Impact of Extratropical Northeast Pacific SST on U.S. West Coast Precipitation

The rainfall over the U.S. West Coast is known to be highly influenced by large‐scale atmospheric circulation and tropical climate teleconnections. However, the role of North Pacific oceanic variability is less understood. Using high‐resolution regional atmospheric model simulations forced by sustained positive and negative phases of the extratropical Pacific Decadal Oscillation sea surface temperature anomalies (SSTa), we diagnose the precipitation changes over the U.S. West Coast during 2010–2020. We find that precipitation anomalies are up to 60% stronger (weaker) for the warm (cold) cases, especially over Northern and Central California during wintertime, and Baja California in the summertime. In both seasons, precipitation is predominantly modulated through changes in the water vapor flux, which are directed toward the coast in wintertime and away from the coast during summertime. These flux anomalies are primarily driven by large‐scale changes in the wind associated with the atmospheric adjustment to the strong ocean SSTa.

MHW (Bond et al., 2015;Di Lorenzo & Mantua, 2016), the strong 2015 El Nino (Jacox et al., 2016), and the 2019 Alaskan heatwave (Amaya, 2019).These extreme events have been linked to changes in landfall precipitation over the U.S. West Coast through atmospheric teleconnections and are connected to the dynamics of the large-scale climate modes such as the Pacific Decadal Oscillation (PDO) and the North Pacific Gyre Oscillation (NPGO) (see review by Di Lorenzo et al. (2023)).For example, during the multi-year 2013-2015 MHW, one of the worst droughts was observed over the U.S. West Coast, especially over California (Griffin & Anchukaitis, 2014;Seager et al., 2015).It is still unclear whether the extratropical ocean conditions contributed to the drought or if they were merely a symptom of the drought.
The wet season over the U.S. West Coast typically starts in October and ends in April (L.Dong et al., 2019;Swain et al., 2015).Rainfall during this time of the year mostly comes from extratropical storms in the North Pacific via the jet stream.The storm track position is influenced by SST changes along the western boundary of the North Pacific Ocean (Hoskins & Valdes, 1990;Kuwano-Yoshida & Minobe, 2017), as well as global scale atmospheric circulation.During summer, as the heat builds up on land, a high-pressure system forms over the U.S. southwest, bringing onshore winds from the Gulf of California with a strong moisture influx contributing to summer monsoonal precipitation (Kim et al., 2005), called the North American Monsoon (NAM).While monsoonal precipitation is strongest over Mexico, rain also falls over California (Mitchell et al., 2002).
In this study, we explore the potential influence of persistent, strong extratropical SST changes on precipitation over the U.S. West Coast.We selected the extratropical expression of the PDO pattern as a proxy for the sea surface temperature anomalies (SSTa) expression of MHWs over the Northeast Pacific based on the studies of Di Lorenzo and Mantua (2016) and Xu et al. (2021), which show how the dynamical evolution of MHWs in this region typically evolves into PDO-like pattern.While previous studies have explored the connection of the PDO dynamics onto US hydroclimate, here the PDO pattern is only used as a proxy for warmer SSTa that are used to explore the role of extratropical SSTa on hydroclimate variability that is independent of the tropical forcing.Moreover, there have been suggestions that the variance of PDO may be increasing under global warming (Di Lorenzo & Mantua, 2016;Joh & Di Lorenzo, 2017).The fundamental role of the atmosphere in driving the PDO has previously been clearly identified (Newman et al., 2016;Schneider & Cornuelle, 2005).Few studies, however, have addressed the potential feedback role of the corresponding persistent extratropical ocean anomalies on changes in precipitation over land.We investigate the role of the Northeast Pacific SST on precipitation over the U.S. West Coast using regional atmospheric model simulations designed to isolate the potential impact of large-scale, persistent, strong MHW, here represented by the PDO pattern as a prototype for assessing sensitivities.We aim to identify the spatial patterns of changes in precipitation over land associated with the warm and cold phases of the PDO, and the mechanisms forcing those changes.

Data Sets
Observations of precipitation are monthly mean computed from the Climate Prediction Center (CPC) Unified Gauge-Based Analysis of Daily Precipitation data set [0.5 × 0.5 horizontal grid] of the National Oceanic and Atmospheric Administration (NOAA) CPC product (M.Chen et al., 2008).This data set goes from 1941 to 2018.Wind data for model validation comes from the Coupled Ocean Atmosphere Mesoscale Prediction System Wind at 10 m, 4 km resolution (Doyle et al., 2009).The PDO was reconstructed using the PDO index from NOAA (Mantua & Hare, 2002), as well as the sea-surface temperature (SST) data from the Hadley Center Sea Ice and Sea Surface Temperature data set (Rayner, 2003).We used the ERA5 reanalysis product (Hersbach et al., 2020) with a 30 km horizontal grid resolution for the boundary conditions for the model experiments.The variables used at 6 hr time steps were zonal and meridional winds, temperature, relative, specific humidity, and geopotential at 30 levels from 1,000 to 50 hPa, as well as surface winds, soil, water and 2 m air temperature, and pressure.

Model Description
We carried out a set of continuous 10-years simulations using the Weather Research and Forecasting (WRF, Skamarock et al. (2021)) to assess the impact of the Northeast Pacific SST on U.S. precipitation.WRF is a state-of-the-art mesoscale numerical weather prediction model designed for both atmospheric research and weather forecasting (https://www.mmm.ucar.edu/models/wrf).It can resolve small scale and mesoscale dynamics, with several choices of physical parameterizations including microphysics, cumulus parameterization, planetary boundary layer (PBL), land-surface models (LSM), and longwave and shortwave radiation.For this study, the Thompson parameterization (Thompson et al., 2008) was chosen for microphysics, the rapid radiative transfer model scheme (Mlawer et al., 1997) for longwave radiation, the Dudhia scheme (Dudhia, 1989) for shortwave radiation, Noah LSM scheme (F.Chen & Dudhia, 2001) for land-surface, the Kain-Fritsch scheme (Kain, 2004) for the cumulus parameterization, and the Yonsei University (YSU) scheme (Hong et al., 2006) for the PBL parameterization.
The atmospheric domain comprises the whole Northeast Pacific Ocean, from −180° to −110° longitude and 5° to 67° latitude (Figure 1a), including Hawaii, the southern part of Alaska, and about half of the North American continent.The horizontal spatial resolution is 1/4°, which corresponds to 25 km around the equator.We used 44 vertical levels from the surface to 50 hPa.The basic state initial conditions and surface and lateral boundary conditions were interpolated from the ERA5 global reanalysis data set with a 6-hourly time interval.All simulations were carried out for the period 2010-2020, with daily outputs from which monthly means were computed.

Control Run
An 8-member ensemble was generated using WRF for the basic state (observed SST) conditions.Each member has slightly different initial conditions to assess potential nonlinearities in the internal variability of the regional system.The first timestep of every January months of the control run was used for the initial SST map.For example, 1st January 2011, was the initial SST map of the second ensemble member, and the year 2012 for the third ensemble member, and so on.The resulting ensemble spread was very small, indicating that the boundary conditions largely controlled the atmospheric flows in the domain for fixed SST.Since the boundary controls were so strong, the ensemble spread of the control was assumed to represent the ensemble spread of the SST anomaly runs and was used to assess the statistical significance of differences between cases.

Validation
The model output was validated by comparing several variables, including temperature, zonal and meridional winds, and precipitation, with observations and reanalysis products at sea level pressure, 500 and 200 hPa (see Figures S1-S4 in Supporting Information S1).In general, our model data was in good agreement with the observations.Moreover, the model spread is relatively small among the ensemble members, which tells us that the boundary conditions dominate the internal variability in our domain.The results were deemed sufficiently realistic to proceed with the perturbation SST runs.

Sensitivity Tests
We conducted two separate SST anomaly experiments, one by superimposing a warm phase PDO spatial anomalies onto the prescribed time dependent SSTa surface boundary conditions and another one using the cold phase (Figures 1a and 1b) for the same time period as the control run.The PDO anomaly pattern was reconstructed by creating a composite SST map of when the PDO index amplitude exceeded ±1 standard deviation.The maximum amplitudes of our warm/cool patterns were then scaled to ±2C to simulate the occurrence of extremes, which typically of this order in the Northeast Pacific.Note that this technique does not produce perfectly symmetric PDO warm (PDOw) and PDO cold (PDOc) patterns due to the retention of random SST variations during the construction of the composites.The SST patterns of the PDO are shown in Figures 1a and 1b and indicate strong similarity with the PDO EOF pattern (Mantua & Hare, 2002).We refer to those two experiments as PDOw and PDOc.

Analysis
Using the precipitation model output, we compared monthly averaged data with observation to first evaluate the accuracy of the data set.We computed the monthly climatology to find the months where the warm and cold experiments would exceed the ensemble spread.The months of March and September showed the strongest responses in precipitation anomaly.The anomalies for the PDOw and PDOc experiments are calculated with respect to the ensemble mean.To determine the mechanisms by which precipitation over land was influenced by the SST, we looked at changes in the relation between anomalies of precipitation, winds, moisture content, and integrated vapor transport (IVT) where IVT indicates the direction and amplitude of the water vapor flux over a region: For the analysis, we separated the IVT anomalies into a velocity (U) perturbation term and a water vapor (Q) perturbation term: The terms on the right side of the equation are the water vapor anomalies advected by the mean flow, the wind anomalies acting on the mean water vapor distribution, the eddy fluxes associated with wind anomalies acting on water vapor anomalies, and the mean eddy fluxes.Differences in variables between cases were deemed to be statistically significant if they exceeded 2 standard deviations of the ensemble spread.

Statistical Testing
We applied a significance test for correlations and regressions in this paper, based on the Probability Density Function (PDF) method.The PDF was constructed using the Monte Carlo method, using 1,000 pairs of random red noise time series with a decorrelation length of 3 months, which corresponds to that of the data.

Simulated and Observed Precipitation Over North America
The 2010-2020 model precipitation is compared with observations in Figures 1c, 1d, and 1e, averaged over three key climate regions of California: the North Coast, the Central Coast and the California Desert.The modeled precipitation timeseries compare favorably with the observations with significant correlation above the 95% level (see Figure 1).It is also important to note that precipitation is not a state variable, but rather a prognostic variable that is dependent on the choice of physics parameterizations.
Inspection of the timeseries reveals that the model is able to reproduce the seasonal cycle, with precipitation peaking during wintertime and nearing zero during summertime.However, the model produces less rainfall than observed as evident when comparing the ensemble spread (gray lines) with the observational data (red line) in Figure 1.The years 2013-2015 show reduced precipitation in both observations and the model, consistent with the multi-year 2013-2015 North Pacific MHW during which atmospheric conditions tremendously reduced precipitation over California.Large differences in model versus observed precipitation are sometimes found during the wet seasons, particularly in 2017.Given that the model captures the precipitation seasonal cycle over key regions of the U.S. West Coast, we assume that the model has sufficient verisimilitude to be used for further analysis on how extratropical SST warming impacts coastal precipitation.

Sensitivity of Precipitation to SST Warming and Cooling
The sensitivity of coastal precipitation to the persistent, strong, warming patterns for PDOw and PDOc are shown in Figure 2, for three different regions in California over the period 2010-2020 (left panel).The seasonal cycle (right panel) was computed by averaging the 11 years for each month.We find that the PDOw experiment is wetter and the PDOc experiment is drier over the course of the runs.The wetter and drier conditions are significant when the red (PDOw) and blue (PDOc) curves are outside the gray envelope of the ensemble spread.Precipitation differences during the wintertime are greatest during the month of March for all three regions.The late summer precipitation is greatly enhanced in the California Desert region for the PDOw anomalies, hinting that warming along the coast impacts the monsoon precipitation.Summer precipitation is barely impacted in the North and Central Coasts.The mechanisms behind those changes are investigated in the next section.For the analysis, we focus on the months of March and September to study the contrast between winter and summer precipitation, as these months were the most significantly impacted.

Mechanisms of Precipitation Changes
To understand the mechanism behind the changes in precipitation, we first examine anomalies of precipitation, winds, geopotential height, and IVT.Given that the month of March was found to be the most impacted in terms of precipitation anomalies in the wintertime, we examine its precipitation spatial anomalies pattern over the Northeast Pacific (Figures 3a and 3d).We find that the anomalies in March resemble the SST anomaly patterns of the PDOw and PDOc used in the sensitivity run (Figures 1a and 1b).Specifically, drier conditions are found over regions of colder SST anomalies, and wetter conditions are found over regions of warmer SST anomalies.
For the PDOw experiment, the strongest winter signal of increased precipitation is centered off the U.S. West Coast and over northern California.This area underlies a 2,000 km-scale cyclonic flow in the 500 mb winds, also present when looking at the 500 mb geopotential height (Figure 3b).This is consistent with the warm SST in that area producing rising air and cyclonic circulation.We chose to show the winds at 500 mb since they are equivalent barotropic to at least 500 mb.For PDOc, the strongest signal of decreased precipitation extends from over California far offshore into the subtropical gyre.This area underlies the southern flanks of a 2,000 km-scale anticyclonic flow in 500 mb winds, which is also represented by the 500 mb geopotential height (Figure 3e).This flow is consistent with the cool SST in that area producing sinking air and anticyclonic circulation.March composite precipitation anomalies range from −1.1 to 1.9 mm for the PDOw experiment, and from −1.8 to 1.6 mm for the PDOc experiment.The strongest anomalies over land are found over northern California in both cases where changes in onshore and offshore wind directions associated with the atmospheric circulation changes appear to be critical in determining the sign of the precipitation response.Southern California is only slightly impacted.The local 500 mb wind fields indicate anomalous onshore flow for PDOw and anomalous offshore flow for PDOc.
To further diagnose the relative importance of the processes contributing to precipitation, we separated the mean water vapor flux into three terms QU′, Q′U, and Q′U′ (see Figures S7-S10 in Supporting Information S1), and found that QU′ (Figures 3c and 3f) dominates the other terms, which have magnitudes of about 2-3 times smaller.The local surface IVT anomalies over California (Figures 3c and 3f) indicate that there are onshore moisture flux anomalies for PDOw and offshore moisture flux anomalies for PDOc.The IVT anomalies impacting California are dominated by changes in anomalous wind, and only weakly influenced by the anomalous water vapor.Thus, the mechanism controlling the enhanced rainfall for the PDOw case is enhanced water vapor flux brought by the anomalous wind driven by the large-scale cyclonic flow over the Northeast Pacific (and vice versa for PDOc), During summertime (Figure 4), precipitation anomalies are surprisingly found to be much stronger than in winter, ranging from −1.6 to 12.9 mm for the PDOw, and from −7.8 to 2.5 mm for the PDOc.The important precipitation anomalies over land are found over southern California and Baja California in Mexico, which is where the NAM region is located.A closer look at the  Q ′ term over the U.S. Southwest demonstrates the enhanced moisture flux is mainly influenced by the anomalous wind which is onshore for PDOw and offshore for PDOc.The PDOc anomalous wind is clearly a result of the large-scale anticyclonic pattern brought by colder SST over the Northeast Pacific, which drives water vapor flux away from Southern California (Figure 4f).This explains the strong decrease in rainfall over southern California during summertime.

Summary and Discussion
This study assesses the sensitivity of the U.S. West Coast precipitation to changes in North Pacific SST.We used a modeling approach in which two different scenarios were compared to an 8-members ensemble of a regional atmospheric model.For the two sensitivity simulations, we superimposed a PDO-like pattern of SST to the pre-existing temperatures, one positive and one negative.We ran the model for the period 2010-2020 and looked at a climatology of precipitation anomalies.These model results demonstrate that there is a potentially significant contribution to precipitation over land from strong, persistent, large-scale midlatitude SST anomalies of the North Pacific.
Both experiments showed that the state of California was the most significantly impacted by the warming patterns; the north in winter and the south in summer.These differences in location can be explained through the different position of the large-scale pressure system anomalies created by the warming/cooling patterns.
During wintertime, studies found that precipitation along the coast is sensitive to changes in SST and latent heat fluxes (Bartusek et al., 2021;X. Chen & Leung, 2020;Persson et al., 2005).However, our finding suggests that this is not the main mechanism by which precipitation over California is impacted by large-scale North Pacific SSTa.We partitioned the thermodynamic and dynamic contributions of the water vapor flux anomalies and found that wind anomalies are the dominant driver for the observed changes in precipitation.The warm experiment demonstrated that by increasing SST along the coast and decreasing SST in the center of the Northeast Pacific, this creates a large-scale cyclonic flow barotropic to at least 500 mb.Those wind anomalies thus enhance water vapor flux over northern California, causing an increase in precipitation.In contrast, the cold experiment demonstrates how colder SSTs along the U.S. West Coast and warmer in the center of the Northeast Pacific decreases the amount of moisture transported over land, thus decreasing precipitation.This change in precipitation is due to wind anomalies from a large-scale anticyclonic flow over the North Pacific caused by the perturbation in SSTs.
Summertime precipitation anomalies over land are strongest over the southwest U.S., and much larger than the winter anomalies.Composite analysis reveals that in both the warm and cold experiment, once again, the anomalous wind is the dominant factor for the large increase in precipitation over land.This dynamical adjustment also happens for the cold experiment in summertime, in the opposite direction.IVT anomalies are large and directed away from the shore, thus greatly reducing precipitation over southern California.
The Clausius-Clapeyron relation, which relates the increase in water vapor with increasing air temperature, would predict a thermodynamic adjustment.It does play a role, but both wintertime and summertime precipitation anomalies seem to be primarily modulated through a dynamical adjustment brought by a large-scale alteration in SSTs in the Northeast Pacific which changes the large-scale atmospheric circulation.In both seasons, warming along the coast caused more landfall precipitation.Pascale et al. (2017) investigated how projected changes in atmospheric warming due to increased levels of CO 2 would impact the North American Monsoonal system.They found that Monsoonal precipitation is amplified by SST warming patterns, associated with decreased atmospheric stability.This is consistent with our findings, as we found that a dynamical adjustment happens when warming the coast, which effectively increases precipitation over the Monsoon-impacted region, notably over the California Desert.Moreover, previous work with coarser coupled climate modeling explore the relation between SSTa in the California Current upwelling system and the hydroclimate over North America during the Pliocene, and found that warmer regional ocean SSTa lead to wetter conditions over land (Fu et al., 2022).
Finally, it is important to recognize that warming or cooling the whole Northeast Pacific with a PDO-like pattern, and thus adding infinite heat capacity for this whole period, is unrealistic.However, this study does not aim at linking the specifics of the PDO to precipitation anomalies, but rather to give an insight on how extratropical SST anomalies can potentially have an impact on precipitation over the U.S. West Coast.Given that midlatitude air-sea coupling could be a driver for changes in precipitation over California, this motivates additional research on possible impacts on coastal precipitation associated with extreme warming events in the Northeast Pacific, which often have oceanic patterns that resemble the expression of the climate modes such as PDO and the North Pacific Gyre Oscillation (NPGO).

Figure 1 .
Figure 1.PDO-shaped warm (a) and cold (b) sea surface anomalies patterns (a), (b) used for the experiment.Precipitation during 2010-2020 (c, d, e) of the model ensemble output (black line) and observations (red line) for three key California regions.The gray shading represents the ensemble spread.Each region is indicated by a box in (a).

Figure 2 .
Figure 2. Time series of precipitation for the control (black line and shading), the warm (red) and cold (blue) experiments, in the North Coast, Central Coast and California Desert (locations indicated in boxes on Figure 1).The left panels (a, b, c) show the full time series during 2010-2020, and the right panels (d, e, f) show the corresponding seasonal cycle.