.'a4
C
66
- 0
0142
1,' 0 L 0 0041 C4 a 4-,1;4
100.0 - IL
0%0 -~. 6 -0 4J6E.4 1 6 -

V.- &O Um00C)t. 66 o.0 21
4
4WC6 1,. A. 02 a04 0 - 0 04 C611
2 >' 0.*0
5
F4)- M 0010r 641 %
60664 E1 ) . ~ ,aI O0-.0 1.4 C,0
4166
*. 41146 -
6 .441
'I C
64 0 -0*0 IA
.-)6
4 0 0C 6 0 4 -0
-06 6 4 1 0 4 66 V n 0 1 01 . - 14 . "D .0
,c 0C " 4 0
*6 4. I 0 &6'S
6 C~0
00lowI0 0 -0~Or 6 -..
N046A .
0 n. 6~~
~04C) .
.-
E4-.4-4
C4 6 -404-4
4N 4 ' 4 2 64
0664
Vc C66-.q.0
10642
r4
04IZ mCG
. . m00~
C
J
6
06
6
0
-40
CO MO
4 14
041 . 01 404 0.4

a.4 ca. mo W00 0 -4 4 412.4 CS 604-4 -4 p41-
0. O%0 f
4V-> Qt a 601,0-I . 1. 66 0. O4 - ) - 4 0 V 1 6
41 ON 0 6604 C0 OV0c600 0 0 E , O 6 6 C 60
14 6a 06
t'"S 0%41 11 I "-C
.4 I411.4
C0 a4C60 > a 14qC
-40. '04.4 6 > . 6
Q$4 0 03v-
- c CO .1.- 06
404-a a6 i c1-.
1 4 0 0
z V-4 0 4 V 1-414
4, 6 .
~~.-44~$ C
*EM-
410
6416
--
1 a - ,46
V
1-1090 111
-. 0 1 a'
66 4
14CC) 064
E-4.4 In 64 4 $.I%.
1 64.4V. A 64 6C1 P4 -4I '4 k 410 c t C 061 6
4j
X~
40
~ 64
-400
C%
kI 4 1
- 1441*01"~ ; '
6
N
01 ' 6 '4-
W )-, C-1 0a .4 - 4 c-661
1
1 . In4 4M 21 -. tI. O42-2 oz l
146)'044 k 10
CO -O 6 6 66 0NC304 0. IVC6Cv-r.4- -Cw >-j0 0-46
114 16 Vp 4.
OC'- V1 14 $4C 0V0 410.4 . .z
No 1C 4ci ) 0 1 a41
.0
1
0 . .- 14 6,
16 -. C1 140 600
a- po41 C .U vO 41w
V1 -4 en0.) -Ac
60 .. 4 000
U.~ A .0 a'-cc.,
06 -4 4 14 I4 44 N0
..1I0 4 0040W0
0. 6 .. j 6 C 44P
t0
N440 co 6 -4 -0-- X4 - 65 4 "1 2 -
41.456 a -~j 6 v~
41 41.4 S r64 .0 6
6~ -0 6041 -4 .. 41
0C *
0I
14-w4 a
-Po
a. S
.41- . 0
-
6141q0 14 Q0 1 o 0-6-4 j 6
0-0 6; In 6- "C'm~n1 aC jIso. , 4 I
r140 -. 0 96 V4rn6
V~*6
j 4:
4j $4 0~-. ojj0 44.
$4 j k. 0 .0 0
t 1 410 0J 4,4 I cc 9%04 0
o q04:I!,-. 3.f I
6046 o j ,41-
ipI41r 004A.Rk'o .I I1 V, 5 - 4
0 -4
h- 4 B3 -A c S I1': 4
14~~ 96 .40 110

~
-4 56
U C
j -u* ' 144h k ~~! N
B- 15
0 N6 w .4 A, 2 41 , 4W, '0 0 0 0 5 4I4 6 oAU

5 04 1 0 0 .0
0.4 .4A sli 4 0 k4 k ,
@4.c
'004 0416003 X. 0.
.04 -4. 0- r 0446614
0.1.4
- J.4e 0 4- -0 2 0
0410A'.0
1 a01> '3415
" .. 0 0 a~
aj t40o0
0 1 41 0 v1 6
0a
64 1100 .N00 -.11.11Oo~-'0 v0 4 0 0 -I . 6 6
a4 0 p41
p4-0.c s o a 0.1
54 ~ 66 .14 0
0
U201j
'- .4
t38 .! 02-410
zoe 50l'
0 006lua a .I c. -m 4354-
o44s
a- .0 a21
-.44 044 1U 4 6 64 4 0t0
-. 4 6-00 041 0 0 OW
14 a~- .02 0 4 -6o 0 - ' . . . 1V.i
w 440. 3 .. g.0tr 14. 40. 00-0 4 006- 14"'00640go 40

04 v60 c , 04 0 6 0 'asQ 0 Z 4
l 00 - > .0 3.46, 1
o:o 664j1 15 '0 41 000 0 -4-G' W $4W"0
0. 0rV $ at w. 400.cuk...>40
66A 54 --
1C 1060 0W4 -..
00
z0 4 a0
4f2
.-
54
4104--4
61: I >0000-
_:4 2
6-. a4 4 546
1 148.62
6
-.
4-I~~
0J 641 16 34 416 ~ 0 Q-1 Ts
0 1624o~0
020 -644300 ~ ~~ ~ ~ ~ ~ 0 ~414 ~0~-4-
~ $ -1 5 cm
6 0 6 41 40-. >
06 t; re0 1 40 66441 44- 51r 0 054 0 54 154 41-010 .- iO4WQQ
0V 0 4 1 4 6 0 0 .06 v>.24 005. 4 -4..() 'A*# 0 0 1 C C
641 4)00 6
0. 01 0. a r,-00
0 665 . 46 0 0) 0 Il ao 0 a01 c0 0 4 a 6 .V
o
f.4- E6.00 ,
0-160C 0-to v5a.4
0..3
o gol "IN 654 000 6 ! W 606041 0 1.3
Mq 0 0 w 54 15 544 01 1 5 05 4 . o U z 41 41S 0V fa 0 4 ' 0 0' - i -
Co 0a 0 .4.-
046 0 v- 04 . 1426 410i 6- 6030 -4 McE- 0 4-0 aO 540.9C 04 v 06 41
0.0641-i
0 4x
044 0o 41 0 .0 .w4%.
0 144
1 V.0 40 4 41,
6 4000
2. I 1o .
6 6540 6 W4 00 4 00 rV1.C.
0404
0 000
$c OQ -4.- 0 00 -2O 10 E40 5%4 6 V-0U64 0 M 4 54
0 4.- -4>034 0Mc C4 0Z 0.544 62 Is*1-
0 R
4.. 6 -4 1
41 41Z 0 0 0 6a c0 20
41 0 5 0 62 4 0 .0, .4 6 0 41 1 0 4 5 -4 0 4
La 0M0 v 4a46
1 .a - 0 0 0 54 .aO O 0 64 .- 0 21 .0 4 0 - 0 0 4 R0.
064-4006 544100600>.~~~~~0' * 6# 1
0 4 0->4 Z 6 - - 6 41 000
5 04454 W44
60 004. 04. .4 00 6 4 0> 5 0 6 4- 1 C 6
t.000-I 41 "4C4 - 04- 0M 2 --. 0 to>540 062-.
Zo IX -4 0 zCDC za
"'W601
0.6
$0
2-4-.-eU-
0
.0.C 0 t Unt'0.6..
40' %-.-4J0 U'
40 0
D a
*4610
V1 4 044 166 ,6 t- C0a 000
Neu0-
-~
64 6 06
410 4 41 2 1> 0 2
61
54.2
6 0 6 06 0 06 0 0 6 5 4 5 4 . 0 * 54.2 .4 020 # 6 60 . . -I 4 0 0... 0.04
E-.0 > 54 002.06*6U. 41 041044 C- 0W . IZ
25 6 0
vN .- I0o1-46v2
-C MC v 6%410 6 #2 *- -4 #C 0056 # 0080 O-1
0. 410V4 0 x-. 0 40 0 C 0 1 54,1 -I.-164 1 6-4 -4410I
1.ca0 1C14 0 62540a 4100$
Or. 41.1--4
0 100 06 44 vON %. 41 5 0 NO
54.a 6Z 0 41 0> -4 0
-4 066 000a--
00 6 'A 1-0
540,
#2 E S21 0 4 0 -4 01 10 06-S#06410 W 040 01 v
410.5 c240 00. -D 0v c>-M.0A0*52
-404 0 V0 I.0 : 054 -.441
In400,
.- cu 414
o0 010 a'105 r-6-40 054 6 4 6 )#rX2 g
w 4 4 04 CP 04E -M
- E..2 o0641,
40444 0 V a0 " '0416. >.o #2
0.
2M4 .0 054 -. 02. 0-46 v 4 a4 0004400.00W
41.0 441 - %2
0 P40 041410 30 6 jJ
0C02 0 c-0c0 Z30)
54 41 050 40
1 540 >0 06
> 0 c 0 Z
a. 00 a 11 04C040w
-. 465a 0- 41 - 0 012o54s 000) 26
.4
E60F0 00- 540n2
40 40 0 41 041 : 0;r. 0 V 0'4 >
A. o 00 a .0o a. E 00 60 #2 3.0 c 54z0001w
22
60 6410 OW
6Q4 :. 1;. 0 -16 066040
oz 0 -4
601
0a1 0.'
05 44
v40
Q
0254 -'004141V 0 0 41 00 41 005 5 2 5 0 4 1
0.00 0< *, 06)0 541. .4j54 Z. .h..5

Me4.0 641c
30 54 4 04 V 46
4 41
m f '00 64 110 0 54>0 0 410 . > 4 -220gW
a101 (, t- 0 6, - '0 C 8 k2 , .
v5'
-40 0060 24%0
0 *U
' M 6 00 0M V -02k 0 0
0w41
!.j 0-~~ c a54 54C54010-
060
00 4§. . I 01.1W E, 1
'-4 0 Q; 5400 4)%.

0 0 01 5 5 0 4 we$4.~
'0 @4 0 4
C o4-a Lo. z-~ o
.g~~j -~- 0 0M41 '0
0N6'A
4*0
a g. 0 0~-JA 4 x Vwai . 0w- 000 6 O
xx4 .10400040C
3 1 . toA.01.
'4 ..
a..4 44 , %c ...4 04 !
c 00 41 MRA 0-
B- 16
0
-
1 M4.
I0
54V 00 4 6 CM
0 514
ak .4
6
%4
0 A
- tp
G0'Vaa
4 )64)
0 4a46
t 44
51acc0
46*4 44 0654 a
'46 4.1ov a044
54 *1gg 04 4)0J
6-4UN 4 4j
60. 0 1
~so Vu1- 0-1A CA as. ~ iS 0'4
V4A a 0. 0
S -4 .004910)4
"A0 660 . a~5 1
is%d ? M a so .0 .,.r FN 0
4j
0' 40
0 C04W4
0.1
E
~0
4 .6
6-45 46
))
4.4 4A a~
P13 a.4~
**.. 4)6 64 0.
low0S
H>0 -4 -4 0 >,0 o.14 0 -,a
-.4 60V4 OS & 14' a 4-6. 0% ty. 0649-0466 . I f0
0 90 > 1o
I446 4 %4-6.
6 0 6 0 ~4)-
:0.20 1 low~
me
-.44) 056
OM06
4) 4-
04404
~6 6-'
0
1~
0 46
4
10
0146r
c1.0 - 645
64-j.0
-4
.4 -5.9
p 16 6 .- 0 >))1
64 P3.
0%A -.. 40CA
05146~ 1.-.4 04 6- c- - 4 2 0 104
03 6,00
4 0 4 N4 A4.4 4 V
V44.- cm x)64A.
0 0
'4
U1
I N4 00
J
' 0
0-~4
E00
-(*104)
04) ~ 4
64...S 49.
0-0 GA'- so; 4*4 C040 04
04
a3t1 406 0 04J 0 0 C54 -. 44)04'-4 0040
0>- -4 4 --- 49 )2 64 6w 0..00 4 5.n'-. 4)4)
4) 066W -4 '-. 6->' 446 -0a * 4.4.46i144
01 00 *.-40 0"Z
%-4-4 0. 0
2444. 10le6a9
So14
0
010 2 -4 3-. A4 4"
-.4 6 6A.
0 6 0 0
515404 - 6 64
6 ,)0-060 a.4
01I 360 " 01 00 01
t . 4 $ 4 0 02 so 601 66
Q. -0 C-1- 66
140.0 A*010V0' 0 M.' "I 'r63'a 06
04- 4j@-. 0 4) . 44.-l 056 04 c .04) 64j
3--60 0 00 04Jm6 .- 46-.4 266HC CL%60'64).-.
.0 4 0 6t $3- 1 In M1 06so -4 6.0 60450 4
31-44)-4 -4. Q0. 414 1 60 a. a-66 6 01t S 6000 661
-1 ok~- so OCL A0Ia VN -0N.r 6 isc'v. 455

%4
C 4-4 00-641
1 -- 4 0 1~
0~>
.
4
5 i0'43
Jj 654)r4404
'4465002 0-.
. 0% P4.
10-4
2204) 00 410 M44,
V444)4.
go. 08a 0 )6.0
.4
SO-. 014 -6044 :s 1i6-- 1v4 .
0 .0 AO
$40 0.
?40 ?9.44 U- 1460) 9 0 V6006 uscl6 *4
0460 406 0 r-0)ow-0 6" 0 0 3 ow4-
-054A0
04. 0
0
4)0
.
535 2
01 0
41
.14
V
04-4
00
N
4). X
A
aG
0-4NO.541-45
0660496 54H0
04J4404a0NV1-
14
$4-4 040 4 OA0'
A-. 06
44-40-66 V1 64) .4-45 -x
-. 654 01 4 30 V rL. ).0 66 84~410i
V1.
F8-4
6
0' -0 V 5 0 6 03 0 0 u--5' 0 046.0 14.-4C
A VC 05= 040.5.0 5A 0 6 .- 0 R 0
ocM 0 4j4)
4 $O%4-4
6 -4M 4 r 141440AC NO 410 -4 94 6 6664 0004)
'-
OD46 ;4r)4)§40
I I0.--0% Go. 6614
140-
60 .4to
.-43- .I
000 c
- 6.06
5-4 0-4
as 1 $ .- 40
063.
4)0A 3 0'4 W W 0
66Of0 0 >14' 10 0 tn.0WE: 4 m0% 0 4 &0 - r k- 10-
ON0.r4" >4 0c~f 05j$4 4 1 0 41 050c34-d 14 so4o 6~ >0
In0 41 P 9f g4.-r 0 6110- a 454 0 N.4 4-4 3V 0 050. 00.0
M'-4 -64)6
011.4 w *6G044~6 0 44.4 055
10
4)C566in1-.-V650a 6
0-644-6 64.6504 4 1-406'. 604)4 -4 -U
w)60
4*4 04 46
666.00
o14 0. 6-4 4)60364-Q 00 6 -C 44 . a001
5 c
4) 4 4)4)4)06 ,
0 ,6CL MO - 041 05 .0 0 (0-.4,60-444x.064)
M 034V060%M 00a
E- C -C06
0
60 6.5)6-4
'i )
~ 1
6 4 6
040
-
~ 66
560404 .3-E6066to 6V6
* O 4
0.2
4 A64001.
0 6.4j 06.
N j5 4)N0 14.43 V 0 0 0 0 0W 4-Q0 4 0.

10 44)4)4.
0 4J
C0
C
60)
0 M1-4
V> C
0.4491,W0
X
*.4
064OP0.400.6
-. 00 &-4
494)1-6 414J ~ 0'
-4
400
in4
c5 06
6-I0)
140
0-4to 0 5 06.660 66 14141490 6 0 0.4.459-

Q. .0)09)-
N 4)14-44M60
44
6146-' 64J
1404444
64A01
14 0 1 .
10 6
A4-
6 66
5 .
01kk
0
4064 4J
04
-5 0)610. us'..0 405 V 6I- 0-C . 01016 490-P4 0.-. 0 6 56
"I-064. 5 14444)
a0N.4 5 0 4 .0
1z 3 -.4l> , 0'~).1. -4 0 a 0 . 6 644a9 %0. 6 01-w )0 60.N
00 1461 .4 0; 6 01-00.6 0 4 G6 to a 0640
06 aa 44 c Q , a A 0 14 r cr 0%4
-6 r4.-60
$4 0la 4j )14 - 1 n .40 'Alog
.0 1 4 0 .25. 10.46 0'1
0 MI a $4 1 : 0 64 0144) 0 4 44. 0 0- 64 41
00-.480060 105 0 006400) a W 04 05 'a - 94 -0
0 - r0 -40'V
440 00 0 06$6 c 4) uu. 4 v a4 r-5 - f6 664
4)4 a 1 4 -4 0- $ k%4 4)6 41 . 440 65q-0 A .4 6 i
0
64)M24
6-40 4)~) 0.4 3 6 14 I4. 146 1664 4)0 W.4 46
06~ 5 8606 1.0 06 440 0 Q064
41 E
) 1. .-. 4 N*
60' -. 0
40 5460 a C-4 I 4 RG06043 N
44-4 A60 96-V4 0 a044 .- 4)2 a o0
4) E.5a433
1444
N
0
M%
t)00
14
.
4)6-
M -46 a
90f1014% 63
VM
'-.6* V -44
00
5.4M
'.
-
4
-
4 ~a Vw9I.0500
Ni
46A X 0J0 -4006V-0
Is M. c cq 4) 0 4) w4j0>.0 . -. -41464
.0~~ 440 1-. )4. 1464C6 4)040
A44
N5046- 6W 3
U -0 k ~ 8vjV%418 54 r .0s
0 9 % . )6 4-.--- -4
4 0 o k 1- 0 S - f
>Z 00 l v04 x1: fa

0 *
o~~
*-.~~ .- a
6
.4 i5*4
4
t*Ii 0 e
Do.0. aq6..f
4949
044 5
0
.454V 1W . P.406- 14;
04 0 % 4 . 3 r -ii- -4 t e-
Sj $40 0 0 3 a
-4 He SA kdj
I ~ 4 I
B- 17
01 0 *Mn 4j6. 0'0 6- ~ 0 oCr-Q

a- AH
CC 046 O
0110Cu 0 00d
4j i
Aj
3
g4
1M W~ w~ -4 4 0
24 -43 j4 W 114
44
'-44t4.114W.1'
44. 06
4 -4J
4A
4j0
141-. aIZ~0% OfHI- 00 qW- qu 14.>44 0
414j.
00 40 V% ~~o
00-"
1.4 HJO
6041
04-$4 0 -A
'0
1 r. 44~f4
0-110
.
m 00 n 0 'a 2 8, 9 '0 0 a, 41icv-M
s w4 -0 .4 U416.O000
0-4 0o aI -10 so aN In 4 4041 0 0 0 -4 41
.. ,. 'I 04 S I~ 0~ 0
0 4 04 00 01 O4-4. 01. 00 0
0,-4A' V0414 30 In -021 kIO.-O
a6 c4
-. 014 cc000 014 1,0H 1.114 0
U 0 0 6 . 4 0.0ma -9> to60%I a 40 1 1 0- O C
001 2 4.40-' 0 W ZOOM 06 N4 41'0
0$N0 6E40 1 ' O
0'0 60040 0 4j 041-r 0*
>1 000 0
k40. 0.0 !-P
40~O 1-r
a4 > I >14.0 4A 041 09 HN COZ0 60 : 0 . 06 0s .4 41
4..
06
0 > 44.640 Q a 1v .6-- C m
U0 0V' 0-14 % 0 C 410 -..- '0.-4
s 0 4 V 00. 16 k
-4$4 N 01
00 40 6 %4 1104 00 06 v 41 l 14 S0
0,~ 0A0 '04. H4 0--44-6 A.4j
*~~ IC1 a 10 40 r-
3. 000lo Ia 4j > 54 418~4C0'-1-0 k 0 04-4 3.04JN 0 IO)0
00>4 040641
k-0 643 00 C'
*.410-.a0C 0044.. 041 A4 L .
u
4- 0 0040C4c 4 *.c 0,5 06 A24-' toU- 01 00 4-4a1 4 U* O0 Z'
z0 I0 aO C 41a 0 0 1 0 0 '0 H(n to 0 4 0 a4. r1.0' 0'
1004. .- I
HC0 ,4
Hg a0-'-- .006 006 - 44 1 0 00 040 1-.40.
0 'ON k 0j 0 1 410C 0 6j01M0 4 0 0 0V.-4P 0-4I2* 7 146 Q.-C 640 C0 0
'-4. Cc0 a-0 J0Ca .- O> 4 96.-oi V - 6W-4 04J 0% c 4ww 6
-0
00 .0 11 0k >.4J0.4 4 , 041C4 ~ U144' V M 000c 143.4 14.400 MOA.
(A 4 140-'a 4 H
46 w 4 0H01.M.%14.-.
H44 14 C 4 003.00 010 4
-41 06 s 0000 0->"43 4 4 40 $
0 4 ' NIkCO0,4 4104to0 .-.- k O.0 0 c 14
4- w10. am C' 0 1E 0%4 r1.-IW ('- 4 O 0 1 >.- V0 a - r4 04,0 logkm r. 000
-V - - 2$ 0 0 14 1 4-
-H 0 . V
O-I004 .C6 0%C 6 01D4 34 0- 0-4 *'14I- NOaNN >c 6 '041.44
NO-40 0 01 0 1 - I % 006 16 > . 1 . -400
V
>00E-. 0014 0-16 C% 'aA 0%041 r14 XH U4J C 0- C ('-.4 010000MOGU 40041
6 610-k40 0-A 0-4 W%0 A1. 44 k4 0 .60 CD060
4J -GO4 $6040 0 014 h 01 5 V00 'A~. 41 . 0t 6- 11-4a00o>0o
k01-.--4 1- M 4 1 c
0 t A000
0,4 c% a .4 -4M0041 0 0 -4 0.060 4 0
.. iIc 6 k*:-400A 041 '0M0 >, -. V
40 .N -4 -1P20 r 40 C 0 m 41
4JCA30r,- CJ 0 00t - IC0 0 - kA 2 r. 6n
a1 1404I-0 04 1 0 0410
0 totatocx 0 0.-.4 000r0 3..0 k 440%C
043 00 0.. -411 -4 0-0000-4 04j Iu
4-na 4006014po 0006a 00.O H04> H60A04101 -4 a
k41..-0> 4k
0 A.- aUO 14 m13rvm410.40W14C6A, 14H -N0.. -44 0.4cs, 1. 0v'
0 4 jH i 4 w40 144160 -: t, a00
'0 4 r001 A$ 4 U
0- O 10 0
210 41 0-4(-' 0 A 0 Wr .- c
0%3.A010 0. 04 .0 1W a-* - n 4j1aOM 0 6me0-4 1
CH JIc '0 -P %'0A C 0-46 a044 0.NPQ ( U~ $4 a
C >N 40. 04 W% TA 0 1 1 C 0I 0Q0 §0 C.4 00 00 NG 00N4J 4 V 6 0 0.
14 0 40 0 AJ
'a-I 20 a MM 1 0 N4010 a'0 04.)'a6 02 11 0 S IO 4 0-G-4W ,a
-4-I0 4 40a 0-4 V
E-24C->- 144. 0; 4 0 3 k-0 r r.0V5 - 4.6 0 400V0X0.. 10 -0 -- I" C -4 341--I
01H4 o0 p0X aa 0 0 x0
0 k.- m4I 6 0 463.40 >
0% 1-1V
SO C4r
Q,.- 2 kV -.0 0-4 0 0 0 0 0 u- '-a c2c - U3C
NO - 0 0C 61 3 4J a0 0 A003k 0it . ra0 H 0. .kWDoc 0W
z W 40- P,400-
N O IC 14 P 0 IO('4.1 >, m .to w 0. In 0 -lN 0 .C0 0 .04
4O CO I0- .1 00 0we
V..-44a
40 W. 0 .40 0%.-1- 44 0 c.A0.c
0t 4 ONOw'0003-4066 c 41V'0 0160IV 00(>('41..IC411 1-a .X04 0-A OCO 01 000
.N4 M6-0 N O-..-O0 0 to' >10 *- 0 1444-.2N--- 0. 4 4 141C 01' C4.00 4 CM141
0 0 -14 N0 -.4I4 0.041.-.4 - an 00Nu4 0 '.
r 4 -A .00 '004 014
044
N01'D
04
0
0
0
14
0o.V6
4 0010
6006
030
NNc -
to~
60
00
to
-W.1
41641
U) -41010
-W
H41S.
t- ) 0 60.14J 4 0466
O.401 6
-40>"
0
-0 C460 01 OQA4; 4 40 001A 14-. C -- 0 6 6 CC. 04 aO Ic 0
0 a.C 44 -4 4 N4a60 a3-0 G04.3 E-la,
CO4, 00 5 'r N>o - 0r E0 c .6 0C10%00 0C
-0~. . -v-0410V4 U1-0044%) V -- 1.0 01 r -I

(1 *40so0
013. -4JW M41036-S--I IUN 046o1-4414
'A A 0 W
30C'0 60'0 0 0%1 , w Ov - 6.03C0 0 4 C 4
41 k 41 00%S.-l 0V 4 c-l0E000 N4 4 H01 N4 z V 0a0-v
V
C0
WO
01 14~
-oo060.~
544000.-45-0
~0
W10
'A
A0
OC0 0P
-I
VI % k
W040144J
4W, - T OW4J00C0 -
OS
4C 60
MC' 0 -4 0
44. WM 0-41 4 Z 4 1 40 641 0.0 -i 1 0 A.-
- -o0 a40J 1 c
0 40C- IX A
14 00143 1 -4 -. 60 W ia
C' 004CO VA..
0.c
-0
010 4 6 ~ ~
IM 01.-C0' 0 V%04C)2 0 41 4 1 A ' 1 ~ H Ct 0004V 14- A-.4
'0 041-
.0-40 0aW 1 U-
0 qtocV
Wm 4 10>>- 04 t 4A
0 H24 C0040 a-0
14914D0 66-
C0 C0.A. 0,v0 0 14 0> 0 0 0904.4
0 0-I 0
> to* .4%4'0 r' a 49 403.C -00
1 0J4
-I O4
60AO 20CO 4 in 14
V E4('H.4l .- 14
Q 4Q- N40 r0r 06 64 a4 .0 0
0%0 1W
I~ c %4 0 cm - 0 0 -0141
O O-.
C 0
I -2C 0
04114C 1 6 000 0-
6 Mt101 41 *O
06 a
4j C4V14
40
a.0 04
%4 0 '0 0 0 4 10 0> 6 0 '04 r0 46 4
mm 0 014.614 0 a060-400400
V 0 P4 001 0-4 O 41.
O. 00
* 6a, .. % 1U-l43. 0, 6 C 0. a0 a 40)
4 CO a10 to
446
NM
0
~ x
60 '
a00
0 0 4x
c-4
0 S
404.
0 a40
0 0.0
v0V140-.
> 40
N
'0
Nw0000-
CM14
-4414
410%*1
6040 4
r.-0
63
0 3. o4 0' 0 -4 AC0-6404
r%6 N4M 3.
02. a 0 0 .4004 2 041 1 00V 000 w a U
AS w
0. ONJ 0 4 00 en 0 0 0 C40 0 r.0u0 004 a
.0-r 40 0 CL- 4 - V0jo0r G r-4>% w V0
44644 4 . 641
4 0> 0T -4A 0% H4I 1100 .- 1 0 141
0
0
N
-41
0)
a
0140- W
31
0 a03.0>
o4 01 400. 6%4
~0C 60
41 CO am
0.4.
%040
0.44w0w0.-0
0 -46a04 V.4 . .-. 10
00 441 0>.& a 0 . Ap 4J A 0 & 0 04-6 .0 00 444
0% 1410 w9
6 S0 014 oo m 14IO g'4. 0 414 0.- c 0 4 -44a
I
.W4 C414 f r6 04.0g.e 14 me
1,I~.-.
4 W
6.-40A
N 0% COAI'l~~~
14 00 64 6 4 - M -4A . A 40 41G
0 * .010 0 0-414-I-c I t066 40 c a00
-4- >41146 0 I
z40 Z 4 N
00149 t -I04 - 4 o 0-40 0%:.-I 0
0 0 . 0 jVx oaw- . 0 t ).0C00 J W1
6-l44 Tt - 8 -,0 0140 0410 g00in k 014
6 4 a 1- 0 . $4V$060C$4A 0-'l t, S ckI
0 '00
0V 3 0ka
*.W is. 0C0 0- 14010
.14 t- a 40.00 0' 0
-4 NA
V14M W
oo-
A ,
.
r A
-0
0
0~l~ A
"00w
.4
0
~ 41
i. V1t
64. 0
%
4Wo
Sor. N'a4
0
$4! 0
14
0
0 $0
01466 -l
0 H1 .0.
- M00.
4*
"
>H I S 000
$ .P0-4 1 W40 off 44- 6 1 4 -
t8Z - ' ' 0 0 C M 0 0 444-40
0. 0ob w
.4 .0 -. >42 U, r",4
8
"40 % .00100 X9 0 e* "A 0.'0 -00
J,'0004
0 r-1- P, C4I
-4 a .W 4 I.01 410:
W 0 . w & o u..
-- A !. ..
d .. 1 %4.444
4163 a c3
06E~
B- 18
14 -40 -w
4
-4C-. A~-6 MN 4fi 3 4 0 1 6 C
0 ' NH 4u-1 ~de4 03
4
02 ' 4
..$44
3 1 .
,I
n .. . R a CI O
G~3
3=5 Q
..D3.~
4 100
~
9 0 to V 9 0 -
o o4
44,
1z 1 ' a L* 01 -40 0.
ac14 04 014 0 6 t140
404 6''
.414* a6 4 60 0-
0 8 4 3 0C
-- - - -21-4 0043
a- . 45 M a :4
- i -s o1
0 -1ClIA .'- 0
643. a
44to>
4 05 14 4- >'.14
413 10 4 30
Coo04...oo *oa-i 041143
.-4
g o .4 4. .1 0 0 14 0
.4G4 00 O
41
-cW. I %a 4j4.j4..aw s 43 06lZ a 00 64 -4'
00 040 COO0: 1-0
140- .40 AN~
84.0101 04
cm a -4c!;
o:3., tp 0 IS
isC 0
C. a 4 300
00 4 0 -034 a4~ . 4 . v LO.4
CJ.-. . 4 149 X 0
'84 cc . . 0 6 0 4 C 41 0 -.0 0684
$4 CA -O 'j- 4 C40. 0 -443C)S

-4-40-40
00U 066
006 '400 .0 'A6 0
W4~ C. 0o~
414 0: 403
6
ERR.6 1 0 0-4 ..4- 0 '040t 0 . 0soi CC a0.-4
1-
>1 1J0 .0 ;
C
mr
0
'
- 40 84o
-0 v
--
"
444
.41 w
~ a- -
r
0.8
1
M 0
~ 144
24 6 414 Or-.-
IS4CS
' -
04_ 430 0. U+-4. 0.4 040 0 4 0J
k00 A4 0
*.~~~~14~ 00
". a z 14 1 4 4 0 - a 444 0 0.
0 >00 00 046 4 6
C6 0 1,' 0'8 4 J 0 i H-.4 0
4
V-01 k..4 .~ 14 4 M4 ,-6 43 . 0 1 0 6

0 44 44
t
6
4 C 4 A
0
H< 5
a3 z-V a. ~ 0.
00
14 40
0 30A
0 4-44 a 4- O r - 0 . 0 4C
0, CO a?-L VO 0 0.6
t 444
01 44
0 0 r- 11 00 - .- 0 C440
0 4 j-~ 0 0 0- Q C0 M4 a. 0 4j. 0 0 0 - 0 *I4J C O-,

04 X .00 6-4 0 o 06 C 1 NC)0.414 0 -am 80 4 4 .-
0 g~ 4c > C6w u4- c 04 6. 6 M6--4-4

... 443,V
:3a
00 ~ 1
1 00 J
14EE-I
6O6
4
-. C
o
0 . 1400
a,6N1
:1 H0a~ 03,43.
0.614
EC' U).. P40
)E 010 M0 we .4
ji~601 w
4006.
-40om
W0 a -
C w~
'30 . 0 S414% -
0, C C .0 CM r. 0WCC *6 *0 3 1
43 464 43*.f43 - CC4 043 CO 0.00
S 0 .0) J: k x 1 4360
to 3-.. >' 641C
to S 0 43 6
143 6 43 IC * 1 04 6- 0, 'a-1
C) 6 C4W 0 04a a 6 0.~ 66k 0. 003A 41 a 0
6-- -4 3C O3 014064 0 -
A0 J 4J'0 C 61 .- ( 0 043-C.
00 14 0 0. 6
-4C C 66 . C430610 C 31
=a t 0W. 46 0 -.C
- 14 0643 4 3- 4j
-. V 6 k403 00 00 3
41'A .0 W C4J
-40 0 S
-$ 0 43 .14;21 a0
4 a M Jw 'a 0 0 114 14
U 6 40 31 -~10,O 00 k OCS &C d
14V8~l4' 0k
in4
0 -0>4U>1034
C 1004
.4 04(U1
>,, '0. C14 1 A . A 0 0 4 - CC
C.6 1Z 0
06 6 I0 64 - a0 0 4-C1Q0 - 0 Cj
3 0 4 w 4 3
6 -I C - 3. 0 8 0 04
140 6 4 .M -6.00
3 13z3.4 0006 40 I3 I 4.11 0 Cix4 3
>. 3 0 . 00606 0
00~>- 66)~
43 -- 0%0 IVV O-4 4 4
04- 1W
0...0 C-1 0 C -ig,4
C. C-0 4 k0 00 W.0404 t" r.-40 0.Ci~ $46a Na q043
0 0- C40 '-40
a40 04 C 80 1 C. *063
80 le 14 NCl
41 C44 64I4 60
46 43 0 C4
j4 6 4. 3 0644-4UNW
(4 Q- 0 C4 S 64W4 '0 r4 4343 04
8
tp84 COD Z 40 -46.4 M43 C 0 0V00
$43 4'4 6 2C 14 04.)144 on C -- 4 -6 ' 43 06%D6..
GO 64344 06Jg 44UC 6-0 4 40A 1-0
6 4.06a C 31 .4 3
6 *0 0-: U) 4 . 43 40. 4440a 0 0.
H 0 c140 0 -- - 'p44 100 12 0. O 0.
%D0 8 0 CO 444640- A
0
043 0.3014 04 '
-'414014 4 164 .-
'- W 60g. Viet%4-44i
000. j~ 4
.1
%z.3 5.6~ 441
O:-.44 QI Z40 to '4'j16
ccC u ~ .- 44 0.0; w 40 0.Ls0 .L V
4360 ~~- 1 41146 -. 3.06

0* >1L1w
: 0JC
14.2~~ 6
43 4 a~
0 0. *
0 im 1 m o o
Jq
4A %4
'4 0 .' -' r.6 40 1

0~~ r A~~' 0
00 4 a
14
0 04Iij Iwo:;.4u.
I~~~~.
l I. - 1 j. - 4 . 04
;JO
vi 0. t
B-19
4 - 3- r4 V40%04r
>04(00 40.t-.4J4000U0O~ 0 0*-44
44 ) 4
04) M 4CM .30 0 m 14000 04) a
'0 0.4 0.4
141 104 1 440 c 0 400 4J
10 *11 4)~'0 04j 0 00.44 0. 004j0
4 114-0 t 10 c)&O AC0 0
0 .0b. -A cc> Q-
41 gc 040 0 00 C
-1 M
41 c 00 0 0 M4 41
c01400 .41
C 0 00 061 mo 0 41 w ---Q0'V
3 C 00 E
1 00C-4 0 W -A4-M411 C100 Q4 1'
0 0 6.0 'a x 0 a
'0-$4
w 4j1 k 0 0000' 64 4j
~
0 0'.40'0)0. ~ ~ 9 w4.0.60a 0'.~to>.
4
0
A
0
~05 C .
~in4 q) '
1.0$41.
04j 0>C4
M
4.000184
040 0
04-' o0 00
0M 4 0'
4jw
C
1W
4 0 6 4
0
0
H -0.4, 00'.(D G, c014HOOCCC'4
R0 o' 4'j 52
0, 4)000.0
0 4)*.
ICm400>.l 4)4.
ON)C 144)). X 0~o' >. o4
(04)0-N. 0 0 C A 40 0r.> '0440
0 000 440 404.)0 .
-4~~ $4 -4 0, 0 0 4 00 4) r)44 54
.0 H1>.C 0.0
W1. -. W1-,4 *.0.4 4 0'.40" 00 CO
.I)0. 0)0 0 01.A100
r( JJ 0 0 0 00 0.01544104 C. * 0.40 > C 0 V0.
00 41 0.0 A4 H
H u.0'
05 4)0.
.1-;o00
1.4 00k C 0 0O .24 ) 0 01 M 4440) m C6- 4'.
00 0 (04404
z 04.4 044 4
4 t00 0.A4J
40C 1N -40 ri- 0. 0 0tow 1 CO.AQ
m
0CC - 0aNO l
.0 41 0
~~~
4Q 000 C'- 0 000. 0
'A00414.4 m --
4)1 .: 0 0.2- M'0-
0 1414. $.40
4) 14 1 'AC 0.0J30 CH000 41 4.4L '0 0.0 0' 4000 c L '0r
.040 ' *-4O W 0. . 0 0 C 00 4C .--. 9.. 4 0 -4 - )0 U
r-0 0 0 - a0
.0 WO go 10 1 'C00 a1 *'.4 4 (04 co4 0
W4) 04
0
WU
8 W1
C 0' >C
0 0
0M
44
0W!
w. a4
4 4.a z c4c 400
0
00- 0
- 00
04J
44 04C
C4 Q
00 $.44 0r 0 O 0C)
14 r0 0..4~ 0 0 004.u. IV *.00*0 -0.c cC
000 0>0 0
0 414.0
01 4 14 C 0'0 00 0 0 0 44 4J *4>>N C014)
O4O c)(000 .0C rOOC'H 400 5,V.V
-U 0 l400 0000
*l w r 44 00j a, 40C 040 00
r'- 414 AH c00
140~~.
600
44 to 4)00 0 .4 0 W .
C 4)0"0 01 042 r.0). 010
01~
00 .C
1- 0
0..0
m.1~4 c 0 60 & 4014.0 c. .041 1 (O 00 I' 14. 01'
a, rNOON c 0 .4 .06.0)4 m.CO 04-0
a> 304.
= . H cc M4 a-to
2o 444 40 0 4C r.400. Z 0 00 4 >.0' 0C
C0 '0 -4* 140 )> C'
-1 '0 <.0 -04)
0.
00.1w r)1 0 .4
0
; 0 0 a410 4
Cr01 1 0 C
040. P c0" .04co0 0 .4 0 a vto.0 4OC. 4 010001 Cr04)
04-M.0-4 O >. 04;0 4)00 504) 4
X'0 I0 40.4.
- )a 4) 0*-100 toOC 404J 0
00 4)O
w c
-'
.404
0 N4)90
, M4
*
~ ~~ 00
01 Z U
CO
44
4 0.00
.0 0g
0-
00.CC
4C-.0 4
000
)..---0O
r..4
'. w 0; 4.) 00 .0 O>.0- 0 0W40(- M4 W)O ) W'-O440
(00 Ho
0: 00. 14(a-44 O..4
c -4 04C 0)0104
1-04 m 000 C -40
'm 40 -. 00 0 0 41 c00. 0 0 W 00to.>
Z-40 0
w<4. 0 04 0 ) 0 <to4 tCor.4 (a0.0 VNO
41 .40. 0 Vr0 4 L
4-0 14 14(0-4 O.00)H
r 0 4C : a - 00H
0'
c W 0 0 V0 010 C X M 4O.0.4.%
>.. to CA CC0 0 M4C-ca 1400 to)c EC;V40w' 000. 41 0H
0 m040 r r C 44. 04 t00C' 0w000 . 0-4.,o

40( >1 Ul - 00'C -0
.40 .O(0N c)) 0
C0COCl w .- 4441 4)0
r
z 41 )00.441 0) 0) 4J
0 A 3a 4)0H 0
004
C M a,4
.4 00.w04>00 4 0 .0 r w c 0)
Inr -H N4 4)(' w.41 .40 . 004
0 c)*4-. t0 00
4)00(0>-l 0 00 S 4)
0.00S30tlc .0L '0 W UN :3 4 0 Q C N -
0100
0. .0 w444'4 0 :, 01400 O. ar
00 '4 mo 4 000401a 14 - 14 00 - 0000.0 4)0
i040 W0 -4 c a00. U 0 C 4 0
)-N >14to1w-O.0'3t 040
0. H
4100 4.e4to'.00.410
Q .. 001 0 V4 W404 H4 '0 W 04) 0.4 *
014 0 * 0w-4(.0 0 000.0(14
w 0 ( M0M0C.4 .0 40' t 0-4.444
4',TWo E .. 11
a,. X 0 N. a 4)440440M m40614.0 4) -014) '04
-' r4 0 to 0.
00440'U->. EP ' 4)-4 4)) 0 4 -. to of 0. wC 0. 0 0
4 ) C0>.-4 1 .) V M M0r.40-44) 0 4) 0 '014)

1 0.000V
P400 .400.-dO 0.044 144) a 01 w 3.0.4.0.4 00
00'~~Z -4000 I~
04).0to 0 .410'2 4) )
-. (0 '0440 .00000'4 0 0. 0 014 0.0'040 4)
H0 (00 .0 M) 4 )0 03 )H0 (0(0 00
(04) 0 m 40000>0 -0 R-(04 0 m 0,.4-4-00 r 00
4JrC-W -4 >4141 m0-44 0 OC-' ,0 0 NO.4u00.0
0
r- ' 0 C4 . 'HOU 00kQ 040 (a) 4C0'0 CL >00> = 04
a .0004 V 0 0 00 -A0 a -40 NCk 0 a-4004 001 0.0 - 0 0.
04N . >O 1 $4 414' 0 N>. 0. 4
Z 00a 1400M >.4)01 0'
0
0 -'00.1
0 6f
40(
0
~ 4
14
0 .4 .
A4)
a 00I 0 0 %44
'. 0 $4 &C.4 0
0
w.4
Ct r- '0 .
04).0
H0) U(
-0
0
00.00 1 .
0, 4)4)M 43do
~
z20 4) 00
t0"'
L14-.Q
H 0414 YO N0 VCCO 0 0 a441 ".0 101 0o 0 -400a .4 '4)
o $4 A.
-00.4001I4- 4w00 0
0-00- 6 go- 0 >.3'e, 010 .0 4 'a I IV6 0
00 -4 0 0. (0 000
6.' SA14to
0 'H 6> )
-4 .0 4J4-A J14J 0 54>.4J a,4~ Im D 0 .0 64 4.0'
ilicI l:4cIIC- fz ij ogjj ! j4;j U) f 4:c7k
w CA01 U 4
111 W 0 4 m .c -4000. to
%4~~
C-4 C-
04
toNh 0 X a4-J$ P1.4
0 -1 010. 0 Cho 04-A- 00 84a)0-4w

04J0 H -4) 4) a c)1 N 4j c-
-.4 -A0c B4Q 0'- N 00 0A co
B-20)
-AN M.0-.5
% 4 4Jy 90 44j 1U0%, 004
ca O $3906 aO..v :0 tyO:0 v 0 = e k4a I
C4 00 MO 0.4 k6 aC 0.04 aO a nMO 00 a00.
5004~~~C. 0 000C.~0O 041 w.-4
IC O-4I.441 M OC cc1r .sgO-
aiw4-4
0. ~~V . c ok10
A~ 44
Ci-3 C A0
w 0 e18 0
1 !.0 SA01%S 14CO
-3 0V0 C 0 0 awM
16 1 4*
40 0V4 0
3.0-
- 8A -0U>1081
C *6Vs
44w N WO ON44
6. 00040 1: C 9 61 60a4
00 1 .
6-144~6 M 04Z 0 % a0C -4
M04J-'
0.- -
-4,04.- C
MU 6
0IM 4.400
tiM. 060 MOO 4)iM>
1 4U -"M -40
. r 4 4 0 2(0 COO '4C 0-1 0 41
V .044P60
a-Oso4 gMt-046 69' 0.00.c
$04
Q U)64 1-' MV 0' 0 0 04 k -43 l
0 CO 41 41CM
a V.0 f 6- M1 1
04 MO 'Oj
0 4 C00
.'4O~~
1
.
0 c ~O ~
-..
. o
M00
0
04-
W0i Mr .
04 0.
4100~
6 i 4 0 OO*
1 6
~~~~~ OC 6 4 4 -o0M ~~
41 0
164 as666..C10
-I~ ~ *61C.. 9.C 4 64OU
M
0
CD -4
06
6
CD 404
C
*
441
W 43 *0
4 OVi
1( .8O 664 -4 gCo0o 0
M % A RVi
to.44
H0-
0
H0 C I0014 .45
4V C
1 0O41 00 M-.40.01
1, a c6 0 w :Vv -40. Li--
O4 ~~ cc0(
60- I
C650li
M O O-.0
V1 61 00
0i~0-U
0m 4 W~
4
0
0
. .440
r0
MC
r; (A 0
oat..
0' 4-0
0M1 0
00l W0
0060WO-a
30
~
10 Q $a1U 0 OC
6
01 4$
0
0.
C
0:
'
,
1 r.ar
VsOX
i 3a-4 04$
0044 *U -4-40
.0 Z011
*j00
4d c0. 4i -4 0 ki V1 0 Uq0 000.
C)
COT
UO 1
-s
0-4
0'16i6
A
6
occeM 014 A.J1~.a)-
40
~
w 4Ma
41 60'L001 41 4. 0 1 06 C 441 0-:. 0
60, 314 60 04 4 C40 A4 -40 0 4 *n 040 p00I
441
0 044~LC
&-4j>.& j w4.40 r 10 aM 0 C-sM-SO * o41 A 0 p6"0 0'.-0
mY z aW 6
A w 4C0>-41 0 4 0 . 0 4j -scO0C e 50 0
V 8 s .- 0
CU)!w .f0 jv COC04 k~~ 1-.
as C0 06 41 M4.4 0 - . "00
$4..- C3 . 6 6044 OV0Cso ko. 040&a0 * 0 4-.5 .0 6
c' :3 14
0 0
r 0 V 4 a >.X6.0 1- -4C 3 oCOg 40 41. ow4414
n 4 t VVCS 0 50CP. 0 0 %''Ci6
0C Mp 0
V0 C 6 M-0 O W *O
4-5-54300M
14 A~C
004
~ w
1C 6
4 6
66.
10k
*0 .. V4~404 0604 C
4~0CCQ'6 0
6 0410440
i k kM
4
00
004j
-A 06 aS 4
00OO 20. ;> 0.40-a.40V4 o
.00 C.-AC.-s NO 0 -SC - CM3.0 0 -0 .60
H kg4 .. w 0N -1 Mj0C4 41COMV 0 14 a H .-- w0 $S46101Cs
.40~--2 0 06 40 %eO C 4 00-0-A. -5 CMM S 0V0 mo10 -4. -'4a
z1C , 00z . 6.- , - 0 -00W10.0
a0 041 a-O 0 a 0.
((O.44 0 4 6 0Rf" a 60.3
S (0.
0 0 4 C- .0 14 k010 .
U0'-'144 4: .%W Li604JM411 4100 O.'-- . 0 - 009XM u a
6 ~ 64. * 0 60Li0AC Mc 63.00CV -0 3 1 0 QU-04>

x0 A 000
4 1 6 -- 6p Cj
VA-4ZC 9 0C1 4CC. 4 1r4
I 0(4 O4.14x
C.- MS
500 0 3 60 00 0 ' Mn 0.-4- 40.- 0
4.
$ g 0M 4 -40 4 04
36 -4PC0%0
,I E'.> *r 041.
' C.04.
0106 VI4- -66 CC1
005.4 C)6- M 6LOC 0.1 M
Va'
0 0 41
40-OC
0 *11 > 010
0 0 44..
g k no'
40u0 20 NJ0 COto 00.iC
0.- c16. USC00 M
ZHO
0
140 6- 04GCa
04$4 -4>. 'C 4j "4. . CO OMO (4 -44 k4 toC100 Aj
41- M4 USC MO- $4
440 61 U .' 43. 0 O P 0
OM-'
rC' C 6; 0 40 uC0 04 s
a 02 01 1 . .0 00
aCEn-
0.
07 0 004
04 0L 0D30 0 .40 -- 01 041
Ca
,-0 I00 4 C a4--OO
M10(.
E i C. 4 &15 - ..4 13, 0MI 4 -1.04J -M M V44V
40 0M r. 40k U0 6M a- A41 Cd -4000

0i 0 - 45 04) 4a1 a1 A E.0 -' * 4
MC CMa
r-. 0 V CA4 .i0 0 - L 4 1.0c 06.i-o40 3k.
N0i 0C
v041 0 0 a0 6go C- 0 01 N.10 004
-40 ' 4 a 401 r- M M 4&
: . - 641r.aI. 0 COC -0
00
5000- V6 .0Lij 1 k 0 4
Z.29 . 0 W 0 i 61 04001 vuu 0 a . L4 - 0
M4
C.4-M
1
0 w6 0 .V6C 4 41 04 w cCCI 4
-'4. 0 0 43i.. q -4~4Q
6 S - 4
111 14L 4J(L
0a e44 $.4i-'
iCc80 IC %.4O.-
IC "43 Oil
"A
3. C 4
~
1 0a
Sj0L>
0 ..0*,
~ 0 4CII
0 4 0 - 1 0 1 pa4 0 $.4
0- M C
i4 -M V1 0 Cs4f M"N004 -0~ '
6P 0 1 k -1-UM 04 a. 0s-'.4
6 1 .
$ 0 3- 0 A $.444a
0
10 a0 ~~ so .0a.:4I46S104
~ ~ ~ 0
~ 0
41 r- 4.
a iM4 S~fffp44
C~ ZVM .
0ji
4CA
1' ' 50
v
I
10
ilfn 0.60,-. -0e

$-4
B-21
C
C0)0 0 14J W4J -*C r. ID *00 0 4W CV4rV
W C.C0
041.1 00 0 0 V' 0 44o
1ol~44 1.. 0 0. 03 0. 00 .- 000
4)f v0i>.o-,
aOC 1 044 0 -
u o 01--.I40 41 41 , 00104 04
.1
c4-1C Z-C 0 0 IN 1.0 0 r0 41 0 . 4J4C 0
*0ECO 0o- Wr ( ) 03 DM(D1 : V04$4J
-4 0. 0-4J
4- N .. H 40.- t-A 1404a O .4 r- w 30 [
0
4J W Id , :-
0O0a>0CpI 1U U 00000
.Ca OV"0W1
V 0 1
~ a
OCO 00
C .o 0 a0 O w~I 4.11 0o0
4 Cr 00oo
0C 00-
.A-
toI.
04! oC .M0, 9.0D
9.:3- U)U * 01 4-IC 0 P14J0*I 0 0 wo C 00

01 1-1. O' O 00 -I 4041 0.-1:4j-V
0-H .4W
00 . *004
M0. 4.I
C 0 w0 20" .0 0 0
0101 w 41 E k j .C 'A 0 m 0 C1 w4 0. 0 W0 1 C
14
04 W4000-sOE C to 1411 0 4 L 4 ri 0, 000 0
00 0.a14)0-4>. jwVm 1410 A( -w.C0. CC. -43 OH
t0J (a w4 -0 C1 0) 44 rOCO .w o MCE..00 0 0 M0
IZ
w04 041 4 0- 0y4)0 114U, c041 . CIO d; 9
Do m. .0100 Lo V4) - 10Cj n1 U1 4-1CD $4.
. I rj 0 .0014 A 0.4 CM4 0 o:Z ch 0 00d . C-0 0.10
0 0 Cow 0 000l 0 40 C -0 A0 0 -0
44OO ' .C C04
- 14o Cv 0. 4 0 4 > 0 0 - 0M 0 0 M 00
0.H 0En4)C
0>, (A1- 01 0*4 0 .4.-4-J0 0 900w43T0 0 C. .4) W CO 4 :.
U,0 4)0.-WO ra1 0.4
(A4 C0.l Oo$4oO W0Jr ) -
a) 4 4Il 00- U410 *) 01- to.-14~ >4100
C %01 m - :0).to,4 .C N0 to 1 0 'H 0 4J0.W0 4>.11
Mo w0 aC) w. w 1f0 U4 0 00 w 00 A00. l4
w . ,1 Vfl 01C .0 10 41 to0 0
to vu a) *1tl0( o.04 0140 4) o00
0l a-.N N s .0 . 4. d0. 1 - >0 -0...4W0W 4 101
toC40" w~4. 0~ C-
ZOr4 a,4 w
ro0 0 0110.I 'a .. >.n 41 004' o-VW
0 C C 0 0
4l-.C 0 00
> Ln
0 ) ro 63 0 CC-W a) -A 0 t 401 z 0r 1 0
OIC
a) 0 g0 4 1 to 0-w ;3 a)0 w 0 00 00 A IU mUN ) - > 00
01) O J $4OIi0) *
(A o. OR n Ew 41 C 0) 0
N0t 4JOI C,0rI-0. 41Ja g a 0 0 -H 0 0000
S.14 w0 w- W00 E -10 14 I 0
rC)44 0 l 0 0.I
4C0 >0.w .0l
a0 4 4. ) orp~ C -.>40.0
E14 F0 000 400'00 OP. 4 (
0 ETI 0 . 0~ 0 0) NIl~003 4400
aam0 4 0 0 .041 1410 C ) Z to 0 1< .4140m 44014
02 0 -010 41 01 4 00001 0 3
C 0) -4M0O0>1
00w
r- uC
-10 >~ 0. c
.0
00 (D
0
CUa
W
. 41> 0
rll0J
A41
0) w00).1
>. 0040
)
N10
U r Z W-400
-0
014
v .4to
00W0A
0 to 14 r
4 4J4J 0 U4J 4J 0 4J 0j) >4) 01 -M1"4 124 -C 0 0O~ 0V 1 mo%
-
r-01N.CE.-44.1L COC 00N O f O w~
0 0 4 to - 0 C
:3 OC10 ZIl0 C 04. 04 ,C 0-
C VF 0 1A0 P4
> a:D0 ro l C *0 *4J $4 0o C)>
C. m 1 0 D0 04.1 40 4.1 0C 3p c
- 014 4.. 0)- (4 C 0 V~0 IC 4 4j 0.CC0 4014 00(a
-H .0 r0tn(1 0)1 0 - aC 14.C40 0 0 4C0P 0
$ .100C. 0004.
'0 Hz00 .(Z0'0 :3 0 C4- 4.1041.to 4
z r Ca) 0 .4 IC0
(a .. O * U) : C 0 1 IL M4N :
4>00.) .lo r..
-h Io a) a~ -z C(41- o' 0 >. 4.104

W4(. r coC4.1NO U >
14 .1(4 r 4 u-l.-4
' 041 - C4 k41 U, 9 4 1W2 d0 J 0 t

lo0.C 041 '44 0
~OC t4
0
C
0.000
0 04.14C1(4
tC
I- C v>,I 00 = 0 0 0 0 M
000)QW 0 W 4M-l t 0 4)V 04 0 0 4J 4
o 0D m1 E--I 0 r M 00r.4r. 10 $40
M-t % a 40 00 -0 m w0 w "0V40
C 0 , 000X 10 14r- I 4C- )0> 001 N 0 C41. 0 0 0
E C 0C1 w' OC.4j
(A .14OCn l13 3 r0>090 44 4 J
C) 01 t0 0 C Z 414J 4J-42,
V 14
N- 4) Wo.014 0 eq .- C4 0 .-
r 4.1 0. (40
10- 0 C VI4
0 0 0r 0 4
0 N 0 w 4 U 0 C C00(X
2. C.1 0 01 0 r.0
0 1 . to t o s-o C 14to , .4 II
CL t-00-40 to to0 0 00 l0a
x < 0 .l 0 0) <4
0-40 114 00
ca 0 004-I 4.1 ca0 0 'a0
0 co0
-I 01 4 Cg
10 0001.EI
0 0 0
r 0 041 01414 0-l 0043ICw01r-
0~~ 40 0.0 04 a1 04 4
14 co 00144 $4 n to r4 w 0-IV-
V0 1. a i .00 ii '0.I&V1 W010

(a00x 0 O.C 0 r04- 100 0100
04 C v w44 r.
-L w
Co H -40. V
0. to1-
0-I .0 01 to 0a1- -. 0r. 20
12"w. 430t 000. eq
4N 14 10 00t '
cc' .11 ,- C 0 000 Q 1 mc 4.1 0. 0C-4
OCO 1~ (''o0OCO
044 t-JO
1
OW-I0
4 4
0 C 41i
00
0
C.-4
04.
>
.- !~
' 1 0 4
00
0~04
4 1
01C 1-
0o 0. 03 00 N1414--'4 o-- 0. 0uL4 C
L)00 0 44i 0 0 014 o 0.4.0l
-. T0 100-2 a4114 14C M0qa(A.4.0
0 4j U
LAC41 000 W 01C la 4
V,. * 0 A0
.0 C
0. 014 0~- 01 6C144 0 5a 0
1041 00-10 0 U-o*-H01-0 Vc 0.)- 941 -
IL 000' 0.-A0 -0-4C 0. *C OO01 A
14 - - F~- C4-.w 144Jm2$44(>. 0 aII
-0 04-
0 - (04~-
$.41
(D- l' a3 E,4 4) 53-V- 1 00 - 14
, C o 0uw
$.~~~10COIf 0 4.1 Ngs 0wc
4)c Im. &, ~1 -4 0o 000
4 I 4W4
0 0412'(0
1400.
'0
4 0 A 4J AJC %
IA014.1
0 4,- r I a 0
a, r10 I1w (4 fa.- 5-4 00041. 0% (P.0 C 00
u a
0 ~~ g
0 1 A
140
0A
Cc 0q ty40 q
0 r. v104ON
0104.
0-03
4.1A
>.I
U) U4rIfHa, M in w O4SO .- ON
0 (2 A 0 .CC-I SIn-
0 -- 40v NO.- 414 .4. 010 0000-0
- 0 0 -0l5
-1...-
0 u..000
. cn041~ 4 - % 4,-
ONIA
H w.00.-
414~~~ .-. 4,4 0U- >,0444
B-22
0' 4 -4 n o40 0 % 4 0 9 4 00 0
*.4 00 -4 -l 0 4X4(0
- l
0~ . v4 I.- C 0.
C 4% >%a 00 0 N 0C -044a
0 O 009... 41 .4 C, 4 '0.1 0 0 4
1; 9L 0 W04
0 ,
ON M 40 0
-.
(-4 En Do A 0 0 to 0 @ ..4 0
4) '- 0
~0. '~c-40a 00 -a w
0
44 0%, 0 0 00l V14

b 9r -.
.
0. 40 "-40
0
. to p k0 0 0 0 0s41
lb to 1.
"A.0
.
0 0
3O
0
0
a.:l
ci.
re 04
00
4 400 -i
4
lb,
~ ~
.00 Ub W
4J0--
.
0 0 0 -. 0
'0. U4.-4
4)L .
t V.0 C 0 0 >. S0044
lb 34 -e0> O"0 0
44 t . I
.0 N mu
04-4n.40 0 0 0440 -.4 004
4) 430 0000 14 e M 0
G
c4n' 000)4 v0q n -. .69C
W s
H
*. 0-0' 0 I lb 4
44 0 60 -j 00 COe 9wV .
H0 M 0CN 4V a 10 -4 4 r. 0 ooA w
a - -C 0 0 'C L
0 1
344' 1
0 N K~-.. sI
vi
04)0-M044
to -4 -0
44 C .4w0 4 4144 44 0 w 0v0.C V4 .a
04
41
0
.4'
0. c4
0l
- %40.00 a
0
1040 0
M0 4J U 4M0N W.4a
00
-4 , 0 4
a4C a
0400:4)
gl -4 0 04.
C004. 4.444 -4 .s0
a, .- C 2
CD
m 00 0 0 444r Oo .4 lb 0 x09 1 lo
C>. 0 44'
0.4 0 4 . a-0 3 0- 004
a0>
(00
0.b t0
00 4 0o
w Ok
44.4.
~~ ~ ~ N 40 ~ '4
a
$J044-0
0.-
0
3a g
00
0 44 04Ob.
00. 0
a,4 u 0 CC%. 04-0 4)04 . 40-H. '-..C 4 ' 0.C
V.C) p la 0 4 E
-'. r. 0 $4 $ ?00alb .0b4C) u
4 -- 44 00 4Je)04
U-0- I4-4
Z 0' -0 -
0i M*
4. 0 4Z'. 04U~
44. (AE-C1.0
( 4- 0 9 4 14 ol 44'
-. 4 0 to . 414
r.41 wu.' 0 0 a
v4)-O .C rn41R 0.mi '0~ , 0 N- 4
0 a ( *a
0. 0 4) C-.0
.
Z- ty-H0 . a400
40
4 0 z 0 . 44H m V01
0 0 u
0 . C0" 4.
Z4O
0 - .4U
0 4 4n v 0 0oc 4
044W V H000l N 0.0 0 C1 to .1
0. 0
00
m
OZ-4 P00
b.,~ 4)'.. '
040.0 AtjUCO4 00 j -4 . 0
020
c0 0 O1i- 0 0
044V 0>
410 m0 *I..
4( *1'44.w0.40
1400 M4 -- 4. .. O04 V4 -4 -M
00l0 41rw4,g4. 4b V
4;b0CE Hp41iw 4 kH0
00140H c0.
00C4 )C4
0)a uV f0
.03 G0 41
C4141 41
>, 004 NUN 0 .40 -440
E. OON.04 0 > X0 44 E. -4~
40 U 2 x 4
-4 44, 0 a -4-4.4c 14
Z 04 0.440)40.
0- w)'S a44ta04414
>. 00 lb.4~ jcb 100- M 0 .4)0 041 00v0 60.501
V .c 0.Vror
0 r.
44 XC 4 Cto
44W
44 0 >4 r. U3 r.0 Q 'A 0 $4 ubIn4 4 C$
tl0W.0 40 04 0A'
u 0C
4 41 1 41 .0 *404404) 404
-3 %04 V k
M 6 4-4 4O0-4j 4) a so 0 '1 M 0 4.0 0 M0.10014V 4)O 0 0 P.0
44 0 9 41O O 0 C IV >1 R*'-0
4 V COOUO 04)4)0
0l to '4
kV0 00 to 0 0 )o
93
0' b0 *3 4 00.0 0 4)..
V0 400 0 r 0C 40$440
0 z .0 .444) o .>a 0N 4J 0
044-44000 I
0030 0> I04 04
44~ t0 W" 444404r. 00 0 4 0
U0 0 V4 lb 0. 0 0
02 'lbm-4 w 00 ( 040 0 a)
At00 >0 k4 00- "3 0.4
4 00a I
4004U4
04) Ad 4 0 0 0 00a .a0W0a 00
C440b440 > 0 w0 90M4CU 0 0
4 3 00 ) 0Go 0 0 00 w
M%0 4 ... 0 0 0
0C C 4) 0 0 O40.-q)0 a ?A C 4) 0
0 I C3 0 *- , 0 4 ) 014 - 4C0 44 a 0: 4 0 J. JC 00410 4 0 H O0
04a 0-4V V L
a .U 44 0 04J %%
4)40 .4,0o41 0,4a u03400
000 4 A0.-4 0 O14 -4440C 00A 0P 0 0
4J
4)). 0a 40 04) .- 4)
E0 Q 4 JfMU 0 4 4 00
02 C V 4. 0 G O' I- 0 14 4 40 0. $ 4 04 040 0 0 4400 4 0
000 43 44 00. 00&04 44 4 >b
04.. 000.toC 4_ 0 00004
0 >00 $44 0V.4 0 40O
D'- 00 $4440.0 w 4)1)
00U 01 0 w . 0 0--0. -0.4
020O r. 4' 4A4 C 4)40 11- .0
to-4.
044 00 OA 0-4-0 '0 a 04) C) 04u0.
0. > 0 41) 4.%0C M '=44OR C00 lbA4 0-4
HO -r4) V .00
44.
0 0.0 ".m-4 0C.1400 -4'a00 0 a
a 0 20
o4-0r, 1 14 4 w so 0 0 4-4 M '
0.4 *OM 0 0 0 03.s4) 4 0O. 4.4 -
C4 0
a$40 5 3 -4
m4400
.0
00 0C 0 r. aR4 i 0> 044 04
044 a 0 0 00e U 10, 00(00 0414 > 0 .. 4..
. 0 k M.000 9 .0 41 a > W00- 4141
H 040C 004 046,0.0 44A to; p% 0 004VC40 0 0 4
' 03 c 04 $4V V0 r6M0 - FA 044I 040
M44. 0. 414 w0444 4 00 .44 r4 0-4 04)
)-4 0 a 1 -4 04 4 4 09 '000 0 MO)--40.40
w0 %(04 0 a4)40-
H
-4)0 444) 40 0 so 00
0 4 0 0 W1 0 0 0Cq3..-41 4)04) 0 0V 4 0
0 0 4 a 2 w 0. 0 w 0 E . 4 0 4) :5
'n*0 .. 0 0- 1" 04) 0 3.
C>0. (-. ) lb4 4.0041 0 P004)
-41-
Q1 40 .004 20"a 1 0 -. 400 043
00,1M 04M 4 4 0 4)t0a ,1
0 - W 0 10 .' W-b.-4 444 0. 4J 5
4 U00 is 0 04 01,20 r> % . k -440 0-
(n4 1- 10'
0444
044.
0) 0 OA44 jb0- 0 4 4 W 0-C 0 , 0;j '0 1 4 I 0 'W)00
Oi J - So 0 044 0 -0 m- 0 a.
00.4. 0 V) a 00 4) 4)
0400a
00
P) 0
U
00
- 0 -4 4 00-43
00 04.0
k4f
-a- 04 g 04el0 a.
00-4
U..044l4.4f 290.-
. 0
c444@-IM
a
0..A u -
0
0 ,no a4' C a
0 0 0r
-4000
go 44 .Qr
M'.3
0040
*4 Z
4-40
4
4
0
-400
.
4
40
o ~ 'A 000
4"5 o% - 0 00-0..
4.
0
0 u . 0 .82
04 -
0M 4 -o n 04W4 04W)O4J50
V -M0
'4a 0
u0' 044 o 40 3. 90 %49. 0 40 0 4
a- k00 0-. 40%
0O.41.0 1 41jj0
14
00 106 4 0 .4 *a
-W m4
00 0 4I 04)V0 04Q
to0 04 -
4A) 044
0 ."4A
>0.
O -1.4
Q 1"
In4 00)4 .b 4 00 9I -
(n.- a 0. 4
U000aM1
4j c)V04)
%4 Mf~4.b
4 0.4 1
a0441~W~
.4 0 R
B-23
04.4 JC 0.04 1. E4 0 go 0 31% 3 0.4J 0

In.0 -
Il 00 C (n
U41 4) 54 0 %. 3 >M 0r4wF
0,
0VW~ 9a-.4 to 0 00COw
toL.C 441 -~J W OWD O CWC
4401 > -. ,4 CO-SC
4 A0 N44-40 v*-S%1 -ML
L'4C* 0 C ). 41- *.- 4) .. .2 w4 -04 0
V 4 0 14 4) rA c4044
WS .4 05.
U)-sO 41.. CO4)
06'
.8~
W 50
.~
w N >Ca050
a)
C)
..4 1 0000 .
'-404>
0 0445J-s0'0
O
_W~
a, t0r
M v-.
ma 04. C a o-t-.
054..-4~~~~.. a)C W 'C34 0)9WJ . 4
1a045
0 C .- 4J S.C - C. 0' 4 0 . --. ,oW) 4
0 0-0 c MW La 4 5 0
4v
a N a r 41 w 54 u
C, 5ix *-4 0 n .- -S 40r
V'a - 0 0.4 0 4 0' L
0.0-.c 1 > 01 0 030 .- C
C. "a54 VA U 0 M.) mC 4 054WxXC30 )V,-
0 00A4 0, 14 V.40 0 >.0W >'@
>0)
0.0 0 .-4 VJ c CID aJ 0- (A.o 1 00 , 0 1
wa r.0 U) *-.-SIgI 004) 0. 6' O4-)03S. 00
c
V 4J N-.4 W !L0.
A 4 010 I 455-W
NUV4 so4C.-
.0 *W 01 M -4 C)WCr- 4- 0.0 w
5O .W .0CH4a 0'-4
V~ cCSC 0V.S0tQ
44050 r o-s
z 0 r. 0 . O- 0-
4" a) 11 r~ oil o'- 0
0 *C)WV U - 0 r-
(A p) 5 0- 0'C -400 .MI W
H 0 )0. 16 010
0 40 0 lz r. - S41
V 0 -.44w
m~~
c 4--ao ma) $ Q'--
0 05 14
501 V4) 4)1 05 10
V " $4 _
C--40 0 0a :
*Z
a) *C0 2- . 5 40WQ 4) 0 ra
2 L04) 4 Cw
r.-4=COv4V
En 04)CH r04 05 1- "do4 0) 0 a W.4> 0O
1-4.C CM.P.)-4
0Lw 4 10 54-fn -
0 I m5 Q
00)- OO3VC)m
cC-3. 4 0.0 0 50g 4r-1 > 44 3-IW 00,-4r. WC 0
-0 r- 0C4 1 4 n -"A 0 00 4 1 - 0100-
0-
41. 0.W 0- . 0 CO4 a 0- 4C-4 0w44 54 4-0( 0
a wO0 4>
I~.SJ
a5~ c 4 c -C , 4)0 4 543 -4 u ) '0%$4 N
- -0 0. ) 0
(7 0V i 0 m--4 0
1504) . 0' ) w NA
w 0 -w- -4 0400
u w0 * - A 3 w. m0.
C- . 3)
IC
c550 10 W- 9-0) A >0C5.. 5.
-C 4 0 H --C Oa54) 0..2 0 , 540
0V N,C >0 0' 0-40
(n U L -4) *-A- 30 0 o C1V. -- - w EOnw).~~)..
OW -0 C. 044
j( 450cww3
$4 O) m*
50 V1a -- 4w00 -4>CU)v
0, m M Am 430C 0 .>1a
030.- O* a) 60() W)0L> 4-.5
4 )C W2-314 0 0 w()Wr 04-c 6- LaW -to 3Ca)CIO40WC 030 0
S0 D0.I w 4)0 I.s '3. 5.J 1-rC
:61 > l 3. 0 4 U 44 > >.V50 ) 064'-n
0
u0 000 to
1I 0.
0.0-4L)c to% (
z -- OC V 4
.I~ - 400Wa) W 0 <:% - / Oa).w .C - 5 0.c (a>00a
>0W.-C 0 4 0MV.-. >1>4)0 U V Z 0 :-Hl.-C ) @-O
Ca .WCO.4JaW
245. to 0
V 4 9 to - M M C -CIoW
Z 5 0 V >- ) >M0 -. 00 La
0 z.0Wa) - r- 0605. :6 A 0 z $4 W w24 cOC'~ I50.
a C o 4414 C()
.. a030w0)0. La'-L0 >.m . - 504>g m4LU >0 M
- -( ) 4. La4
0Z.4--W0 z C 0-. 02 to " ma
.. ) 0-L 4.5 % La
0C l0-4 Ca)
10 0 CIA - W 4 r- c) 0>43 --4 -00 0-p.Ir4 0 -4 )-W
La4) 2 ---. 0 to : a 0% ..
4 Z a .- zv> > m.'a0. 0OV 0 0'-4
$ 0 a . 0w 0 a)4 0 (Y' V M. W hC.0Za 0 W)5-4 1 4 3 0 4) r
6.O-42 0. 4) V)W0) W '-064) 0~45
0 VW .CCO
4J0-4L. M V 0 01
W~ t;0cZ.aa
134 C
-C V V-1< 4U V )V )o0 ) 0 o4) c$4
'41
--- 41 c 40 1) L0. w 10 M ) - 50
_J. - L a r-CV0 .. M 9
44)W j;
4'--4W 0 V)V 54W L 0 M =>. 44W05 V04 0 W)-Cr 54-.-O-SW>E 0 0
w Y . 4)U0 W-0 .4a0 nxs'1, 0)a, E@.-. 0-0 m a E Ic a)0 o 4j WWOO,-.4)
=WOO' 50 :34)LaS.2
% . 2-, 4 05 4 = U) U0Jc )
0WW0.C 4)W 0 0
c-04
CYz mo 4. 0.La-
M~
InO U)w0 E'-za W 0 ig41La-V400.Lmaw cz>4
a)- U) >,a)t( C0
r"40 Ow
.4C W- g 9 w 0Wm WOO >.C
42~ > XC54 10.0) -HV.)
-NI-I--
m 35410 -100--
a) S44 4 oa C
a)
0 4
0' W
0 15 .c 0 V r0 0W> c:co 0.19
u
NO 5.4 )- >.C -~W to31
La-- W 0 a k> 04)1 4 ME0
m0IT to
041$. w- CO Coc
*a.C 0.44 ccCI . 440r.(
0L
0 04 04
*,c'a 41 r.0cg-4C $> 4)5in)
4) al V0 .4--'C4 > V C-C Wa)
0 c V' 440) *C3'- 4.3 0' -4
a) 0*44 40 4' C .- %.. -.45 C
,,. _4~a OOVV40 44..-al 04) 0 0c 0 0
Q cm41 0 9 3Ct 0 -4 0 -- 0 Cw W
4) 0 41 w
0-a 4 0'W1 01' 1 .C W-14
4 0 S.C
4)4p . w 0W r- a)W -44 44 4)
to20 54 0
$4H0 -4@-.(a0. Z, - 4) 1 0 C) W ".
0 N-S514 ,
114 -4 La 04
go OV00 AoC
W-4'-
WW *4)
A. 0Wa0 %-.C * COO0
0 :1' n aON 0- O 4) La V UOW
WO 4)V1 5. V X0.aW O04 0 U
0>1. C $4
Laz)' ()5Wa 00 o4 a 04m 0 4)V
Lw (y,4~-
c V -s; s01 0 L0. - .44 2WCO ' VC-
tow2~ )V0.'-0 00.541 "1C 0 120 ) W o-4A a) C 0 V c'
0 a. 4)m4a- L)a-.0 C. 0 -43 0. - C0 OA
a. wC60. V) W
WV V L 00 05.- -50w
&C 004)00 ow~
54LCH a5 00
W5 CONO 3v
41- rW-W.- C1 OW
0.3> 0114 0 0% 00. -'40 C)
U*- -4 S.04)N0
05.N
0a 5 -4
4C4J kr. -- OC-I- 00 4J54V4 w ) CD0 00
.- c .N w -M 0049 " $r 41- 00' c, la W
-4C 00w 40a 0' -A 0) 054 WcI
-'O4au 0-s v I C.1 4) -0&* C 4) 1,-
Ic0 9).0 41 r4c4 a .0 (S %
00. 4)0 OC 000 q w 4j - q 0544
-0 00
-- W 0 0410 i -. 11, 0.- g
2i V) to go
4)) 5 Wu s 1 -5-)0 @2
N 0 02
14 05 . .4 atW )0 0 M 4o 54 Ix
m tW
P4 r
k'@ w 0 -*141aw
00 -C>W. ,
z~~~a
4 IM0 V4 km
a' W 4
al ( q 0 0 aW l 0 0-W,
Ln0 Q 0 m 4 kfm ( A 0 4 - 11
B-24
-00V A W0W k - W 0 - .00M440) - 0300 0 -M0 t7--

so s0 0 0 0.0 0.404.-4 W 0 0.>G.C000c.4ao0
M -41 044 0 6-11W4 'M0Q0 in 0 444 4 0 j-.4O.-46 -P
0 410 MOC4JCO t7440V6 0 0 w0 6 V.4 W.C00 V
.A 0 t 1900 0 wV 00 .41 0 . 0.-1 .-
0 4.4 - 0 -10.0Q4a 41 1- in Vj VO 14 a4
0
06 '00.1.41 4 1 W 0 '-4 0a .0n r 0 .4 0 00930 S.0
O0A
0. -1 g 00.a. 0 'A 044.-40 00 E
0o - -H 04 0 0 44 031%-1 *0 r, 44- .4 0
16- 00 04 00 4-044 00 H 04 44043 0 . -H
0 w 0.- %,0 01--l A -4J0 04 '00 3 .4 . 4.
o 4 0 6000 4 03r00I 000 - 10V- 0 0 010c
.4 0 0'a 0o 0t 04 C 0A40000 t,00 0.
0
0
in O.O1Mc tr 0 0sa-w
.43
0 q 441
4
4
'44 00
44
0
P.4~;S. r0 >0rto44144 4
a,00 0 -4 0
O 0 144 14>.I -- 0 4
804 -4 I- aOOM 0440O01w
04VMx V a00014k04300VJ A In. 000-
00.~~~~~0 0. 0 V.lOa4400In4.0.9, 0.-I
0: 'A0. 010 C 4 064 0 0 0- .- 00,0 C .- l C4- 440 0
0~ 00004 01 *.40 u -V. 0 04
OS444144j r4-4 0 is
> .04. V-4 004. W44 V0 4 140.- -4 M> 0 O 0 -4 0>
0 440 0 r000.- 40 0 0' )010UN0P4 >41- a040 -M04
U00, -40a 14W- 00U - 4 0 1"-1 V 44 V AlM a 0 1V4 to 0 0 t40l4j
04 0-we 000%3. 4 4j 0 44 .- 40 W444440 '4
FA440 SMCw 00ka 0 Z 04r4C0t 45 ,60 V
14 .490.
0 0 0- v1 O0I.I04 0 M* -2 0000-1 4 0-.-4-IwIa
14
0V W0 V4. aVo o C -. Q r a4 0 W.0042 0. 0000%4 0
0 E- 5 444
ED1 0.W00 0 0- > 0 004 -A4 0 01a404 40
j 04 x1 0Hgo0044 0M0004t 00-4,G 00 -40 A0- W 3 " 4 141
Z014 1. W V0 0 c>0 M COP.00-0.- Fz 0.0 0010 04V4 .0
00 W-C; -4 c000 k000 000 144 H- r.0.4 0.-C A 00 0140MO00
0.0R - 0 - 44 4.c> 0 -0 0 V Hr4~ 000w 10 a4 0.0 .0 0 0 .- 4 V
H-0N IV4 41 0 0 0040 0.-4.4jV .4 l< 0 C V.4>I0M00.4. 4 0414w4V0
0 00 w10. 044 0 04 we,
0 014
U.4 100.0r010ld001W
*D0 0040400 a V 0-
0 0
0.v14
w4 M
R4 0 !.-I.14)
G0140 00.h
4 )X V 9
0$40
C000
0-4
~ 0V 0. 04
-A.-
-P4w
000
0000t
14 4aV
1-43
'1
4400W0 .04J04 W 0
M-414- . 'A.-
0 0- -41
0 MW
.40 40V4>0'. -VX 0 = -4 0$44 4 44 0 X H r %4 X o-M 4-4 AJ O.C I 00C

44 0 0. 04 -0.I 1420 -4 0-406 4 014-40 0
44440 040 0 -414 0> 0 0r 40 00040w a.w..
0 IN - 0 DAa 0 0a cwMAVw.00.0041
10
0.-4V~~
4 1 .. M4 C
4~40
04-. -4.0040
40 01440
Ma 0 $440 0
14 OV
0 0.0
04
00
W 14 - 0.-
-00
.
0 0
~& r.00
o .0 0440.0
r 0 4 j 0
0140 o .4M>%4
M440- > 0 00- V0. - U-0 0 .40 10,
0
*',21 4 00. .4 ) .004 N rV00.401 > r-V 4V 0 0 Ny' '1 04 0 "0%0- 01
00 0 2 4>"a. 00 14 441 . 0
A0 000MW.00C 4 j0 0t C C - .00.4 r 4a 4 0
4 644 0 V-.00 0 -. 4.0
r.O0
0 ) 2C C0 v 0-.t .44j0 VA140 C1 a
P)kr4 L 00 1.41 10O V 0.0 0 4J 4 440-A0 0 0 .
0-4 00 C0 044 w6.4.000 01 000 0 0 044000
V)00064 1 4 -0000d1 0
0414414- >.W00 $4 C-4004. tO 41H.4 0 14 .~4.)0 >144 0 0 0 0.4 W00V0
40 0 14 1,.0Z 4 U 04 0 00"tA'0 16..X..4 00-
w440 444M 44
.0
- r
I- .40 M 4 0MW000 0 H0 i 0n0 0 41V
1 040 00M
rz c041 En 400V44A0 0 0 14 0 0. 0404 0c06 04M c0- 0
0 W4 WWr.n >1W4V0t31,4 0 00 :02 0) .00 w - 400000 006W 4
*-I0~ .. 0_ V
O)44 -0 V 416 0 14
0 I 0 14 0 0 0 060100 00 a. 000
00010- a, V 14440V0 -44 0C N 0 140V
. >rC
c4004W000 0A 40.-, -40 0-40 04044 0 0 M 000 a 44P 0w
*-04.->04 0440 0J 04 044 0440c a -. 0cI02000,a'r0.04"10100 004444
-W .to0 0.0 ty 000V10--400 4 0 01 4 0I -40 4 000c. V
to0-.40' -I tM 14 0 44d 400.14004r V- M 0 0.. k I la V 0i 0 040.0
O -01oa 4) ( 10ctoa -40>C.-4 0 c .. 1a0%WC N r. 4 Z Q0 Q 0 000 14 0044
v u %D0 4 0 1r -4.-I M414 3 >,-I A A 10 1 4 4 2 0 4400 M0 *
0044M 000 .06444 00 0 "401640 :14 0 U -a. V 1 4 to 0V6.
0440 0 FA-4.
44040 041 044
14001.4
0000 OV C, as 6 * 04141.30
M w 04.-..02
4$--
00> 14 0 0000440 w.04~0.
0 4000 40 0 40100
W14, 0 *c .04 4 .00rqw .0 0
% 4 k'"0l< Z 4 J W 0 4 - 4 -%4 0 00 0 - 4
V0 0 400 w v 0 0V.. 40V0r-4 a c
ul0 v 0-4.01V044.4J 4=V
w h 4-4 04a400 v0 >44 0F V( jH 0.1. 00 0 V4 0M--0 040 U 0W
-. 03
0-0.0014 .0 00.00-4a, 0000.40~404
NOC 0 t 4 ItI400 USu0 o0C 14-440 4 ,0 04-4-44
10 2O 440-O
00- 14-M4444.4 ' 0 0 .. 100MU0W0 .- 414001$4 W4X044

000
010 0.1 0 9w. 0 0 000C -4 04 403W19V0 0I 0-.0
U00 a-- V.-40to 04- 04.V000M44 0 -(4 00V0 4
4-4 04 0 v00 0 Il .0 0 00 04 01 4 404414a C0 V40C6416
0 102 00000 Q 0 .. 114. 40$6.0440d $4-40 IV 00
in1 t) 0>100 in 040 00.04 V10 0 0 4X 4
00- 04. - 0400144 V0 0.lO, 0. ->0 0 00c0
E- -40 144 4400 000 040%0 O
$4A0
0 l01
(4 00- 0014 1400 l4 C0 C1 1.0'0 44A000 000-
44 'I-4 . faH., 0 (a4to400 in..
.- W o c,. 0 C00W440.4 C00.I M 0 0 4400
0 0- a 0 01 .4- 14 0-4-
) 40%A4 0 44 0 0 -4, -A 0 10 00
0 I W r ,
1- >. * 046.4OW 0.0 00.40. 0-4X0 W 041C4.0404
00a 0 00 a0 0 04 0-40> 0.Mea00 0044 0VJQq r a440
(A W-40000V0 4L 004 V -10 04 4 Aia00 4 r -04 - VI
0.4 CA
VO V 01 0401 00 -4 C61 0 030r 4 M A060
.2 C0 0 >10~
1 M%a 40 -444401 4 0.0 -40 - 1 V
44 0.0 tp 000 0 0%40.4
-~0 0 -000 0404J ONa V H440
04 41
to .
00.
N14~~~
~ 00
'a.
0
4161V--4
No
0 ar
03 -41V00 2
M0' 0).4 1 I4M1
04
1.001~ r0
0 4 14
XO 6 asM0&
(a0 c044 v0 1.-I A-0 - a . 00 0 0 V04400442
44
00 040 0 W0.- 440400 00U .I00
04 41 00 0 1 1$ .-4 V10
0-0 0 $4 0a
6MO O 0
01r 00 1H- 40
W31-002 M004. 0 EI0 0 .04 0 0 0 W4A 0044A 00
014 4 , 001 44 0 00.00 . 040000 r..0.114a -13M 0M
1-4 0 -401 Ai 00 1-40 00 0 o.o014a 0 B N 44
01 400 4J V44
00 440 V -. -V V 0.04 0
0 1- 0 0 040 lak$ 4 4 IS'A 106"00 e
44ix
0 14 0401 .4 ' 004 0 0..440014 'A 00 004-1- .0
440>- U.0 000 0 -00 40 %-4 4l r0' 011 4-A 4 -. UX4
04 0
14w 0 to00V 0 c4. 0 00 a004 0 40. 0x4-
z,0 0 k A 0. 014440 04 acF0.40 04.4 O3-4
040 0 $4
In~ 14.0
C4 U.40.4 00 .64.1 Ip &A0I 0040-03400
40.4~~ 064 0 00I.4- 3- a414 0$4.40 .4k 0 4W0a 00
. 0
t4 W 00 0 00 0 t0 14 -004 4 - q%-0 .0 . 0144a
414A
4029 00."0 0 14 a.00 1-44 4 44-0 0
V00- 0 0 to 01ICU1400.4 0314V0
$40$I 1410 0 4 a. c
004a-0 0-40 0. 04 o~ 0 go
0 V 0 454v 0-4 .40 02
240' a- N 00w441 4 0 'A04 .40 '-40 04
a 0 0. a4 No r
U) L .0
X 4 . W40 A- 0 1 W t 0 .40 14 960.44 00
go t
U00 00.I40%.00I0 a. us..
0 "1
0
Is~~4
a. 314 00.0 6 4.01--0
a 0062
-
-45
V4440
1410
41
0 -1 -W V- 14a044- 630-4004
G 044004 1-04; 0L00 0 $4 0 4 40 00c
'0 4 -V
00i V 4442.2
1 -
U.Qa0 4 1440.4 A=i t a4 .440 1
01 012.-. 0
00.00 009 .0 0.Vk1icu46 0 44- 30A00300-01 a4 0 00.10.l
00
0V40.0
41-
2$4 3l V004..4 >00. 014 3 414.0 wo t O 4 4 4 0a 0 tr
00 0~ 0 0 c a~ 0
.4146.a c 4 0 040404 600
.0 a .00 C00 0A 4 A Ma " a -41 q0 p6 0 40 k
go A40-0
v0 VA.444 so4U
OE 4 -0>60
046,100 04.0
A 4 00 46 4F C.1A
. t6*.0
5 1 u N- 0 0
4000
0111 4j40~~04 1.V604
z g 0'a0 .4 >N 0.4 0. w %..
I. E- 444a 04014.44. 0 a4
.4 *01'ccc40 A04401 -.4-- a in 4-0. 44p0 0a 1
40 540 -4 A444 440 444 4J444 V 14-44qo0901 04140
A.04160
0 I;. W SE0 ,
B-25
. 000 C
0C W~ 0
>,Z4 o0 A0a
H .0C$0
V 4 O-4 0 0 .0 O40 MI
M0 0 44000 0 0V8
-4. 0N X 400.
C 4 6 0 4!
(A $4I044
* 0
'-> 0 , 01 c 4r Q-4 $0.-M a
0 0.0c 0 p ma04> XE 0 o
.4)0
0 im 0 01.-0 V C M 00.H4
bi- ii CR Eno 0 Sto ,A 04.0 *606
0 04 02 9.4 0 0 H H. 0 t"0 -v 1 '
0 - 04) -40 V
* 0 -.0 .~~0 X.0 04 > .
H0.
H
IN0.4V
$0 04
0a
0 0 0.
NN 0>0 00'-o. CO.
0'. 9 E 0 r.00 C> .M 0.440a
000 0...0.r"0 0 4 H"
0w 0..0)0.-,40
.0, V0 0400ON u 0
0 r.-.- 1 W$0 W >' tE.. 94,40_4 0.
00 C0 -4W 434. 0%40 0 ..-
0.. 04 r. v CL
HC C 000.,.4 r. $4 N .o 0 a
a. 820 110.C40~ .4V 0 rz
>4 04
r400 tr Ho 0. v 0 . 4 0 C a
00 W 0 094 aC o.. -42 >0wCD 4$0 4.0
0 1 04 0. D $4
r4 z
CJr c 0 40- 00W: 0 04C4 .0
OC .400U 34) o >4.0 WC -4 M
=40S0 0 0) 44 0 'A W>
0C- C z.00
I0.ON0
C M
0 0
C0 r
0
~ IN$
(43
H-4
HO.
3-C
>4
0>
.,
C
v 1' 00
4 0)'4 to 10
VCCOM N
50 .. 4 00 3$r 400 :00 C 0k 0 v u IN ..
a, U00 0$4 c- 04 .0 0-4 0 N~4
0,0 0 X -41 W. o Q. 0.o
WCV 43 0) M N r. 4)
T 4D0 I J 0 0 a 0, W0-4 k a 4
4) to 4 M0~ 0.. >CJIN40 O >>V 00I0..0
lt
Z0 04 E-. C N 0. ..0 0. o' IN 90 to
I
0 C 4) 4 40 $0 W0C- o4 0.S z 0a ' 0 4M'
a, 0 o) 0)0 0440 0 0a) % 0 40N Ma44
go. =
,C 440 0 QV OH4> -. 0 . " $ 0.~-0 ( I
14
40) E
0 2>
4
11O$N $4
V0 :> 00
W00D --. 4)
1 41
C -
0 ~4)- 0
0
0
42 a
$4
0>
9
N
>..HO 0 r000a .0
. i> 0. r . 0 U D0.
4 0*--noo
0
04- 0 Q INN-
0$ 0W -.$401
:5544*0f4 0.
0~ ... I
r= 4)
zj
"
1
-
A
V
0 -4H>
.40V0.0
0.0
. $4 0
4, HO
I ~ 0 04 now
-4 (44. -4 4 0 04 A
t0000 04 4)
~0 CMO-44- i4)
z aC C44$00.
0 . aC
E4 M10.000-4.)0
0 o U4.IN
000>.ooot o
0 00> 0,
000l<. 00
.-
0
M r.
E-)4
.04V
0M
$4V00
0- :3
.4
-&4 0 0IN0. >4 N) - 0

IN40to .
L-40140
C .(0 *-V C$40r 0Ca W - v0 to 004a t
00 C0 .),4. 0 $ V
(a W0 0 00 C M 4.q4) CIO
0004I.)0 %4-N0 X 1-4 4.144 0 000
u 3 t -44. '0 4 L.C.U4".IV 4 H~ Q- 40.4
a 040 -. 00 >
110,.,4 .. - 0 $ r_400 00 0 H N- 0
V( 0"4 >0
"40,C~ cc 0 0 -4 4 04~0
V.0 TN N40V0.--,.0VC .4 03
vza0 S.. .00 0U
.0 00 V440.
A C 101 coo 0 s. tol.0-.040.0
Wm0C00 0 -42 a _344
o 31toI 0.
40. ---
0 0 1 0 0 0
V
,X
04.)0
0.4 O
E 0
a 4)- 0.00
o.4.$ C00C
0 0.2
w414
m0- %-K W4-4000-044044
v 0 P 0n 00 > .V
C-4 of X$ 0o = 40n .0 0.- 4$--
r$4- .).40 0 4
440 o 04 0a; 0401..C -
C 000 0r.
. 00.000 . -43Q44 M A0 540
0 A0m4e 0u 9 4. - 4% 4N -0
0'- 0 4 to 3 16 4
.-.
4 V- 0 4.) o4-
C0 A~00
$40 41
U o4- V41- 1.4-4-0 0
0C--00 0 1. 0 V -40I
V-4C$
04 C.
$44 0IN. 014 -020 r..- 0. 0 0_ $
-4 0 r 13)40:3 .
00 0 0, M0 0 ,0000 V0 OW$4 "- V 0
m4. 00M 0. x41
5 C0
u N0 .0 0 0.4A0 $0 w.o"404 I0
k 0 4 04 - - D04 N4 -44 $ 31 0-0. 0, .
00 W4-.
M0"0
90 CD0.
0. 0 0 2 0 0.40 0 - -s$C o0NP
>,4"v$-- . 0
0' 2 >'4.)r 0n 00 .. a004 ..M4.0 00 0 0
9 U va C40
A v 00 0 tc $4 04) $4 :4 004) IN
4)0 V 000 W0.r00000in. 014V)-.) 001
$4 00V 0 00J4.044
A 4 014 .M~ 0 0 41 C .C
0~ x 0
C0o 0 = W V.0 0 t4) .I
.-. $4 V 0 0 44 0 l< in 0$4a,.--- 0 0)0V -04 <4.004 4. IN
0 0~5
.4
w
0.0' 00-s o > 9
440000.-)
H0 0 :31 00.
I0 0 as4
0 M
-44 (N.
%D104
$4L
.0 40. 0
-40.
00
$
00
. ~ 04C
000 0...)
00 0 00040
rO .0. 044 to4 o00 14.
C $40 0 a' 0 H4 0 3 44 -It 400 $04 t; o - 00.0.
C40.4 000 9:~0 V
'- CO0 WA4440.
O a0 W0 t 4 4 0.4 -0
0.J a 00.& r 0 04014o0 .0t41 0 Ca00 41
4 0 9V$4q0 a V .
>..*.4~. 0 0 0.0 v4'-14 00N 0 I V4A4.) O 0. 0 x 2.1.a
0 a 41oC 1 H"4 1 $40 r. 4 42tom
5N 0 40 044a0.)
00 o A0.00000 0. 4 04 0 0 W i . 4
0u.1 0vv.4
S..0 -40 x.04 001 00 4 HO IC MIS
$4 40 0$4
00 02M - U 0 0 (-- IN40
41 0 -. W .~-4
CC 00 t0$r-4W0
Ma$ ow 0
N0 cH E4..4-
00
I C
0>4
0%4)0. 0 to
4-....
u0
I Xtp
C0.0 0 c
04'4
I 40 Ai M
(4 kV
-
0 0
0 4440 C1..
0401
o.040
0 IN0
0 CID
" go$
W
004a
a-)0-10
0,
N~0
r.IN
00 C
"4.00
4-0
V0
"a 00.
4)
V a0M
o0
~ 4
-fu O CO
10 04>,
4 0-4 00 040
% 0-
00I
14 0C
-
C4:3.
$442410 >0M -06~ - 0 04) -040 . A W z

4J1
v0V .O 1 0 0
"4V4. 0 * 0 4 **$4
00-0
cIN C 4 C
.0000S0
0140.000
Z ~4,0 00.-.
0 0 0>
$414
-04
0
4-
.
. 0, 00
a' r"4
4. o r 40' 00 0 UC.-.a --
4)aI
0o0 H 14
k!..
C M~P 44 .4
-40 05MW
%.444V.UM)-04N4 40 cI
%0 V 0 000. 1a4_
.)o 4 94.aa-) 0
$4 i 0 4 )
00
141_
Q. 004P 4 >t~l r41
-.
4 0.4H P4 WVe 0 02 V4 0 41 W QC-40
0
021
0.0A! A4 1
k4 - V 2~ 0 0 &~
V 00 40
40 10444114)44
N1 %0Wo0 *0)
1.1 a 14i
-. 0
1 00 0 N 643J~ J4 044 0 >, -0 D04 44.00 0 V MO. a a4 14-
0,0 4 901 0
02 U5A040 .wW V I ia > -4 H 1M 1V~ 1-0
u0 l 44
0 V
441W4C 0 4 $ r M-0 40~
14
0.0 1000 0 -( 0 as C C01 0
'-04 00 aml" ON MAN 00 0 49 1 r.
100.- xx 5 O f4444)41. 0 . 0
- 9 l P
b0 A$4. a a. 0*A> -.4 0 .8 o4 0 A MID- a 0O4-
%4t00W C 0 40e0
0, N0~-~~~
U)- 0 600 1444 214 0'"H-1
"0 0
ZT
30a
14
V
0
r4---.
S. U01 0I
S. 0 I zN4 0 00081$0.0 WOH-~~f 0 m -Ao s ok
0
C 42.5 Z0 a V COO- 4; Q IN tP A 4 ()1e 4a Mv bi 0 1 0 -1. .4 41.4
1 4 01 1
0 i C O N
gA 40 v en40 "0 00-
0000 ~ '-4
V
40 rIN4-CN 0 v~'
a0 0 8
1 5q 0-4 9X*4~ 4$1>0000
or0140 &0
0 "4
0 a
0 C .0440
>.&~ 000
-4
VC - 1
u. i 0 A m 410k h-. 0- ) -V- m
0. 00 v -. 0-4O0 .0
B-26
o~ ~ 4i14
a' r4 0.g0.
S4
~ 06SO.U4ou
.4 .0
* ~.4V14W. 0 OS
* 3 a) a.
* -. -- ~ o AJ-4 c4i r
o,*. 41
a 40a~ 0' 4.4-6-4 W
-of a4 O A4- 0%0Ut,

VU
I ~ S4J ot 0 =4j
HA 4 .0
-4 'aa *U,4 -4.404
.e0
0 1.4 04 .00 -4 -4U4r 0 1"
0
oe .-0
$4.4
$4
1.uo41 a 6
* - 0
V.
.0,0
U) 4 ~ ' 0
.4 MIX. coo-oomoo4
4
- Aj
$4
10
k C.
3 a 40 0
410o .4 0
c-4C0
014r
~OZI04 4U *40 *0
41 ~ ~0 040"' r % 0
c 4U 0.1
49 IV .0 -1 0 a -4s.
a
C'% cn 41.1% - 41;410
N ).0- 4 -Z 1600a
00. t* 4'oN .0 'o00'0 $.4
-4 C01.4j C, 0, =
j.0 .oamo
O r- 1014
a *amna .
10M
10- a
04a 9 0 41
$4.UaU.144.-
to 0. 0 a0V$ ,
.4Ci.-a0-4 Qa 10.0g > 301

.4 *0
14- CC t 4 -4 r
CN0 4-.4 4'~
$4 . 4 k44 14
.W4 r. - *1 O 41 4U 4 0'0 @0I
~.t a ao Oj ",
a01a a4
0u. 5 60 4'0. ~
M 4 11 41 4 0
41 W 0 11 a 0J-46 r. 01-'
0A 0
.1 00- 41 a
Of* 0~ g0- 0 d J " 0
a r00 '-0 0 4mOc 0414

41 Aja 01 0. a
o 41 '-
4'6* 4 0 0
04 o 0s ae $-10 c
to _4x 14 0 14
r- 04
01 2c0 014' S4.O 14 r4300 -.
a. >u 0 o a 0
to U4 C.0
0 0 aCO$4
Mta C4 Q e' V
at.4 03
.4 4r 00 14W .
00. 0
(Aa% .0 a
0, 0 N0' '1
a 4 'MO X C rCU J14 0a- 1
OD.4 ->0Na0. 044.4
MO 4 80 0 4 0
0 t 0 )r C. -'o.4044
.a.400 3 c p.
4' 0~. ,
-4~~~~~~0
I c
04.
'4
4.U
0a.0-
0 40m 4f
-0 0 $-4 4 )
144~~~ 2,2S
inr4a
4 4- 10 ueV4 c
-4sI. 601 0.04414 s

B-27
C 4)004c004-.
-4 0 H '.4 ag 00 0 0-4 A
IS "I a 1 4)0U4a) 04-' 0
4400
0. 44
a *.-I>0, 04 0 a. 0 4 40 40>
o C004 V 0 V *-4 3.0. 0. 0
44 0 v. 004.0
4)0
4
Coo0
> M0 0
W>,"- ~ .. A
1.,
m 00a .0 A C
o4 0 00
O N 0 ( .4 01
c 4 0 -1~.40 V4 O
0 0 00 4 4
00
41- lo*% 4 0 .1 4 .4
>4
t(2 0- . A 0 V >> 00
41 1 M5- M9 >
.1-4 44 0..0001
CJ i a~ $O 4 C m-.4 4 004 , .'
a C0.
0R44 00 4
-404
40
00 A
*4
(n no
0a 4 0
f
O04 .04 -f 0.0MUO0 WHO 0 w 4 0
r.4
14 0-.4)-0 C
0- 0 POC)
Q
000 4CL#140 0'0 Z04w.0
*'44 0
H> 44 0 c0.4
a.4 v)))04
04
44 go000 r.
0124 44 00.
4 0% 040 010U -24 w l a4 0 0 10
a , l t 0.40C 0 a-.
44 0 044 4 4c 0 . N
n V) %
OC 1 0-P ., 2.
to r-
0 c4 0- 0 4 00.> 0 (a 4 C40 0
41 0 0 0' 0 004)
0w 12IC 4000c -4400tLWC '300 0 444..4
4 04J
o 4414 1 V c
44C k M
10-4H OmC Is(o'4N0 4b.o 0440 0 .4
14 4-'& 9
404 w
IC Q 4
~0 4)c 0 0 ,004
a O006V0 -k
('0 C
-0
.
VU 4J "-
0 to cV
t
W40
a
-4Cy-f4). nIz .4 44 00 %.0 40m 0
0400 -(Oc 1~ c 00 C 8 40
14..' 44-0 0 0 *.C 0 4j* 44V0 4 '0
.4k- V1 A 40 .4 k 4 4--
041 0u 410 $4IOO a~ = r
kN
0 1041 0 $4 ( - 0 4 4) 3.0
00(0 00 410 0 4 *40-4k 400 44
.0%4*4.3 40. 004.. 4) 404 0
V) 41r0
U0 Ua - 0 Ic4) 0.
00
.Z0UX
in(
4) 0)0 awda JJ % 0-
m$4-
.
0~
000.0
0 '
'1 0 0.
U0~.' -0. 1-A30M
, 4~~l
01 00.m.0-E..A,-4 so
4
". C ?
U04
4)
0
CC
.4441 . C 04
84.
.C OI
0 a4 4 00.04
c.. 0 (- - 0 >-4 -4X
w
-4.4 6t w 0 M0 U ' 0 0 4
M 0 go V
4-1 0 40 1
0**-mt 440 -104CC-0-4000 100)0>C

04 r W . 0 0 4 *00r. 4co w
c 4 0 1 . .0 40
. Itn.. . j.0-41C C)I400
V N40--.40U .- C 0 A Xt .4.to00 0
tm4C- 04004 0 C4 j 4 C404-C 0 -cCOa .0
4 S V,
4sJ(rlk 0.0 44.)44) 0 10 4 0444 0 4'C; .0$
(aIC440
-=4 v -. CL 0 0 tn -. 1 41 0 -4 CL-I0go000 I
1.0 .44. k 3 4 -00z
>C000 0. 444)040>4
00.V 4424) 0'4j4.44 E.-I0 44M 01 04)UI
0 w00-4 . U k 9 9 m40 0 ct 44 a-
> 0 A:2.44) .C c
- 4 c , 0 H a 944-
C4 . 0 .CA 0 c 0 -40 44,
0 0 0 a 04 3 4J44 .4,44 £1 C 0 0 0 4 )00.4r03 0
I0. 0 00.0
Q.-.- 4V 0 6 4 -0OV'-4 t 4 Q M -C 04 0
r-0 0 . 0J-0 >040 0-
A-4. C . 4. k 004.W04V
OOfO 04)44 000 -0 k Qk 0 Wl 0-.4 0 00 4 40 044
r. .0 ck .C 44400'a04 0. 0 0 vI .ICO''44a 04404044.w
z- V 044. >'4
-. A2 V 4 00 0 4J4.0 0 C
0. C4 4J 00 44
0""4 M4 CO 44r 0 4 0 4) . m
44.- 000v4C0- w
10 0 >0V30C 4)0'0 C 4A 04C' >,J 0V 0))0
0 0"'
0-44 11. a 0 0c acM 00.00
(2.. (44 0 k4 4QC C.-
K4- 00 0 t4V 00 -40>.
t C4A00XMw v k % -4C
0 a Oa
04
0C
-
0Ck0'
.0 0-4
a a
6
0
' 0 0
go
4 a4
MGM040
acm
2..
I44 01$ 44a
HC
.
0 04444 1
0048;14I104)
-
o'0
0444)4k
)40
Ma Q%4o0%4
4 aC u 40k
£
V0 0 c0.4 7 0-ItS 400 400
U00-44)M404 0k
00 0 r4 0J 0 J40V 04C 0 60~04>. w 4) 04 4J-40a
r444
%4
00448.44U
a4 0 O .4 94 C)4 0 4)4) 4
UN) 0 0" %4' 044V0 MO 4A C UI a c a
%0404. 17 >
0C VV4 0iC 0 0U4 4).4k
4 0,, 00> 0 005.40 00c0 to0
C 0'-4CIO .4a 1 9 k44 V"0 Ma - 4.0 aMk CO
0.. 0 C00 .-00 14.44>)
044 04040
-CHc 0 0 a Ct 00' a 40.4V 0 '1 0 1 0 4 r-
04 aN£44K
04C.. kCVO4)04
4a
'0,-.40(00P2 0 -0 k 0 .04300~ 4) 00
.44H
0 a 1 02k>a9rV4m -4 1 AO !V
H'~ 0 c ~o
0i-Oa.44
00 a k
M44 04M 0
-. 040
g .
4
0
44> 00C- 4
c 04 0414 000 4
E-
'A.1.4
0
S. 41 ~
a0 4 C4
to4)00lgo 04
*)4
i04
0.44
*. '
.
C4.I . 3 0 8 1M c >. 0 . 0 " 0'
o. 0 .- 1 0 0 3.
I2
.4 0443 311 4 cO544 4 go
'0~~ .4*40 4)0h~
-4~~~.Ts ' *2 4 -,I5 4'4A.~C

'4
0I
IC 449 . AV A -0I!~~
lh.0 4I
B-28
t 0 Z1 0 Z&' 2 N0 1
0-
re 0 54 00.0 a C 04)0
.,N.Co 'k'o 8 0 00 08 00 v 00-.
0~ 0 c
05.0' 0
000 Qc I's. :2,0 0 . 54
CKW-Ic ct. ~ .0 9 :i 0 *8
00 0 0 -4' 00 .w 5o.- 0 J0'
0 0
a'o1
c45v
41
M
a' 4 c.0 0
4)
'0Z..km
.e o r44o 41
0k
V
w N"
'0-
w'
00u a40
%,a
5 1.
0000I0v4Z
CU
:
0002004)R
010141% 4J 42 0 9~ 0-
0a. 000
008 '00. O 0
H ~~
00.
0~~~~~~ 00 0
0-
004
00
ov.4.
~0.-eH0
844
0
44)0
000
.,4
.. 0
kO.40 to0
4
04.40
0040 .0
a.40)5'004)0
4
0
04 0~4
54*~
80
.0
054
S a~
54
kO4
00 c
-400 %8-4
'N
00''0)..
00 a0V4
54 0-
W- Woo.. k0
0..
0- ).
.0A).0
.140k
o o
0'
4) m 54 '00.k
0 )0
~0 01
5.
C.. 4
5 00 0 0
o
0
00M~
00 0
00-8 ~0> zo
04,
0X0
48.4
0 0400
1 0-'
r 0%
W440
NO ~
S
4)c.
0-54 0
~-4k).
.
$4
00400 0
-0~-o
0 0'400=- a "00 08
%40_~
0
-4
k.0O0 0 .
r
540W
&r0
>000000
000tO
0.k0
0'-4 0 aV
a01)0I4054"0
t1 0
10A 0 W0'a0r
a
40
'00V0c
00.0e
0- .4~
28-40~ 544%44'
.4, -'N-45 I
30
>
003. 40
0I W04.
~ 00
41
00. 0
I4.-
0.
'
owo5cwos
0-4
00) 00.4 0.0 60>. a0F4-48 0 to00 *- 0.0 &10 00.00 0. 0
o 10 a to r-4 0540 o 050 54
8-. 0' ~ ~~~ H4
0 0
42 c)
54
0 0 5
1 k areo 0> 0 41
-0 a.4 ~' 00. 0.
00 48 44i U010 005 .- 4- C kIt 0.0... 0 0
C 0 14 a0,40k400 g48--
0~~~~~ 0 v a
0
0t
~
06 40
0.40 044000-4a 0, ~ vo 24
0 4 0V 0400
.QCO
)41C 0. *0-. 5 004 00 k0
0 Il. '. c0 *0.. 041 'A 4 -4 03
o o o o oo-., o0~ a W 10 a. 0 00 we 00 0 00
0 k k a 00 3'0'S00 540. to)- 6.C 900
4 000..08-. o-0
3
to
4 0.
80c 4)
O 00 00 9 "0 J M5 0.4 C- *00. 00 a
0, 0.0%.4 4 in
4)
'40 000>. 1 0 81.,0 l.0
-
I a
0
0
0 4 0
~ C4 0
k80
>to000 -
-000 v0 g 0
r0.
0 0 V o ij
aO00' 8C a~ a'-0-a 05a V 0 00540 Q, k w4) w a 8. 0 *.4 0% 0. 0 c to4) do V 0
00.
41
048H
e4)--
C 064
0
00040
-. V 0.4 .004
4
0
0 0 ~ 40Nc0
04 >
00840go
00a
99~.
0
- ta
4 - 4)
in0 0
) k'.0,
0 00
.4 00-%4
a000
k 0 0 4)2 41 0.W0 '-0-4 .0
0.04 0 0100 00 0'.-4 5 .s .0O (1448 -- 0 40 0005k 4
0'
'40 02~ 05-'0.000 m-Z
'4~ ~~~~~
-4) -' - -4.0'0 4 0'".00
0 F 40) .010 0
0 44J4-000 E-r o 4'J
0 .n0 c05
AM
g 4 4 0 >,a r.0 04 u0.%. 0W04)4) M 00..-I RC
a 3- 4 V V0.0-to 0A oo054 -0.50 4) 1 6 _g 00zQ ->-4
0M 14 00to 0 0Ik3 0440 Q. 0 Xa 3
zC04~ 00 . )0 0-000 06r 0% aAZ 0 a 0 r C0-4 W 0
c 0 4 1 >.00 540018In4-0
o kJ 19 0 "... 0 0 48 0 0 %4 3 0
2 Is0540. 001
.- C)) -00 054-454 . %
80 =-0 - 4 'X 0 4 03-4J-'00 3 0 0.4 .
0 44 -0 4 0 4.) w0 4 j 0 0-0 4 044 N V.0 m0am T ' '
0003
C
'0
44. m m V 0000 q04 ~ - V.'.545
a4- "Us N5t4)JCM
4041
-
04
0>8
04>.
.04 o
a0C 2
01
a H .0 r. 00048£0 go 00% X - 0 a M 0e405
-0.- 00 0.0 goo0 '00c' 0M 4
0%. 44 0 1 E004A003V
0. 0 o 0 0tr4)0W
1.0
a. 00 0 0 -A
0L540 0 0 Nt 054
0 o a0-. 0 04 U E0-4to .0 , 84a 0 t" 0I 0 Z
.48.,2cc V V
-0
*411
0 V 0 aC. 2 Z. 0'05400
-4 00 0 0 4Vk 0 00 . 404 0 50 0 4J 0 0 0 4
-'4c 0 01 6-0-0 0 k.4 0 -4 S4
4 0 03 0 *00
00004 Z 0 0
0 M.--*.
-4443
C 140 cM0 0c - O 48&
GN 2 H . w 00 .0 -40 'a4 ,0 0 goa54 m 448 0 .0.
0 .r.k0 Ole$40003 . & 0 k 0 0
40-'~~ -.. )4)4 054 So 0 k- IS04>
n0 M.- 0 c0c-4a0 08,-05 >.0' 0 000 00 0

E. f (d U4' C004c .0 0k-4- 400 *.to c
-Z 4 004
.0 = 0U0V 4 . r84 -40 k.4
a to -4 0(0 0 04 0 k.0 0 654* c4 oft 0
000 c0011 .30 0 0 54 55.8
0 0 4 .44
,Z- 0 2~4
> 54k 80-.
k.CU40to 0. o~ .V' -
(v0. Q VV.X -o( 0 0.
..u: 4 V4 40.4 0 4 -' 9 4 4( c 4H A5
0.05
O4H0 05a4 I 0 >3P6 41- v0 .0
0.' 04 ~ Mo .0M
00 1- wo4j 0 n 0.4~.
1 a4 C40 C 80
14 X (n 4j 2C 054. a a~ 514 >"i 0- w
fu-Cc 00 45J0! 2 3' 08 0 0 000 4854"
-4
0-.494"% 0 so O- c a Nf.0 r44
8! 0 a'
p .~ -4 'a40~.~ 55
4u0 V m 0 ma
so4
48)
0 ~-4
V o
4
r k
0
54 04
>In -4 kcv 4 -M4 %

w 0 M o 'A~ Wo0 V6r4 4- 0 t
B-29
- n 0 ilk-CM1414..CCSCOr C (0 0 $140 Is .0 00 ;V'O 06

CICIO
NN. 0 a0 a .C O04w0-4 0 0 N -4 c > .4O4a
.4 1C N44 VV4 44J4 4) k f4 A0J01-'O -84.. 0 -A.4IV .f
Ia MMR0k4, V4 14
M . ,1 4 .
0C - 4 U0.413 "1 00 0 40a4
ic 0
0441
01.
1tr . E 0$
no
~ Z.1 V- 0$
14.14
0 00s c
ISe
4 6 o x
3
aca r -
V0
04 V
0 Ob 0 W0 0 0~4 N 4
01t 6410 640~*z. e v
04 0-x
r.C 0Mz C4-4 OWO 0 D.C 0*. a0a

0. ;-.1 o
U)
. Q Wa Ca
44M .. 04.
S.
ou..axj o4J-.0
a4
aH1
aae C1 M404U4.
a
.MaM40~4
aaaaa,
t -4COA-
00 MM 00:
4 10C .C X 44 W $4 4) to 8 Cto 0Cc)j 1 0
as~~ oro ~.-0 ao S.a 0 - c 4WC M a
141 j0 0a4 -"-0
0
V
No0 A ~ 4Wt& -
0
a4
'a14 V~ k.4
>..10 C0
Ma A4 5 C- 0 MWUA 04 a 44 92 4 0 4 0 4 Ma-0
o a 4,4-.. 0 t.C %$ 0M . %41 a C1:
00
Crr
O.- m 0> 4C 140,aM V
-a
MOCW0
44 .0. '1 C 4
Ct
aa4 M
. M04Ca4a
1mo.. ~ "4
44-
MO000- 02a
'DU 14 9 C 04Jg1 C. 4) %Oaa a M 41 4 40 e 100 1
4)c C %.4.4V 0 ACl .C COa a'4 V 01. 0 %4CSa
aM4E
41 .. o 140 140m a4 N ba.4 ap VOC 41 WC
0a.4C 0a
Z.. '* 0-) 2a I>.a o- -C 04 OM r.o a- 01 aO44) 4C4
.)a 0 I..
Z. 0'.4)
a V to
0 aa 0 a..-
C 01 c4)o4 . ~ us. I00 was4 '0 to -0 a -a
ai1C01 013
14)0 a 0 U0%40 aa4)0 241 a 4 aa 0 *.4 0 44 aO4)3>.00a& *M0

(A o r Q14 .Q n .0 10V
r 0 va o4o.C.0t0' 000 64
a.-4*-.4 1.o
-o0 . 014M4C Ia a o
C~ a .4M
00-
4~.
0
Mo
oao 0 c
J0
1444
a 0
C C aC4
a
0o 0 C00 c4) - OC mow 0-C 01 .4 4)0b .0 4J 0C>-4 -A a zW-
0U4- 0' 0CU
. 0 C 014.4J MUC 4-4
M - 400141. C40
*CC C CO0 4.j ao C ab0 a1 4))'4')W
0 4
C
C4N ZO-00 O4-.460
W)1a 0 :
0 r4# .04 .Co1
40 04 *..
a
o' a a4 eo
aa r44
-4
.4C - >r. Ia a0 4 -.. U u04W00
Wo
OWh-a) a0 I
-4)1 00'o.AC 0 a 4Jt 924) to
.A C >0 049 41 4 0
-004)0D oxa.~ U 0-.4a1a c-a a 20w 1%- 4AJ 0 , W4 1 '4"014 )-4)
C 0a4o
4))04a. -. 014304 4 4 V-
$ 4 6 ~0
ZU')-0 L 0 4) 0 -.- 4000 ,0A Z1%00 10P
04400 CH04 a. 0 0 0 441
.)00009 a&
N<-0aZ - "z 3 1. 044) *--. 10 aO000
04J C FA 4600)
-4 C-.-
010 a ).- 4 0 00000M 0 00 4) M- 44.-1-4400 trc a 0 C a r a14 -4a W3
4) 0 NZ- r C00 z 0)ZCo>. r 0.a NZ. 1-40 $4c.0a1. 00on
>.0
C 00 A0041C04
to0 0 0 .. C 0 z
1. 0 4aom
o~~t a -4.- tr
a14..
0 1 1. N 4) > 0 414a C44 E414
04 NJ 44 0 4 a
.01l c>1I >.0C 011 lU- 9". go >.U 0 C& 4 .
0' 0: aC 4)004 N a C Q -4V 00.U) a0 64-.c a
ozic C-00 0 go T4M0 V0 0r0o4O a 0W-Ia
01~~~~~0
. - 40Us0 14 04 c0 V V
V)4 >W1 r. 00 00 0V 4 0 M 0o a 0-14 V014.Of. g.1440) C
-C
a 4 =- x 0 j 4) to-41
'o 0 a
W-
0
O j
~ -- aC-
A A 2 0 0 or
0v - E
% WIM->I0.0
S 4a * A%04 4
M 0 0
0 -4 0,4
0 4MI.N4 .> i ' 4
-C
a 0 V
0. > v-Z U .40 1 0 W0 o 0 to 4 -40)0Z W000 a ao
0 0 r O 1 0a 14 0 M 0 0 ,C4 aV1 4 0 4
k-a 4) -4) c 0 00a% 1 a 4J4. 9M> W ,4)0
-. 000.
40
Cjr E '-0X400C.1 4)V v 00 c o a MO4u 4-444V0M1
u za - rj00C1,M 1c04 0C CM4Jj C %.4W - -1.-4 V 0 0 M
.0 0 L. 0W U6N0 0 -4 0 V4)W 0>0 to440 0314J 00
< 0A. 2 do 341 4)4 Z14.1 u. 10 0 1-CV 4a*4. a 4 ) Ha j
C. - .0 v.- c o o wea c0 a4014 um %4a
0.lb 0 r. 0aaa>V0aC
0'-
01z . 0 - -. C4410 t 00 .U)jz
4) ,0-. 0. 4 0 4 400 r.0CZ->. W
40 ->0S
Wo 6to0 0>r- l 00.0 v-v-
100 44.- Ca1 -I4.a0'-414 -44J0- 0 6)A
W -C40 04)41' 5I 0-''CUA .M2 "I9AOM0.VM.C . U 0 0 XC.63
0 1.. 0 0 0m >tt04
C $ $ -CN 14 0.0 0 09 4)0 0 2 M 4 M
0044t
4)0 a3 - 0 30- i %v0 000I - l 0L a 001 E44 040 1 14 0 WC<
* 4. 414;a
40
12
a
0 C40
oo(4 44
0Oa4M4
*01
..
jV)0
3 -4430 V)
4
a4 J= 0C41-4 02 0)) C k0aa
0 0..4
.00-
00
14140
0. caC04--4 10044>0 -aa
30 %
4 %0~4
040.a Q
OOC
MV
COM
0.14 C z0 a- OW MO 0 E 0, 0
an 0 OH &10 .0aa 5 -0 E 4n 0 0 04 q4 4aV0
0
4.0'
0- T..0 0.. . rU)a OCC
$4.01.WCE- 44 $ 4
4 0Na a) 0.4-404040
901 4a0 t" 44, 0
0- 14.4 U0 c -) so
aN a,~Q t
>C001 L) ~~ ~ ~ 01 014 V0- 4 0 0W
CW-o.4 a 4) 0'
(n 904 N4 0-4 O 41 1
U
C'4 CE 0 43 to4W 0 -414 Oa $4 4
0a*4
-a@ 4-4Jaa4 0 £04
W0 0 C4)%3U 0
NO-z C.00
0 144 44 Na>
0'0
>
00go
0-
1.4
~ ~
CS wo4
40 r-$.u0$4>.E.1
a 3j 0%-..0-I
a CM
0WV1 4j 04)3
Wa>0
s 1 0>v 0 -04 04 0
014tV4 U) C 04104 .. 44 '0 .4 *0 0 0~1a
tO a,;,
60 a >14C.kC0
,4 C 00 M.
Vo 00 00 0 a0a~t a *04) 9
0
00
-C
W 0-E
0j.0-V)Ic a 0 94
0 2 4IO.14 aI0S
-4
0. 0'C 04 H 24)
0O 0 00W
4
-A
k0 4040C
0OIII.
450-0
U0
140V0
9 V.Q ~
A; W~ 44 W "0- U
0a -'4 4- a' 0 M ' $0a-0
a W jaa 4 r 0 0 u -P r 1a a .
N4
4 0.0V 0 014146 ,CI 4"4 4M *4>40
2. 4 q0. 4
Q 40 W-0 0 041 .4r0- 0 . -A--V O 02 -4
4 4)
0 c N 3-.
v01000C .
1
14a4
~40
0-
44
4)04
a
O 44 0
0 6hr
00-4
3.W..4
%0 _V 09
14f% 00
CDQ3
1 r 4 n014 a$ 4-0 31 C .40.4
B-30
43; N N 4 0.454 -4 c- a 0 fI .0 064;.6 4C2. :A A

Of 0 430-1>4i0m6a4. 41
CW4'004443
64 0c14 36 0.030060$.444 3 .
I r-. as4031V0 1' w
V-00 44 0.4 44 00
I0k . r .0mk
044 0 00.0 "4 (' 0 64360 0 0 0600-14)00
0! 01.-4 000 C43 .4 40 0 6a 0 0 0 44h4.06 0 %4 0- a4j >.
Z2 0 .44 0 0 a r.0r N0 I0 >3..a- 4'0 414 0 .@ 414
04 c2 C 4 V -43 o 14 .4I
4 40" 0 - "
ga ~ 4j 00 .4j
4j CA0%.
-4
. ~4M 4A H
""4
H 0
Q
44 "N0
i0*U-
4 0
0%
.64a
14
01 0r. 0 0 0 0 4604 6
4,4 4
V3 n' I do U0
M04 0 C 4 0 0
644. l< 0 0 4j6 wf a
a8-. go0 4 9
4314 >G
4j 0UN .40~~~~~ *.
go0 .04
6 0
~
>3~~6~"04 .4&~4.'
I44~C~1EVg
k4 0 1
66j 06".4 0 ' 0 436 4330§ 43I "
43 .0 Do 0430V." 04 00a 40 404 -.
-40414 0 4j344O 1406 .41
0 8-
Iota-.
H
4 -- S.-(
~~ 0 43z00 0
~ ~ ~ 0 0 ~
66 04 . 1 00 40 1 40 "0'
3.0f 0 V30 "00.-"~. 23
40M05- 1444
.46; 0' 0 %60 " V404. k .440 w 60 a"4
0 1.334
10 00- 0M.4W3 00 M en 0c 40 r 0V304346
C d
4,444 o*j
V1040.430 14-' A'4k0
$4 A0.
0
a .414-6.0
W0 A
G 00 0614c
00
tu*
$.40'
0 0j 04 143004 40 6.4 6 1-;1 C00
)4.0.
4) 0 140'.3 ilk 440 014 44j 0~ 4 - V4 4
NO 140 14 to 04 0 CC400
14 0 a 142444 4 w 0 C4-4V 64 "4 14041v4
a
N0A 0 41 22 0040
u4 4CO0 ". 0 It06M Coto 4 w .146=.A 10 r4
-4140 6"9.-0 0 q1400'V"4w Oa $4 '0M " "4.040640-44 .00
00 i r4 1 46 "4 U~. 6 06.4--0 0 0 0"4 v4 r.a
A'
4-.- 4 00 0. -000 0E M4 V ,' "6006'"4 4.
014-0 40 11 0141
c.6-4 6
inO W4O1~ -4
14 a a Ow0
4
10
C41
.0$4-4
3
C30
1-1V0
V=0
4a a S
04.4
MV
C043
6
0 0 144 0
40
06
a2
M
41. 140
60 00V%
I 6.&
-0- .01Mc002V64.44 m WClN4 of. 0,e a 9338"
CA in 43
401>0
4.
24432
043.
0
00
0A
~ 0 a0 .4j 43-441
1
>. 0 44
ai
00-4
k'
0, 43.
0 00
"4.
a
to6 6-4"
we100
0 40
3W00.14
0
60
06.6403-6-
M(43 C 43 1 0
040 -ix4 .0 a,0~
034 to404W .- '4 43
02. 42 40 0 40 .- 46VVOR
020044
r034 a3 w60.0c 00.6 1a .0 0 in & -40' 004 M0 '3 CO 1 606
4 0 a44r.0 -ZW 0 RM 0 0-4 )S( 14 c0.. 0N"Mf C
43 M"4 0 4006.I-44306n
4C20 0 41 0 v .D 0M0 cOA 0 w4CID r-4.-410 -to OV43a0oo43 . U
034 j 0c0 0
M 2c -'li. 140 t o.x0-j.4 a U $604 N.46.A 6041 . 43A 44a 0r6r CA40 406
N%4i 14 0 = -40 4 014 w4 N64M61 606M4C4 014 '0 H0.0-4 0014
4
(A0 00 C4 4 0 43,q0 >,04 00 40.44 q00 4 14C.'4V3440"
4000 '(O N 4.r .0 030
0 4 0. 0 4 444 *4.0 .4 .44..0044
$ 6 4 4 0 140
440 sr. %0 4004
0 : 41 0300
4. > a M-At;= eq0 '40*.0043-4 v 4.14r- 3.41o4004.M0460
0r30'
4Jt 49- - M 0.4

3V6 J 14-4 0 > 01461M6a . 0 a a
44300 0 A.4M o 0%410 0 .04 0
M 104300 40I00 A.a
wpm0 x > 0 0 -0 6 .041 14 '00 .40404.x. - 4 '
so 4 MCL 4 444 to0040 0a 64 43 '0.0
M0a 0t0 alN 43 A0 4
GU3'' 094 q0p 41 0 "0
H a 0 In 14J
c04t;0'a 0 x 0 400 0 k U"42X>4 .406 00
6 1 =06 0 0 0 4 - - . 4a.40 64066.0 ca
4 04 S . 04 0 4 (I.c
0 H -4 **0'0M
014
C4 0-U W 0 0 a 4 4 -4W0 V 04JC 06.43-
4) M4' 00 0 0 4 >.$.QUt)
.0 a8 .4- 040-4044
uj14 04 W-) 0 .4 -1-44 0 Wv m -Z 0 0=146AC0.4600.W
04 a 0 C< 430 a04' W 0~ 4 "443.J46004 0 0 4
0 0 0.. 0.404
. g .4 04344434. 0;0
404 J00400
- 61404 ;W36
00
0 1 A010.. 0434.406V1400k.C
VC .e 14Z J . A- 04.440"4r 0A >0 40 4 143.4J14%A
- 0
0' 0- 0to
0
Da* 'A4 "4 3A=43 .) A 0 1 0. 4V4 0&400>40> 6001U00044 a 0
.4
>.4 z14;00040
06>a.rrAw
& N E O 0 0 0 44 4J.O 0 0 :
0 0cE 430146C
0 c 0 04J44t4 00.444 l o 0 'A W
.4
60
4.V0
1414 H 0
0
~ ~0V6.4
~ 60
$
004-4
H9
0
M;3 0
*244
0.40 k'
61
g
O$ 0
c4 4j
0V.0,.4
U6 0 N4
0 00 0
0 $4 6- V 0 0 C UA a440
A. 00 43 "4 J.0
000X4k 40M 04 0640
I 43N 84 1400 .4 0 0 04C0 0"W$
H44 t 00 04X> 0004 o a (A
41 04U4 $,0P 0443 vf
04330 t, -" 400
0k x00 4 01 M ' 00 0 - do4k.414.0"0
>14614.4A. 01 0634360 430
14 [ 1
0
tM u
a 0WM:1 49 . AZo.
az
0
43c r le0r0N
a 04r: $ I~4j 4
k00 014.604 0a .4.
40' 0 01440043 0 a 4
44160034 09 -4G 0 .0-a
. H64>
ON 4 a.. N60 M4 HoaI 0
4J a00.04 I0 0 .460 O0141 N
NS c' 3 00 W~0 09
144
4404 A .m1 4 . J k4
C M4JX 1 rN43 r
ata " 4001 ,aac;
OW 000 0 t- 0$4 0 044
1440 0 6.14 0j0 14114N J I
0 64 CL4
N0 %0' MO>4 640 036495
a r lp s a 0
to
1~~~ ~ ~Ma
H
~
4
pa
01,~444-
U0 1.4 '4> 0
c 00.0 0i404
o
0
o 340k.O
4
c 'U 0ao
0 20.4 0 03,140 041 4 4 0,% 40
43 M20
N 40 .-
"4 1014092 4J6 4 3.
. i4 ".0 Is0- NO 4349 0. ff43
in ON 0 6 64 r.
94.)00464 0> -
. 'a .40 0 a a -4 4J06 tOM4 4
-4 k-'4*
6 - .c 4 4 " 4" Id 0
0 Z_ a c $ 00 a6-4 00.
$4 1 0 0 4 9. .00 012 0
c' j0
~
Z~~~~
4
4
A4 ~~ 48&3 .
0.0
40
4t qcn3 U- 0 04 .4
, 043
%4.4
0 A
B-31
w4 000 000 CO4 -. 40 a0 c0"c f -M 64 0 r

A.) j4 W '04.
0 1 V' 4 VC~4 a tp 4 w000 00O.N 40"440068Q-0
ac0 @54403 .4 C.0
aa CC .(0 430a0a' 44.q -a4. 0
' '.0 IN 0M 0 4006A 0 el 0 w0.... . x~ 40a
1h sED >' .094& 0' 04) 4'40.430 0 ;oo >A 400
>,> 4000 064
I0. 04 U404
*.0 r C 004 -M 0 v 4X
M 10 'AV 3,14. 0 0 00';~ 4 A V 0 04
00 MV 4)0 4wtp 10% c0X. r.
C00 t6
0; v'4"46*404430 0.4R0 (0 00 '009 00
4 0 1.0004 09 09'94* a0
0 0'.' 1
M W 4 ..9 A k0 xC 0I& *4 55094 %4 C2 0 ma0
0
0
-
A4J 0
0
0~~~~
. 0
C
0'O
-4: 0fW
$40
M -4M60643-
4c6cIMc4
00''
go ,4j0%c
we0
0-443
'a0 >
0
42 r
s.0
C4
54
a
-4 -443600
M9.4 6..0'49w4$4.0%
13 0 C043 1>0-444N4
"a I W400.
V40 w W 2 'aU1 6434 g-40d
43 U20444c 3 %n z 121.. 13 e -40' we- M.434-
-6 0 4 D.
5400r. H >, M0' 6 s 01WW
CW V)
jC*9C90 41'-'>6L.40 0 4544 A.
>. V2-4,-4> 1 C t, a k .
O 000 r. 3.6.4.0w V -AC >kMa00V2Q
005.-615 043a44 C6 m0 C.' A
l
11 IN 14 rE4'V0A0 4 G0-40. 0 0 0 ',M00' C 'a '- 344 50'- 05C 4 1.40-41 .
c 4. 4rM04
-, = 0 c .6 4 1 0 04 30 00 00C) 440 4
0 L 004 '400
0 6 043006~
->,oa. 0 CF
O0W4jV0W
430'6
54 4 431a ~
$00
5400"
-4 0
0
4
G44-C5
043
U4C9 w 5
! >34
00
4. '
0
W 04 004. 54 . ,0- ,a' to 0 0 & C -'.40 00
0 410 040 0 V0 ON 6-0 '01 0-4 k 5HU4- H - 4 P W 'W0
0 04 o 4ar4
C -
0
%4 0
9to4m.;40.1
-
%
(40C00
0 to00
4j54 W 00 0
443 50 0 000 43jr
~
S00.4 Q W.004
-4600 40 -41000
5 030
>*-4 4 -
440 14'
M-. 604m0
9 09430
m
0
ov.~ 04
44
0 o8
O
31
N
090'0 ,0 .4 -3014 4 W606.4 0 0-4 -~ a-
6054
'4-4No H.4-4.N 00 4 4 1 4 4 >-4$46 0 OC '49440 Z *-4 M 04 50 40 0
a'~. 044 . 0. A43MO N0300 0435t-46950J 4 540.0 0 '00
CO4j. 435 '4j H '0' z
19O 0 >C 4A 0 >.56
i 0n - 0 0i j
>.0WC 04) 000u % 0 .4 cHM "4 $444 A-t a 0 5400 W40
003 N.40a _4 M00M 00 Q0 00,000 k0Go
0.O'14 'N43 0''0 0 c. 0 m04. 43
0%400 43
k .C 04 0.
tC a4 .1.0 0 0 'A 0 &00 -40 440 C 0 -ik- 0 05 E 00 '-4
MO1O00U30- . 430
0 0043M
N-0 4.a 0540 4'.>. A m -00'
446ID 6004 4 J0 004 040-3.4 &_: Wi * 0V$ -t 00 -
0 41-2C
0 .
000W
06V>
in ix 0
.
.)w
'rM
0 .40
C
'.9M
0..40 4 0 t4
C2 144'0 M-1
000. t0'
so'
0 M
*-9'C 0-4j .4
0
050 - .- 0 0 Q 0 ,
00. 6
'ar440
50 r
0
C
MM
aX
Oa4.4-4
0 04
o0
004
U
113 .14 V0 m 0043 C
"4 454 C mO 40'0-4 0 0 00
'0 00
w& 0
00.
)1,
. V
0 401-0"
0$4 >0 4
:0 -H0 o4 .41Qn 0 0 r '
0W 0 009W4.4 X-4 0
C u
U)-0. cc 00.4 054 0 4404 Co 0 10 0-' 41400 0 a 4
n
H -44c 0 54 0 9CC;1! W k,6- 09. .4 H 0.0'-. 3 4 08-4 044 Ow
0440
0 0 ma aC 0 0, 0.00' 44) 0 0 0 OD~40 9.4 '044.4
04C-0 to4-9 01> 0434 6 0 404J 0'4C4 C 0 4. 0-14
00.40 6-
0.4 0N a4> 0 e.4 V
44C4 t 0-0(0 - . . AC.W 04 4.40t
040Z a> j*
QW
c j C40 , 04 0 - 6 0
r U)O "4 C
aaC
c54
N
014
M -4 -00
toc t
M0
0.'
t Na 0345' IM 0
N
00
'
0 0 03E 0 4
450 640
14~ .C4 0 a . - 4 '.00 44. 0410 44 0)
t & o~
0430
0 0 000 r
0 ,
O c 0*
7jC 0v 0m
w40.400> M40 00u 04
4C C0 0
0'C0
%
0 -A 0
0 10t
0-0
.0 to4)0 0 -0 430- 34 0 1 04 5400 430 -4 00 0 03304 00 t 4. 0 'c 00 t
vuwv VAV 0 430CD :05C V43 0 6-1 0 9C -98G000430 430M40 43 C > r&)0 004
N4N-4. V.4 U3 0000V4 4334340g 00o t 043 -
M4.4'00 0- 44 C1a 0 0.0
.4~~~~ 000 4W43'--
040 C 0-0 (A 6 .0.'-0 0 ,4 .45436
0C 00 *.444 0.34 0 06 0 643' 0k M 04306054 4 40

0 0010 6000064CA -454 04 14CC r 00
.05 X4 VC
0000000 0042
019 6 -4---a54~
0.tr _ -. '0*000300. 08 -4>(40 C4 M0040C W0 0A
. 046 54C 11WI C 0 00'.- 0 W Q00C Q944 0 "1 04.' 3.
4) o CM 0054
U 6 0'
N5 0 W 0 '0 0 FA0' 0 0Ck0 E..0 6.40
C454.-
'01 MC0 0 0 '0E 34-
a'a$ o
4 . r0C 40 -4 M 0.4 WC-W4 4 C
In V60. C r 0ar04- 0 -m - % VWUC
'4 044 )C40M. >
0'I 00 34 0 0' 04 300C
544009' 446 6 400 146oNW 04 004>0V.C
44w,.0 00r.-44r ,W 0 0k,4(4.0 0 W V0 04' 4 W00 0
cC3
4j0 0 C03. V0 )Va0 0. t 3.440 . r.04 0'00C
CD94 0 00 c. 0 g.JM 54 " 0-
343 4000431 0 '00
N -30.0 '.9'4'450.v4 w0'0t 5443 04 044j0 0 00 C.
t 443Po00c a0.054 16 44 CO '0 06a 43 4 5454 0
10
m"4jM Ca04C3W 4 0 5443W NI - V
44.0 I ,40 0 00
4c ON44 . > w 000 54031 .00 000
44.43 0440, 0
paz0 1C00434 0x 40 0.0 0' 4403
-. 005 54 C.004$
0 C
00. 4'0 044V ' 04) 0 r 0r 5
M
40 94' 0 04454
4440 C $44440.C 0
43 435N
V4 r0043 f43 0. 43 C -.- " 0 X4 :. 004 446 J -'454 0
'- 04
0C4 0* & 40404 C 000-'64
V4C 0 43 aOW 0j
4J 0654at
064 .4'0 0. 0P0045C4 0 43.-I C H- C k- 004r04
A30~
064. 4 COCOON- 4 44~ A990 -0 0.4 00 0 - 0 0 C5V q; D
,'4 09" ~4 .~404' 064 0.0"C 5.462C0 r. .430
0'.0 . r0 1C.-
T 00 ;.,4 A4
0 -A V . 440.0a0
0 0.4 43 0-4 04.(
4330.
04
'0544
. w'Z x60-0 a
M0 0'0 0 a &
0".4-m OO 50 ;4400 x SC05 030 .O-14 -9.4O
.0'0 C00 1
or.4 4 13>434 90 43 a 00 5.0 0 00 v0 4 C 00 1 -
NA
a450400* .4 0 k94-44 w. 09r4'%. M 0 = kQ
000 A 93 140%. >44.4004
V44 mu0P C 3 m a4(9045
0 6 0 >.40-
0
ON
-0 00M . )0t
u40 '.0 340CL t'
2 4-, 1 n ) %
.4
054O
44
r6
00
a4"0
~0 0
00. CZ
9 446 44
-11
00
r-4%.4
O 0'4
4 540
*'04
04354 0k
0 0 M00-
5
- i-PCD 054ak 0 00k0 0 00 40 6 0:0 09 1i
V300 X-6 004C 4300. 0O '3'40 C c-44A40 43 540 60-QA 4 I
"446 0-44 O 0 0 43630 V 0. 5443
-00 $4 430a5-40 0 0 43 a 0 0 435k44 4 4-4>I*A3.640
0.4 r 406 5' 40 0 004 C JV 0 V
544
04144. 1 4
.0 0
M.4 4 A4V3 40,C 0 '7r 1 5 -44%4 4 4 04
In 0 0 0. 0 0 .0 4 4 .4
rz V~i- *3~4
~a ON40>
-1
4 4000 900
05 0-450000a'
6-OC
M
'50
OC
0.1
-410
'0CV
"lijzho Ijt 0, 4 C4 ON4 0.4 4j 1

'6443 .4 4 . 944 0
34.
04' 0 0.40 ON
a~ .4 54 0 0.0v
0 1 0 , U..I'w.- 04-4
I6.40,40 - 0 54
[1>0
4400 NO
>1 10. N.C 4
-4*-04 f 4 3 6 -
LOU IN 0-4Ac
,a
05A4
M
54.40a
0a .0
.4'~
t I
it0 a4' r. 04
a.04 C9
4
621 0 r, 0
- 0
3 40
0
-4U 2 41
3''03ru 2 00 "40 50 0 4
I ~Ii~
~ ~~ EE.. 4 3OJ8O4
HOC....
44-4~ >14 0V... 0-4.)0
C >1.. 4
.1 .04-40, -110-0, 043 V )00
00 C.CC0 0
004 00
C '040'40-
04.0 00C
~ ~~~~~
>' 00C
0 k4-400.
0
W- V1
IM00
40 A)0 S..0%O- ;4- 0 'o c0
C
%0 aC 1 C4 0) &t~ -4 MHC W001 4,a0 Ok4J
e0
V
43~~~j
C400
*.-4C
C4
0
000
4.CO4044m344
M-~
-.
0C.0
140 0u4.4
4 w
0
~a 10
o
0
OC
~0
04 A4k 1 m-
*~~-
* 43
00.4
-C-.0 ~~~ 00
.4j>
01
>.000
001)
0 4,.4
C0.-
0 WC
1.4 C 00 004 r:C 40 U44
0> 30 u3-'
0 00 0 1A 0- 90. c* M0Q
04' %40 04 4
0. C0 - V 044J
44 em.4 -0. MX' 0'0 00
2
k0 CV Wa 4301 oot4 o k 0~ w C .50 w0C0
O 4. 0 40 C C13w $44 0 4J4 0 34OCO4004 1
0 4,. (4 A W0-J0
0k0 k04 0 C 4 w 0 0
0 .20 o C0 '0. U 4.-
) 0
1"4) 4 -MCO 4 4E 0 04) k
0. 1 .4
z .0 0- k ( 0040 0 0V0C-
~-OI
k0
I 00U
0
W0-40
.004 4-04
,~ C k
J 040 %-.4.r 4C
Cx .10
0
0.
O0E 02U1o4 0 - k 44-4.404 0 00.-A '0 0 w . 04)0 4.
00 VO -0.OC
k0. E4 to0 OF. 4) Ck A 0 --
004 0.-. 04i 43 .14V -43 a 3i 0 .t

co0 '0 CC'404 -4 0 0 0 C 0040 4 C430- CM4k
40 C N1 0jaa0 w34C.Ck -4C43 '040 00 0 0W 040 C 0
0 o0V040 r C 0 a~ 1 0. 0 044~-
0 .0I 0
.C.xCo
444440 0
00 W
t 4
0 . U) C0>C,-
1C0IkO0
4- C OC 00
0 $ O4
00C V 0 01 ,0 024a V.. 0-IaO MCW 4 04, 0 ak4
a- 0. 00k,
m th 440 tE MO C30 0 0 C 0n C0 :3:3 A 4j AC 0 000'0'040
C-.
C O 0ow $400 404.4C 00V4 41 41 1
-0 0 a -4)-C4
H C -4, '-0 -. 400.0 M.-4 A.-
4 rUW WC4C4 0a 00' W-0 4'0
4 V.C
1 00.
. .0
u4 . 4 .00 k J -
E4..0 .C 0 -0 x00V 0 0 - O 14 Ckj 3 0 00 W0
-
$4 Z M 0 C4-.0 W0. 0 0I
'a W.A 4 0 WO V
C-0.0*C
01b r4 4z ' Ar
.Cn 4 440
. -0' 0) 43 4 430.00z r' 4 0 0%X4
uE. a -C .09 0. Cu 0 >0. -0,4a Q W V 010
43- ,
4JX
C..-0 4 44J 4 -. r-4 4 0-0CW

.,40.0 .4. 444a0 04 oe0.0C . 4- 0U434
O 0''* C a030
I04V 0 043w O,- to43 to-'40
440 k 4
Aw 0CUV - X m0,C Q CO c0 . 0e0 M 00 M 241
-. lW 0 x4
No 400 r
Cl-IM V'r'04-4- n 43 C0 4 w
CCX *4
- 4)
ID- 40 :. 0IO 0 -4 w 0 w )4>.
000C4 O CW Ck r'. r~ .
4J
ZjC0 b-0 I.. . >. 10
4J 040 w.4.c 4 0 0 .C OCC -. 0rI
)0 m w 40 4w 0 0 C.0. 0 430 0 >--
OC E- 0 -40 W
a'04
C - =OV0 C 0 0 0 r-00.k4 430 MC4 0kr= 000-.
E.-C M a4a~ E40O C0. 10Cc4 C--=I- 0 I-. 0 CC4X C4304)
O0 40 C000 .1 C
QOU OC. 9r-to.CO QE0a0. < V
0 0.-4
0~~.
Z4J E.0
.4. 43003-.40 0 -4-- .C4304-40
04 fl.0
A.4 C $4 HJ C O C O O O C-4wk C
-4u0 10m w 04 0H 043 > 0 r-. 0.-4M > 004C 43-.O
44 m Cz0 00. 00 , 00 4W4 4 0 , k0000. 004U
En k w 0
r-r-.
~
C000j
o 0 &- )
0r
t .0
. 0,
0 '.
C
0004
0-44 u
w U)k-0.
g
0-0
o-'g
V 0
0
aO"(a>>.
'o0rk. I .%0 -4 z '0 Q0OC 4'V0.C C.4g-) 0 044
4- 0 2 i C 00r4 3 4 .2 4 1
C40 )g 4 Q $4 W-0 C 4J
OX
:J. C C
0.-r-
E :3'-4 -4 . 0. so 0En4V 0 %DU - aOkO)414
0 : 0 00
300.0 eI c r?0 40 0k C w 0.i n >.-4. '0
00.
W >P*.4 - 0k
4J J040 >.Cj( 4 jC. 03C
j k0 .C V 4 01
F 4.H 4 -C 0.
* WVCr4OCO~
4-M En4 r tq 41-00zV ,- 0 0 v a 0x Ck4
5 0
O 414 $.4--
4-) CI~0 04- 1J.-040
r4 C 0 C 4-0 C .. >.0..0 U W
040.44 -. 1OC00-'0
w 02-C.4- C0.4J . t C 0 w 01.C4 r3Ok0 k0.40V C
k43w4i-4043 044.J0: 0 Q V'0-4
k.44-- 00k3 0 1C40w .0
P4.0. E a-.40 OCO
0 o .4go.t43.I.
k k- 0d 4- m 0-C 0- C O 00U4 k3
w -44 0 1 .00q nC O 440>11C. 0 4 ) 4 0 4ul4340--
? 1 0)4) M CO-' -
00 0 C0. 00 V a1.
0 a>-4
C0 0 43
NO-W
w 0.
00,
) t* 0034
04 kk40040404
00) 1 W 0W0440 4C003.03
Or 044'4 4 00 $ V
C -
k40 0 V0 0ow43 0>0 3C W V)
4. co-C E- WU 0
0 0 C 0w-a C A WC0 aw 0 C w 100 OkxR00 r
4)' 4 3430P0. 0--4-
k-
043 Of 02 400 00 w o w E C**0 - H *04
U4 2IQ- -0004304- -. OkrikC<
C
0.0 -0. -4 &k0> W0r1. 09: 0 C .0 4.C Cn
04) ) "4 . 00 to 4) uQII 1.3
H .1- 0C 0 E-
43' 40040'4 M44)4
0 0 4
0O4-.4 04C4b
0 400
44j0 w4
00> $-,40 r 0) kOO v 0 0 9 .V C 04 C -o
4
CD~k~l
0)a4 -0 -. C 4. a ;>0" 4 040c to 0to go V
0~ 0 (a V C 0 rXk0 ,00 4200 S 5
Go C 0 04t I00.0 10k u3 0 U4Q) 430'
0>. 0-4- ,OD r 40 a >Oq 04-C a0. 00 COCa
0. w 0 .0uM- >, 0 0 0 > t
043-4h.-040S&
isar wO P0C.-O0U--4I4
-. oro m0 40 C Cs . t o0kC a
> V 4Vk -"4 CC0 1 0. 00 04
-'4 o0
ink l-'' U)- N1O >.H 10-0
R~g j W0142 1'a

0 0
WH4 4 04 ..
40
*I)
.C
~ 0.C
0C soC
04
OC
a 3
toa
0
w " v- . 0 %M W= - 0' 0 0
it >VV00VN
V) C 4 : 001.4 0 C w
0.0A 4144 w 0dC
0
1
Cr--4
3 kU J0
-4)4)kEC10*, -4
oC C
S 0 4j4 w- -"4 ' V) U *.04) 4 k-C U-4 300
1
: r j-
NqI'-kw- 4 N o4.D-4Q0'-41bi$ 41.
144
* O4j a4
410 d) V I4u w44.40- kA- o10 0E kCkJ
0 J Q= 1 a( 3. k - V
to0- >3
U.k kO" kIVUVV
j1
44 C thP" - 4w 10_0 0 C0 . 04
B-33
.4.~~t o00440000
kzV04.4
4)0 ~~~ >4 -'W ~ - a40.VaC$
-
C$4 c4 .004 0c4. N C
0 41 .044
aOco 0 9 .
0) > 00.- 0-k . o a .44)44
NsV 0 'cc W
0C 0 n 1
a O rlN a0 4) 04 2 1 *."490 11
0.-S a in r4 0 .~>0 00
.C H A44-
OCCwf r40
00 0-A 0
-4W- 03 . 4-.4> . 0 -4
<4) CME
0H-0-
~ to C 1
C c4 0
0o CM c -.04 4>O 74>
$4 HH-
&0 00 0 0 04)
, 4. 0 :CC-4
00 '10
*.04>
4)0
-00~4 4)
-,5 4 0001 00
0.0 0040 . 0440 O V.
-4 4) 0)
1%4 Cl tCo V44 "1 4NC
4j0 04 v)O 0 a c -4 A MO-404
0.$
41Ofl$
0 to
C4 0-4(a44 .-S0 *0 00 a 0 0 .0)a
N04 .) 'I) c4 0 NO CMVAO 0 6 -04 )
O'4)> 00 OCO 00). *j 00 00.
to~.4.4 0 W 0OC .e -a 0 0 9: 0E
N 140 .- >00 04J 0 M
0. -(14)'l00dA
0 $4 0 04 a0-0 41 0 44 0 0 0 0 -4 4
HC-4 W0av C* r r4a l10 i
) C)r CV 4CON 00-a 4 O P0
44) r0' V 4
-
CO $ V,<$ 040
0> $0 -0 r0 w.
. S W
4< *O'a 0.X OC--404J -SC
-40000 4 0 0 E--HC 4) 0 0
C-0. 00 0
.5 H C
*44.0 00Ma,$0 00g. -. 0 00 4 Q,-s
MOE-.-- Or- 4- C
4 - HOcM 0) 4 0 $4 N0-4 00
$4 >-H CD -.0U 4 ar
Ca 0 -40 . OCIV4-1
0400< 0)4) 04 sO HA V 4) 44. 00OMCOO
'-0 A' 0' W.-SO O.-,-SrZ8CO1 004) 44)04W
E4 -- >0$40. 0, 0l $44 CO-M O. 0
(is
VC4.- - 0> V 0r.CC4/a2 004 0 U-14 C0
<$4000.>. 04 4o4 0-$O 04 r. V4 a$OO
0s
0 c4- 4) * l QO o'D 4Z31 4) CI 4) 4)0 020 x C t4 )4
WO-0 0 0 0-4 4) N 44 -040 ,m 00 06 V-M M04 $4 -4
1- 0 00-4
4 VOWa 04.00 4A Z0 . '0 $ 0 4 44 V00 -4V
0C
4)0 >4-
04
0 >44<
V. $4.-
VC
04
0C
4
00 rH-_44
.0
r 0C m)0a
00O 1
~ 0. 0-
4)0-
0) C
05-403.F)H - C 0 0.,60
40 O 41 000 C 0V. 0 010 V4
00 0.0-0 $4 OCGo
$4 04V 0 = Cm 0 0-e4 k m c000-
9 0I-V)- 4.0 0400000.V N'; .- c O v Cv4S- 0 4 -40
4Z<E-0 0-s 0CO V< V1 -v- $4 4) d
0000 .
_ Q>.040 1420- CVM 004-4a .0 0
= 0
4 44, V,~40w4) w~ 00> 4j
<0 0C04V w00-4
OZ 0) rS4 00 4Z0$ -- o- 000 00S $
4,0 CH. 00 0. 41S W-SS0)0 W 0VS 0 0C0)
44 C44
T0.0 0 N0 0.-- 00 40a 0 to 110-
.401- 041 ,)0 0 0 a 3
.4400 t . 01% r.$$44 44
00 --4)0
z "-C 1 0I- to= - C'0.4 -to N0O--S0 001 0 414$4 M44 0M C4)
-
9 cO
~
0-4----44
l
.
V<)0
- .. V3)-50$
C
0
- >~ <W4)
zOOc<
O 4-
12
u >
0 $1 4
0
4 0
V 0
4-CSV0(
to w
00 z U,-- -S'C I U2>-4.44to>t <0 a000~.40 4C' s 1.W
W
c e . 4)4$
0'0 C.
VVC00V% 0 0 w 0 000> .0 . -C :
0 $ 00-S4,$0
4)11. 0 0) --C 0 00- 0 4 NO $4O4.50 4 )-00 4
0 04 .0 0 4)0CM-
0 $4 NO'k 0$444J 0 W -$4 040
IX,
$4 ID
-VC $4 40N 00)4 0C E4 04)4
4C ' 4. C 0$V0
0.0M $ 00m 00 $40. -4 -4 000
cc0M'-usC 0 C 0O a w4 -. a'0 $4 C-4 CV ,$40
w4 N nM 0 00.VCID
00O O 0>. 0 k0 $0w V Cg $4
4
C 4 00410 $4 C--S>.M0 JIC0 9 0 34 D .5 4 )
30 4JC0
,4-
0 4
4
04
OCD
r
r
Q,
E 4)-
a,, to0CO$.c-
$> -
$4 TW N C Z 011- M.
$400 a044c
1.4 00 1.2jH 4)4 0 4 $4E10 OZ -S
= -OS--0 00 a 0-C 0$ >-00 *CW1
0 0V 0.~WE v00 V a> 0, s 0 )0

0$4 0 00 0.
00044
CZ U. 020 4vC V C: 00'C1
V0 .0 044) 444 V'50'
$4
00. 0$0 9 4
00 $4
H0 0 CO0-A-S 4)0 4)0 4) 0 o1 V 4-s 2-2
0 L
0 ,0 4)4)0 V0$VW
00$ $.4
4 04 -. C 0 0k44.
a4S
0' v 0$0S~$V) 40V 0 04 kU - W
~~0
4)C$4
44
0 $4
0
0000
000.-P$V m
4 4)0
0
000
-f-$ C
00 00
af$MO$
0 - 0
04 $-4'4-
C4 0.
)-0 -a $4IA$.0 V V C 04 c 0 z $4 0 0 $4 0 0S4
* 0 a00$
4)0 N -4 1 0~
004 4)0
ON40 4 .0-4.0
r-N> 00 0- 4) o'+.
U.. 4 $4
01>.$4
-1%Z
~0.0
W'$40M4
~ ~ >. 4 0. O 0
OC
0M4
0Z 0
$4 00
$40
004 3 0 .0 a m
004)0>
%C
4-4
0
0) C~0
.$ CO
0-4 0C 0 -4>0)0 0 $
- 4$4 41 0
C,0000 W0$ 440 0c 0 5-4 0 CO 00 - 4 00 04 00$4-C
0 C .004 -0004) u0 $4M4U$ 0 00 W 4J00 0
4)C
W. (a0. 4 0 jc0 ZC 'a 00 .V -0004 0 > 000 %
a4)0I. ) 1V04)0 00 0) . 4.- 4 'A r$ 0 -5-4 w so$$4
-01 m 4U -- * 0 0' V $404
C 00. 4.00 c0414'4
(a0 1MC 4 -.S0 0 - 4) -.44
0)(AW500 .4 0IA04)
$
4J 000
0 .00
004 0-4CC
, 4 $404 O. k 04) $40r4 01C 0
$414 $4 0010-40) 40 -C N 0 0.S 1-4r
040 OC' f. 30 .w 410 4 0 c0 0 V 4V-
4$4'0-
04V0U40
0000 - 0
C
0
-- 4) C
. 41 044
-S1l$ z.40
4 DI
.k
0-sV4$
o0
4)
w,
0
0 0
0 a
1
4
.. 0 -
0 00011
400 04)
0000
mea
4r C
.0
W.l
c
S 0 0 0O 0 4. OC$4)'0 W r0004> >0 4)0 V

00 00 $4 $4.. 0
00C: -(0' '- V Me W)0000IF a >C0 0.
o -.r0 0W,- 0 40 8 04)1
' 0
u vto00 -40000 1 H04J 0. a60
4)-
o os
(a w0 0 OD $4I
. 0uA 00
C04 "4 V-0 00
V V C01
O
41<.4 -- 0 41 a$q
A4$0 4) co) 00>
go$4 0 4
C O C ,v00 0
% 0$40C $40 0 c >. 4 4 0 V~M
W- 4 $40 0 a
$4 $4 m
ww 00 )' w0' 0k 0
054
NOW
to
k 0$4
2 0', 1SU010 0 -u
~
$4 .4 4 0 - 04.w N4)
_C
02
T
C MO 4 04 " $4-.1
~0 v 0
V0 $4 0. 0 4*.1u ~ 0 w -.4 - 0 W04E V 4) o' C
so6-0 0
N<, 4-O
1.20040
(<0.
04-
00
0'u- N
0m4Cz-a
0
> 0
1140O
k44
*4)$
a4
22
N cc-04 c 0
q0 a 0-- 14
Nl-S
v 5) V "41.h ~ 4 '0
r< <0 4) r. 2 4J44 V4) ,I 0$ 1
44 E6 $44 0 $
0 0 V .~. S A< -.trsM
-4 >$ ,v 0 -!. -4
1-00. CmE0' 50 044
-40 0$44 -C4 1$a b49
f0<aleS.m0a Q$0 -. 0. -04 00S$4
B--
-A 9 mooU000'14'- 0 c14 y40 -AO>

H o 0 .4C a.14C AaMV O 0 4 =0 u m
a a rC'J.
tQ-A- 0T0 0~0 >1 a04a
ollo- tj$9 4j0OS0ua4 0 M fsat O>.'4Cr
w c14
00 8 4 xI
R1A
G m
00-400
1 J "1WA u . OA0w
: 04V -b 0 0
co4pa ll!'0 4j~4Q 0 -. 4 - 0.4 -4
0~~~aA HU ~ - 1
).8Q-44 43'9 0~..0 0

r. 0- 00o -.
-0 0 0 0 C >O 4-)
V4 g4,) 0 % 1
:3 0
0'to '4)-04-4 0 1 0 >o0
11 S'4 a 00 C0 .4 0' a (A 4 v 0 a 0
0-4 j-I -. V4) 00 44C 4 -. 0 0 )
go aC a0-A
u IQ4 0 9 0 CA0 0
W41
04041 V0 10 >I MA0 00 I0 04-.
00 0-
0-..0c4 0-4'~0 x4V A0 3 44 WI-
c C>
4C 1 t 0 00%1W4
-- 40 0 ).4.
M 2 -0 04C1 z00c 1 0 0 j
41 0 4J00r4.00 ko. 4J44 l 110 0 1 100C 00
o'.'00
.4
044
0
W00-
0
4
04
tPr. 4
10%4.d.A. 140 $4-
40.401 =1 ,C
CO
W
in 4414-.00C
.0.. O -4a
4 $0 01-.00 1 . 01- M.-04 0 0' 4. 0
.0Cr*0
O4-.04 4 -40 8 00 $4a-.4
00, 0 C 010
G u44 C
014 00 4 00W14 14A O4-W C
U- - C-4 f 04
-,I 09 C:
q*- 10 40 - U ) 00 0 WO9 3 r.40 WK 0
O 4 0 0-4 J>. 0 $10to r. W -. 402 C.
a 014 0 * 0V-A4-M >O a v-.0 .% Q r- 4i4 0 09( 4 0
fu0 -AC9 0 'a01" WCa00 W=0 0C 0 -( 1 0
-0 c -U 0 9 0 $4ais0.410: -4 r. 1 W 4 4) 0-,t1- -H 0(1A 00
0, 9 V14W 0 0 01 N
ri~ noo 0 0 00
6 54 C n-o V
0'"'~ ~J 4. AJ~4
-W 0 t.0 0 41
10W01)4 0 N0
3 ) )4-
.4>.iA 40 -04'-C
0 0 4 40 V r
0-400 4 Q CC)GON'40M O0' '
Z 0--
4A'.40 kA00cc01-1404 I' z4 WCC 00.400 (4. C-
40
(/2 "4-44)40
*CO -1 .00
00'9VO
0 'C -too 01.-4 *0W" - 0 14 rz 14 M~ A--4o-m
104- (0 0 010v$ '.-04 0,- 9.0 00
Q4 SA..O04 040
M 3 $4 u (2 0
C.CU-40 0 -aZ.0 a0
P0C
4a $4 C 1% 14 -.4 1 0 0 44W4J 0
U,40 0 0 r. X40~
W V 0 FA IN .4J M40 0$4 4 '00 0 14 J > M- 4-'
r. c In - 0 1409,4 . 0 V . 00 14C
0 <&MOO c 'a4 c U-1 0
W Z0 N 0 0- > 4 9 0 00' oO0OOV0 0 Q 0 0 0 *0'-M 0
j0 04P 0 0 -A win. r. OA.
Pi WJr -. V-4v a 40a9. ac0 4 4 0: N440 440 00
q54l
0 0 0.CD4 X. 014.00 m 0 r-40 mw $b'4C0
a oe 4 . o0 j in 14.0- w n t
1% c -49 4 -M V)~a
OC44J 0 01- 14
000-'o VI 41214 31448> W170 0 -1 44 0)4W . V
- .. 4 '00-A -1) V 1 ). 0'Z441 M 0 Wo44
.. /C0 0 0V--4 1.0C
4 4 2 C140 ' W ,410 00M.. 0 jn -- r0 . r. ..0w0a 0
nWU
C4- 0)' 9 CO 0 0 0 A0 I )xHWCk M" 10 a00
.1 ( S.0 xO4 ") -4 W04U 0 004 -00
t
az 0 0c 1a4 .4 .0 OV0 $4 -'4 -49 "-4
W 0
Z15Cz 41 r-. 4J 0M r.MCOC % t arc
0 C 00 CCf, $ 0 -400 v oWI 4 41 a va c
u o - *14aa
ME.0 2 0 Q- a.0 W00 S00M0 Cz 4> a . $0
0 0.4-'C 4 .~ in1 I~ IC'-W0(40W 0 000 01-D4
ga 0. cmco 4
-0.. 0 0 in0A .9 W14 0000
04 IM0 1 .OC 0 0 14 wNo094i
0"
00
14140 $0 -4 0)M* 0 4V 0 04 A. CC 0 0W>.1 4
W "Wa
4)00W
*4J'
0
4
-4 01
44
0- WC-) 4 we m*o
C'O.
z
'4j~ .
W0-44))W
0
0o - 4
4
000. a0
00
004
0oOW1- A4'o-44
v >.0
0.14~
4)
~ 0~
>,-'4C 00 4)
A
Q
(
$4A 1
00W
Cf
- z $4
Cu - C-Wa0'I 0 0 0"0.-A00 '4 '4O'C
00 - CLA >.. 41 0W 04 Di 0 0
0 ' 0>.6- 004 0 " 01 w 100'W(4 0~r 04C10
00 4)..cc4cn
0
A0' 04
fu 0
0t.) r-'
CO'".0 OC
S4, CN
q00-
D
"
0"to4
V-0 Co.4101- V900 31 cm f. 150.
a
MO~9 0 1400a-A14
0 0 > 1 -OOO
CO0 j 04 -- 490 -41 W
-40
0014
V- -(D"n-%a4 09 r i- >I 04 a0
Q V 04U 0 M 0 0, 4011-0& toc
*-4
4)
~
E.0.-~ 0
1 0 41 0 0
~ $.
0 4-
i 306100
00
9
(4410
.00 000
0C
C0
-0 0 0I 0 4 01
C vl%
.
14(12 0
('. O0 .4 4 004
0 C
N o4r.
10 AC S-
0)..-
.0~~4C&&W
0 C-
Ua. Vv-Q
" .0 0,jS 0j 4 It
2 .0 04C
4
1-
1 U&
00
A 14-
C u-40
9 0'
P
'vO
> 0
4
IE'44v
, 0 0
I
10 V~ a -4V 4 0.0
,0
to go 00~~ t 1> Lo 1 a t,( v 'a T o-

U.4c c 14m
'004
I'-
W0 V A 000o 0 r.t, 0 - 0
rj 01101 -
to 0c94'l0 % 0 0 N0 0' oC 4)dr.

4, V 0 0'. a4 V
k1. 0. - 1
C. E- 40 1 0.-1A9-.48r. 0-4~
O" V 024
I ~!V00Z j~~oo
a r o 0a0
CL0114.41r41.9-
~
40 0 .
B-35
20
ro 00 V 10 m'
0 0 0 'n3 400r0 . 0
a..
-a . *-u04 I , 14
-' 40.2
"4(-1 10
'a4 ar
1:.I' 41 -4).:01:4 '0r 0 12.0 01Ma0 .11w0.
0 toH0)U H0 *-4001 4 100
W0
0' 41-
100 0 4 04 *4 0 r401
CA 4* 4)f
0c 00.
A 0-O'' 0. '(. 0 0 41: 0'-)04JW - 0110
0. 0 440,# 0) w0>0
0 10041 w tr w w 0 C. 4) r1
0M1 00 r.141 -0 c~ 0'0 0 1) 0 H O .
w 00 a 0 a0'0 m 01m14) 0. w
0. *-).
(D0 .q 0 0 O00 >.140 a- 0 411 W w1w41.-.
0
0 0 Q >&O >1:::1 Q
0'(V$41 00 1 4 0a.. 'D00M.-o-
0'- 0 Q* a E, 0.
M--140E: 30 to 10:--00.
1: l. 00 0l *-
m )0.-C -2 D.1 40
r=4 0- (a
0-4 m 0H .
k.C 0'h Um0 , W4 ), P. )
-m4 rO G 4jO 01 * ..M. r~1 4w . 40:a 4w
4 0 , C)' : 10 10 M: 00.4 10 00 1V:> 4
'(I) 0 "I . 0 - 4 :0 4 - 0
O 0 ~ 'N .10'4014
40 ON0.:0.9zI
0
0. -04
1 )4 00 04 00t 0. C-1 11
0): o
044 4
~ 0
to :1'o 0
00
1)4 00. 400112 4140 0 1:m U) r10.10 0'a00
.- 411 *.o0i. 0. fa 0012
V- 0.(A W 0- 0 -
41.-It
44 14 .10.1-14 00 Ai 001
E :40- w -c~' .-.o-o o
04 P 0 0 44 'a >:1~4
o,-4:a,
00 w 41 w- 0r1 u0'01
-H4
V1: 00N .0 V 0. .1u 041 D 14 0041 w v V1..- c V110:4--
w:
00 041
$.I r. 4)4 m.1to C) 44 1 0:.I.
r 0'.- w4) 000:10 24441 r) a) 0 ) g4 H0.
.-
M 0. 0 41 1::00. . 0 W
*.
0'' 01 (d . .0. a Q AUN
-4 0'1: 1 1 'n0)O
411N -1C-)
o -'
3.: 0 00o 4 M 0 000' 0 c w 01 L,00--Sim000.0 .1 . W.m
000 0.4.0'.0 "14C11 ca 0
1' -C) $0.0 *-IC 0 )V - 0.
03ax 14
cc 20040I 4 r. w 014
0 - 00 rW 0.4 c E0-4 *..
W.0
C4I)-I. 0 w4' E. C mX00>E. 4 UM
w4 0.41 1..- 01.0 4
iv ~ ~p~ ~ ~ ~ ~ ~14~ 00~ E00
~ 0 ~ ia~ -11
N410c4 n2 .'0:0
a).
-;-U
014..1.1:0
r041
04
. r
40
HN >.0 C( W0.C: 4) 1: r. 0~ > 04 0 0 4) Qc 41 0 000.0 4
11 010
C c 4j-' 140 41 411141-.u 1
V' 0 0 x0 1 141
*.- L U~0 W'
410 W-4 Z'00 - 0 1: 0 4W UN
0I10 0 W 0
-- '21 -0 :040 a10>01:0 V

0 0 'a00 01: "-4
41 .01 041 WQ)t
*-4n01 > .- '00C-A144 ) 400.WCC
A
0 014400 m a,04 r 0 00c00Va.M'3N
m $ 0) 410 00 44
l< 14 0 V0Go
000. 0 4 V0 :0 1:1 -00 1: 10:C >
v m 41 ."a00"
c 1 4 9 -4
) u 511-3 10 (0-HO 0041
N00C.- . :j 4 , ..- 1 4 1.-I : 4 0 C ca4 0
H -0. 0000.V 410C 0r -- :4 r0. V -
"- t WN4.
41.tO 00r4.
4 :41) 00C00 u
CNl-1 1OflO-'1 040W-4A C 000- 4-4 140Q0 1
<141r 44-
(00. 0 - ' 41W ( - L 14 1 41 (
N0- 3-4C
:01 04 410 41 (a :
to00
)100'-'to 0 C).
.0' M.0400
Q CC .J00a'0.CO M10.)0 r.10 Q $.I-r-041
0.0r 0
D.00'---C10.4-)-1: 0 :3 2 4)' 0 00l
01 r1~. (V
£14 0 040 .( 4 91411
00000'M
C) - ---- 041:OW: 00:0" 144

41024 0 . - 0 -)w0 1. 1 0 2 14 0 1 4
10 0 01 -1.
411 V
00 '1.:140 4)00'
41 >Q . C) 1 40- 4> 0 )4 14J( (D V011 U 10
0~~$ >0 11 041-
10'- c.V0'-40- 0 0H V- M
<1 _0. 00 n14>10.4I-I 0)0LE4:0 a 4.40C:E 0 3 C)44 C
41 14 :0~1V:01
4)10 0 410 C 4W0 0 1: 004)-0104 0a
>. 00. 09'
- > V 0040
VW-k40 140 0 F.0J 0 *-4 M M 0 W- H M M
r1t0-H 4 >0 W.0 411141 00C' W0
41404.:3 0
014to0m u t;H.-.41:04. 41 r>. - 01 01141011: .)r
03 01 10 0 0-40 Or 0. '0 c V 0 41t'a 0 w w 0 r 0 44 .4
E- *.tjA *0~ 1 141
c1 0).
'a0 00 0 '
mz .- ' 0' 0 i0004 4 1.
En H 00 5 -4000 410..m414
-n ur w4 41- C100'0. 4 to) W0 00 4 V E-00 0 41
D0 00 to 000 V2r
0441 1: 00 " 00: t
41000v
Q, 1 01 0. 0 41 0V 4144 4 ., U-1 00 r. r
0 0 0 4 R 0 V4140.0-- '40 ) a, 414
V V0 .- 41 V0 0r
<C10 W :1 W 4-0 4 V 03 0.0O 00 )4 d- to0 0 . 1'0QI '- - 4 )1 .
1:0 0) 41 411 4 414
04-41
0 V
~~C 444)W01
4- H'
M10:
1:4141:J
0E 0 0 WN>
.41"< $0 30 c 0-20a,"0U1
<Vc4)0 V..~
w::> U 0.212)C4
) "0 W
41 0>1r 0':414 40 : 0:00 ) )4 :3(c
. wv>
0 01: w .01$4 "1 0
r-11- -0 1::co
044 014444010 r-4.-'0004)2 I C1 0' 00'0' C 4 Q) <I0 17

01 r:0 (a0:-I.4 (0 - - 30>V 0 ..n ) >00 0 10 > En t
00 0 0 1'0 04 >044 > 4- H1 X1' I Q., " r=40
00. 4440 04041 -0 0110 W- 44314'4t .
41 01" r 01 r.4'014 14 04140 C1- 0 :>w 1 0 000 4) a)--014
.l1:0 I)W(
0U) =0 0 00 14-aV441 0 0 R .0 41 (0e0
00 0 4101:1400'100.
> 41 00. 19 00 : 0.>Q . 21X4)4 , -r
a, V)
1:-012 0 0 41 r c V 00 (4.0 -D1 00V14 00.WC1co440
00 0 0 ma-c4 1414
4)1 0 to1: C1
4 u'-0000t14411:0
'4)
04 040--I 0 0 000 $.1 10 a0 4 En0 0 1-- 0
a, 1
40
141:
0
0.0
010I(1
41 I 00'
414w0.
.10
L
'N 441 1 1340
0
4 'O
.
C r010
:LnP.01-4 0
~ QV4
M
0'z'
0.4r .- IO(A1:-> 1
00to4-4
0 0)
41 (4
040
0
-I
0)00
004-
>
0 0
I"W
-9.
~0 u 4)1 P
H00
0 .01 0
$00
-- 0.101 >014101 w 1:4100 0-00 -0 0 4 '0 r- 04 0 V
9 (z- ' 041
I c40 to1 (a>V -.41:14 0 ' :o'I WO14 '0 - 0 4J W
'(:0 '04t
x1 - 40.4 w 1000 04-H 0) V' 4>..M4C
Z4 441
)140
0'40
* 0441WO 00 4 41 14
=- (a000--to -4
1
40 0114
*4~0 $ 0
0 014V ' 41 I)~
u :0. 0 C 004) 0 0. 00fa.
00 3 10 w 1:1
q 1)2
1:1 3-u41 0 40 M C'- Cl4$1V4.r- ) 01 :301
13r
4 CID
14 C *) 1: 01 C:9400M 1-40 0H 10
004t0(a to *1- 0.5 NH0
0'( .v '4M' U M 0 (C Q W 0 4 004 0 0 k4) Ln0>4
41'4 00 -0.-I C0 r' C1.:14a)1 0 0~
4)1X100V00
UC 0
0.- 041:0 0
(D '0
M--i 4 00
o o 41)1 0V14a) .1 00
O VCZ
0 4 c14%- 0 r _ >CH 1 u: ~0. m014 l 00 0c >41 maw ~0.
0)4 oo( Ito >0 Vt 1 O I V . a 0ix 0 0.
14-00 . *0 44OZ 3 ~0
0 04>0 m 000, a
0- :0 0014%-4 0~ 14 & V
M 0 0 CO '( "O
X X(P44 V 44.1 t
"0 0 0 10' V00 -0 33 w & 01
11w4
C 141 m4.
L0 - 12vCI 0000 v0 m40 0 E .1 4 0.10 4-~ w 0140
0 '.~' u 41 -4 4j0-: r..0 w1-wZ- 0 1ak41.-I 0 M~
-qO0 r 'loxI0 44 m0 tr1c a 000 - 00 m14.-C ( 0 : gx .0r4411C0
v. 4 13-
0.4 '0N , 1 02 $ 0)10 N4r..414p4-
W 000 -40004)
A-(
Z 3 4 I4:
10 '40 0 4'4'I0 ; 0' v.44141.4
a,0lU 0 14:1 C90 M a,.
410 40
r > 0 4,-1 0 0 VA040 0 on .
0. W4 1444000' 0 0'0
J 4) 012>0411-0 OD0 14U )- wr a toaa4 i0 4 C -
140 ~&~
00 a1)4J04 Q& 4 9 0 4.-4 0 M
B-36
rn
.4
.CT300'
41 C44C
M r HQ
f 44)W40IW
(C
U41C V 40--E-C--
*C 0 )C
rI-00.401nVU000
A3
-. 43
u
-
U
(44rA
0 0 ~0
r 0>01 W
0
A4) C 0
0
40
0U0 0
4343
A W00W
00:>0
w. . A H,40 .C E0 130 V rCC-
-(a
43 ('340
01.-i w-41 U-> C-i C01
43143V043W0.43
14 >. C, UO'3. 4 w000
>-'4 a a )0 o- 0 40 to 04J V -U W 4 C. 0
04 4) ' -> 0 - 0 3 W0 to 04 $ c3"0 430 ,U0 ..-. >1 4 a04)-
*43
:301r040
Wi C 40 43 'a0
OC11CCA0
o
*-Il
001)0.lw
w0 14 0i 0
4 .
JON
I r-C
S~,
04-43 9 0
01 V 4-4-
aJ0>C410000.A&JU
W0I
W C00 - 0
r
0 0U (003- -V , 01 0C010 C30 C Ca 04- .003= 001U- a Caa
x Q It04 0 z 041 -to 04 k- r4.W0
0C Oz 0 0 1)--10U0 00 0 00-.$
o- n 0 W OU--i010 004 0
a4W0 > 1 -. Z W C OV VC *l0 143r0 43~ 0
U y-0) 0m* 10 00 0-I &.H4 w2. 4314>0--0-- 4C .4
*-i 0. I 0 4 4040 0 0. 400In3V ow014 -- 400 ' 0 f:r>.
v(l*J
o g
4U2CC V 0
I-
0
Q)0443.
1M
04j 4E*
0) -i
.44
?4C -,40. 1 OC:S.
0 44-30-.10
4e0U0 0-I .. 40.0
0 -. ~4Xto
.4 -(M C
4i-H H
CU0 0 0 -4
W0
(a4--1-
4 0) 4C
0 0, 0
V.- cc N
WX04.01'->
W
M
14 -C r
-r
04V3430 0.
4043o
(
N-1
0-.
z
.1 00
N4->.
.-
0 04
'D-4303
43010 *4m044-40 p0
4 =00'
430$40
:(>a~30
4
a2
0) 4100(D-
> IA-A4 004
0. 0.4 w00040
3t3
34 C3
114
00
4
Q.
go
43
:1
0-
0>.-0.
Q
-.- I
W..
C 1
0404
0H
C(44
0IM 'rUl 0404 '0.QU jC
01V M
04
0
U-40 a
43-44
0
0 V.3a
30
$. W.
04) 3)0.U 9U
.0 0
0 ow0 0
tp =u4
000
0
0
000
0 C CL
> 0 >.WGo0 0 tr4)14 4J U0 1430 0l4 V3~ 1.40010 ar0
M01 0-0 10004
,4
H .- i)O" M0
04r 0
-4)
Ua
01 03t-4 a I40
0(D'
0 ~ 34
00>.3 . 4)01--l
41 M 4
0
,0-
0-40>0
404.
C0-- 0 a
T)
'434w0rx
0.0)
w 0 0-
EC y 01 Q) 4)340 040 0 .0 0 r 4 a 0-.40 0 0vN 0 UC C- O 0 4 -i03
4)~
414 000 ( , , 34J- 0 04 C~3 4J 4j: ) C 0 4j M U0 0$
zo:0
w < zE-4
0EI rzv(y, 0(a
0CDC
VC 0 a M
>0.M-1 4 0 do0-0
a 0 k00
0'
0
44
OOZ
rz
4)3.
410 ~ 4) 0
C
A
43004.) .(a
0 0 404
00- 04)-
W NJP :3 4 4
= n TO-
W00. 4) 0) 4340 W 1 .1C 0 W -01r * - ->.M4-i- 04402)--l0> >
4) Z W C0 4 ' C a- 4 >. .,>'0 , 4 4>.C
W Cp 0 40E 0- 1c=
00 0 04 mU4 - V;4
04 C w V0 0-4 0, 4 0 4 04E0 *--4 a544j-0, ww
403--4 CO 310
4). 043C >,W4 0 4j U
030w -. C-I 01 0 101 043 0 0'04k0430WOx w34 -00 IOU)010v.. 010w 0 () 00>1U-m-.
0, 00 -043u 0 aOI 434r0)0 -i0 4C 010 =
CO -0 - Z:3:C S40 0-i.C C - fl 043
0.0 H-4 m IA a, w0 3 0 >10 4-C0 -4 0-- a1 to*010 0 tr'C40M040W4V4V.4J 4 044
M 04 U Cs3.)
- ) ) Q -4 b'0.-14 0 W 0--4 W. $00030
4--U 0 00 t 4.- > C-
a) =O 0 -.40- C0C3~
0 0A43 000-.I . 43-40$0 0
af~> 0 m01.44 0a03 43
0 >~ U 41 0.0 4 WU1' -4 0 30-i 0-404V-z 4-4.0 0 40 C3 A 0'
N4343 4
.01 C 00. C E '0.- 4, R.04MI (030 400-- 0 430 ->4 4 0 0 MC40
cZCr.M U) C0 -).- 001--4. -C4 3 v CCw 0 03 V U03VU>444 U304 E-(A
co4C.0-4~ U)AH 4CDI 43 C4 *.0 0a 40 o'04). 1 4 uCDo 0 04 403CU4 E0.-.-r CC 04 0 0t
co0 . - ' lC 4U-C01U 04) U310. ) 40 .0 4- Ar C4.- V - r CM -s04 - C C
NU4 4 n4) 0. 7 C0 M30000 0 0 0 0 0 -
M. 0 1 4j -- I0m3.w13-.3 43430Q3 C-i 0343
Z 000--r-0 0 Q 431 0 *H>C 1 0 1.< C C 4C0 -I.4X03-0 -)I T.U)0v4c 04) 3
cy 4 ' *w 04) 0w -00 (1) m 0 (4 a,
O 40 0 0 ' 4,r 14QO0- M5--CC00S0 0.Ow
co 0 0 'a. 030.I 410
OC .L
0 wC 0 00 04 w
U.0400'-
1 4) 003 0 >- 0 .4
.00n20 4 0 : 3 4) 1 a4 a( 1 aO010o , S 43 '0' 4.1.1, 0)
0 v > Q~ 0 4>1 r U43> Ai 3t C. > )C 0243
(D- 0 0w4."- -103
j )0) -HC C 0 , C4
0 0 ) 0 43 N - CO4 4 3-(04-I-I-I 0 O-
04-C -.4 0 4-.C U 4 --4 0-. 4 04
L4
4 0 0: V) w. V''--4 N(3 001
41.-
CO U 43 - 43 4 m04 r0V0.0 0t
40'0304U
C.(EWA z 0).0 w C -IC 040 *. 4 304 to0a) C0 30v -U C03,-4W u -
.401 0) IV 0--14w
.OZ-, 0 C
0 CU C$. 0 000 W
O> 43 VMC -10 W0 W W
00C 4 .C0 0.0- 4) n 0
-'- C n9 - 03-040CL
0-->00r 05Z 0 -0340 040HU0 4 4 -I 3 430
A 04 4 MZ-04000 0020'A C 0 sOW *0C)~4 40 C 01 4 CD
4 W W M. CC4..002 -H r- $10C SOC 443.0 MVr40 (.44 L
43043 0 0-.i41C0 W ;CUa-. t0C 0
0U -
U( Pi43- C2 40. 0 )020 310)4H 3 0 N <-U 0 M- 04>,C (U M0 0040 4.)
ta Q
-- 443- -I 43-o-l 01 4 iU 0 U40O(U 4 - 0040V - 'A4C--0'40 CD 4-. 0'j
1.0WD4 0 r w (aW 43 4 C m04r00-vI.C w M 4J0 M-J M U3 Q434d-W443CC--100'.10C0C -
0) N(n :--C 0 (D 44 - UC W .40 0 w C 1-Vl(4-30 - M'-X- rN-4- 0 , 0C W-r 0 V
40-I 43-M-V)-_W 43
u QDI 0 06i 3.-4j 2 - 'a V) M Vi40M10U--
M 41(4.
M 4U.1 U E ra r3- 1 0: 0 : r- 00004040
v U003. OCO 0 ni 4 0004)-
-- 0 0 3.0v0 04004.<.4-402 03$4U4000A00 - tr 4.) 94)Ur430a
000--H 3>0 1-I.)C 4.) to -A0'1W 43 0<D0 - 3--I-i 20 40 00.00a0.040s 30403w4a0M>w 2
z HV 4-.0003000W0~~ -40 0 2-1 l< w 1 0 0.433.IC0 $4 $ O-13 o-:3.=0ED
s-sOU.g)I
- 0 0U0
o 0 0 t-iU4043 0-04o A 0 >0.U0.4000-I4W 0 4J0C---V'
E- O HCi-w- IC 4 C 00X0 4 14t ao

000W4 C .E--
430 > wC 4 30uf
>. C- .0 4j 00-o 40>

l( 0 400.-
w a
04-0L' 4C0C00 - -5O>W M-.CI3-3 -.
0040 w 3-.C0r'r4J -1 X r-IC- 033- I
0V
0 C 0) - 40-. 4 4-j w--s000 S t A0)0)
40
-0 (a43 U0-..-s4C 0) r=04 .4 0 u 143 01
a, w' . -- V4.3r-- Wfl .0 C.0400--i--.

4'- 01 - 02040C_50Q" M4 04-4 1004 0.
0 C4 03- C )0 sO. - 0C-I04 0 . 0w
C - E caw C C 0.CIA 04- --4 -- 3-C V t 0MO 41
--- En"I
'1 4 .) 4 -C0) C 3 3 0 j 0-.- 0 303
14
3-(>. 03.143 F-. 4)E-'40U24W0V3V>. X M M -. C
0.
-"-.0430. ' ..- r00.0
V0-3 0--I00 01A40 "0.
.0~~ -4 0- V.CWo0Ua, V-0 0 3
> 043OD400 2w0
O o30&A40--Q43 WA'0434304w0Q0X
0-. m.-.0(03 C0 3U$4 0 00 0 30 0.) 43 4).Q
NOU C0-'N 4 3-- 0CM 43 0410W 0 0' 434_ 0 --
00 V- 0 GoZ430 0 0 4S. 000 0 C 404, a m40C4S.3m .0W

4040A - CO4 0 .. (D $4.-.-0.C 0. 01 0 0 43 > W E- 0)
(n x - -- ,0C R 001.- .) W
-30 U ta 4) 0.
0 U- M0-30.4'- 0V00 0,>3.> w " P0 003
0(0< 0C1 a -- m -0 .0, ot - 4 4)s-A V.013
.$.
40)4z 90-, 041 4 0 0 QH040 4J 4.C0-'P V 00
0.0.- -043V0 0' 0 12 V3w4 3C 0 U 030404W40
IL04 .uc C-l01. -M-0 A=
.0 r W0 4 0 '0-C > 0
- ).0I4> w..0 C V4V343444 C43- 343)0 43
>U
-- (N-0- m4 -'4- 40 0--40 =4.
>U..4(. 0 w0000V*.- 4
I000--- 04j -00 -43 01-I4. 0- 013C.I.
to03 -ClzA< 040 ~'. 0 .0) .l 0 f0010040CV
243w 0 u .
~~~~~I
-.- 00
"145
140 4
x4040
.. I.10 >4--
4
toU0C00014
U-I--
0'--h
0
-- 4) M 43-M
WJ0-iO 404301
7- ..
U '0V000-W'
f0..
U.443
C
0.-
43w
04 ~CUG N4I0100.0U
4031 S.034.100 0 10a
E-.4V CW40-3.0-- 43
l 01mix0-IC, 0 0-4 C : 140 .0 01 0
C.>0 -M 0 a 0 c 40~
0 43-C 0 V3 V. a30U4 V 00 0 43
F (0 -4400 C 9: 0 10C C- 40l2
W 0-4H3V 3..-
r. 0 . 0,u 0, - v4304 4 04 0- 4 A3M.iC 0
400I-I 040 C Q 403
9.64- C - 0
V)~qb 04 9 w J go V V r0.41 0M42 40-s
0 0
ZA4-if P3 - a40 43
030001A. 41. 0- 0 40 0.8.4 30
~ 0
9%444
-4 D . 0 V U40~0 43-.. 1 00 4 N 4J 4304O
N 03-
V) C 4 0 -- iV V 00W Vi 4 0 N 00W
0. H q 0340g. 0 a 0.4...
20 M0 3
Hce43 04
§ 004 3
:.40 041004Qm 11"
U
t- B-37
r* 02
IDr J 0) -M0 rW MM 144344-4.4M
U.N
4- 03.> 0 1 0 W
4J440 02 0 0
>1 r-.4. &4 23 00044'>
c 0 4 W 4 00 4
.0 "---0w 4-1 R~ t- 00440 to..
04-4Id 4)0 0-4 -a4-)4
o 44 t, o04 4 00 ) 4 4 - ' 02440
24) 440. r_40 04' M $4 0 444

-C < 0204
* N'- 30 44 44 4 4444 .0 4
4J 44.
0 c4.0
.0 -C3 0 w
g40 44.0 4- 4 M4
40 44 0 4M4
t 4-402-0-..0 3p 44000 to4 0404 o
U)
44~002-
440.
~
02
.4
42
4.4
43.
v02 m0r 0
0
in
r2~
a. >.10-
00444-
-A4m444444 r
>02 Ix 20 000 0 W2_40 44-044 a4 ri'00

0.444 0. -N 1 w4 a;2
4 . -4 10 01 41 0 44020444M
0r 0 44 a44 w 04
>3a.002 440 3 4
R 10.04.4044 0
4 0 V4j M4a
V)-
0 -0
3 440
440<~~
.0
4 m0:
~~ ~ In"o
4'
00.
404.44.
r4-
14.
u
4- 0 440o
'i
co444044> (a--04
4 M
4-m 444)
2Z -. 4.4 2 ->00
0 0 41 0 .4.14-
a.4
4)02 404
4 4- 0
a)
4.'
2- >. -
C440
.40
4)-4.<
0.0 MI
1:6
020'a4
$4.I0,
S4 444444
u
0 0- m42. 0 A4
"- a.
3 2 4440 4 V
4400 0 0
0 a) L.A 4 .=. n H-1I 4)> 1
V4 2- - 410
4 0 w2-.4 4 0 400 044
4
4-~ 023LV .. r 9 w 0 o w
r- () w m4
4444s 44 04 44.4
4 0M 4 . . <' 0V4 4 24 44 4 4J4

C .00
2 4j
ix42.a . -40 2 J
2.00 0- 4 eq o,400244
'-04> 0
W 4 0
3>n4 4Q 'a<
442 a44co0 o<102
cooo~.. t 024-4 O
:3 VX F *-00 .4'4O V-40
w0 a44 )00 z U, 040 w
oo. W, .-. M 2 - 0
V4 02m 0 . co
14-214- 0L $44-0.4 Z $4 W, 4...0-444 0 0 ( 02 W a 0 3
'a 44
a- - .C3Go 4 c- W4 144 4 -
0=.4W - 2C00 <mr- 4>-
0 0200 0 2 4 - 0 a)~4 0 - 4 . 4) - - -0 .
< . 04 * .4-
a0 11 r,4~- 4-40 320' .0<4 t02

C, a) M 0 w W.,4 v "4 >. 4J .00-'
W _00lr ,(n4
0JC.. -
z 0 0 ot.02 --..
4J3 -.
E-4 0 Z<0 co - 2 $4' 44.0
Zj -' 0 0 - 0..0-4 w02R4--r.--
r0 M 4 0.20
W4< 0 . 4 0 4 4 *,1 04 cc 0 4
m4En4W044
a2 0 >
443)4. (a a- 444MJ~ E4. 0 x 4) 440 H -'4
. -4 0 .<-0 4 e 0. 0 2 > 4 4j - - 0 4 0) 0.
E-20 2 02 E . > 4 02 -
r
-
0m -0 <<
-
~
,r 0 Uf
0 2 W
0<2
0 X0
~
' Z
~
41)
040-
4-----..-00(.E
- 0
-00
2 0.n 00
. 4.0r40 4-I
j 4M20-W0a00
00 4 < 2 20 - 4 .X0
.0w
r
4-2
E2
>.0 U .24.
4 )
0 W.
4444
V 0--4
4 0 4 0
)0404
$.4 4: E ' 40
w -1 4o 0.
0 a2 44 1 W W0 .0
VI
2M3 444M 40 44 4
--- 0 2 0 W O.- 4- M4
>,04 3 4 -WN 0>0 44 . 0 .o1 24 4)
.0 '-4 X
0 m.- 0 r c.02 2 . 0 to0244044

0 0 00 44 - 04 2 ) 4444
r40
0 02 .D
0 0 m a Z w--M
0o 44 44C4 4I
44 44 4)
4) r 44 .4 -4 4 4 z r r_
02
<020 -4 2 4 c440-'-
0 0440 0--
-41 .O1 4-0
---. 2000 m 440 _ Q 1zc 2c 0rA:
-44)0 2N. .40 4
w M a,0 003 44).J 04 0
w
0v.
4 .4
442 40 --
204 U.0 0 4 00w 0 444 0 442 40
cc> C 2 0 0 o-200l-
0 04-4140=9
=0- w 44 0 0.4 44O
t 44 4 >.0 0. E-0,M E- 414 E-1 U 0-ZI to0-4 v0>.-( 244
04.4 w4-4C; 0 )4) . 0140 0 2 4:N 4

M4 W-1440 - 44 la
2~~~~E 044 0 04.4 022 34V42 N0 044--0-4
a 0. 200
~- 4) 42 4
4 0
440
4)
0M
-(>
0Z
- . C.0 )
C444
44-
0-44 4
444
r4 10 4
044
-4-> 4 44 0 41- 0 V.- v--'. 0-.4-- 04 - 40 44

00 4 x2-0 44)j. < M. u.444
44 ' 40
0.0~~444- 4 ~ ~ 4) ~0400 4
02 000 41 04444.3
4 . 0 . 0 0 -0
4- u-2 24 a-44 4.4. 02.4 to 0 4) u >.4
0 0 4 - 04 4 4 4 04 0 4 -MQ4 N 4444440>. 4 444 4
4' 00
4) Z 2 ., a-4 0.0 u0 m- 4
-44 444>- 44 .C 0.
4-0 344(0- 4 2 W44 0 -444 20 0V4 .441 " E4 ) "-
02C22. 02)2V 4 r> .0 0. . 4 40 2 0n c 4440 a) 4 4
0.2 0244 u CD 0.N 444 M404 r4.>
(1.2 0 0.)
4404 44- 344 010 .4 . 4) to 4)4 34
44
40 3 .-C)4.'4 4 '- 4 w40- 04.-4443U0
<44 4 0 0 -0g 0 --0 003
4 04 444 -) r 4
Q)0 4 4.0 44 .0)0 . 02 <4'o442 4'4 0 . 4 3 4

(0024-3-3 l 0'4- ..-4 -- 0. 14'44
0s 4 004 34 4 4 ,- g . 4 4 4 4 44 0 0
00I44 4.-4-
s04404-4 . V40W.04 0 to Q
..1 :40)) m4)-4 U) 0 .002 0 44-'0 44 4 .
044I. 02 4443D44 ';4- 1 . V)34
-4 --.- 444-.
3<-) 0 .4. I. 40
4 0 w<. .4 44r M 4v4 4 1440 4
02 0 2 4 0, 44 M4- 02 -444 -4 0 3-. 0 4 a4 cc u

0<0 4Q 4 02 < 4 4 4> 000 0 , -4404M444. . 4 .. 44
-- 44 4 0. 4 4a 4 -4 4 40 ' -4 0 44. 0 - <4 4- 04 4 I .
OD04024I.
I J-
u
u0,U0-4
140.4 00 0
-M 0 04-444 44
fn44 0 V-01 4-04-04.-
0, 4-44 N<) a.~0jC W 4 ) DMxa0w1 4)

Z 4 1 0 0~Z. 4 44 .04 )0 - 4 M440to40-I W2~) o42
o44 4 - lI j0 M >.a44c 44 02 4 4 444 4 40 444
V444.
a.. V, 0 U 4 .4 -t 4 .%.0 0.4)3
U &Q1 X-
Mbj
0<0~ 0 ~~~~~~~-404444 r.0 -9 ~~Z 044.0-
004040.44 902
B-38
w
Q P >0(0 a a.u 0 .0v 04 w442 -. 0 CO Q .,a 0
4 mm =0 1 0.-4AON4)40 V.V 4 V v'0CC WC.0 A 4O

4 C o 0 m 4 O 60 4 . 4COO.000
a -4 A0i"4f0 *C .. E-, 43 r.-4 '0 0 04j
6 I*V( 41J
$
£3364M aC Di
4,
~0
V M 0
44-s
E4.-4
.0O.24ft
4'
0.
we 60060044w
C'a 0-6
5440>
O-4.csm0 0.0 00
0 r,46C
v
i'04 *l
.m4
um
0041 WOO60.
£4 0.44000
r
V
MC* W
-1 -0
4E-4U'4- . .4
0 . r-44
46C.44
0 -w ?
_4C
~ £40444
C 0. 4j V a U 0
14 tm0
0 0
00 0 0 0 04 06 0 . O4 :0 000 Li 0O
co .'O'0kA40.0
-. 4 0> -144-
44 0-460 -VaCO a o 6-no0 a 0 aV C .
44600C£4 -4-4-C 4 ' -1 0 0064A0 C0 £ 60 -a 1 - r0
41N0
r4
0-4 M ~-4 C6 0-4> 6
k 4
0140-
j1
- 0
-,I V,40.
-. %.4
0 0 a -4
C£r4k0
44 > 0 r.-C a 4.-0
C) -M o: -0 CO40 .66 j00

006
ow O0.ka 00 w t0 V 4wo 0 0
44 0 C-42w'C Co44-. 006 V -4r.
0 W-lA C 4404
C-0
* 0 4 E - ' 04)4 0 46-O 0 -A 6 It),' 1 ) >~.4 a to
10
a40 go04 64
(0 0 (06 -40
.-4
>04 1-to.C. -4v£40 06r.-4 M404.4
6.A
U 6440 - > -
to4 6-.4U-4
0400 0
~44.4.4(44 46
0C%401 V0 3
000 64
04.4.4(
440 0>0
lo
04.M6 *.44> v44 0-.6
4i 0. .1'0
(Ti 664(4-4
004 . 0 S. 0 044 I-M2 0 OO-40C.'60
-604 A
o C-A r
0
z1
1 0 zw
00
~U0fi.0"600 >44
4-4
44
0J' 0~4
'C6.4- ,1
6
M
046600400
006 k 0,= -.
4.10.. 0
4w4444 V 0 p~mx044 44£4 0' V 4 V V 0"4 -4 > 0 00A
w 0- Wo 0. 04. C 01404£404P-4 4 'V 0 1 -0.60>O6 44 1- .4
0. 40C .0 01 41 A, 6 00 00 o0
to Er 4
0oV0 .6 06 m
.1 r
0 146 C
06 44
4)Zm C -4 X 04w-1 4 0 .40
W Y)C 544 UNC .O 0in
w r' wV CO4 -. 4 0 0 w.
0
04" w 4C4
64 0 4
4 4 4444
404k 044Ca . .6
6 344 CD 10 C (fr 0 0 '0
>6. Q(00>U 0 -1 £4J 1 0 104 V- % C 1.:;-44 -0600 01 0 g40
0 tEO. -0 C %4O0 $ '0 4-.4 0(040 '0 - m24. k4-1
046 6 0 0 (n6. 0C( CO
>1-46£6 0.6 £o4 06A0 >..
40>-ON4
0.04 'A60 06 0444 w4040 g-566414o4t
to44
-66
-MCNCO
co0
440C
w . 04 0 O04 4'
&,4-
0 0 0.W0
V r.
is.4
-440
U .4
000w
~a Q do
0 W.4'£
17 A to I-
J; 4 4 4 0
w4- Q) C0 >i 4 r0
In046 0 m'-4-A 4 6-44404 ON oI C -AmUV 006 H0MN 4610C M .4
V~24
N r (a (0 -CD 0 £4 C
10 waw-
01000 1 .4 orn -0i40
$4 00
'00
-4MCX$
ir1 44
0444
N O
-446 >-
aO 44 (ND4 . 66 W ,4 LC0
4...0W
o Co r N0
>0 0-434.0.0-
0co$4V '4Jw
0
w
44 C040 4-400M
1, 06. 3.0 - 4J U£lW 19 43wr..460 00. 60 4460
Ln( 04
9-0- 44 6 6 >. 0 4 0 mm 4.i 0 41 4-1 -144
(a-- 6 AC' W_04 k 0440.> 6 WE-. 1.0 H4 =' 0066 A66 > 0'00
co0 m 44)0 60. 0 0 0-.500 co0U~ 0
0
-U
m0 s 0
$.4
0
4 0-0 to 3
C4 94
..A.) (6n , 0 Q*.-.06 to0 A 3 0 -61 4.. V 0 0
wf' VNLD , rg 640 0 4j. 0 COO * 0-4

IN E. W 0-6 4 -40 .r.V 4)-41 V4
Q)E. 0 C0 to t 4.) m4
0(434. 00.>W. H6 Z-£ 00 s0 a0 a 4 O
06446
r. 4. C00
06A w0 C< w44660 -l- 0.6 60 0 k
0.46 4J
EZO 4)O0' 0 ,1 4)0 0 23 >4 H 0£4 1
4) $4 0 '14000
zo .24 60- 0. -. 0 >. 61 6.0 00 -1 4 0 w4 0 0 0 w44
tr~£wU
004 m 0
w 044--50 w N 60
COO '0
'-O
w4 ~
.. £4
CO 0
C.C
01.0A0
0 m r.
£4061
-4~
-A 0
34 -C -.
U0u.4 6<a 0
z mS £C4 O 139 0' % 1a .-..- C4 0 >43 4044E 0 - ,0-.4-Z-40,
-444 o 0 $4 INCQ. 4 0 4>C00 CM0 'M On W
LO 0.$1 m "O'001-C
m . >
0 1 .. iC.4
(0n4F6404(a
wr
0 04 CC.4 rgn
0
,
'-.Mw
WW
z2Wm 04'0
,
Z g -. 4J
> ."
V 0 0-100.
0
-) 0 >. L C V -5
(n(0£ 4-- to W0>104 $ 41.34.0
(v u Z.( -mQ1 0W4
En 6) 4) 0 r- £444
4
u > 0 OO0 4:3 >0 C4 $4 .-.-
60 CC
I6t).'( 4) -. 4060>
;6 4
4 V. -4 1 1.13-44-OD
P09 >0M6 ,0'.- .1 o
S4co)w 0 0 ' -U 00 -0i *4
C> $4-4$l0 -4- 10 4
x4
*.. 0 XL 64 r0424
4) C4-4-0ma 440 "1 0.4
306 0 k
OON'010C x 1
44 lr 0 0_ 66 Q,--4
46to44 046W 4 1 0 0660. M1
-
0)C 0.4 U
044 0) n1 a0 L X
OD40-s 00Hrr 6 601-UVgEn>00 rCr I
.6 . 43 '06 0066 06 .6t4'
w 0 0--M
E- . c CO't
41 4.
00 L, 3OIX
£064 39w -,0044u0Z o
-i
c. 1.-400 V -)'0-4J- 0. 04 V4z-V.'
'44
CD 0O-00
W4X4444C k-1 W C 1t N>-
E 04 44=
R £4 0, $4 0 '4'0 1-1 r40'4w 4) M0 4
£4 0 04w41'C 0 0C CO01 640. 4 -4I
w
00v 0'0 6) 0 R 0 0 -. C.0
u.
- w0 w430' 4 r.Q. .
0 1 U4
00W440 600
: -4 0U
6 0 0 HN
Co 0.U C3 I644
44r -4'LaO
0 10w -D _A6 r0.
4'66M4-
44E 6 Z40'
-)6 -l 0 36
t)4
£440 a 00i.-'
0 44,02 4 0 044. k
£44400
441 C 6) h4 .0 = 4 4 OCOEl -4'- '0.10
.4 . a606>o 304404a4X0)44 $ 6-. co 0-ame4C
4 u= 5(. -1 $. WQ -w 460 0 -2 l 0I -4 0 - CO 0
O1 C -l0 £4 ) * L 0 IV -4 0 o
CM -4
0 440664
0 to4) 410 >w v a0 40 0 -1o06 0- CO
O0 (.C4 V( 44.60 00 a >. CC-1o'
5 4)
04 06 0 0 04-44406w -40 004j- -4 V 641 44C
$* -4Z 0 w 40 444V 0 V 4.; -f4. -A
).l a 0 01
COC6'4C A4 0 0 0-0.0 a, mi OCO k M 0 60444460
0 0 044.) ON0U6 m. 0 44 00 T r4 g.6-1.0 -44430-
044 £4 C--r'0 4 . w (a5C w 0 ON IC 0 44 A-.
4-4- C6 3 0-1%0-4HC
QO 440 4 C 4 V4 44
C1 -.4r
.0 m400' 0 $., 64.0
464 .
(6442 044 o63 lid £4. 4C(n 44 0W '0 W44k.
41ra.a
40 C 00 04 ,a r -1 o m A -0 a0
60.9z C-
4400 44C 46C4 0
.0 062 44
J og -4
W. U Z4 f44. 44 66 00
0
Mw' 04 ,4-1 ,4.1W
04M 4 r. H0V 1 4'0 C
00V0 vU
-0>4w.. 00-0~04 44 0(00 4
04mf> 0
10 44 0- 0 1 04-44- s.4 90CmW 6 6 a 40a 0
060. CO 45 0. 04 44 4400 a',0 g .6
W I 0< V64
0.C4 - m 0 304' 4 0.10 C C1-4 60- 4

(z00 C, CSI04k4 0-
H0 la m6 0(0 .4 4
Olt V9'
-0 04)rJ0 £4 r.
Ci t 644644V
co r 6 e n00
m-
r4(64
. £444
wWc .4 U'0 V6*. M0 )N 044446406
00r 0
-.44444V CO 0 1 Nr 4 4) 040 -ON -
cokC 0 640W 00.46 0 a .0 44o 0 V4CD 10 04 -l
, 4 1 VOU 40 4 0 4-l4 4e 04
164OD'0
U '0 Cn 0 04 400446080-
OD
w(t- 0"-0 44. Hi 4
'0w M V.00-i04
010.400 00 0 00C l. £4-O.C.4 64
004444
V0 0 0 6 £46 00 445
Z 44 4 6 a6 144 4
OZ,
'0
I0 04 0
0*
'4~~
0
~
6
404>
-1- a4 2£ - (.0. 0
00~04N ~C6 04 -1 41 66
a0. C 0
:, w<066 >.
46
0-2 00
-. 40 , '4£4 W 0409 M.o-2AC 1,
z,< x >v 0, X-0 a CI0 610-4401zc - 1 0-oN
V( 4M 64J 4
w6" C 0 1 6m C1
o0 00
U00- 04 w44 6414 IV4(
z.5 'A>.6-4o a
'A 0 C00 8 1 04 >=1 > 1
a.44~046*
442466F . k0 40 L. ~~ 04 4- 0. Q.4
06 so~V444(
C4444
>.>
03-
6 -4.4 a A.
. 145'E 00* 4 -10 t - 004 ga0l4
2 - 060
v F.6- 4 0 .(
0 vU "w 0m0
u0 M
CeCOOB Co4jv0 0 oe C
Ai0m 41'e A 0 V 0(
0 4j 0 :3 r- _414
B-39
6) 0 V r a 440 C 641
0 w A 0'a 0
0 1 -418*.
2or .J 1,0 1 ( N.04 bw 41 40
V44
-0 ". E ...
06,H4.4J0 '44 '04)N
N0 - 10
60 0 r
600 01om > 04w ti. C4)a) 1..
10 0404
U
k,
~
a 00
%4.0
4
10
c
0J
0
oC
0m
o
r.0>
. g-o
0'V
V 0.0 004 c r 60-.I 0 A**- W -0A.0 VW
V VV 0 1.4 IS 06 W
0.4) V) . N 4to.V 04)
C3 -00 04 40~~~~ 0W -E 0O0 &4C C VJ60I 644 04~ .0 -100
0
-(CA3
0
>.0 0C a,
* 1 QOCI O4 0.0V-4 0 4) &0400--4 0 mo 0~41 ..
S..E0.4 0'4040a . t4) VC0 0.140.4 64.r_
00 C.in 0
41). 00 04 0. CU0 00 -'in 4, m -
-- 0 0 Nto 40.0m4- 4 .0 toa U -AOE- 0 n $ 4)
It i G lg 000 0 80 -- 04 0CC w 0 ru 40 0 40 A .U
r00 -4J 0.W M60 t)'0I 41. 4 0 $ -k sM40o -0 C L4A J 4
4)0rco 0 0 11 34 C1-6 10tli. 6-44- w C0 4) go 0 V w 0
a n' A- 0 64EU > 0) 444.40
C.. 4j
.4)j4 0)'
0o 4)j ' WC-- 4 W 0040(
0041
.4 0 4 C60.'0- V 0-- "41)>.4.
CA 00'4-
*44 -. QO 4 -44.. 0 44 4
6 A . 4) 0 0 0 >.6
m 10 604wa) t m. u W4-- r4. 0 r
-
00 I 60' H u 04)000 Ac Co4 V
00 14 --. 64--> C-- M.0
C 4CCO0. 4-0 w 40~ r 00) -4
4 ) C).6 4 -)4 30)-4.OC- 4'
O E4)404~~ A 0 I 1.460 ' &0El.40 00
Q- vC-4 001.a 0'140 14-40
0 :3 V) A- 00 .0$4494
:S u0' (440m )4
.9 0J.))FA0 * W V.
w.
toH0 4) (n 0
*C44
04 % 4
4) W66
4J
CO
V
C4
0
-4)
0 . m' 441
-0
1$ 0 0 >*.4
4-%0 .r
4J4 to
t M0 o
00
0
-
10
n o; I
46
4 4)0C
140
?A6)
0
m0
0.4
H0
C
E
63, 0k-4
6Q E 24 u
C4 C 94'4
0
W
.4
01
a0 0
V4.-14
V66001
03.4 c
c
U
014m
wU r. ) 4 '00 M-U0 V UO-A 0CIwe%0 30 0 014) C6w663
.0 V0
01404 )wI V0Q 0 w 0 94 H-4.i4I
4U
Or--0
1. $)4.VC 144
a46 >-4 04 0 .4 -4I r.0 W0 0~ V 4 4C) 00 0 40 &-4-4 4
6)4 M.01 4 0)A4 4 444V0 a ,A4 0 0.' 0006)4c I.C r4I0to).c
40
614
.04 1w
4) a4)6
V 0.V 0
I =601. 0 0.04 *f0
U44o-.
-AOUC
IV4 80
)4.-4 .40 a
V
0
)04
C ]0 4 -4)
6)X%4 0
*-0CN
r_ 4 0 00 14 4 4) OW A*- W 0 0 4M' N :) 04- ('40
E6 1.044. 4 0'-.'W
CD4 r--4 OD.4 Z() 0 $ 4%4 -4 ON-' 6>1 04Vt-0 $4CA Z4'a,4l : 0 6 4 > .a
( l 4 -I 00 o04 60 00 l< U 4 10
00 0 60 0U 01 0 $ 4 0V M 40 $A 0_ 1 4 01
wQ = 4 1 .4 A.,U Wo . ~.
0. 1 0 0 0 -0U46. 0M 0.0*D*_o 0 4) 00c00
W:Su~~-
0 AW( a -0)00 m.C 044) C
$4E.
H W PU0O.E-0 0)W44ed0 0 4J O0 W0.-4 W )4w '0 -04 t" E.0 0 .. 40
0 u0 *C! 20a $44 1)W4 4061s.m4)E- 04JCH
C Z'U 10 010 3.4
0 *.4- c D. 4 0 ow 0 to .Ocl' 40.4k0 6 6 4
)4.00N to w4) 0 a 0 0 0 C -0110 0 t,041 1dC 0
40, .. 4 - 0-A U 0 414 k) 0)$4 -4 :3 0 n ) a m..4 060 E- >.N
z4) c6 w6.44)1 4 0.0
MA(a24 * ) :1 -4)$4 03 0 .-.
4
.< 01% U4-.46. /M0 4) 106&)40 AJ-4A4 cn *64) -.- 460 .4)4 Q)0
c z40 0000.0 -. 4 ')-14 100044 1 00m 0 r4- E.00 0 .c 60 w z-cO
0 Hf*. -4 0.C0)4 0 020 )Q>4
a, V0 60-I
0. (D 0 0 44to()0 0 A-00 J - 4 - 14 4 J. 1 -4a,6 04
40tako0- a00 2 g 0 V)40(3-10E4 ?00 N440
- z 00 0
22
A. j4n)04U 04 600 0 . *0 4q1VW0 - >.0 60)
.46.04
6 0.-i ;* 0 0 C00 641191 00 vt -040 0 -1) 60.
u 0.
V0 04 . - 4 0 ,$008 .O I *0a
c0 it 4 DE 4 -C2 10C.W -0.0
4mw M-A14 40 - W o -.
t 0 4 C. to O- Co .6140)>,( -Ar C 04 0 -44
640 :3 0-.4 M40$4)0V M 6.4 1)4 C4 -4 CO. .0 4
4 Z 4) 4J- 0-4' 6)- .4 4) 0 2,4 0 0 4)2-. 6 C - 4 -0 .06)4W> 440
M404. )r( 4
CC4)' .0 066 000W42 01A04t 04.c w4)0) 1 4.40
0E-4-C--9HCC0 E4 0o0 0 00.00 .0 0 04 a 0 CD 0 )H c cc 0 x 44 > AOOm'64
w661w 0 0-.4-.0 4-4 W4 0U 01Ca > 0 w61.0 -A4001w4)0low
I- c f C4 Z M41w 0. 400 4 'aCN w E.0 of n
u)-Z 10 0a, > 0
C ) A4)Z
- 6)4.W4)-IJ
W w4)Z - W -4 W4J r.
6 0 4 w 4)0 4)60 0 4J a d
0.4U 0.
41 w FA~E w) cNE -4w6Q.
60 ~~E
U146 C
CV
46 N Cli 040)0
04) 4 .C U0 4 ) A -. 4J M4W4)0
w0 '0) H 40- 0 4J) - - 0 0)4
41 >1 C 6'> 0 . C 0>
4 w0 0.c &
C 0' 00W -4 0. ,0 4-40& -00M -4 4140
6 0 V'La-V0 4)44) CV 'La rj44.40"4 4V
0~~ 6)0 0 .4640 OC00 r.4
S 0 00 4 04)000 u0 60 0 %Qw0)4 - 0
w4 c c m -W 4 0~ 6nc3 tom 10.4 >$4 0
V4) 60
W C6Ln 641 -40 v 440C .L .014 '0
0 U -. 1-0 04 Al 6o10
a-4.-04 0.0 C
C 6) O
6.4- 41 cc24) '4 60U OD
-4 4 - CO
E- W 0 410. 0 1(A0 01 0 C0 V406 V W .>0
00S6O0 44-.0.)4 C -0C60 00 = V41
6 0 64) D0 0 ?A aw-I .46O0C 0 - V)
> c 0 > _ V)6 14 0> --H -A4 '> 0 I
0 go 0 -40 c a.04V0g 0 w. Nw0 464-M = I
31 -4x A-.0 0 'S)C -4~--. X 4 0o. OH
k4
0 C
u
s , g 6
r'
r-60 '0 4)
-41 004
-0 -4
V)
C
V00
C C ;-O0
~ 44~0
0 -. 6- 6
-4 w4 0'M 1 0$4)64)$4- Q 041
04 0.40 C)14040H
H 0 -4C 4 * .4 )4 a 04) U 4 4 ,-4C c 4 -4 4)4. 0. rC C-M
a a V --.- 44 0 .4)0> 0 0 tP44..4 4~ U)M-400 .40
4) 0 W00 (40t - 41 -4 10a 06 CO ( U V04 04A c--4
0> 14 4' 06 A c00o0
A 6a4 4 4 '4
0-.-.-6 O 0j 04, vv460
6 w0 0 040 0 4)
u4 604 C-) 6--4 404H.4 .1 HO w.0
&4- Em4GOD, . 0 )-' 144) &
)40 .4)1 I .06r-Aa4
66 146 f 46i64 $40

0'a46c44.
0 I a0 6>.-O.a40
06 666% b. N
CD
%11 6.CA
4) 4)C4-4 0.14 14 4m440 6 w
60604
104C04
6--4c6 4 w 04 A -4 016 A C.1N4 $ 4a %
ODr0 in
C6)44 0
M
0604no0
OW64 04 U) -.0.40
c o6.6 00r.
in0
-. 0 6 0604o0
-46 -'-0
w144
0 o)1 0W
0 0
>4.-
-4 I I04-4 )6 -. 4-4j '0q aD w0 %4
V
D400 a- N n 1 1-440 WC 0 40 1 V w C,~4 0~ 046
60 044 0 40 0 4 e 604 a 1- 0 1
0)4 -H %4 044 04 ~
o0
-60 04' o o 006

0344 0 Ckt.
44 ~ 1 - 4-
60 's 0)%]C 0
0 .4- 14 k0 0
1.4 14. 1 ".3
14 141
B-40
>. 041
0( m.~*~~1414
004. 00
-0 IV
0V>
.04 00
>40
4C- 00
.4.
4) VNwW4 34. Na
0 0300 0 0 0. V (a4 &0
4A0
c %40 2 00C -. V
0 044D 004
0 0 41
c 0-- r_4
424
-44. W T.0 Iff
.0 4 0) W.-4 C0 441).
n ' 0 6 41-V -00
:3 4 '-
41 0 a,
0n 4) mm4>4 90'00
00) W U:1. C
-.
r,-- c4
043 01
00'ja.
C.0 W
-W0 ~ Y m -
0 0
i 0-5fl
00.>
aO O-> 04>M$9
*. NWco >1 C04(100V-H03
NW W>041 u . (c" -
* N 0 rW0 .- 4o 4'
0 44OO. 00
01 1 0 00004.0v10.0 $4( E1gg
M$ u-4 2.. U404
'4 c Q.0)1. 0) ta 0aE-4 WV V 0
W0 (o $ 4 0 4.140 V04l0. 0
CV:3Of.0 0 a0V..c0 40
*41. 4 v 44OW4
0 41 _ wUn1J 00nZ 4o- Coio4.4X-I
C0m CL 4
41.-4-0 000 j0 14
50 U; 4w .. o 'V...
.. 0 4 .k% 00 tP
.400
W n 4 a$4 (a 4jC oo 40
;-4
M44. rm. C low0 VO-400c0W
W
0-o A10.40 . V.~0 094 0 0 .00
to A,4JC 0 0
0 v x.
-- 000 0-.>tV.> 0 .0 4 C:
H1 04 04. 0 0 (0 . 0 >10 =aU0 0
-4 'a >. 4 - V0-4410 W 0 0
0-0.4J IV0)4-
I0 a)O 41 04 U0 . 'A
41 W~
0 -0 0000 0001
-., rzO10(1 M-V3 -.. 040 41440
C 0 o)4
0
04* Orv, - 0 W--HgO0 0,4
to*- 0 . 4)
I O4.-W0
o< V 0 0 J0-C0 0 4J W 0
4D4 M 0tomV
4:>.U 0 >
" ur.04
c It% a,00 W4:-W
0 0 0.4. :5 4
t t'.041 4) -440 0 000.
0c - .40 W0-4 V0 0 4
40.4U)0 r-
N 1 V0eAiW
4- 40 4 L
4C V m >0 04a4 010 V 0C) 0
*a_ 0.t
C 0 00.00'2
.o Z3 V 00P ?AI4 ) 4 0--(dco
03- to - . 41 0.41 0. '-0 0 41 w 00
*-I 0-0 3 o-
wi4 a0 rr 00 N
to4. 01 (2 00 0 3C. , 0~.-.-
9
.4: 01 mO-m -4> .-4 -- 0 4
00t
04
. 0Q o tn e0.
404 00to M U4
00. 041 to 0 . o0o4.0 g
0 0> GO0 $4
0c331~.V C:1 40
-0 ED 4- C co 0, 0 04:0 0* (a0.-O'
4 00130 0 44
0 W0 -4Q V M
0
003>
1 r.-4a,-M.1.4
W1
-~43003o
0'1C 00 C4J
0 4
0ix .m0 .
-0V1
(10
3
-4 31
C,<4 0 W
O ~ 1 :3
o
4
nV
,0
u
04
M40 ..
N
r c
>
4 ~
m z to03
=1.44
4 >4.04 '0
0 V-4.J'0- 1 4-4
r .-0----.-4
E C W U0040 N.:
. ~-4
0
(1
to
0
04 V
0- E fa0 0
4
-4
0.410
) 0 r o m00 m-4000
3) 0-. 4-0 4 i 0 04) 043WM.M4 to0.0I -- (a00 0) LV4 41
M
- N. f-m . VV.4 a-HZ 41
0? '4J W 00L.0c
c3
L"J m W 0 0000W0)W 0
040 in to 4 -)0Z MH 4v00-o
4 . 0> 41>00..J. * c $>.
t-'. W 40.1.4%0M0104
O403 0 Oulu
. 01 .1.-.
'Da m 0. 41-40V X43 a 4M
E fa 03-'.-4
tWo
= .)toM 0 a
10'M 04100M 0 -.
44. .0 Z ff 0 CC 430 41 44
0 0 4-. , U V
Wo( w coU 0 >
Z m.,
.
03W
430%4 0 Z11,QC
00
4Wa ~ -H 4 .034J
3t M440
. V100)4V0
400r. 0 0
--40.4
L , 0-4 0 404 0.
040>.01v 0 -Z ' 040
ta 0 0t-404 0
-0 W.-Q
000r. q.. H
~
V4 > o n.. 0 4.4)2W.. 0 .. 4 a414 0 ~4
04 0- 0 (n
m a a n-W. 00. Q c1 404 >. - W 0N0.M-MO.L4 0 0 -. 4.
00: W W0 40.0 m400.-
m 041 o >.3040 -4so
-C0W Q V 4) c Lo5
0m -4 t 'p3 0-'-- 41 W4
WC 00 $ 0 .- 0 WI=.44 0.4.4
00- 3tW00 .. 00C .4 0 VO 000 M> 0'4 0) W04
0(a .04OW10 eq
10) >1 0. 0-
--0.- > 00 0 00 4J 4 r 0)
.0e 0 0 04-). 4 Ai0 zO 0-
00u c 0 %r C40
C.
44.- 04
fu Wmo (a .0-00 U4.. 0.0
fu 4J 040
41 (1)4 04.4
W .44 0 0.0
M
L, r''4~ c 0 C . a ) W.. 00z
- 0
00.-- . u0 003.1. r.. 0 )0'oj r- 4 _4
0 A = 4 0W 4 VW. 0 4 0 --0 000 W.~0
m-040.. -0V 0, 41~0 VN'
fa '
000 0.m 03 E.0 W- 0.4L)M. 0. 0Wt 0' k
fA u40 '-- -4--40W000h<41 0-1
001-4 4 t04 0 4J 4J 4440M 0
c.....40
0 6 .40040 04 44410o0.0--) 0C 41 00V04 W 04m
0 0-0UN10 000-V 0 41. 00 vi a 00W. 440 = a.
0.4
_..
r'
W.j 04 >0. V0 0. 0 00.
00t 0X
41
~ 01 a0U.
1o -o0Ul
0 - V 444W10. FaV041-V 0 0
4--
W0 0 V
40 04
E4404. fn
CW .. m 0 Nay 41> -4= 0W
C.C.:4 Q40 W 54
4 W.40 0~ Q, At00 00064 00 00 ..00U0 00
4)T040
k00. 0 c1 A 04W V0044
4:0
W oV4) E004EnU 3. ( 0 1 CL 0 0 0
0VC ?-0o
0 1 0 ;.
0 M 0 x0
4'0 %.44 0..00.44.
90 mI3Z4
m XJ.4 04 >1to
4. 0
tP 10U r0O -:1 0400 010W 0.4 4.4100a
00 -4-- a ''- 00>40 4) - 041 0W 0 010 (4041z
loc1. 0. .. .D -to 40 *&0.
-4 0 coa0441W1.
ON a>1 0 4? 0 c404 c
c0 a .ev0 =V4 01a - C4. 44C0Wm
k0.) 004. 0
ma4 r.0' 00 .0 -> 04W
4.) 4
040 440 4J 0 4.4 01 W 44
CO:4 4 ... 41
>WV.-.-2-
1%.-4. --c44 O~ .4Pr 4, 4j.
AC120 a4 V4. 1 4. 410-. x 0 000
04(0 f x 4 .4 ON
0 04 41 0 84
a0 l a.4144 14
oo&
0
N
4
C0-4 04 M
~
% 0.4.4 .)& 06. -4 C -4
- 01 1
r,00 4 C U 4Aj A i
N0 44 xa00 c
14 4 00m 60 MW4MLN WQ
0 4 1. 0 .. - v 0 AA 0
113 'A a' 01
z ~ - 0-
j)1
0 1 . - f=
a01 0d 0.0 4
-4 & > 1 11 1 C 1 4
W 4
CD44 (q wm
0. 4ff o0
- ,, 9>
4.p AJ
0.0c4-4.-'W fo
U.4 0 1 -4 C 40
B-41I
~0 301A0t
jZ
0 0-W W
VCpx
0
C
00to
A
' 0
4 0
. 0 'l43.
04CI 4.)A0
41 Z
0 4- N
C4j 9W4
0 v 40C
Cfl
0e
A
EN - -4j - 0 0.0.0OV r44*E0M

000wS
I
IS
M .. to0 - 4 -o U 0-.r -A -'.~a .
a,' fa Q wCw4 43
A H A 4j l00 0 H &%4.3.0 0 0 '4
w0 V44 -- 0 0 0O)X 0 C >.4.-C '4j -1 0 C.I 446 1
r N1 .NLn o.. 4.V C -4 00 0 *In 00r. 00 U01"
4j M uN > r O-0 0 .'. 20 4.0- .. -0 w > - a a4o > 0
N 0 -0. L..0-- r'. W 0 MO .43 - $ 4 0 -4 0. 0-4 a 0 a--,a
0 oi.o;
4 I .. 4 0. c A =a 1-A.oa m3 .o0 M3 4 jW0 o O q3o:
Q m0 toaN ) 0. 4 tou ur.u
w w040 n0 a xOO C CusC Aj
c 3t W 4 w.8 m A 00 u 04.w0 3'nr. 0 04) La 00
x - oC Af rz *40-
w F.0 > n0 00C 1, 14 0 17 -1
o o 4-04
E C4. 443.0) wOamO. c E-40 0 -'4CM
a 00 0 )0 0 c4j AC > ! m mo43 4 0 0$.4%44. 0
0
M
4
4.
0 tOCOO.-C
N04j
)w 0
m 4.4
.
Q
(
)>
w00 r
in0
4L0.4.3
0 4300w.0 3. 0 C cc a0
. A 1
C tp0.4 00Q o -
c w 040
a4 . 0lu4 044-4O OX
00'I
m 4.0
aM 4j0N-v w C0'3.. 0( 0Cw0 I A 0- 0 1CCU3; v 0
Tfic 4J &;Az
430r000M 0 x .-. 4 0 0Q0 00 000004
WA
Q
r v 0 r in-- vC0 0 23 >.'34
(0 >->0
O0. 1W a0all 4.4j 04. g C0 :X
0'344
00 ON 4 N 4.~.40 C4 (.. 00 COO > a
000040 0
0 E'0.wfl01.C -4- 0 a W0x* C O.0
to 4.' O
4 0Ud o N
*I'-
' >.0
'4C4 00 40'
w. 30 e 0
0. 0 0 Nz .1
3- 43 -c9 <0 3> 0 0 C .0>I 0 4I w r.4 r.- Wo a ONr A $ 004
ul 0 $0-4Q) .. 0C4 OC 6000Mm4T.0 M.. 4 4In 0 I 50.0 0 0
4.3
Q; a)440 a,3-
ECn' C 0rc0 45.a4.w. 04 (na, m o0 0 O.1 m 0
4jwz 0 N O > 0to0" a'0 0.0w'0 0 0 .4 00 0 w4304 - 0 0
0 0 0 , L r= rf 0- C O 0O m u - 4. a -x 4 v k 0C0 0 4 4 . 0Oi
0 4 w.A0 400
'
3t 4j 1 00 40
w. c3 -w . O4cO00w 0.-'>
0.4 000 l
0 0 A w 0 0 % 00 0 - w0 >.0304434w
0 0t
0 c1 waaZ"a- 0 04'443
w04 0-'4 C l CM.4
C-.
c0 tN >w 0.44 -'o0
co.OON' *- 0C (a0 >CO -04.04 04.j* a34
0004 C N
'-w'0Q 0 z
-0' L 1
00M>'a04 00C-
m C-4Z s z w iM
0 j 0O w
0'-4 -aC-u
a.,- m o - z 3 &.m ( .Vu.0= 43N3 3t 40 4.E :1 0.M- CZ 4w >>0dC w -- '--4
4)-wz.- w.''n z0-*0 u) w-.-.'N 4440P) .40 wx 0) AC0 04.0W-V .4
-0--D0, 00 C-f0.Cf -04 V 0- 00 C -40- 30 0'4
C'C00000'N ~- 2io '
>
0.
.'- .. o40.-'-
0f03. 0 0 0
cm r.C
09
4.4.44
0 N
OC
.04C
'0ll 0
00 - O 0 00' 4J . -0..4C*0 00->0 w3 w.4C-O4C
a
-4U0r
~40
N
o'40()
Co
-.. 0
mN
.C v
a
-N
C ws0
00
-- 104
a4.4 ~ 0 0.
a4.)44-wO
0oc
cC z .. DI-C4j
O w
M.-U
% D
0
3
w
04 2 >..0.'40
E- 04J0 a
0~0-'.-. - I O W 0 4 ..0') 0 4.0 0 C0'
O-. u01-C3>.4 Z
l 0C( -4QC 0 a-t 0 MO
.oN. -.-- 0C0 W-w 0-40400N
a
w.'.0 w m C AV0W r.0 4
0~ .'- ~ w oj 0 00-NV
N-C04.w. C u
tP
C04
to M
ME 4 U0.
0
9X04 000.4
V~ m0.0.--
a,<-~0 .4 0 0 w---
a, .00 >..C4
0 E- . 0 Z. '00--f 03- 00C
(0 >0 0R 0 E 10C00 *- a w V00'0.4
NO" X C3_N 4) G)4_ )ON 0 c .034 0 CI A u) w . -UCOr. MO C (C 0.
E4 V -v-~0. 0 m u) v . 0v-. r( 4 j0N3 ( C4) 430-- to. 04 0Q0 4 0 0
0Z3 2C0 4N > - '0,00- -'n.t .0 3z"N~ G 2OZ 01 00>.. -4w . 0 CO l0 0 u3
0 ' .0 4 ( 0a 0 00 . -. 4 .. N(a 0 > '4- f 4 0W 0 0m 0Q 3 0 0. M OC
430 u4 0> a)(00 03 = j3W 0a0( 0 > 0 0 .C0 00
NO <0 r> 0 N>00 4--C 3
W0 W 1%w 0- O 00%C4Jol 0 0 0 .I
U0 X0C Q) - L, W0N-' C- M.-W 4J4.~4' >0 000!000,j w b) 4 0 >0 0000
<.01 0>.- a 0 m-.-.0 0 *.. w4 No .1 00'>Ao. 0 0.,04 0 174j -w00
- '--O 0.0- z 0>o- -'.44 &n.000-- a :104 3 '-o -4.- r CM ErC0-. 1 >. 3 4
NQ4 o , Z. w-
ICN0430'4.l.4 -4V0 -
A0- WQ W.4.
ia 4 N 4 - N' 3 0) p> f
Ea W-4 0' 430..W
4)O (
00a -4- $ 0. W 0 0 0 4*3. 0 3rr.) L
NN x '0- 3 I0 04. .12.0 U--.J Cj 0 0000

a, ~-
C0 0
.4.44-
1
4.40
EW
000
z u)W4-X
3
VC
L.02-
ON-
w to4-4
0
.. )0M.L.0
004
C -- 4C
430
00
40c
go4u
W 0
a) . M
N 4.L 0430 T . 'j <4
>.10, 0 4 m02 :j.3N - 00 040 00 00w0wv
N) Q 0.4. 04 m 0X
0..w.CN a u TC4-30M040 0 014Q
.4..'Nw 04 >N0 O4) 0 004. aNOWo

3-300-. 4j N w 0.00 4 E-C4E-0 U0 0 C OOV
44d.-00 4 00N 0. 00C 00 fu =,a
0044' 0. '040 0 0 0 w 0w43Cw4.1
CO a) m- r>0w404 lo V34 I
-C -'4 -40 flC- O' c'00 -
0N4. 0 04. --- 3..4-' aC -, = 3-a 4j. 4
4)V 4' 0 -4j 2
w0to (n z
43 alN4
OD 'a ) C4wa3 41 40 0C
mN w.. 04 >0 .4.
OD.
-I00 C (a- E 0 00 .40 a3 w4.0-.a..a
4 C )
C04400 43. 9OwCON3t 0 a aCXI.4 000
W0-C40 CfN 4) O.-. 0) 4-w0.-
?A4. w a, 0-w
0 , 4j w 0 w w-03.
w300 ON 'lu 00>L 4.C
0 'Cv
004 00 c -- 0 0 400 < .C fn.a 4

00.43 00 >.-- 43 003 c0 w
N-4 w) 0 w .E0- v) M Q 00
-0 faOV 4.0. E.n- 0C M CA 41A
U 0.0 000w
"i =. '4 4) E a 0-4 0444 w0ja
ON 00 0 l r = 4
<
043 .0-0 u)C0 M (
0.Z4. 0 &
0r-- o- 0vC
r-N0- 4 00)- u >0 ) O'C. > 044

43 U 41.-
u00 0 t.14 u0. 3
-'40
'-00- W v) 4J
4-
0. 0 00
4
4.4 ~0 V4OCw1
m0. VMC
cO
1
L W4 0"' L, l4)
-'C 0 4 o0 4'C 20 r 8 43> 4
'v a) -- NO 0 00 4 0 -W
C W43 .O( 0 0 4.
0
43
to4--- 4j'C Cwm4. > 034. 0 r4.C40 <0C 4j1 .00
44 .C 4 00o3--au
. 0 430 a lwo, .0
-4 0 M 01 0 &U, 0a
000 C000<0
04- 44. Oj w0
0~~' - I 00 V 00 $1 Cw
0 3.- to v) 4 w-3 W0'-0 04.,u or V l z M?
,a
W 3 c -.
cCC
0 -C
0' . 000. 04.c
C14303a 0 4.
m
.
r
ul
00.0b
4j
0 1 ~ "4
w
C.003.
0 0 v40
4
4. c0 4.4. 01 0o 000
-u >.Vz0'4
L43 *Ca1I200.
0 0 04004 j u ,I r r>
1 w luICOn l4CVt 3. 4. 0W ''- 4 W aC)
S 4 w0 Z0a 1 l 0 vu 6 rwtaG60> 4.
0-'M U 44.4. z. w4 j 4)0.3.C
4 0-4.
D4 . .0
--- 8 . C u
>0a00
m- a
00c
'D
w 4) .0 UF V2r
A07'!-S
.0 o0
0 1
0')E
vV- A0
XJ.-0
jc*( a w
000-'4 r 40 C00 x3 s-
,a 4~ a 44C-U- 4 f A0 JUal0$ .% ,vC 4430

9 r, , . a0 4 4) 4
3Z43a 4. 4 .4 C IOVIC
L.0 --- w
C30Q0-L C m
r.43-
v
0
-4 0
0 1'-3
v A
-17'
~ v
w
4
0 >
'- 0>. 4.3q0-0,34-. M-Iw04 En- c'
B-42
- 0443 4- 0.4J
044j ~ in 0 0
0 .0
4) 9 U~** . 4 4104 02 4. 0 4 .-
c 44
00.00~.0
CW 0 0N.- as' 44
m E-0 0 W.-4k 0 -'-4444V40'0034j
0'44 U)- 0 000A - 04 cU 4 0 *'04C X..
ooro0, .4C 0 VV V4 V V4.4-

~ n -Iz 0~
Aj
o
CCc
o o ~ a
$4.1 4Z0
- r04--4-.0
000 0%400 00. r.- W
a m.
.. in.4 . -0000 0 04 M4
0 .w " 0 44 k440 4.. H4V.'100 -A 0.--
to "3 Ol00
04.4 %D0 04In O O 4 04
00~~~~ 04 - *- .0 00 0 ~ ~ .4i
>40 00.
*-. IQ .0' 0 0 4- 04 0 k 04 00 k - .0 p 0 0.-
224.
H 1.2.4 0 C tr 0 0 3.4 04.4 -
0-O . 40 0V 0 .4 q -. 4 4C44V V40 0 aloe 0
-4.>0 '~~'0
0 1 OD 0)
00 0 V0 C34440
a4-a 0 r,.oJ.
,--.12 9.40to.0
12
I
r 0)4.1
00
40100
0004 Z -A CO'C
-V0 t~~- I-..~ a c ous. 0 -4o
> t 004- 04JA 4A
0 *r4 wo.000 a0. VA V. 0

-4.4- 0
.. 0- C. 30
. 000. c000r
.0.-0
*-.. V 4j C'a flU2 0Hm C. 1 HU$
11 44-- O 4 0-. 4 0 C0410 0 fn*- 4J- 0 0 0 4 0
CO 0'0 0 *-4 00 5100
I0 wo,
COO-. Cw- W 4 0'4 a04
0
- 444V4 N
0-S.4 H 0 40 N ; 4 . r.'0
0 .20
.4.0>4 0 0 040 .>4J "4CO0
00 M1H4 H v.f -C' -WJ4 00 44 .. 0,1 42 -. 0 0 ~ 00$

0. 0. 0w 00 00 c 0 0-C04Cf O 0.- 00
- - 1.a hc W z 0 W00>. ~00>0 -4t)4
U 4
C4-
w-04
-
C ,f
o
0 0 0 n
4 C0 4 . U W'0
44~0'U w'' 443-A 4 > .4.-'

4---L. 4J Q LM0 $4W0.00040 C C 0.0.-02.4
E0 0--0 00 CIO
=N.- to t444 -- N.4- 1 104 0 ' c
0W0A C
'O uIx0 0
CO/ E 4440to41040 0- 10 V3 0A0 0 W M4
z mm o 00.1 000
00w .. 4-4J H 00
IE441HVCCm4 D 0 (-go4
a444.V -.
a, , , 0 lp
, Z0, C 0 .CH
4 0 L0 w 0 4
c 09 '0 -0 u00-404. Om.4 0 m-0 N 0C 04..44-
wC '.f 040. 0-4
9.0 r-C '0 0~t 4C '. 0
Cr,
E-L,4
00
) 40'CL4E. M
*l
w
0 -
a -
< 4
m0C
t" 0. E-
co0 0 k 440-
4
900 90-4W'
0
'0 0 V30- '0 -HCw - E-0 . 0 C4) $ > 0

0 --- 040to
C t-. C0 0 ~C40 CO
COW 000.
0
'-
O
a- C
-00.
0 4 0.4vC
00400
)
it' 14
0.4 .. 44 4-0 to0 0--4 W 0 4 $C 100
04. 0 )W 4
0a
0" ~ 0C0.
C
0.0
0-0
Co4J 0W0 . .~ < C 4
0044' M0
CO'0 ' C.C
-
00
0'jI 0-4WL3. > CV .40 40 - 0 Z 3 0 S-. 0 4>4V - .aM

C
06 0oi, -C0 . 0CW W .0. 0.2i0. &N 4 C 0 C t
co
U00'0 000-40
%4 z 440w0020.00.4.W .3 r..*-4CV z w r
4.40. 4) >0.
44 00 0.-
4)OM 0 CL< W - w0t000 M.4-44 to40
00 n. w C.O 04t to2 0' w' 'O u 00c'0 W r w CO a-,C.
u wxL> 1 0000 4C).Cv2V M 044j.0 1C'02 0C 0
<z m 0, >U) Wf O C w4 04) In14 WW0-, >.W4
M0 4400 40 rl-6R
.
Nir 0 H, Z 0 c0 w N 0 10440->14 X
I-ZW 4 404Z a' >24.4 Nz -40 0-4COI -.
l 4. 4P$0-.' i o- ,$ r r 4r M)
0. 0O . 0 . 004Q440 0.' 0o0c402 E 0. v 0
4-Oi 0. Q> 4c0MC C1 0 440 w V 00W c Ho o x(
0 0 u0 4 3 00 41 -. 4 E E U.400
4.Z 000.)4-.
0 Ai-2 3- ,0 0% 40a 4o0 0 H4 1..- 00z0

0-Z CwJ--
1 af .04.=4(I .. c c.. A.)tC0"1- 44
- 41 0N 01 w.204 4
00 -J 04 0 CO 0 0 . 0 0 ' 4 4
0C1104 01.4E-4 4 0
'4 .1- 4 0
00 0 4 N 0-~ 0 44 00 4 0 N 40-V
~ 4.
0 -0
00
v
;U
0.0
a o w~- C
00.4 0 Q
0
6WV
.- 4 .c i
1C.
0 a>r
0
0
to4,0 U . 1.2444'g 0 C 044 0
44 0am
4.I0_'0
ow "a20'
-
.
-.
41 w
0,00
4O40
m 0c :4
00K
~ . V C0.-0.44 Id044.
0D 104 0-
c ,-44 CC to
4c M1. 9 04 v4-.2 "4 0.-0 Z 0C.
r 004 -.
*-.4 CO4 >~~ 00 4.CC)~0-4 44 NCe0 .
u E00 O 00-. CO 0'. 0 0 0,2 .
C 000
0 0 -IOI 00.4
0o
44.44 0 1 ~4 M 0 0
410 ~0 V4
l
0-s 0-04) 4. . 404 0 4 0.
00
0 4 04 004 0
44
fu.-
0
.~40
: W
1~a-C0
0200
0. r4;464U . -. j0 1.2 0 or .
4
010 0-.0 CC yU 3 . n~~8
000 -IC rC
44440004n0$
-' 0 AJ 0-4 >1 - - C2I

4 B-43
4Z OOCrW .C4oWM -W ty r 04Zo . 11s_ 0 1

ID
0 ( W)0Q -0'VOO0 U.-- 0
.. 04'Wu c 4C0 OW, L
co 0.-q W . Q4'E--4 A4 04 . E UM.
0W -U o 404
CUW 0 COU
.00 W- -C OW ) (U (W .0'(Z
A4- x r >I'D (0
-- 4c
ic
(
.0
(U
m0IW4iWW '-0
A.go' 4)
000
~~ 0
4--
toV a
~ ~ ~ (0
0
~ .0.04
CC
VU V4
~
4
-0% 0
$ 4W04
.- X 0-
C-V4 04
0
WU4444
r.
C V *' 0 to ) to r 0. 4004a 0 r
Cyl . 0W4'W
a o-4U W v
40
(W > 0-0 0244~0 > .4 4 > 04 'C -4-W(
4' .4( 4 C to40. -4 4 U C'40 N~0.04 ~

0~C 0' *-O (U0- 0 W 000-,-4 ) 04 c C'1 W
0 'u,.WM_4
C 0 o V V &l. -'4C o4 OUC r.%)O-4X-.a to4 M,0 ty4(UC_
.c0 (U31 t -4 r~ 0 V A
4440.C. rW'C00
(U34NC W04' .44 41 WU 0 W
4'- 46
0
WU C--M 90 0 WV 0 0 4 0 WV '-0
*I(U
0-4
0W
( 0-IC 0 W' 0- c
0W-. A (0( 0 4
U4. FW OV '- m4 - -~U 4W 0 .44(U.4
0(.C 0 0)( I0CWC toW.-.
(U440 ~ ~ 0~ 3 -4U.4(4( 4W c(00--4 0.0( U0'-4'-4'

C C04 o v. 1W'2 0'N 0r0C-44 4W4(4(04W
44W a' >-440 .- 404a4-4 0 0 > .V W4 H4 4003 (UAV a (Ut 440r M.X
(U. 0 V W U (UV
WVC.
(U> *W
X .0
m 0A (UC
4 )
0 M." 0
V0f~
4) 0> 0U k--0-(4
44 a o
0WV4a
m
r.0
W0
V
c
r
W40
C(
-40M
0'4M
0 (U
a4
04 (W(
r. u? V
w 0(U 4 (W 0
0 W (0 0V0 - A
g ;10U 0 -4.0
a C *W D
O'W4 4 0 C OU( a W44 QC.V WO V ' OW1
r.-C V 04A WU.40C.
0.Er.CW~4l w
o4 4) 41V44UWV
04-~W '4 0- V 0 4(0. Z4 0 (0 44000 >-..--

(0 -. CC(U ( 0 (
0to . > W
c C444.-4~0
,00(C W0V(440'0.
9 04 V 0' (U 0(W--
(0.4 vC' 4~
*0 041< N
0
0 t 3>0)U _W4 0 C V 4, WU c40 Ucm U) 0-4--- 0 , >.(U-4)
?A3-O0'
w.4 w4En a, N0.W$4 r ) 0 C . 0C 4 (U W o .- 0 $C
(U - (U CU) 4V-0-
V Qwo 0 4 00 (U(3 o V- ( O 300 0o a
a)01 *.- 0-r r0 >10 . >4 k0(Ja' Ho 1U *-0 0 .4-.a (W(U--a.ca >
010 c4 0%(-4W
0-4 to c W to Q..444' QNW (p- x *0 W -40'M - 0.l-
W 00 mV
Z4W - W 00 M - C - 40W k .4.0W oZ 4 U 4 - W 11 IA W 4'U.4444 --4o
C4>44 *4 C
0Z.kU0. C W40 C 0 (o C 0) 0 W-4(. Z WO. U(W -- ICC
a.(WU 0 N0 Z V.04040 kW440 0. CWCO :b oNCOW . c
-4 W'. c-4 rVr. W4 0.
(UZE-.0-4
0444 >I t y (A W 0. 4 W1
M(-I En r.>W MuO ,
(U.4( C0> .XI .0 C W4- 4 '. 0 3.-..
D4( V 4M'4' S4WQ
Nu~ w fl EUW V 00120
*.4VOW U $1(n N.44/ .) *-4 C(U4'A4C WQ w
r-c Q D
0,v
0x 00CEUIC~ 40C(
tnC>- OU C"
aM > 4-.n 0>.W 0 w0:
V 04( )
3414.0U''U( HC0(>(U-. N44 MCO>U- 4 'A U 4 4)4,c 4U
1% r.0 W'U.040J W40-C.~4 >U
443W0-402
$-4U
0)44 c Da4 $ Q )E W000 .C *> VCOC 44 l4 U
W-CU404
0
z .. =- 0 4(U-4 0 0 kOUOVCO4 . >00 a, r, uCCCO E 0-AXAm > W0(Cr
ar 0 1 0 0 0 444a,.C4W.,4xU) (U 00, a 0*4%0r= O0OW-40Uc
CD0t - 0UcUM 4'00 11 W oV CD 0 o- 04 U0- *431 cco mu0 . 0C
0 Av
CM >.0 0 t 0a0 Qv0m0 O) 000W
(U4'0u400W -U
> (UC . *VCW . r 4a, En ' 0 0 A .. 40 W> .
0-4 (1
E-
40
(U. -1
0 .0 mC 0- 0 X0
'.M 04
-W>N E- 4W(a -- V (UW09
( 44-
'04''N r..
A004E4C (U-WV4r 00 44)X ww fl(44 10 ZOC4(U-V
rA' C0.0> Q WW. n g- C r0 WUC V0 'A 4 'A 'M 140
4V40 -WC( oCVN 4V4 -03 AW00 -=0
z r-<w ) .14 - 0w 00z - - 0)r W 44 V40 01-Wo4 a
4 0-
r
4)0.
W
V *--4
>3Q.4
4440
o 1 (U ,
WV
w
v0x(
c 14 00 :3
>0..
.. 4 04402 >V
* 40
0 C*
Oa.r 0
C 0 40
(U4(UC44
-4 -
cC.W
-r a
0 0 4 -0.
W 3 -(U 4 4 R1 W4 X )aW- 0tP r.1 0C (UC 414 044 V :
UU SO2C (U 114 0' OO C( -
W. 00-)- W04 . arW00 .0 0 z c x-cV40Na)44
0 A> CW40 > V W 0 W0 p 4 D W-4 C.0Z 'S > 0 V0 W_0 =3
C)-,DV0.-4'- 0'W0 0 9O W0.04 V i , - .5 0) 1 4 -4VV
a --4 (U 1z
0 M 40 4.C W 0& UV)o W 0 '-40 0440 VI C 4) 1
0 ,W 0E4 W4 n . 2A 00O WC4 >
r.0- 4-ua 00 U3 2.4
.- WCCW.COOCwO 1 :30
-4U) 9,0 W'
' 3 o0 C 0-.- .W
-00( ,W *.4 Q
45a'Cn4 W0$
0.4
0C C OIc 0
0OoO,0 . C4 0 0 0 0 _ 04M u
C
w
.:j0
W 4-44
Hr
0044*
cc
U)F C4 -
0
$
1
r ~-0
WC
m9
0 U)W 40 W40 -40 1. wS0V
W 1 40
0) z
W
V~ to-
,
IQ~
1 414 ~ W)0
0 r44
NW 4)
44.E-4~
04 a v4 r jo0C4U
N E.4 m4.4- N4t r4. N~ E 44 00W .40.4
*a "W ( 0
W0(1 0 0 00u
V C cc34
4* wOC 04 4s -4 D
4* -04oV30.Q
r .mr c 0.-44*0 C 0
WO 'a .
44 v00 u -V40r
VC 4
0~.0 ~- '
010
'IC4(
11.-
im. 'an0
- NO .0
4 '
0 1C
0100 WW ' 40C3-
mnO0.4 Orl(
X0.4 U)rOI 4
Z4 0.o40
-40*.0 -,0(0 C00d -r00
C--E4~ 0 W*09
4 .O 00 0 E 04 :2.C4*W *0W4W
W -V )
4444. M4 0-1 a C 40
(n V MOMON V W-
44444 44-U IA)I40 v CQ.. 0 0 u
(10 0(4 o 0 -40'0-4% 0%O 0 4*o..
c 0(- .4.4E *O4'-
>44 14.S 0 4-0C a4 >ON
On00 40-U004
000n
-c 1 N-41 0 r
W r C0 0 I m). 04o4 0*
I~ .O.- 0.4 CEaO 0 C6 u 0 C44 440

04 0(- 4ZAo 0 IM 04I4 0 § A
-. 44* ~
41 0 0c
0 1 t4" 001 4fn 44 W t
413Wto0 4Al k m In X
.e044I
t-' -. 4-
0 .4,0 t bU,4
r' .
00ol4 VW 4'oni 0, 0% A A
. 4 -. -4444*Z 4*
1A~W z W is4U
U ~ I4 ~FA-*0
& 0 - I 4 4 0*1s
B-44
L4 - -V -W40 -4-c. CX 4)0'-'

V44'.4r4 -. C
c 0 0 0)>.Z In - 0 m0 M0 mcn2. C O0)0>.10
4 v..J4- -0400
v4 0 c W 0 5
J-0 00 0
0 04--0 - ), 0 00
o '
.0 0 -. 140
U CIO4 C44w 0 00000
000'0 00U
t $ 4.. 04. v o4..4mU4w00
v) 0 :3 M .Qt
M 4.) 04-o* a *-. w0Q M0..L)
a rw 40 00I..4-
w 4to) 0 -W 0 or t
00X. 2 04 vC 0 La0 0, 0 0 .--4w0 V wwc0
Crz 0 04 W .. 500s c
c .
44>0 )00'4) 1WO % )Wt
M . aw 44-4
00 0 5424 G0 U l< ' 0 a,
M ' -0a12
Qo~ 0 w w .C 4 t04.43 4.000)
L, m ~ 0 ~ 0
C 0 . C 0 4)4. OCc J - ' 0 4
> V
C1 004)4 wC C. c.
4 54c
>44 s~ >~ -eo u co-D 0 c w a0
mcv 00
w0 -44. 0 0 oo (U 0 -4Q4
04.300--.04... 440'00~; 5.C VOC' 0 540 00.)
> z - v4I>300 .C4
w00- 01
. o%0-4 00-44)6
0 004.-4-
-3 C' 4.. .0 W0 4 0 05
1C43 1.NVC C' C W.
4C4~ a,)Z(0 00 440 wv
0 *..W4
C... 0- 4 -
*0
M. 0-4 0m. 4. >10.4
V,4 1) 00
04w. 0. j 4
W)0.C 4 00C - Z. 04 0 0
0W > W.C 00
r4 44 z'~. 00o,-
m Iv- 0
m-m 0o0M) 4.4.C 'a 0W A. to 4 -E0w4 .
o0
2 r r. 0C-M EC4 04
40Mmt Z44-4 0 r_ w0 v 0 *.n.0
*.) 4..0-. I 4)40
cW 4) 0 -4 ) .. 000 c4cnV 1.a 4 I 044 (a
En
:3 EO 0
W004 )0 U0-4:Ow -V 0 ' V..4 r-0 0 4J
W A u.
00 -4> CC 000.. Vw. 0
0042 0
00 . Q) 40 3 4) to M4-4,
0 W
144J-W
'$*-44
m 4 4410 V r
no4 r.:. C.a.U F4. w0
*- a w> >r-0 a0
4 0 0 0 * ..- 024.0
- 00 2 40 0 0 4J
w) *04-j 0 0 4 004 0 4.) M., (1W -4 14. 4J M:0r
<044V.P. (2.- 0 m4. C4 r<.4
04) w >. C 0.
4. u I o .
42 . '4 04 0w w4.Q x o~ f.a0
$.-40
r_ W
01:3~~
0 44
~ 3!ta47o
~~ ~ $ Z : ) 0:0 w a)
$4
a; M
a) u 004 oN a)
$4. 0ioo o 0 - >0V
=0 .V
;'-I o .z rC w044
5031' E C 0o
tr 0 0% a) 0
w.-W0r.Q0w.V 0 a - .- 0~4 H 0ww 04
> .. 40 - 04j
t3
. w 0.04).
Dw c = c- 0
m-0. 0 -4i 14w -~ W U .. ,.- m0 0
.410O v 4 L,.'4 0.- 4)
X.0-. 4 w .0.0
t m V 4.0 M 3l-
c
00 Ms0~0 Cf,-a,4 4J0 004c) w m00

-30 0. U CD W $2.4)444.0 C-0
4. 3 0.0 -.. 4n3w0
s. 4. w 'A(.4
0 ~= v ~~~~~~~~
044C
50
>(Nm)4) 3-4
4
> to a,
0 0 0- 0'0) 0' CC 0 0 . -440
l<- U 0- $4U. C 0 0 0 W 0 0 CM 43.4
o
4.= $ Mc 0 $4 0 D-00 Q. w. 444~ W V.C.
.: W04v r c0>4m
4J 0 04 5044 C040 < . I w .-
C: P .W a 0--4.40N 0 t0a
mCWi 0... U.
4--
,4"4 54
I 4t3)0Q oI40
C -.-. N 0 0- r0 0'.- 4.543a C0
40 4)4) to 0
z 4 w4-0-> 0. w 45z 044c 40030
0 .0 00 4 ) C0 W' 0 4j
M4 , 9.C w fC o.C 42 0-
m0 C 044 to 1-'4c00> 4..Z044004 0 1 4 0C4.4 P0--
V
3 4w0 _ - W >4)0-4.Ol
W =
. 0 .) - 00
V 0, 00 m 0 *.c
4-' -0't l-$ 050. 3 ODW m>-00..- 04) w2- W M04.- 0.Cs 4
tn MW000 C 401 0.O 00 4 I C - 1 oD 0 '0 o 0 440 e w >
W 0 0' r 3 M44 0 444 0 W40. 0nW4
3 c 0(Nr w)m)0 w.w4U44)0
M
4) t .00 06 0 V AJ 0- 4.04)U0m4 '40Zc- 4' 440U2 5c.- -0 I
.4405 N c 0VCOc
M V to o 44 mC a)0.<0 4j w E (a c 0
4 .. 0 N 'A-4C...,4 A w.. o 0 w- 04-
4.4
oc> 4400) 4J 0C 3 100-40.-4- v4-)44-.O 40
0 w C444Q0W0
-w m.U54-) 4o
CDW 0 0-OD0.-j 0W0 0 054 000(-
E*W DE 0Q I4 d)' 04J C 03 wV W-0 4C
:3 - ) a,40 - 0 0 W=a
0 00V =. a40 044-: - 4 0' 4J W - M N $4
4. . 1% (n- 4V 0C .0 05 0 0 E. at40 I04c)0wo m ..- a .
0 tr 0w ;. CN. 4. V W 'a. 0C r.... C0~0 W.1 0

MIDWQ 0C
0 00
010. 40m444 0 0
-4' r 4J0-r50 000000
4 )..0
0*-~ ~~~~~~~
~ m..
.j .
0 I..
00.4 0 0-e 040
-. .. w 44 -4-w
o -4 .0( 20 w4) n 01 W r 0 a:C4
V - 0.
Q0 00 o4 w4 0V w4 toC~ M0C 3t44
V)>
Q) c 0 CC44) 002C 0 2 m~.4 l 0---4.0
.4C
01 ' 04 r; r4~ 0 .01-. Z CL>. C 0 0 0
17) 000000
(a M 00(0
0J ws 00 044.V)0
Itw '4=0
r_4j -- 4041 04)C.c0C ., 1(-O
4D4 0020 ~
.r ;a
(N (A40 44-40V 4)0M-4 r~ r m G0.CV04 M 6 24 m
.0 c 31 -4 to.4.CO0)
C0.5
E.4 4 00L
004 )
r
254
0
3t.
4)
0w5'405 4 4C4)
0W u 04)0w
0 W 40
PN
m. mC.
-,a)S 0.00
(n
0
40-4).
o G0()4
, o .0 ) ~
4 .u
w4-0-
0
14 m4).m a w w
50 0
.0
0 U
c
0 w w02 l 54V w... m.4 $-1 0 0 4444C
444 w
0)
oo0
~ ~ .
4
4-
4 j-0
~
4)
Wc04. w >-
.- CCa ~040 t ~ ~u CO ~ 0
<
:3 r. 40
.a
4)-.
0, w.>0
0 >1
0 n V 3 -W
0 0
w a>c j a
U04.- o- m(24
- c 4 - u 00y , mr w j0- 4A- 5
(N4
4'- 4.0 0 .0 42 5
0 v .0~
1. . 0 0C 0 0 tr
.44 4~4 1.20
0. 4) E-0. V WV 44wO.
.0>.>.>. M 0o 0 no0
.0-4 A)0
0 C-4
r-. 414
0. wW0 0 0 -'nV000a0
V00.. 4 o5~ 0- w- -- V.00-
0 w
0W0 r '0 "4'.44 0-4
4).~4 a. to w040404
2. 0n 0- w w U -- 0a
0 w' c r Dwa :1 n cc 00' 0 4)
4
st 0 0
0n4 0
to0
>04)4
M54-. Is
0 0
tn r
ll~~~0.
P r4 ar
w. W
m- 4 0 c .. 1 c w-w0 c >-%
11 r w mA4E
4)- $- a:54 0 to o- A0-40 > a :,; M.V c
m I 0- oo.
)- o 0 =)0 -04) wt j o r 0.
r 04 -341a% 0V'
0 -
.. 4) 2L U c .0-4 r-. - 1 1 0. a0 0 04 0 A
co a o w . 4)2.c 0 6V0 0)Vmc J 4 .0 4V00VV
a. o.5 co
0 i .
r45 0 00w
w4500C 'a
0
J.04
0
00.4.-
M0&Z 000H
:5j.
QCV0054
W c
....
.. 1 m v- 54NW4.m4
0
j w
00 m
VS>.-
1
0 0.5 0
0r
a,2
04 tom0 ~ c3 -4 J4 a4(010
$VS
4
. 0
04J4)-to
$ 0 Z .- V 060 'A 004
44 445.0V a, 04ix0w 4 00
4C
at~~. .0 c..4V.4.0: s j 44 0>
V>
m c4403Z I54>0 0 0 1'. oco4 00 **W-
J a
a<l-
0 0* 0, 0 'A a C 4a.03..400
-'a004 40A0VC J>54j c 0 X t4)
m" r to,V%.
Q VM0.v
r s.C :..L. M2 4 4' tno V0-0 ly M 4 .0 ..4 .2v
B-45
1 00M 10 000. .- 0 0 0 4 =MAO0

m4
000to
C: 30.00.4 0 H r 00V W .
m.4 0 4r 001 WA-0- . 0 4 00 0 0
44C L41
w. I a~-w 4 0 w w a- ' 14 04) V~ -41 (A4444
j140.- 014Iw0 w0 0 0y ol 0 44 0 c 1 0 4014to0
Ix 10 (a-
o'<(0
00440 40u
0.44 ... W-- 0000
.4. 1-0 04 W -W 4
4 - .11 0.-. 41 0 4'0
*-o x A n ) 4 00.00 iW 440
0. -)
I L 04 00. - 40
*1041 .. 000 V0 1% 04
0.-O>
) > 0. 0 0 >04 0 ) V c0> -I0
> 0 044-40
co L zw A<0 m0 441 14 0.0 10 k- 40
4440 040
0.00 w 0 '~00 0
14 441 W~
00oz a,' 1a40- 2 1 43 0 ) IV0
M
.4-A 0-w 4.0- ' 014 u
M %a0
I--- H 00 44I.,001
0 a 0 14- M
-'
0 0 00-.,0 .0 -. -4 0 r. *V0 MM0 0
ooo
C< 0
--.
40.00a1
a)0
.0-C.~~~
ooaaooo
0w M1.>
0m
0
~ .m
1 -o ..
a,4%0.
'1
44
4-4 'a
01 0 0-0a..041. c< 2O-...4-
mo I. > 4 10-~-0- . 10
0.. 40 >4 . ,,a-0.0 0 c0 0 4-4 0 0 a
- CL- 0*' '00 0 C'c0. ZZ 4
a' 001444.0C '--0
:0 0 4 Q; L4 3--r- M 0 0 w/ 4000C0E-.1
NO .0 0'00m> fl 01En-'
04 .. o U)- $1.-4 040 0
4P40 *0, DC r0N -0co0)
H4
-- 4 -c-- 1.0 w
0 0 w1414u 44
004 4r- 140- 300C0AE
-*.4 U -i- a'0,CA41J 41
.U)0'.M to 014S E-0Ma)41 U)
> -'N 00
M0 0 14J 0 L40-M 0
.r, 00) 0)-4 a'r 0 ) a14 0
'01.0 0 00-4E0
Q)-4. 4400 00
0O O0 - 0 000 44 Oll4--.0
4 )Ma u00. -4.-.1 41 04J MA.0 4
14 0 0(a (a4--l 0 44 A 4444A 0 uI
-1 00 ,4 )r 4
0400~& -'.0 CO tn 04 0
to-00 A444 1
x A 0 x 0m
N 4 10-.>.
m4 Q;2k -a). 44 0 W*a)oa) L)'r 4C 44 14 4 4-W 10
4 C
0--4-
--
0 0>I 4 r0 tr (D
414 1. 4) -444
1.:
00'u 00aE: 0- 0 0 41 -1c .- 44 0 O L 0
--- 1< 00 0 ;3r0 a 0, a,.00 ' 1.4: E
0.o 0 -M M 1040
.u 0 c !4 r 0 v0444 )A -- 0 0
--00r
-- M.~0. m4014 4140
(D - 4J0.W V04 4-. -4n- 0.004 0000AW.0
00-Z - 44 - 00 000 l< 4 j>w0 '0044(D 0 -Q) 0.
0 0 00---0.1--*40 O 04 040 CO1I-.44.-40CM
'-044 440c z4 00 c 4- ) 440 0 0- 410 (a00440
c< - 000 0 -0 14 Ln M a)O~
$00
1. 0
-L; z 0 W--0.00
x<1 .-. 044-140 ,l r 0 $*4j44 0. 0
0.0 ----a 0-.0 0
V 00 c w> . 04 M 0 O .0 u004 000000) 3
OI (D'1
0l
0 0. : 00 4
aj >,
1 A
Ll a) I
-4
<
0.-4
)0 0 a>.0 0 rqt -040. r00.0
0 44 CC/I 4 14 'I44 r.- 4 J 0 .
0-1A - 0 4 4 0 0 44 4 . 0 04 a 0 M<.0 4 4 4 0 0
144.40~0. 00 00 0 0 0 a' 10-4 0600
F-.-.0 0
0-U -14 41-4 a,' A1 M. 004m.4 4>m0 .r
C:~--
00 0V 0 r- 0' 4 4 0-.0 4 U440-0 040 m 444400.E
- S' E
C.40.
.
01 1 4MCC 0000; 4 1-o.- -044 ) ( 440
4 1 4 . 4 A 1W40 0-0 -40
- U) N00 < z ( o0 ( I ) o a - u
-- 44--'-- tn 14c 0 4 00M4 - (n a-:4.
m
C4-o-- . c- 0.1,4 I40z4444
00 - 0c 4) .4 0
M0-a)41/ 0>4 1
0.5a -0 0 0. 0. r, M m 0 1- r 04) 000144401CL
tr
0-- W ' 00Tr <00> '4IW000W -0 - r00 'a w
-0) r'0
le 40 0 > 0
00 414C 00-- 40
E0 W4) 00 1 44D
>4N CL 0 00 ( jZ 1
41 Q)
00404R404> 0.
1r v4 .-ONN 0 -0--4- 00a414/~ a4 0.4
'4 0
a ION---.j41 4 40- * W
-' 4 ) m4 44 4 0--
- 00 4 0 41 4)144IV4 0 4 CC- 1.w00ato0o0CU)00X
$ 140044140>
0 . i 0 - 4 > 0 4 0 1'0 44 03 V 440 0r 01 4 0 4 4
04 .0 ' - 0 1 4 -- 0 - 0 0
. - 4- 4 Q)40 A-- 0. 0 > -4 . 0 04 W 0 4 CC4040 4M 0
4J 0(0 000
-4 .- 4 00 4 004 "40 OZ : .4 4-)- 0.40.0 0 0 0
c0 1 0. 0;0. 0.44 a 0E>4 0 0
0 Y "407<E 0 C >W 0 L4 4-.4-
4 00cCu <0 0--' N c0 43w0< ' 0 - 4X 9

0 00/.
0.0 400- . C-> 4)
14 >40-
0 a4 01c"0 O 1 0
)0
004m .4> 0 04
t40'1.00000.'0 -. 4n
00 *10~~ (N N -W It444r-
0-404)m E.
440 0aW-- . - 0
o-to >10
0 0<c.... 0 -. V4401 40M 00W-C4a W-2m t nCO4 - 4-0- .- C W V0 04)w00
ZO CC 14 0'0
0 o 001 4 to
"O -- M. V (a 3 (1001243 04a to0-4014''014--4' 'o0 W 0a) I 4 -040
v. r Ia'rn 44044 > o-w E(Uu004
C0
f 140m 44 14
01 0. v ) 0m000 41 >--I 0 =00 V
00 0-.C. 0 1 >4 0 , C00
14r04<a, 0 '044(a- -44 > >0.V a 40to4CZ
0.-M- w0c'--'44-m- 04) rz E-O 14 (00 0, C V0rn 0 jI10 -
m0). 00 00 0140-, M0) 0-'104-4U--o4I0 W 4-C-4 M
440
Wm U)
0>1 - a- cM000 0 M4 0W 0 4
0- 4 E 00-. y O a 0 .0-40. 3 140
0 .40 00 U 0E 0 -0') . 440 t 0)4041 . wC - to r.
0 0.- 04) 1C OO C0 4- 4
0 U4001 1401.40E-'--....D4A4M'c-- 0
00 fn-c40' A4 v- U0 X,-4i0 > w - >4 a a,
UO.- 00. M..(D j Q 0444400 0
-(0z 0.- "-0 000C (D4- En 0 0- 0.>0W
j
A' ).. 'o
w0/ -. 41 toMCLa-H 0a -40 ( 440>&0 '0
< 31 W0-.44
WC<30 44U W N0 m).4-CZ0. a10 - L,0)4'-r 4-.v
CL4 0 0 0 410- '--a)0 00 4 0) 0 wa'w4m-
.- 0 to4w 0.0A w4 (a4Q
0 0> 1- c0 014N r44 00 0) a)0 c0004 011 >) 0 00 , 7 0 400A>
$.-' 0 4- SL' 000-444 .. 00.A
A --.w14
a. 0) to 04 0. E 10 0) 3C O40 4 00 0 14
V00
MM A0( - C004 -- W 4J00..40 V0 0' Q. 00
00
r-< 0>.-4004
C4 a .4 -.. OC -or4V4014M. 0ol4444044
0 -) 004
go '00E.00000W
(/I'a tn4 w -00
E)4 $40 4 014 U 00 M '0-t40M
nZ. 0 41~ 4(
0----0 0,-0 0014)4-0
g .4-40440:
C u am
< A~
'-040 0 44044
w0 00-.0 0444m- ----
440 E- nlCI..- 0. M IP.to
0 W4 0>
00J10c 3 1D4.-.3
W00--
0- 10 -4 40U.a
0 a v 4-0 0w --.
W104.4 4 0 W . 0 14
0
1-/Z 4- 4 w I W -41 W4 0 4400 '0H 44-.
4 0100m-0 0 Z->1
00>Z Ev
044 9:< 44004V4w0m4 c0M.U.-4 *E-
0 04, -- 0 0 M W '--04 0 04 IT 0 V- >
a -. 400' 4 41 > A mt-4-4 t w w
01t 0
w0-440 0. -00 - 0 0440000' 0 .. 0 CCL44 4 0
<044
004)r4x 1 0 L. 10
".4'4 4- o-
44 1)4 ; -Z/I00W1 -0'0 J 0 MA4) <10 c m44 V 1 v V 0c0 00t
000-.
- 1(
00
(A
00
Wo - w.
0---4M0000A
to4 004-00 0
4)40~
)
000
w0 A0
44
u
000
M n ' 4
0 ~00c4a14>
00 )$1ac
-4. 40
:c
Z a E- u % r404- 44 4- >0 1 4 0 -01 4J 00.
<0 :4 00 T 0(D 4--0
'o1 m --
M 0.4 4 44
0--'01 00 440 Q4 4-44 0> *04 0 044)000.00 0 0
(A 4 0 -10 1 40441 0.4 a.000 'D to - to
00 4 01
4JAMM4J C CLZ 04E 0
0O 40 0 0.'0 0 0 0 4 > W44 .0 1. 0 0 -4 400 4 0 4 4 -
r 0Vc0ww 04 "4 - 4 0
t- . co 0 4) AZ4Q 0-4 ' 014w0.>0
r W
0 014)104ty--14' -41 '0440
0 0 $
m 0 l
4) c0- 4-.. w4 M -0 2'>,0
4& 000 Z140 U a$ 1.46 4C00r.
at y
<.0 --4 00-- to 40 ..-. 0 0 0 .r 4 00 00140w

0CD:w 0C .-0v 0 I 04'- 44IVci0A AwUc 4) 0 U-g0
0 w 0
4 0to VC4'
W<44V04O
c c U $ 0C ,'0 0M
-- (r.0 3 vI- 9 -41 1 > 40 L.4140 14
'44-'~- 0
440'0 0 0'41
0 00 ~
0 10 0
0D4014--4 0 0 '
fI4>
4a 0 0 4>m -C
c40
-
- 0 01 .
4 41
0000
0. >
004
' 41
a0441 04 4U V 4
0
I4~
to>-40>0
404. 0144.14. 1.4 4 41~(A w.00
4 0-0 0 0-0w4I
U 1-00
to)0 0 0144 -N 00'- 401 4 001 M0. *-. 4 0
44044~- 0 m.464.
400. - W10>4 '0 (D w
-0 W Q 4 0 c
x~ s. a- k Ii4. 0
440 0'44(a .>0. 4 'a)
0 WOO 41
4- c
<0044 C--0-2 I 0 0 M* -- 41
1 WQ,000 0 4, - 4 4 '0 14Vtj 0-4c 00 4a6a$.$
0 -444 40 0.-4,1.
0 140 0
0 0E'. - 04000 to9w
4400
4 C6 w c a04 w Er0> . L 41 4 >0 4 1c00o.0 ~ 4 r 4 ....
). V4
0 >M-.0
0.
a44W4O to0 - c
0 u a x
V0 2 E.
14441
M 44 '
B-46
44~-1 ) V '-V Z C) C00~C U - QC~0

r. 4C V
40 0-4 -4 0 -C.C 0 >. .4 rCW - 0 o0- a
0 ki
to H mO-
*-41.VC
31a..0
a 0
c to V
41
4)
I
$4
0
N 0
.
m
V ova
-4
0 0
~4j
c a
L. 0 -4 4 -04.m 0 0 4) 0 -4C 0 -4444.V C M 0
1-. 00a 4 V 0 00 0 0 N N a 0 af)0H a00 V c 400G
o Hn 04 400 0 ma A
.4M 0 -A
. *0 j . 4 l 044 OMV c
0 0 ~ ~ 0 0 -0
V44
-@ >.)4.
4 2-
4 .40. a.
C 43, O
0
S 4 N
0 1-4004
M4.
.0. -4.
a
.44 0W0-s c)0 io LO$ A0 00-
j- M $V414 0 4J M-4 .3 11 13 .-40
U' 4 0.0 41. .
IH N 0 a to040 4 0a) 0 l< 0 =.-0 x V0
u 0 -A04)' '0~
-4 E- C 4w 44
5V ow 0 V R 0 *304 0 -4N V0
0 V V ( V-A M >20
OX4 0
v x %. 0 040~ ol OU) H 0 0 0 04N
0 000' a 09 0.0W E..W6 . 4 .0 C 000 a.0 to ?A
*-I. ).*C 0 4) 0' 0 0-- 0- 4) .1.0-
u01.*r J - C E. C. .4 04) -0 0V * 04 .00 0
0 I.V mM~ $)40 E0.00 0 W0w M o 04 01*4
V Ia
C W 4j~ 1' %.E44 40
tO 0 .. .O 41.W4 4J OD .4-
j.V
M co
.L 4- - -4 w4 . 04 a44.c-. 4
M 0 .00it
1400.- %4 4 .. .44 0.4 0 4 0
0 0003~~ 0_ 4t 4.04 0' E. 4v 4.0oI000
0c ms r C40 04) Q.~ V)4)4
Ca - 0p- V 00 0 00k-40Q
04r.c00 $ 0 4.
w0 4 00
V ,4
4j 4) c H 00.4 w
" H OI. w 0 w0>
to :3 0 044 114 31. aw0~ 04
.10 W V0?tCM jIV.0 040 V W4CLE-0 .
NU0) '
39-HV 0I -O 0>c
e &C 1_V z4 V0
9 x 00
MC.. 4) ;
Nv 4 , UNO ... 4 - 4 C N 0-- 0j m ,- C $10r:
E-
U) CIO 0-m-0 4) w . 0 3
4) " 0'1 VI t 01 0. .0V 0 44 10 . C
>... 0 41.. w> CM 0 4 0 * 43 4
4)-
A 3t 0O-4j O
>1= VV. UM0 4(. '.0144 0)0 4)e
0 14% C U 1 0 40004
to.
0) 0 VN 0 ) - 0 4 1$ 4 a.
000 E400
-C4J0c.) -.4 C
04j' 4.Ol 0 1~ 0 Ctu -0 0.
*0.'C C $44 0W4-440 -V E) V14-. ol w Vu j&. 4) -4 0
0 4 to .0 to-4 V0 4 01. 0 .
91 '00) 4 a CL.
4)44 w-44
W ) - Em
H.- a)04 Go 0 00 00,
4J8
V...-1 M VIV V 0.0 W(
-44.4N 0 0
r.--41-. 4J 0
r. 00
Zo )9 --040 rz V1 a 4)41 .Oi O0 9
344MCO .. 0 1 >.1.
-4 V0'.
rr10r - u0,a W~. 410.4 00 Id (a4t 0 00.. 0c Q ' 40 0 1.14)
N ~04..10
4)0)
0000 w0c4 0044
0.0.1)0000 r V
0 0 WH..
. 04 4)
W4)040
O0Nc4.VO'
0
0040 o'
(1) CL30
0'
j
0414)
) 0
0 04
0.
tn 040
00 a 0
4)
0 44
0 0
41. %4
V 0 O 0 0
V 04 0 0 1)Sa
rU) 0 c~
.0000 004 w 0 V..1 : c C)0a0 -j -. 10 0 *,wSIa0 00F
U4 - 4)
V 4
<ZDO *.)40 M 0V0 0 co1000Go4) 000k00a
0
V) w
0 10
WW C: OD0.4 0>4$A C. C.C :3 0 V.4 >4 0 04.14 00.S.1
V. -w04
-) 0 c4)0-.-V0 09 0. -4C1. 40J.. r_ $01.'.0
VV --
IW00 40C V
%40004 0 W *VZ . 0C%000.004
z .2V, 0-W4 .0W0 Q 0X.04C.- 0 24 ..-010 -01.w0 C1 V-0 41Q t
< N.OH a. 0 : 4 . 4-00-
>4 to) 4) NO= 4 't-C-U.-00C4 00>
c0 L cmrt 0nM
. i40 0a4j0
<.UQ)X
r tl Z
U r .4.0m.
w.-40V
<. 140
MC N >)
4441--4Ia) M 0
s0 0-4 0
Jr MO
VO)m ~
''. w _ 0. 4etV0 0 0c 0 C 0H 0 00 44 04
11 .0- 0, Ix W ()..4-0000 m. 0g4400
[n.40, - 0 0 0, 0 4)*.4) :1E)i - .4 V4 C. wI..04 o0 0
4 X:.. W . v 0 - 44 0)0-
0 0 00 1-0)
U.0
--M>00 c 0 j 0 A 0.0-1 w 1 C0 - S -- 000 N00
'<
V 444 C6 0 0 w4>A .o40to.. .0.H4CN) 40. 000 m V '-400 84
4) 0 00' 4)0w 0) )0 4jr M -V N0. C -004 j 0V
'-1 u*-'Vw
w' 004C4 0 o 00
: 0 .10
Q' -4 OO0 0 4W )0
0 0V 0
C 4 04 V 40
a;co4
3-.zco<
0
H
) 00
.) 00)
O
W
4D _e
* 40
M 0 4
WV10
0co1
C .
41
H
00
Q1 O 0
Go 14
4)2-
.0 400 l 4.4eU
c~ 4,c0
-0. 0' M.0 X40C --.1. - 0 4100 U 0 00
I.0t0 t4)0t m-W ru0 - .VZ OCO-00 mwcC-m04

0 0)0
C4 C CO 0 w4)
0> . 00-V 4 0
40 N 0 0C V EI4-4 -> -0 _ Cr4J W 4.4) 0.VW 400-44 ow
C .4 0 -40LJ) V Cw z r. 4-..
4) V0 WV.l i 4)
004 0 0 CO0H0 *0V 14$.
0.4j00 1-4 0
0 CO 0 IA >0 O-,4 04-c-4j S -4 0 cO C V m 04I
*, 4 (aCo0 4) ra oC'40 A-H00 - 'a4>, CO)4 0 t
4NI. . $H 0 x ta0 _ P 0 NI 4 -I ..
4) w 0 ).. %-4) 0 1 00
-404 > 9
204 00 o- .14-W 0 0 N L40.4 OC 1 . .
004. Capa
0. rA
44 V
>- . C.0CI >4 0V 4 -I CO-C-0 00V
00 1.4 0 t0.4 - 4J .0 01.4) c
C4 r)4I.- 0C00>a .
O MO wV 0V' )0C4I.04 >V 040> c440 0) I 0 c.0 400 4j V cS '.
0 I
00n, lc a)M -IV
0 w0
~~40.C I. 00 04
0 okc0000 0
.0)
0 -4
4)
HO..
4
4)
CZW
V
c020a
W.-I-.-
r 0 t
)
a )0 V
& V)
Ch0) 4) 0 0 %-)0. ) I.V 4.41-' 4 Ea 0
I. 0) 0411. W C 0 ). V -. 0 0 040 w0 00 ) 0 000 c414
V a5 0CO--W W04 -40 0I a >V-C 0r0 e. C
04 -000
000 V 0 moa 0 0 00 - OD )004 r44.C 0 .04 a-v
4J3~' I. I. 00.o 0 ".> CO 0 0 0 44). 8.0 . w'
0004 4 4. -. 0 V C ) c4 0 0 D 43 M 0 0 0 E- 000U
VC*OOO
0 -or0 4 )0.U .00~ C 4.0 11. -4
N0? .l (-4co uV4) 00n 4) 004) 4)
wO ) 9OC .,o-0 000 o
r. OOO 1 CIC -4 00 .001.0 . I4
'.002 02 4)C 4, t
>,2 '44404)4
00.0' H4 j
A).4 .4) I.ewO u r go 001.
N. .02..
>OV t) W1
C0 W-- c c V4 0 .. I W) -4 0)40-40.
:3C w0 r O0a 0~ 4V t . 0 . 4 V00w(0 W C XV )0 4 0ca0
w6000000031M C - V 00 C? .. A. m0 V)I 0-a.0 0 9 0 -s
0) >.of 1W0 .4. z 3.HO...4 M I. - 40 0>. 00VNW0 V W 00 0OC
a0 0 .041.40.4444 0.4. C-40. -.4.1 4.0 to) 4) 0 0-4r
4 a z.)Qb 0 0 *3 o% 0C 4 04 3 060
t;CU 04.00HW4140 A
w0 a0 N aCl. L
10 .0 1O-4 )> 441
w0 3- 0 4 1.4r 0 4.44>44
MO 00. 0) L) 4 41.4- -C. Vc40 z . W.0 .O.C
NO 0. -c -4 SOC
.1.41 0 0 NO 14 3 0010 I 0 Ia
-. 04
44 0 a 4a. )0%) >-4 OC -4~9 4.-'-A 0 $ 40 I0, 4
.0 a OW 4r c %W4>0 00 L 000-- CD4 -40C.41E 4. Q~ 1 4
C.ME-4
M0.ODN AJ4aHIC.
.4 0t 0 00 0 C*400
eq0W-- 0 4O .0 0.&3 04j

I. om 0.0 n
4 0 0O0 c a 0 0.h 0~
Iac4
0I r0u aCDi 0aL
ODi
dc 0TOV .01 004. N 4L 4.0-r
4 a 1 ).-40 4
040 Q.- -4 0QA040
B-47
go 0.*0C Q O >40 '* 0,4 "

m0 i O0 0
1. 0 > 0 0
0Il01
3c
44-0
oV
OO4o c00
0 A 0.C 4 0
000
V0 X0 00 a0-40 'A M - 0 4A.
* * 41 10 .c 00 001 4 tr 00 >.
r..~~~p O.-000C-00 A1O4- .0 0m
rr-.to -A4j
0 w4.VJ-
*Vm0
0 0 .00
U x-I-)-
- M
0
0 mada0 p w ~r v0-w
0
-
1.4
.c0w44
C-- 4 0N
a00 0 0 .- 0 :34r-. C
0 01VOJ01 0FW -4.40 V-..1
0 ?A 0Z 00004 004j -004 04)w0
to W IE WI 4)C t, 4 0 .- 0 0 0 A A E 410.>
40V1 01 WM 4-0>00 a4 00t
o 0.
4-4
; t4 > E-S404414 ( (A~
1444 04 00
0 0 "H W.A 4) 0K4In14
a 0 - 0.
0a o-
w00044.
0 01
Vr N
r0>
c a 0 r.Or1%
N
01 4 N0 to 4j 0
0 -0 01w4
-m 4 V4 CU)0 04-4m 0 . 0 Oh 140.-I04 4 0
0 V 0 C0
V1 044V04 k 4 0 w 41 OV
00.0 04 al 004V.04.V-4 0. 0 1Z 0'0WML
00 l40 a u)0:W 04400 . o 00 04 000
0 0 1440 44 ., mv , 4 0 m11 . 4 .0. 0
4 00
0 4 4 .4.-4 0 h.0 Ok 44 W
400 4EC0=
44-4tr 014
4)0 N
CrCp1 OCO 44 " 4.- I4a-4
.- H0 M ,-011 0 4-' *'0u
:1 0 C ZO .0 V 0 4 000.0 0$4A 0Z00, . ' -
0 10 V m4 w aN0 0 X w 1 4. F E- M C V- 00 E.0
toI~
U 41 0t
, >. 0.1.4
OmVUO>
V0 0 04w
H44r4 N 4 c0 40 - to 0-4 H00 V 10 C0!-
r.C. 0 M-QIC 03C 1 414;0 Q cm a (am04- C0-04-.-
a.-4-1 0U) n4
0 . 4.H4000 140 W- t'o 0nc
) so to
414 44L 0 0 4. 044 0.004 t 1<. 40 m000 O c
''(C4
*.C 0-OL- 0-04.4 0 4 J OD
* 4c40 01
-. r 14 C C40 04'$4.1-0 0. 04 V WV
N 4 '-C 00 0
%00 V000-Cto V4Cp-C
0444
41 N, 4) 0 00. C 0C0 N 0 CO1
c 0
r -0
o 0. 0 OD
r r
0
VO 01>0--.-
0 .14, 0)0c
~0.04
0 3 4; q0V X u
C'D
44o0c >
4 00
0Ij10 H .0 44
a -041 to 0 4) ;00 r 0 NC V (n c
co14 N - .-
44a9 0>03V0r.0
ma V rz.O 0L,40
0 0
04-C--
ODV
~~44)
-40
0 > I 0014t
C
co -5 c *44 0 -. 04<OD 4) 4 h -4. -0 z 0 NC 0 Ln -- 0100
zoIj-.00 c-N n .0 .0 0 Q 00c> W* c-* Z.4 N -0- 0
CU4- %--44V- I- I 0 04 od40 C0O 4) V 040V
0 0 w000 L 0 C CO 0 0 0 W4.>0 0o r 00 1
w 4 .0ID 0-.-4140w 03> -- I4 0 014 0>.
CW V
00 t a 0 0 V W4 0 0 -'- -4)
c -4- *, 1 0
O. J4-" V0.0
411
-414 0n-0
mV
44-me00 04044
344444--44 0 00 M4 0.0.- 01e.w
40('A0C 444 0~- 3100-
:3 4 - 0 0 44 0 0.. c x( 140
-0. 0 e m0004. 4 0 044 04
9- N 4 -H--.-
4440c00 4 tC 0
U
04441
000 V0 C4 rO--U
444'0a w
' =-0
0-
01 4)
h0W-4 0 S t,> 04 4)0000$4 414 a ox>1I0Z0 C 0 Q
-4.0
k 01 .044
0oV -0 - $. 0 000> 4 -.-A0 t. -A- 4j CD C
-
OC
0' 0 O V40C4
.04
VCCO
0
W?
4
n V 2 .
00-toE
040.
~ 00
.0-
C 4
00
4 -0
00
PC1N .
C-4
CC-b
m 44c0r-U V00ma0
,h
0t1 AC FAV44441r
~ WucoE.j4
W - 0
E8
o0
0.
w>0
00(
04-
N ( 4- 00 Q $4 a 0 1j414-
W C 0V >.14 00 0-
414a C 01 0. 4
4CC- 90 jca 0 . 44 -- 00
0. 01 140 00 MA'- 14 wGo o 4 u4044VC04
'a w r4~. j0 " I -0-4- 0.01 0
0x P. P3 t, - 40 4C11- to, om .. 04 0g -401 0H0
0N 0 1 r0 4 0-04r04
2.1c .40a-0 0- Q .004-r0 (A 14 4 a00.4o
0 0.-L'. 4-h0 C10 0-
r-4 r. > ., -40N
rC v >014000
N 0 0J1ANc
00 - 000Z4 - >. 0 4)-.CCO
r- V Ai1CV44.r
4~ W * ;C 44 A0 A) 0 4) 31
00..-40>r
.4 X~ '0 40. V 0t 440
C
44 40 00. 0 0 WA 0 w44-'-10V4a
w~.
0
4) IS-b. 4t ro
0--- 00 04 00 C0 D 0
W >C 000c 0- 1 4 r- v W- ja V
OD 0) U 0V-. 4) 4 -0Aw V40 =I $0 4 MV >
04110
>,c 41 a o-
44-.0 I I>.1040 fA rm -IwO
-- .044J.1440 044; 0N 01
La > -- 0 14w
-400 ) > too 400140W 0 C44-4- V 1

4000 '-- 0 4 hrN04.04-b-(A).0 V-04
a .- U --V4-4.0 fa 9 1
v40 to W. >. 41(
r~
ZI 40 .0 0-0c 0u044-
1 - P U4-401 0 to -- 0444a0 noO044-4
orN004- 0 04a0 0 .' 0 0 O 4 0 0 4
010 0. 1
AIton-40 . .1111WO10401 044 ' a, 0
410 u 4 0-, 0000.
LII u44 00. c 0 04 %0 wt 0 z -4V - .0- ix 4
N a000 V a vea0 N 0. 00.0U04-4 A0.- 4- 0L
W44 00.4 0 > 0, 4440
44 -00 014 0100 00
- 0.
e z I -- 4 00 0
v
4 '0440 C >, m 4.4 0C
to - 0. -. Z0 O 44114>0Q00 0 )I. 0 t 00014
"
(01 '04 0140 0 140* 04 4j 0 0 8 0VC W A40CI X.r44.
0 00VO . c0 ->1 N4 00 -40 0 0 044-414
0 c04)
VN11 4
>40e00400 c04 8 0 6 1 06510I-ON > 4400
-~~~1 001 v> V04~
i0
-- 40 0 011 -0-4~~- 01 4 0 4.OM 0
..*4-1 a .0 mr
164
4 04Ic a~
4v
-0a -0 1 '0
0 w hI0 N
0%w(A
0 V, 44
4. 0 or-
014r .
0
14.0-
0-0 c H0.0 M% r
00e-40U
ji
nI W 2 x1.>u
o-~
oc~~~~~a
w
w~~~"
CID444-5
aava
0IcIQ
r4
o$
1
r4e4.IO
6O
-00 -40 a
. 0
.0-4-at
0
44- n 0.0-02
44
a0 0
I.,-. 44
14
a 140
19- 00.0 c 1-4 a0 41
1
0-40V04CM44a.Q
0 00 a
k44
ItV
O-~4 -4.1 , tFI. 0c - 4444 04 4A
[ $4 v0-
IOI ,.0
OZ*g0
4
0- ~ ] 40
%
p040c
-
-0
C
1. 201
N 1
44 4 0
0
a0
.'W
c 01
04%
4
W
144
0.
co A -0 4 A.U000
B-48
U) 4 42m 0 O.2J4 U .o . 41 a4 0C
to 00,402 I -4
V.-4V -4
*7 &4 go:.%4 414m4I 0 40 C-444.
14
0 0 m H 0" 0" M
4 42 04 0 0
04.V -0-44 so42C
04 - O
0 ~0. ~~~~~~
'0' 04 V4. . V2 r..-.
0' uJ4000 0 4 OC )0 0
0 2 >0C0 4--4
rk442
42
V 44 40 4-4 0 2 .0 C II
V -to 4 - O 0 W4C
3 0 2 0 0.44 0 0 0
2 M23
2 4. -. 0. 4 *- 4 V24 4a-.t-
0 0
00/C2- w .04 0 0 00 ' 4 .-M 0
C 44
0
442-
041 .C 0
4 z to20. to Oa) )4 to4. 0 0 f
'42..0 C104 2 4.042 0.4a,
.4 w2r40
410. 0 0 0W4C
O. 4. a
r-S. . m2 00.0 2 V. V04 )01 V2
0>4. 42o42cc 4 -V fl.- a 140 tr
0.40 s4 a 142
04 T 0 .4 V 0 0,4
is4 4.4 42 421 0 042go04 042 $C* 0, 4 4
m44 42 WCM0 4 44~o0 C >.4f 0 r H c -4.to
0 w 0 0., 314V 0 004. 04 .C 2.)> w w
--444 t 0
4w 41 N 0
H2' ."
1142 0 0
0444 En c2 .. '(24424
mt S 440 .1-42: 0 0 u0 W
r-4 0 F1 U--Q, CQ'44 2 0 ~ "4. w 2 V
41 00.0 4 . 410 41 4-0 . 040 .4.444r.
m4 >. 424 4) 024.
14420420 to42 4400 M 2.4. 4
0420- w" -. M a.
>C44 0 .0 42'--0 0% O 00
Q 0~
4442 w (a-- w. :1 40£D044 VO 42 OOCV
M442 0 0424-4 M.
4 3: Cr 0. 0..404a
$14
42-. -0 . 0 04 4 1 >'V00 ) M V 41 g 0'
4 faUCo. 0 4 -- 40 44 0420-4O QID . f4 0 14 C6
V0 %
0. W $14 0 U s00
m 42 Q'. 40 0.. . 00 .V 3 0 W4E..4
4 0042I
NC1 :)40 410.0 0 41 4.4
au 0 09 CD042 2 0P -- 4'22 4
r 1-i 4 4 V 41 02 M4-. E4V4 444 0 2 0
4 M 0 2H" 440 -4 442. --o'Zr44r
. -4 - (.04 rx 4 ) r_
CO44:3 9) O J4 > 4-441 V. co04
zg2412 0.4C- V'a
Cx.4
10
44.w0-1-M
42 w4.-4
.44 4 0 4X 04
.- 0r0,.22 .-.- to 0.1 I( C 0 >4
44 .- N4404 >24 O -4 044 4
a.14 CL 0042'a 00U -0 4.W42-
r (71 0. 0
1. w Z42 W 0 nO a) cc0"w
42. 00 o2.
~144-...0~42.-..C4 4>44. z
M-444- 'A
.0 to u 4V
4
C.0
4-) 42444
"4.40-----(n 2-x.W 0'04 k C() 4 -y2 rE0H------42c *- 00 4 444EM
4
:j 42
w2C4 V0 0 t0 .4.44 *.aC- wr: . 4220w 4 0 C0 4.403 0 C42 r
0.024
r.0
-420 44 0CW42 >04. 2 ---
E Z-~ 00CtoJ0 4
C3 0., 1,-042 CO..4.00
(a2 '04 C>024.-
0044 4.4-.4 q-0.V t z- w-.C'2.2..
Z)
r
0 " ~rz 4lWL
MD-n00
4-1> '
) 4
0.
- M
4-.VmIC0
$ 41--- V2
mC
0 <4X(a
E 44 -- C
34)4).W2C V-
4M 0.0 . M
42.f.) 04.QC
4' 2 w0 0M420 4. Z024 - C.L04 W
4,
4to V)0:3. - . 42M
0 I0. A 00z v 0344 4 4M O w C ~ ' 4 04
X Q 4J Vl42--.Q
2 'Q)o
4
I0r-4C
V
4J 44 0 m fu u ODM 42 w c N 0u .w'4L
0 -0orl
t 0. m -W o W N30- V4.I4 IW 0 440-4
(4I 4) z244 4 -C 0 V 420444..
10
- 4D IC.V
4. 0 >4 0 0 -'40
go .44
V . >1 44 m-
0 0 ,t a,0 C--N.. 4t_ V C 2 j C4 C .C OW.i
0 E 2.
w . MC '0 C M V 04.-..) - w G " Q ..
J V--4
400 0.4
'Co0 s4 > 42)C0 4.-
P4 C - 4 40 >- V'
444a2( (L4 £0 C>)0 0 w2 0 x2.'Cv0
4.4 -.-
C. 4203 W 4) 041=. 0).4 3 82 44 >0 C C r
. of.r 0.tn 0t 0 1 0m- f 42 4224 fla .) C 44 C
44U
-1 4' 42 0a0 z0.- 44- 42 440 4 4) 0)
4
4-
pi <
W
<-00
~ *-4
..
V4
00M.-
44..
a) a4
3
41
4) U
w
A0 42
0
0'
01 0
424
r_
424..
404 ..
c.24
I
4400 0> 42 V 44 t
0 0>X4.4
'n V
-u42- C0. z4 04
0 . 42444
> 0- MJ N2EC
440 .0>
13 v- 04 44420. 0 0 *-2 -4
42Z4 .) 44 422 V 042 00f 0 0.0 42 -- 4''C0
ZZ4'o 424 U . . 0--r" 4 4 V4.~4 44, 4
0.
CC. zZ MC V4 r4 >. a2C0
04 mN
C. N> 0c "- 04
Ww 0o 0 42 444 4 .4-
41 Q)44 a
C 4. 41-- 440 h 42. V 0 w 04 VOOMUc
0-0-
>' 3 -.44M
v 040440 aC).-
I.2.4
04C4 ' m 4) r4 0%r4 04J 00.4 C424o
. C C. C6 C 0 42> -0 04 0000>0X
'0044~
~~~ 44 Z'.--4
Cr
2- 4'-4
r 4'OO~ 444440 V, m.22

42 0.4.
4J-.-4 u4. 3 0420C02 V l
402020
24 .. 241 "100 W. 44V4, C4 0 0-
C ~4 ~ t 04 a,.. LA
4440> 020'
044-0
0
to
44440z CEn
0a 0
.02 002-
0 00 z" 4144 0
V3
04
002420 4
04 4 4 41 0440
CflIm 0 V3
-4Z. 2 4 o,
C.4 M Z c0
V4044 424 ' a N .5 0 ,2 2t 4j .,-
-004
o 444 .0
0 - - m 4 3 0 030-' 04u0 v
30
000 4 >( 4=842 4 v '.40444244 )
.CZO; 0 .. M24.. )-..'..4 0I2V0~
'4.V.41 W*i
W C $(4-0 44440 0444 C04 1- z~.42 0 -4
* a J440
0 T0 &14-4 4 .0
u- "4 Q 4 > 4 'a0.
0042 .kM 0 0
4 400 2 C OC 024 -4 44244
W24 M , 044.2 '.- o 0) 4 244M
>~ '
.4403,Q
02 z. 0 O .1
'004 42 0n09 Q
1-
0 1
'0"' u-' 6 Z>~ $ -. 0 c
Z44.' m44 V43 Z c .4 04.4 4
4 0.2 42 a E42424201
u ~ .4)Q3
044 44A
v0 cU __1__DH__
B-49
.0 00. ~ .N04 r.kV c0 W>

0 "44j= 0 00
0 -4 OX In
0.0 C 0J r 4 -4-' -N C 0 90
a~ 0
ON 0 c 0
=0 0-4 " 4M4CM00
0 00 0.0~ 3- 0 o u-
Ai -i 0vm00a . 0
r
00 J mf 40001 U.0 ",
40 En 2.0 ~ 4 -.. 0
0Cl
E- I0. 0) iO0 uri2.z 4>
0 0 4.) tr 4) t c I w .40 0 a
> '0 4)~
0 40)0 4)a 0
0 'O0 0 V .0.
0Z a00
f,0 1 Ic w 00 4 . 'o0 *( 10 4
0 ,I .M 1 .4'.01 C 4 0v> c .1.-
to 4 O0
V.- A4J 0 VX* 0 ' 0 05 4
m N>
1_ rr 0'.- 10 4J 11'40--
e* 4)n
0 w C 15E,0O .110--VO* 4'
00 V IA r40 31 0 4 0.C
0.0 0 004J0.404J0.01..
C- 1
,.c m C,. 0 N '4 X C0 . 0 .0
40 r4. 0 - O 4 ( M W.4 0 4
V $. .- 0 *V.0u 0 - r2 0 to
-0 4 N04 S.'0) .0 x00.'A 4 0 0.
4) z
00' ). 00J0 r;0 .4.. 0 0 '
-0 0 U V x z .00 > Q0 go
2.c 0'40. 0 2f . w00 -4U
L40 Z ~ '0 o0 01O4
co 2.D 0'40'.0 430
~00 0'.,-
(N0 a,'0
L.0
0
00,
V 4) t 40) 4
k 4
0 .0.0
4) 41
0.41
-40 O0 04) 0-.00 IS .0-
ZOc .00 to 04 LO)- Q)-
'4
WOO 00. r -0 0000 0 01.00
co t *..A -0 4' ..- >0C 0 04)4j0' to0
0 00K0 41 2
'4O .. tr w
E- ) .U EnA trxOO
4-4 -. 0 >.0'0 t 0 W4
4.4A( 0- m 444000004a) JV
ZtJ40 *- o4)
0a 0 3 . 0to*
0 0- m44.4 0to-4 U m 0'4V
0 r0.0 -4 L 0. u s c On 0> )
4.1 0 OW M u0a00 0 .40 0t
>0 r0 w 0. w $41 >0-0
0I-*4) m->
0 c rf V m0c- r-0 >.1Sm'0 u' a f
'0n U-0 - )a 01C;a0c N.00 401 0.a
N~- ~~~-
00 . 00 4 *1 0-0
0o 3t1..--r
is2~
w4 00
-0.
W 0 M4
4Z ) W0-- -0 U W 10 0 . m
2)o0
(n 4)4 .04)0'zc
* 0 .1 Q-..u0 W)4J010
c0 4 0 0 *0 0.>-mO0 0~*...
P-0
8 -1 u 0i w3. 3 00HW 0a3.C-'
REPORT DOCUMENTATION PAGE
1. Recipient's Reference 2. Originator's Reference 3. Further Reference 4. Security Classification
of Document
AGARD-LS- 182 ISBN 92-835-0683-9 UNCLASSIFIED
5. Originator Advisory Group for Aerospace Research and Development

North Atlantic Treaty Organization
7 rue Ancelle, 92200 Neuilly sur Seine, France
6. Title
FUNDAMENTALS AND SPECIAL PROBLEMS
OF SYNTHETIC APERTURE RADAR (SAR)
7. Presented on 5th-6th October 1992 in Bad Neuenahr, Germany, 8th-9th October 1992
in Gebze-Kocaeli (near Istanbul), Turkey and 26th-27th October 1992
in Ottawa, Canada
8. Author(s)/Editor(s) 9. Date
Various August 1992 a
10. Author's/Editor's Address 11. Pages

Various 198
12. Distribution Statement This document is distributed in accordance with AGARD

policies and regulations, which are outlined on the
back covers of all AGARD publications.
13. Keywords/Descriptors
Airborne radar Remote sensing

Algorithms Simulation
Digital techniques Spaceborne equipment
Inverse synthetic aperture radar Synthetic aperture antennas
Polarization (waves) Synthetic aperture radar
14. Abstract
The Lecture Series will cover the field of airborne and spaceborne SAR with respect to its
technical realisation in order to convey the participants' ideas and know-how on SAR, on its
capabilities and on the technology necessary for the successful construction and application of
airborne and spaceborne SAR systems.
The basic principles of SAR will be explained and SAR will be compared to airborne and
spaceborne radar with real aperture.
The influence of the antenna parameters on specification and capabilities of SAR and the
advantages, necessities and limits will be considered.
Digital SAR processing is indispensable for SAR. Theories and special algorithms will be given
along with basic processor configurations and different processing techniques on a hardware and
software basis.
The simulation of SAR-systems as well as SAR-products will also be a topic of the Lecture Series.
A presentation of the present state of the art, giving examples of presently planned and realised
airborne and spaceborne SAR with its foreseen applications will conclude the Lecture Series.
This Lecture Series, sponsored by the Avionics Panel of AGARD. has been implemented by the
Consultant and Exchange Programme.
,1. .. _ _ _
E' uu E , c
<0 <
AC
54 to
- CU .. .0
'a
-g !g .
j. ~2.
Cum
00 z 00
-u 00 M -2 -
<0 o .E -
ci < ua v 2 E r-
Wu cm >0U cm ' = u0
-0. 0 -1uI0
~CC
0-r r- cr
E- 1.I0
m E~
V3 C.,
13 -S~
~~ u0
-L do <~AU
0 -F -
5110 21
G.E E.~ ~~'4 0~0.5
"Do.iiUo%.0m )
W0
6 2 . .0 .
0--i
-a IiiAl
03Q ,I-0
G..E
0-0av
Cc ?000
1~E
ID- -0 0 0
rv-A
NATO OTAN -
7 RUE ANCELLE . 92200 NEUILLY.SUR-SEINE D39IStRrMONOF UNCLASSIFIED

FRANCE AOARD PUBLICATIONS'
Telephone (l)47.3d.S7.00 -Telex 610 176
Telefax (1)47.38.57." ___________________________
AGARI) does NO0T hold stocks of ACJARD publications at the above address for gehil dldtibutlon; Initial distribution or AGARD
publications is made to AGARD Member Nations t1rough the following National Dltrlbtlos Centres. Further copies are sometimes
avnlahle from these Centres (except in the United States), but if not may be purchased in Microfiche oir Photocopy form from the Sales
Agencies listed below.
NATIONALDIISTRIBUTICONrCENTRES
BELGIUM LUXEMBOURG
Coordonnateur AGARD - VSL See Belgium
Ftat-Major de la Force Afrlenne
Quartier Reine Elisabeth NETHERLANDS
Rue (l'Evere, 1140 Blruxelles Netherlands Delegation to AGARD
National Aernste I oinratory, NLR
CAN : -
Petee11n Flus paid

Spiace
AdmitlratlSjl" ~ ,
-~~ KASA451 .stablishment
E___
DN National Aeronautics and Me BUL1huflaUUMI
Space AdiiialnPensity for Private Use130
FRA 204 OKtor to AGARD
LA.
GRE
~tecllc ,ra rutux oatbc Ankara )

Dekelia, Athens TGA 1010
UNITED KINGDOM
ICELANDtoo ~iti Defence Research Information Centre
Direcor o A-itinKcrntigern House
c/o Flugrad 65 Brown Street
Reykjavik Glasgow G2 SEX
ITALY UNITED STATES
Aeronautica Militare National Aeronautics and Space Administration (NASA)
Ufficio del DelegatorNazionale alI'AGARD -Langley Research Ccnter
A.eroporto Pratica di Mvtc WS 180
00040 Pomezia (Roina) Hampton, Virginia 23665
THE UNITED STATES NATIONAL DISTRIBUTION CENTRE (NASA) DOES NOT HOLD
STOCKS OFAGARD PUBLICATIONS, AND APPLICATIONS FOR COPIES SHOULD BE MADE
DIREC-T 'TO THE,1NATIONAL TECHNICAL INFORMATION SERVICE (NTIS) AT THE ADDRESS BELOW.
SALES AGENCIES
National Technical E-SA/Information Retrieval Service The British Library
Information Service (NTIS) European Space Agency Dociument Supply Centre
5285 Port Royal Road 10, rue Mano Nikis Boston Spa, Wetherby
Springfield, Virginia 22161 75015 Paris West Yorkshire LS23 7110
United States France United Kingdom
Requests for microfiches or photocopies of AGARD documents (including requests to NTIS) should include the word 'AGARD'and the
AGARD serial number (for example AGARD-AG-31 5). Collateral information such as title and publication date is desirable. Note that
AGARD Reports and Advisory Reports should be specified mAGARD-R-rnn and AGARD-AR-nnn, respectively. Full bibliographical
references and abstracts of AGA"P publications are given in the following journals:
Scientific and Technical Aerospace Rprts (STAR) Government Reports Announcemtents and Index (GRA&I)
published by NASA Scientific and Technicalw published by the National Technical Information Service
infornmation Division - Sringfield
NASA Headquarter. (NM1') Virginia 22161
Washington D.C. 2056 United States
United States (aSO available online in tWe NTIS Bibliographic
Daabase or on CD-ROM)'
Printed bv Sr~cialised *ritn Services Limited

40 Chiwil LaAe, Lotlghto n, E.sseX 1GI 037Z
ISBN 92-835-0683-9

Sar1 Agar

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sar1 Agar

Uploaded by

Copyright:

Available Formats

AD-A255 6S7

ADVISORY GROUP FOR AEROSPACE~ RESEARCH 8t DEVELOPMENT

- - NORTH ATLANTIC TREATY ORGANIZATION

Published August 1992

ADVISORY GROUP FOR AEROSPACE RESEARCH & DEVELOPMENT

AGARD LECTURE SERIES 182

rrT(' QUALITY INSPMD 5

This material in this publication was assembled to support a Lecture Series

- _ North Atlantic Treaty Organization

- Exchange of scientific and technical information;

The content of this publication has been reproduced

Published August 1992

Lecture Series Director. Dr Wolfgang Keydel

SAR Peculiarities, Ambiguities and Constraints 2

Motion Errors and Compensation Possibilities 3

The Real Aperture Antenna of SAR, A Key Element for Performance 4

Polarization Effects and Multipolarization SAR 5

Inverse Synthetic Aperture Radar 9

Special SAR Techniques and Applications 10

Review of Spaceborne and Airborne SAR Systems I1I

BASIC PRINCIPLES OF SAR

SUMMARY overcome by the here considered Synthetic Aper-

The human eye, for example, is an optical sy-

Diffraction limiting laws of optics apply to .. L-, AM UAW

microwave remote sensing systems, too. Here,

I \ For AfD = BmIn and AT = Tmx results from (6) and

In connection with (9) results herefrom:

If a tracking filter will be used than a so-called

The required dwelltime for a specified azimuth re- 1 u2 t2

(14) To this oneway range difference corresponds a

3.2 DEPTH OF FOCUS

Under the assumption that a maximum phase error of

DOF = 2AR . (27)

The maximum allowed one way phase error caused by

Therefore, the complex signal relevant for the in-

0.8 Herein, the target is positioned at t = 0 (where fD

X. 2 " The received radar signal is of the kind of

t + Ti the observation time with the respective

For the dwelltime TD Results from (11) and (12):

Fig. 8 Signal intensity curve of the image in Fig.

Fig. 7a Signal to noise ratio over the whole scene in

Fig. 9 Distribution of reflection centers on a car

Under the assumption of statistical distributed

The phase is uniformly distributed

P(9) = for 0 ! 9 5 2n . (40b)

Mean value Vr and the second moment V2 of (40a) - M OIL

The 2=2 (41) -

The variance is by definition:

The variance represents the fluctuation of the re-

Sn, - = 3.66 ' 5.6 dB

This occurs, even in the absence of any other ad- VL = N 1 Vrl -

The time T during which a point target remains

The number Naz of independent samples in azimuth

Fig. 10 shows as an example the histogrammes of N,, 2 .

radar ry is given by:

If NY returns with that resolution will be averaged "PSLR:-S.O1dB

The equivalent pulselength Tp.q associated with NY

pq = N ,y . 0 iam acm aco

The total number of independent samples is given

N = Naz •Ny. (49)

Equation (48) shows as well as equation (47) in 0 ,o0 Tim (ma

This ratio is called radiometric resolution. For

Fig. 11 shows the processed radar answer of a point

Electronic _1 _