Introduction

Southern California precipitation, or lack thereof, occurs in extremity, resulting in multiyear droughts with sporadic years of heavy rainfall and nearly the highest year-to-year precipitation variability in the continental United States1. Owing to its lower mid-latitude setting on the eastern flank of the North Pacific High, heavy precipitation occurs relatively rarely, but can carry high intensities2,3,4, sometimes on par with those observed in climatologically wetter region of central California5. Climate change is projected to exacerbate this pre-existing hydroclimate volatility by reducing the frequency of low- to medium-intensity precipitation while increasing the intensity and contribution of extreme storms6. Above-average rainfall years are critical for replenishing valuable water supplies to meet growing agricultural and direct human consumption needs. Precipitation also helps to moderate wildfire risk7, a costly consequence of dry foliage and long dry periods. In 2018 alone during a severe drought year, California’s wildfires cost the US an estimated $148.5 billion8. The occurrence of extreme rain events and runoff can also have devastating effects across southern California in the form of flooding9,10,11, debris flow12, and contamination of coastal waters13,14.

The mean dry conditions in southern California result in local streams and rivers having little outflow during most of the year. When heavy rain events do occur, primarily during the winter season, outflow increases by orders of magnitude and plumes of freshwater propagate into the coastal ocean15. In addition, non-point source runoff delivers freshwater to coastal waters. Storm runoff is known to carry pollutants through watersheds to the ocean, reducing local water quality, increasing fecal indicator bacteria signals, and posing public health risks13,14. Runoff and river waters also carry other accumulated properties or quantities, including nutrients (important for biological productivity16,17) and sediment, important for coastal morphology and sand budgets18.

Climate indices often are used to characterize atmospheric and oceanic patterns favorable to wet or dry conditions over western North America19,20,21. In particular for the southern California region, El Niño Southern Oscillation (ENSO) usually brings drought-like conditions in La Niña phase and a higher probability of intense rainfall during strong El Niño years21. Other studies have considered the influence of the Pacific Decadal Oscillation (PDO) on winter rainfall in the study region22,23,24. Shukla et al. 2015 found that the negative PDO phase increases the probability of drought favorable conditions in southern California over the late winter months24. PDO modulates ENSO teleconnections, including precipitation25, and contributes to seasonal precipitation prediction skill in the study region26. Negative PDO and ENSO phases have been linked to increased fire activity in the southwestern US due to decreased precipitation27. These climate phases impact the hydrology in the study region due to the latitudinal shifts in the atmospheric pressure systems that are precipitation-bearing versus precipitation-blocking21.

In this study, we utilize sea surface salinity data to investigate precipitation and climate variability. Over a century of daily sea surface salinity measurements have been recorded at the Scripps Institution of Oceanography Pier. The standardized collection procedures28 provide confidence that the salinity measurements are consistent over time. This long record enables a unique 105-year investigation of hydroclimate variation affecting the Southern California coastal region. Annual-averages of the pier salinity record have been used to assess wind-driven upwelling29, but freshwater spikes in the daily record have not been well characterized. In this paper, we demonstrate the potential for using coastal salinity to infer runoff from heavy rain events over the wet winter season when fresh salinity spikes are most prevalent. Here we use a 105-year daily sea surface salinity record to show that El Niño Southern Oscillation and Pacific Decadal Oscillation both have signatures in coastal sea surface salinity even in a region which receives small total freshwater input (through either direct precipitation or runoff). Specifically, we find that the freshest salinity quantile is a good indicator of year-to-year variations in precipitation and runoff, highlighting the connection between the coastal ocean and the influence of storm activity, or lack thereof, in correspondence with known climate phase shifts.

Results and discussion

Precipitation influence on surface salinity

The daily SSS record from Scripps Pier (Fig. 1a) exhibits spike-like fresh events ranging from −5.6 to −0.3 psu (Practical Salinity Unit), contributing to a negative skewness (−2.84) about a mean value of 33.58 psu. A blowup of the record during January through March (JFM) for one year (2003 typifies average conditions) shows that the fresh SSS spikes align with winter rain events that last for 1–5 days (Fig. 1b). Within 1–3 days following each rain peak, salinity at the pier drops abruptly by several tenths psu, and slowly returns to near pre-event values after several weeks. Local streamflow exhibit a near identical pattern to that of the precipitation over the events, with the streamflow peaking within a day of the maximum rainfall (Fig. 1b). The SSS, precipitation, and streamflow measurements are recorded at locations within 20 km of each other (see map in Supplementary Fig. 1). To quantify this relationship over the longer record, we composite JFM SSS relative to streamflow anomalies, which confirms that, on average, SSS drops for weeks following sizeable outflow events (Fig. 2). Compounding effects lead to variations in the timing and amplitude of the SSS response relative to a given streamflow or precipitation level. This is evident in the 2003 record (Fig. 1b) as the largest rain event does not result in the lowest SSS response, but this largest SSS response is the third event in ~1 month span.

Fig. 1: Time series of local sea surface salinity, precipitation, and streamflow.
figure 1

a Daily sea surface salinity (SSS) measurements from surface water samples at the Scripps Pier, spanning 1916–2021. b The correspondence between daily SSS (black), San Diego Airport daily rainfall (blue, inverted y-axis), and Los Peñasquitos daily streamflow (orange bars, inverted y-axis) is illustrated for January–March of 2003 to highlight a typical winter season.

Full size image
Fig. 2: Sea surface salinity preceding and following streamflow events of three magnitudes.
figure 2

Sea surface salinity data shown 30 days prior and 30 days following Los Peñasquitos streamflow events of three categories, low flow <1 m3 s-1 (dotted line), medium flow 1–10 m3 s-1 (dashed line), and high flow >10 m3 s-1 (solid black line).

Full size image

Numerous factors influence ocean salinity in the study region. Coastal upwelling increases SSS by vertical transport of deep cold salty water to the surface and peaks in the April–July summer period30. Therefore, this work focusing on winter freshwater events does not investigate upwelling-induced changes. During rain events, atmospheric freshwater input lowers the salinity of the coastal ocean. An additional factor that directly alters surface salinity is horizontal advection. It is unspecified in existing literature whether offshore water mass advection variability influences the nearshore surface waters of the study region. The SSS drops seen at Scripps Pier are sudden and extreme, consistently corresponding in time with rain and streamflow spikes (Figs. 1b and 2). This suggests that the most practical mechanisms for these SSS drops are from precipitation directly, localized runoff during extreme events, and the precipitation-driven streamflow increases and resulting freshwater plumes that propagate into the coastal ocean. The connection between winter rain events and the SSS and hydrology of the nearshore coastal zone in the region has not been examined to our knowledge on yearly to interdecadal time scales and will be a focus of this study. Interannual streamflow variability in larger river systems due to precipitation has been shown to change coastal ocean dynamics31,32. Our work shows that even small, intermittent freshwater plumes have a robust impact on coastal ocean freshening in the study region.

River plumes of this relatively fresh stormwater infiltrate the salty ocean waters with a plume area dependent on the amount of rainfall33. The fate of the plume depends on the relative strength of the discharge and the waves. Plumes during relatively small discharge and large waves are trapped in the surfzone and travel along the coast in the direction of the surfzone wave-driven currents34,35,36, whereas intermediate to large discharge can escape the surfzone and travel along the coast depending upon the inner-shelf currents37,38. The direction plumes travel in the inner-shelf varies with wind speed and direction, river discharge inertia, and buoyancy15. River plumes over the shelf in southern California frequently travel southward along the coast due to the relative timing of regional precipitation and wind patterns15, which has implications for the study site located 7.5 km south of the Los Peñasquitos Lagoon and 12 km south of the San Dieguito River (see map in Supplementary Fig. 1). The interaction of freshwater discharge plumes with the surfzone and nearshore is a topic of active research. The SSS time delay relative to the peak rain/streamflow event suggests that the southward advection of local river plumes/estuary outflows contribute to the Scripps signal15. In addition, the SSS decay time likely reflects advective-diffusive time scales of the plume.

The fresh salinity events at Scripps Pier can occur throughout the year but are most prevalent during JFM, the three freshest months on average (Fig. 3). The JFM winter season also corresponds to seasonal highs in Los Peñasquitos streamflow and San Diego Airport rainfall. Approximately half the annual rainfall in the San Diego Airport record occurs during JFM, consistent with estimates for the larger San Diego region (Hydroclimate Report Water Year 2016, https://water.ca.gov). We focus the remainder of our analyses on the JFM winter season of high rainfall, highest streamflow, and lowest salinity (Fig. 3).

Fig. 3: Seasonal cycles of sea surface salinity, precipitation, and streamflow.
figure 3

Average monthly values of a SSS, b accumulated precipitation at San Diego Airport (NOAA NCEI), and c Los Peñasquitos streamflow (USGS). The red band highlights the wet winter JFM season.

Full size image

To represent the freshest winter salinity values, the 10% quantile of the JFM salinity anomaly (hereinafter referred to as SSS10) is computed each year after removing the JFM average for that year to de-emphasize background oceanic variability. Similar salinity values to SSS10 are obtained by subtracting the median quantile from the 10% quantile for each JFM season. The 10% quantile of the remaining anomaly array is the selected metric as it isolates the most extreme negative values for each year (see Supplementary Fig. 2 for explanation of quantile calculations).

The SSS10 time series captures year-to-year freshening in the JFM winter season. In years of low precipitation, SSS10 (Fig. 4a) remains relatively consistent at a near zero anomaly (0 to −0.2 psu). The strong correlations of SSS10 to precipitation and streamflow (Supplementary Fig. 2) further confirm that the 10% anomaly quantile characterizes the local response of SSS to rainfall. As a check on the representativeness and reliability of the Scripps SSS10 time series, we compare it to a similar estimate of the 10% quantile from the Newport Pier SSS measurements (Supplementary Fig. 3). A strong significant (p < 0.05) correspondence between the SSS10 time series of Scripps and Newport (r = 0.74) suggests that the rain-related coastal SSS signal is widespread (i.e., not just representative at this one location) and robust in the daily pier measurements (i.e., not subject to changes in observers or methodology).

Fig. 4: The SSS10 metric presented alongside winter precipitation and streamflow anomalies and relevant climate indices.
figure 4

a SSS10 (1917–2021). b JFM-averaged precipitation (blue, 1940–2021) and streamflow (orange, 1965–2021) anomalies with climatological monthly averages removed, shown with flipped y-axes. c 5-year smoothed JFM-averaged PDO index values. d JFM-averaged NINO3.4 index values.

Full size image

Together the SSS10, streamflow, and precipitation series highlight especially wet winters (SSS10 < −0.35) against a background of otherwise dry conditions (Fig. 4a, b). The SSS10 record particularly emphasizes consecutive dry years during predominantly negative PDO phases (Fig. 4a, c). A noteworthy feature in the SSS10 record is the negative PDO phase of 1944–1977 (Fig. 4c) when the precipitation and streamflow are weak and the SSS10 is nearly constant for 30+ years. During the two extended negative PDO phases, with the exception of 2005, the JFM-averaged precipitation anomaly remains under 40 mm. In addition, multi-year droughts in excess of 3 years appear to be less apparent in the first 25 years of the SSS10 record, prior to the 30-year long dry period. While dry winters tend to occur sequentially, one low SSS/wet winter is rarely followed by another. A potential source of this shorter-term variability in SSS10/precipitation is ENSO phase (Fig. 4d).

High, significant correlations between SSS10 and the JFM precipitation accumulation anomaly (r = −0.78) and Los Peñasquitos streamflow anomaly (r = −0.83) further support that SSS10 reflects rainfall and runoff variability (Fig. 4a, b). High correlations of SSS10 to regional precipitation (Fig. 5) and streamflow suggest SSS10 captures freshwater drainage of the greater coastal watershed (see watershed in Supplementary Fig. 1). The SSS signal tracks more closely with streamflow than with precipitation because of the delay in advection of streamflow to the coastal ocean. The probability of below average JFM daily streamflow is highest in the “saltiest” years (SSS10 > −0.2 psu) and lowest in the “freshest” years (SSS10 < −0.35 psu; Supplementary Fig. 4).

Fig. 5: Correlation of the SSS10 metric with US precipitation.
figure 5

Correlation coefficient contour map of SSS10 to Livneh gridded precipitation39 (1917–2011). Inset on the left shows southern California north of the US-Mexico border and south of San Luis Obispo. Colormap from ref. 54.

Full size image

The tails of the daily streamflow distributions (Supplementary Fig. 4) highlight differences in the number of large outflow events during fresh, moderately fresh, and salty years. In salty years, the probability of rain events causing sustained high streamflow is much lower than that of the freshest years. The fresh years that feature the largest SSS10 anomalies (Fig. 4a) correspond to generally higher chances of sustained intense streamflow (Supplementary Fig. 4). The distribution of streamflow highlights that heavy rain events occur less frequently in salty/dry years than fresh years.

To further explore the connection between Scripps Pier salinity and regional winter precipitation patterns, we examine correlations between SSS10 and the Livneh gridded precipitation product39 for 1917–2011. JFM-averaged precipitation exhibits statistically significant correlations (| r | > 0.20) with SSS10 over a broad swath of the southwestern US, with the highest correlations (up to r = −0.79) in the Southern California region (Fig. 5). The low SSS10/high rain winters appear to be indicative of large-scale precipitation patterns that affect a large part of the Southwest region.

Surface salinity and climate modes

We next consider the extent to which Pacific climate modes can provide context and a degree of predictability for the JFM precipitation and runoff captured by the SSS10 time series. For comparison with SSS10, the NINO3.4 and PDO indices are JFM-averaged. The strongest ENSO teleconnections, linking the tropical and midlatitude conditions, occur in JFM40. Correlations between SSS10 and PDO and ENSO indices are weak with r = −0.28 (p = 0.24), and r = −0.25 (p = 0.01), respectively. Despite the low correlation with the PDO, multi-year periods of high SSS10/low rain generally correspond to negative PDO phases (Fig. 4a, c).

Scatter plots confirm that while the linear correlations between SSS10 and indices of the PDO (Fig. 6a) and ENSO (Fig. 6b) are weak, there is a strong tendency for high SSS10/low rain winter seasons during negative values of both indices23,24,25,26. The case for the PDO is slightly less obvious for annual values, but with a 5-year smoothing to emphasize decadal polarity changes, the relationship is tighter (Fig. 6a).

Fig. 6: Division of SSS10 by relevant climate indices.
figure 6

Scatter of SSS10 (1917–2021) versus a 5-year smoothed JFM-averaged PDO index and b JFM-averaged NINO3.4 index. c Scatter of JFM-averaged NINO3.4 index with DJF-averaged TNH index, colored by the corresponding SSS10 for each year (1951–2021).

Full size image

JFM ENSO state provides a stronger constraint on year-to-year drought conditions than the PDO, with only one low SSS10 winter (in 1918) evident when the ENSO index dips below −0.2, and none during strong La Niña events when the JFM-averaged index falls below −1.0 (Fig. 6b). Thus, La Niña-like conditions strongly favor below-average rain levels for any given winter25, whereas multi-year drought periods tend to fall within prolonged negative PDO phases. In contrast, wet winters occur during positive PDO and ENSO phases25, but not consistently so as high SSS10/dry winters are just as likely, and high index values do not ensure low SSS10 levels (Fig. 6a, b). The wettest/lowest SSS10 winters almost exclusively occur during El Niño years, but not every El Niño resulted in a wet year. Low SSS10/wet winters (SSS10 < −0.35) occurred during 17.1% (12.8%) of positive PDO (El Niño) years, but only 1.6% (3.5%) of negative PDO (La Niña) winters. High SSS10/dry winters (SSS10 > −0.20) happened in 73.2% (76.6%) of positive PDO (El Niño) years and in 91.9% (91.4%) of negative PDO (La Niña) winters.

A recent study41 notes that the Tropical Northern Hemisphere pattern42 (TNH) is highly correlated with California precipitation variability, and suggests that a combination of ENSO and TNH indices (as opposed to ENSO alone) provides improved predictability of precipitation. TNH is a prominent winter climate mode that reflects large-scale changes in the extent and location of the Pacific jet stream42. Dry/wet years in southern California tend to occur during TNH positive/negative phases. A strong negative 500-hPa geopotential height anomaly is seen at TNH negative phases over the NE Pacific Ocean, and vice versa for positive phases43. Negative phase TNH conditions support cloud formation and precipitation whereas positive phase TNH conditions reduce the winter storm landfalling probability in California41. A scatter plot of SSS10 versus the NINO3.4 and TNH indices confirms these tendencies, with low SSS10 wet winters clustered primarily in the lower right-hand quadrant of Fig. 6c where El Niño and negative TNH conditions meet.

Composites of JFM-averaged surface wind and precipitable water anomalies illustrate the relationship between climate phase and southern California precipitation. During La Niña and negative PDO phases (Fig. 7a), the Aleutian low pressure system over the northeast Pacific weakens (clockwise wind anomaly) with negative precipitable water anomalies and average wind conditions over the western US. During El Niño and positive PDO phase wet years (Fig. 7c), the Aleutian low strengthens and deepens (anticlockwise wind anomaly) forcing the Northern Hemisphere Jetstream (NHJ) southward, resulting in increased storm landfalls on the US west coast and significant positive precipitable water anomalies44,45. During El Niño and positive PDO phase dry years (Fig. 7b), the southeastern extent of the NHJ is located more northwest when compared to rainy year conditions (Fig. 7c), pushing storm landfalls north and causing negligible to negative precipitable water anomalies in the southern California region.

Fig. 7: Composite atmospheric conditions for three combinations of climate modes and SSS10 wet vs. dry winter groupings.
figure 7

Composites of JFM-averaged anomalies of NCEP surface winds and precipitable water (1981–2021) for a La Niña & PDO negative phases with high SSS10/dry winters (n = 13 yrs), b El Niño & PDO positive phases with high SSS10/dry winters (n = 6 yrs), and c El Niño & PDO positive phases with low SSS10/wet winters (n = 4 yrs). Colormap from ref. 54.

Full size image

Implications

The long-running salinity record from Scripps Pier reflects regional winter precipitation and streamflow patterns influenced by both the PDO and ENSO. Over the past 70 years, the majority of low SSS10/wet winters have occurred during positive PDO phases, and more specifically when the NINO3.4 and TNH indices are positive and negative, respectively. The SSS10 time series especially highlights the occurrences of multi-year drought conditions over the past century. In particular, during the prolonged negative PDO phase between 1944 and 1977, SSS10 was persistently near normal (Fig. 4b), indicating a lack of heavy winter rain events.

The SSS10 observations suggest that a freshwater signal in the coastal surface ocean that persists for multiple days requires a major rain event or a few smaller rain events that result in substantial streamflow and runoff when averaged over the JFM season. Large impact rain events can come in the form of atmospheric rivers (ARs). AR events in the southern California region occur over a few days each year and can contribute up to half of the annual precipitation accumulation46. Based on the AR Rank database, the 1978 to 1999 positive PDO phase featured 50% more days with winter AR events in the southern California region compared to the 1944–1977 negative PDO phase (12 ± 1.3% (twice standard error) to 8 ± 1.1% respectively). AR events previously have been shown to contribute more to the total water year precipitation during PDO + and El Niño years in southern California1,46 when compared to other years. In general, we find that mean JFM precipitation provides a better correspondence with SSS10 than time series based on AR Rank, as expected, due to ARs providing only some of the precipitation to the region. This implies that not all ARs lead to wet winters, as measured by the coastal ocean’s SSS response, and the sources of precipitation that lead to low SSS10 winters are variable.

As the climate changes, current projections are uncertain regarding how the climate modes influencing the southern California region will be impacted47. The IPCC WGII Sixth Assessment Report47 states with moderate confidence that extreme El Niño and La Niña events may become more frequent, up to twice as often, with increased teleconnection responses over the remainder of this century. This would have a direct impact on the severity of dry winters and of the unpredictable wet winters in southern California. A consensus on future PDO conditions under climate change has not been reached47. The US west coast may experience a decrease in the frequency of non-AR related rain but an increase in extreme rain due to ARs specifically6.

The ocean salinity time series at Scripps Pier complements information provided by nearby streamflow data and reinforces the detection of interannual and decadal winter rainfall variations apparent in rain gauge data. The salinity record also provides a long-term context for assessing the impacts of extreme precipitation events on the coastal zone, even in a region of low total freshwater input, which has impacts on coastal pollution, sediment, nutrient delivery, and beyond. Results here provide a more nuanced view of the value of the coastal salinity record, in that event scale structure in the form of low SSS pulses (or lack of these) provides a gauge of the vital heavy precipitation events that dictate the status of precipitation and water resources in the southern California coastal region. Together the measurements highlight the strong influence that Pacific climate modes have on regional precipitation and coastal hydrography. The 105-year length of this salinity record provides a unique opportunity to isolate these decadal patterns in precipitation as it scales with the coastal ocean response to these events. Our findings show that salinity can be used to investigate the strong role that interannual to multi-decadal variation plays in driving hydroclimate conditions in this region – multi-decadal periods of relatively dry winters are common for the region and relief from drought periods may not occur until the PDO shifts to a positive state. From revealing climate signals not shown in salinity data previously, to bolstering proof that precipitation and resulting advection of streamflow waters are the leading causes in coastal ocean freshening events, the implications for this study increase the breadth of surface salinity records beyond the uses by oceanographers alone.

Methods and data

Salinity data

Since 1916, daily water samples have been collected at the end of the Scripps Institution of Oceanography (SIO) pier in La Jolla, CA. For the near-surface samples, salinity has been measured with two consistent methodologies over time, initially by titration (1916–1964) and now by inductive salinometer (1965- present). The current collection and measurement procedures are described in the Shore Stations Program Annual Reports (https://shorestations.ucsd.edu/). Here we analyze the year-to-year wet season variability using the daily sea surface salinity (SSS) record spanning 1916 to 2021 for a total of 38,330 days of data. We also examine similar daily salinity measurements collected at the Newport Beach Pier following the same methodology. Time series from both stations are part of the Shore Stations Program28.

Precipitation and streamflow

Daily and monthly precipitation accumulation data (1939–2021) for the San Diego International Airport (~16 km southeast of the Pier) were obtained from the National Oceanic and Atmospheric Administration’s National Center for Environmental Information (NOAA NCEI). The Livneh daily near-surface gridded precipitation data39 (1915–2011) spans the continental US in 1/16 degree resolution and is provided by the Physical Sciences Laboratory (PSL), NOAA.

Daily streamflow time series from Los Peñasquitos, the closest major stream to Scripps Pier, is obtained from the U.S. Geological Survey (1965–2021). The stream feeds into Los Peñasquitos Lagoon, which discharges into the ocean ~7.5 km north of the Scripps pier. Anomalies for Los Peñasquitos outflow and monthly precipitation accumulation are the JFM averages with JFM mean climatologies removed.

An Atmospheric River rank system categorizes the magnitude of landfalling ARs based on the maximum instantaneous vertically integrated water vapor transport (IVT) and storm duration48. This work utilizes a retrospective analysis project produced daily AR Rank dataset (1948–2020) for consideration of JFM AR events46.

Climate indices and atmospheric variables

El Niño-Southern Oscillation (ENSO) variability is characterized using the NINO3.4 SST index (1870 to present), obtained from the PSL, NOAA. NINO3.4 is calculated from the HadISST1’s SST averaged between 5°S-5°N and 170-120°W (ref. 49).

The Pacific Decadal Oscillation index (1854 to present), obtained from NOAA NCEI, is computed using the ERSST Version 5 extended reconstruction of sea surface temperature50. The PDO index is smoothed using a convolution filter with a 5-year Blackman window51 to emphasize decadal variability.

The Tropical/ Northern Hemisphere pattern42 (TNH, 1950 to present) index was provided by the NOAA NCEP Climate Prediction Center (CPC) and accessed through Columbia University International Research for Climate and Society Institute’s Climate Data Library. The TNH index is provided by the CPC with only December through February values.

For atmospheric composites (Fig. 7), surface wind (1948 to present) and precipitable water data (1981 to present) were obtained from NCEP/ NCAR Reanalysis 1 (ref. 52) and NCEP Daily Global Analyses, respectively. NOAA PSL provided access to both NCEP datasets. Precipitable water content over a given area denotes humidity, which is positively correlated to rainfall53.

All climate indices are averaged over the rainy season (JFM), with the exception of TNH which is DJF averaged, and all are referred to without mention of this averaging for the sake of brevity. For all correlations, significance is determined by a p-value < 0.05, assuming every data point is independent unless otherwise specified.