Upstream Precursors to High-Resolution Modeled Extreme Precipitation Events in the Mountainous Regions of Southern California

Erica F. De Biasio aScripps Institution of Oceanography, University of California, San Diego, La Jolla, California

Search for other papers by Erica F. De Biasio in
Current site
Google Scholar
PubMed
Close
and
Konstantine P. Georgakakos aScripps Institution of Oceanography, University of California, San Diego, La Jolla, California
bHydrologic Research Center, San Diego, California

Search for other papers by Konstantine P. Georgakakos in
Current site
Google Scholar
PubMed
Close
Open access

Abstract

The enhancement of precipitation over the mountain regions of Southern California, in conjunction with mesoscale and synoptic-scale forcings, can result in high-intensity, short-duration extreme precipitation events (EPEs) that are often associated with hazardous impacts. In this study, candidate upstream atmospheric precursors at relevant spatiotemporal scales to such hazards are explored using a WRF mesoscale model with 5-km grid spacing and an hourly temporal resolution. This high-resolution model, once validated against observations, is used to discern statistically significant physics-based signals between hypothetical mesoscale forcings and the modeled precipitation response. Specifically, the role of upstream instability in modeled EPEs is indexed by convective available potential energy (CAPE) and is examined for two mountainous regions of Southern California at several lag times. A Monte Carlo approach reveals statistically significant differences between the relationship of CAPE associated with EPEs in comparison to the analogous relationship for any precipitation event. These findings hold even with accounting for the well-established role of favorably oriented low-level moisture flux in orographic precipitation. This could indicate that atmospheric instability plays a role in providing additional lifting mechanisms, in conjunction with orographic and synoptic-scale forcings, to facilitate the high short-duration precipitation intensities that have been observed in the region. This diagnostic exploratory study provides additional candidate indicators of predictability of such EPEs at higher spatiotemporal scales than previous work, based on mesoscale model physics. Further analysis should examine the identified precursors using observational data with adequate resolution.

© 2023 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Erica De Biasio, edebiasi@ucsd.edu

Abstract

The enhancement of precipitation over the mountain regions of Southern California, in conjunction with mesoscale and synoptic-scale forcings, can result in high-intensity, short-duration extreme precipitation events (EPEs) that are often associated with hazardous impacts. In this study, candidate upstream atmospheric precursors at relevant spatiotemporal scales to such hazards are explored using a WRF mesoscale model with 5-km grid spacing and an hourly temporal resolution. This high-resolution model, once validated against observations, is used to discern statistically significant physics-based signals between hypothetical mesoscale forcings and the modeled precipitation response. Specifically, the role of upstream instability in modeled EPEs is indexed by convective available potential energy (CAPE) and is examined for two mountainous regions of Southern California at several lag times. A Monte Carlo approach reveals statistically significant differences between the relationship of CAPE associated with EPEs in comparison to the analogous relationship for any precipitation event. These findings hold even with accounting for the well-established role of favorably oriented low-level moisture flux in orographic precipitation. This could indicate that atmospheric instability plays a role in providing additional lifting mechanisms, in conjunction with orographic and synoptic-scale forcings, to facilitate the high short-duration precipitation intensities that have been observed in the region. This diagnostic exploratory study provides additional candidate indicators of predictability of such EPEs at higher spatiotemporal scales than previous work, based on mesoscale model physics. Further analysis should examine the identified precursors using observational data with adequate resolution.

© 2023 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Erica De Biasio, edebiasi@ucsd.edu

1. Introduction

The hydroclimate of California is often discussed in terms of the extremes that are characteristic of the region, involving either an abundance or lack of precipitation (Seager et al. 2015; Kam et al. 2019; California Department of Water Resources 2021). It is well established that the largest contributions to total precipitation in California come from a small number of precipitation events with amounts in the high tail of the overall probability distribution, and which occur during a relatively short wet season (Mock 1996; Dettinger 2016). This precipitation climatology driven by extremes makes a few large storms that impact the region each year critical in determining water availability, but also creates the potential for unwanted and hazardous impacts, namely, flash floods and landslides, floods, as well as drought and forest fires (Dettinger et al. 2011; Oakley et al. 2017; Ralph et al. 2006; Seager et al. 2015). Compared to Northern California and the Pacific Northwest (PNW), the Southern California region tends to experience shorter-duration, higher-intensity extreme precipitation events (EPEs), which are more likely to create flash floods and landslides (Oakley et al. 2017; Lamjiri et al. 2018).

Although there are undoubtedly similarities in the forcings of EPEs for Southern California and the aforementioned northern regions, these differences in precipitation characteristics allude to the potential involvement of additional driving mechanisms, perhaps acting at finer spatiotemporal scales (Cannon et al. 2018). The steep slopes and shallow soils of the Southern California mountain ranges further increase the likelihood of flash flooding and landslides once an EPE of this nature does occur (Modrick and Georgakakos 2015; Oakley et al. 2017). Because these types of land surface impacts can transpire quickly and often result in detrimental societal impacts such as property damage, injury, or loss of life (e.g., Han and Sharif 2021; Kean et al. 2019; Froude and Petley 2018; Calianno et al. 2013), identifying additional atmospheric precursors to EPEs at relevant spatiotemporal scales could be highly valuable for use in disaster mitigation and response planning.

To date, there have been only a handful of studies in Southern California that investigate the atmospheric conditions associated with EPEs at the synoptic to mesoscale and finer scales (Cannon et al. 2018; Oakley et al. 2018b, 2017; de Orla-Barile et al. 2022). Although the studies listed are significant in providing a first step in understanding the processes contributing to EPEs in Southern California, there is much that has yet to be considered. For instance, there has been work in other Mediterranean-climate regions that includes a detailed analysis of the atmospheric precursors to EPEs on synoptic to mesoscale domains (e.g., Georgakakos et al. 2014; Warner et al. 2012; Ricard et al. 2012; Viale and Nuñez 2011; Aragon et al. 2020; Sierks et al. 2020). This type of study constitutes a critical step in exploring potential predictability indicators on impact scales but is lacking for the Southern California region. The associated atmospheric conditions of EPEs have also typically been determined using global or regional climate models with coarse spatiotemporal resolutions, while the EPEs themselves are often defined using various observational precipitation datasets. The recent increases in computational capacity have allowed for the development of “convection-permitting models,” with kilometer-scale spatial resolutions that allow for explicit convection and are able to represent relatively fine-scale topographic features (Prein et al. 2015). Mesoscale models with fine enough spatial resolutions, for which parameterization of convection can be “switched off,” have been shown to provide improvements to the representation of precipitation. These improvements are particularly significant in regions of complex orography, such as Southern California (Prein et al. 2015).

The more realistic representation of modeled precipitation facilitates an analysis fully within a modeling framework, allowing for the atmospheric forcing to be directly linked to the precipitation response. These models are not without their own biases, and often show overestimation of precipitation in mountainous regions. Such overestimations are difficult to fully quantify due to the lack of extensive ground truth and frequent undercatchment of rain gauges in these topographically complex regions and hence must be evaluated with some caution (Lucas-Picher et al. 2021; Vergara-Temprado et al. 2020; Lundquist et al. 2019). High-resolution models have been used for such studies in other regions of the world (Khodayar et al. 2018; Chan et al. 2018). This type of modeling study can provide valuable information on the theoretical relationships that may be expected when moving to an analysis of “real world” observational data, which are typically sparse in space and time. With respect to temporal resolution, EPEs in these studies have also mostly been defined on longer time scales, typically from 1 to 3 days or greater, that are less relevant when considering those precipitation events capable of triggering flash floods and landslides (e.g., Cannon et al. 2018; Oakley et al. 2018b; Warner et al. 2012; Carpenter et al. 2007).

A common thread that emerges from previous work on the topic is an allusion to convective or mesoscale processes as potentially important components of Southern California EPEs. The previous analyses that do consider stability in the region indicate predominantly moist neutral or weakly unstable conditions with low (but positive) convective available potential energy (CAPE) (Cannon et al. 2018; Oakley et al. 2017). However, these analyses focus on small subsets of EPEs and use coarsely resolved datasets in determining associated atmospheric conditions. For the latter condition, this might imply that there exists a relatively low threshold of instability which, along with the addition of orographic or synoptic-scale forcings for ascent beyond the level of free convection, results in enhanced precipitation (Kirshbaum et al. 2018). Several additional studies have evaluated the upstream stability associated with EPEs in other regions, utilizing indicators such as the vertical profile of equivalent potential temperature (i.e., convective instability), CAPE, and the moist static energy (Lin et al. 2001; Chan et al. 2018; Champion et al. 2019; Ramezani Ziarani et al. 2019).

Low-level moisture flux is considered to be an important ingredient of EPEs in regions of complex topography due to the potential for orographically enhanced precipitation, particularly when impinging flow is roughly orthogonal to a feature (Roe 2005; Houze 2012). For EPEs along the U.S. West Coast (USWC) as a whole, there has been a large focus on the role of favorably oriented low-level moisture flux (Pandey et al. 1999; Bracken et al. 2015; Gimeno et al. 2016), commonly through the lens of atmospheric rivers (ARs) (Hecht and Cordeira 2017; Guirguis et al. 2019). ARs (as defined in Ralph and Dettinger 2011) have been found to be associated with a large proportion of both extreme and nonextreme precipitation along the USWC, but this relative contribution is somewhat reduced in Southern California (Lamjiri et al. 2017; Kim et al. 2013; Harris and Carvalho 2018; Rutz et al. 2014). Perhaps because of this, those studies that more granularly assess the role of ARs in EPEs tend to focus on events in Northern California and the PNW (e.g., Ralph et al. 2006; Warner et al. 2012; Hecht and Cordeira 2017). Nevertheless, favorably oriented low-level moisture flux is undoubtedly an important ingredient in Southern California EPEs, even if the criteria of AR conditions are less frequently met, so they are also being examined as candidate precursors in this analysis.

Other examinations of the drivers of both nonextreme and extreme precipitation in Southern California focus on large-scale processes and internal atmospheric variability at a coarse regional level. A consistent precipitation response to ENSO has been found along the USWC, with enhanced (suppressed) precipitation in Southern California and suppressed (enhanced) precipitation in the PNW during an El Niño (La Niña) event (Cayan et al. 1999; Jong et al. 2016; Guirguis et al. 2020; Ryoo et al. 2013; Ralph et al. 2006). However, this response is sensitive to the strength and structure of the sea surface temperature anomalies, and hence can vary from one event to the next (Patricola et al. 2020; Lee et al. 2018; Hoell et al. 2016). More recently, work has also gone into studying the precipitation response to several other climate modes, such as the North American Oscillation, Arctic Oscillation, and Pacific decadal oscillation (Guirguis et al. 2020; Brown and Comrie 2004; Guirguis et al. 2019), as well as the combined relationship of these climate modes and location of landfalling ARs (Guirguis et al. 2018; Xiong and Ren 2021; Kim et al. 2019; Guirguis et al. 2019; Harris and Carvalho 2018).

The present study undertakes a comprehensive regional analysis of the mesoscale to synoptic-scale atmospheric precursors to short-duration EPEs in Southern California, working within a modeling context to establish the existence of a physics-based signal between modeled candidate upstream precursors and the extreme precipitation response. Prior to analysis, model simulated precipitation is compared with a variety of observations to ensure that the precipitation features and event-scale precipitation distributions are sufficiently represented. The WRF Model is used to explore near-coastal upstream mesoscale precursors to 6-hourly wet season (October–April) EPEs in the mountainous regions of Southern California, with an emphasis on metrics of stability. For the Southern California region, metrics of stability have not been thoroughly investigated in the context of short-duration extreme orographic precipitation. The mountain ranges of Southern California differ from mountainous regions to the north in their orientation, slope, narrowness, and proximity to the coast. Therefore, it is important to have a comprehensive region-specific study that assesses the impacts of stability on EPEs in the Southern California mountainous regions. This 6-hourly event duration, defined only over mountainous areas with steeply sloping and windward-facing terrain, designates EPEs with short, high-precipitation intensities that are more likely to be associated with the adverse land surface and societal impacts most common in the region. The high spatiotemporal resolution of the WRF Model used in this study is a significant improvement over those of similar past studies in the region, allowing for an examination of orographic precipitation processes on finer spatiotemporal scales.

2. Data

This study uses both model-based and observational data. To ensure that the WRF Model adequately represents the precipitation distribution of the region, data from the WRF Model are first compared with a variety of surface observational datasets based on rain gauge observations.

a. WRF regional model data

The Department of Hydrology and Atmospheric Sciences at the University of Arizona provided the data used in this analysis (Risanto et al. 2023; Luong et al. 2018). This dataset is not discussed in a previous publication, but the WRF Model uses a similar modeling framework as discussed in Risanto et al. (2023), as well as a similar spatial resolution. The WRF Model domain (Fig. 1a) was selected to study areas of the southwestern United States that are impacted by the North American monsoon, but also covers the Southern California region. The model uses 28 vertical levels, an hourly temporal resolution, and a two-way nested domain with the inner domain having a 5-km spatial resolution. As this inner domain has a kilometer-scale spatial resolution, cumulus parameterization is turned off for this grid. A spatial resolution of 5 km is just above the typically defined “convective permitting” resolution of 1–4 km, which may introduce some deficiencies in adequately capturing all convective processes. Explicit convection at coarser resolutions than the typical 4 km convective-permitting threshold has been shown to improve the skill in modeling some aspects of precipitation, particularly in topographically complex regions, with similar representation of the seasonal mean climatology when compared to simulations with convective parameterization (Vergara-Temprado et al. 2020). Additional physical models and parameterizations used in the WRF Modeling framework include the Mellor–Yamada–Janjić TKE scheme (boundary layer options), the RRTM scheme (longwave radiation), Dudhia scheme (shortwave radiation), the Noah land surface model, and the WSM 3-class simple ice scheme (microphysics). With this background we consider the 5-km resolution adequate to elucidate orographic precipitation impacts in the region.

Fig. 1.
Fig. 1.

(a) Full domain of the WRF mesoscale model and (b) Southern California domain and the two topographic subdomains used in this study [the Central Transverse Ranges (CTR) and Peninsular Ranges (PR)]. The Western Transverse Ranges (WTR) subregion is shown for context, though this region was removed from the analysis due to its proximity to the lateral boundary. The mean aspect of the windward-facing slopes is shown in the bottom-left corner of (b) for each topographic region.

Citation: Journal of Hydrometeorology 24, 5; 10.1175/JHM-D-22-0105.1

The dataset spans nine years, from 1 January 2002 to 31 December 2010, during which only the wet season months from October to April are considered. To put this period into perspective climatologically, the Los Angeles area experienced above average or near average rainfall for four of the nine years (Western Regional Climate Center 2022) and there were a few seasons with several significant events, most notably in the winters from 2004 to 2006 and 2009 to 2010 (Miller 2017). The winter season of 2004/05 saw several heavy precipitation events in the region that were accompanied by significant land surface impacts. Most notable of these events, the event that occurred from 7 to 11 January 2005 resulted in widespread flash floods and landslides across the region, including the La Conchita landslide in Ventura County (Carpenter et al. 2007).

Variables from the WRF dataset used in this study include CAPE and equivalent potential temperature, total accumulated gridscale precipitation, model terrain, and variables needed to calculate low-level moisture flux (i.e., water vapor mixing ratio and horizontal wind components at 850 hPa). Using the NCAR Command Language (NCL; NCAR Command Language 2019), CAPE was calculated using built in NCL functionality (specifically, the “wrf_user_getvar” function). The function used calculates CAPE for the parcel with maximum equivalent potential temperature within the column, which essentially finds maximum CAPE. Equivalent potential temperature and model height are similarly calculated using the same NCL function at five vertical levels between 750 and 915 hPa (750, 800, 850, 890, and 915 hPa). With these variables, the vertical gradient of equivalent potential temperature e/dz is estimated as Δθe/Δz using a central differencing scheme.

b. Observational precipitation datasets

Prior to the diagnosis of the relationship of upstream atmospheric variables to the EPEs in the model world, a validation process was carried out to ensure that WRF adequately represents the extreme precipitation distribution and climatological precipitation features of Southern California. To this end, daily and monthly gridded PRISM and 6-hourly gridded California–Nevada River Forecast Center (CNRFC) datasets (PRISM Climate Group 2014; CNRFC 2003) are utilized. The PRISM and CNRFC datasets are based on differing rain gauge networks (see Figs. S1 and S2 in the online supplemental material) and were both employed in this WRF precipitation validation to assess and account for the limitations of using either one of these alone as a “ground truth.” PRISM is available for the period 1981–present, and the CNRFC product is available from late 2003 to present. The aforementioned datasets were obtained for the wet season months (October–April) and for the same period as covered by the available WRF record (January 2002–December 2010), with the caveat that the gridded CNRFC product start date is not until 1 November 2003. The spatial resolutions of both products are comparable to that of WRF, with PRISM having a grid spacing of 4 km and with CNRFC having grid spacing of 4.7 km.

c. USGS topographic dataset

The 1-arc-s USGS digital elevation model (DEM) (U.S. Geological Survey 2020) is also utilized to define the areal support of the EPEs for the validation process at the event scale. Slope and aspect fields of this topographic dataset were found using QGIS version 3.16.11 (QGIS 2022). The DEM was first resampled to a 200-m resolution using bilinear interpolation prior to initiating these slope and aspect calculations. The resampling of the DEM from 1 arc s to 200 m is done in order to account for computational constraints encountered in using QGIS, with 200 m chosen because it is still at a high enough resolution to represent small-scale features of the Southern California mountainous regions. Direct calculations of slope and aspect on a DEM with a similar resolution as PRISM (e.g., 4 km) would result in significant loss of localized topographic information. A bilinear interpolation method was then used to resample the output (slope and aspect) fields from 200 m to 4 km in order to smooth out the results and bring them to a resolution comparable to that of PRISM.

3. Methodology

The mountainous regions within Southern California were separated into two subdomains (Fig. 1b)—the Central Transverse Ranges (CTR) and the Peninsular Ranges (PR). These subdomains were determined based on the unique axes of orientation of the topographic features throughout the region, as well as from literature on the geology of Southern California (Chatzimanolis and Caterino 2007; Matti and Morton 2000). The full extent of the Transverse Ranges was initially included, divided into the Western Transverse Ranges (WTR) and CTR due to the more southerly orientation of the former. However, some issues were found in the WTR in the model’s ability to adequately simulate the precipitation of the region. To ensure that influences from boundary effects did not result in any erroneous conclusions, the WTR region was removed from the presented results and discussion. The remaining two subregions are well outside of the relaxation zone, which is typically set at 5–10 pixels from the lateral boundary (Prein et al. 2015; Skamarock et al. 2008). The hourly WRF variables were first aggregated to 6-hourly totals for precipitation and both 6-hourly average and maximum values for the upstream atmospheric precursors. For the remainder of this analysis, the WRF precipitation and associated atmospheric variables are considered on this 6-hourly temporal resolution unless otherwise stated. Linking WRF precipitation to WRF meteorological factors ensures a physically consistent check on the meteorological relationship between precipitation and these factors. To confirm the appropriateness of the assumption that the modeled WRF precipitation accurately characterizes the regional precipitation characteristics, a detailed validation process is first conducted.

a. WRF precipitation validation process

The PRISM product is used to validate the modeled WRF precipitation at the climatological scale. As noted in Prein et al. (2015), the benefits of high-resolution convection-permitting models (i.e., over more coarsely resolved regional climate models) are seen at fine spatial and temporal scales. Gains in model skill are diminished when high-resolution model output is considered at coarser spatiotemporal scales, such as climatological mean precipitation. Therefore, the purpose of this climatological analysis is primarily to assess the general ability of the model in representing the key climatological precipitation features of the region, particularly the orographic precipitation signal.

Although the spatial resolutions of WRF (5 km) and PRISM (4 km) are comparable, the WRF dataset was first remapped to the PRISM coordinate system to facilitate a direct comparison between the two products. This direction of resampling is chosen to avoid any complications in resampling of an interpolated field, as is the case for PRISM. The resampling was done using an inverse distance weighting (IDW) method, with an inverse power law for the weighting. Mean and median climatologies of monthly WRF and PRISM precipitation are then found and compared for the full wet season (October–April), first half of the wet season (October–December) and second half of the wet season (January–April). To reduce mesoscale-model frequent low rainfall simulation, an appropriate “low precipitation threshold” to apply to WRF [of 5 mm (6 h)−1] was also determined by this comparison to PRISM at the climatological scale.

For the event-scale validation, the distribution of daily PRISM and 6-hourly CNRFC EPEs (i.e., exceeding the 90th percentile) were compared with the WRF distribution at the same temporal resolution. For the daily analysis, the WRF precipitation data were aggregated to daily temporal values using the same “precipitation day” definition as PRISM, from 1200 to 1200 UTC. These events are defined over two different areal supports (i.e., the spatial extent over which mean areal precipitation (MAP) is found): 1) within a bounding box around the entire mountain range (as shown in Fig. 1b), and 2) using the USGS DEM data to further subset these mountain ranges by slope and aspect criteria (Table 1). The latter areal support effectively selects for regions of steeply sloping terrain with a favorable aspect for orographic interactions (i.e., the windward-facing slopes with a similar aspect as the mean orientation of the barrier). The 4-km USGS slope and aspect fields are also remapped to the PRISM grid so that they could be used to define an areal support that can be consistently applied to both PRISM and the resampled WRF precipitation. Additionally, the nine most significant daily events in WRF (and PRISM) in each topographic region were compared to the same daily events in the corresponding precipitation product. In selecting these events, the criterion that the nine events in each region must be separated by a synoptic timeframe (≥5 days) was applied.

Table 1.

Slope and aspect criteria applied to the USGS DEM data used to define the areal support for the event-scale validation of the modeled WRF precipitation.

Table 1.

Efforts toward validation of high-resolution mesoscale models often indicate an overestimation of precipitation in mountainous regions. In regions of complex topography, gridded observational products tend to be based on sparse rain gauge densities and rely on interpolation methods (Lucas-Picher et al. 2021; Lundquist et al. 2019). Rain gauges are also susceptible to precipitation undercatch, particularly for heavier precipitation events and in mountainous regions (Vergara-Temprado et al. 2020; Prein et al. 2015). Because of this, some aspects that might introduce biases in the observational datasets used for this validation are further illustrated–specifically, the impacts of rain gauge density and location. To this end, differences between PRISM and CNRFC in both space and time are assessed, stemming from factors such as gauge density, gauge location in relation to slope and elevation, interpolation methods, and differing temporal resolutions (see also the appendix).

b. EPE definition and Monte Carlo simulation process

A Monte Carlo analysis is then employed to test the question of whether modeled upstream atmospheric conditions are significantly different leading up to a model EPE when compared to those leading any precipitation event in general. For this analysis, EPEs are defined by the MAP of modeled precipitation along mountain faces with a steeply increasing slope. Similar to the event-scale validation process using the USGS elevation dataset to further subset the topographic domains, these areas are defined by a “slope threshold” for each region. The slope is calculated (using the model terrain) along the orthogonal direction to the major axis of orientation of the windward-facing side of each mountain range. This selects portions of the mountain where orographically enhanced precipitation is more likely to occur when favorably oriented moist low-level winds from the North Pacific Ocean are able to ascend these barriers.

After the low precipitation threshold is applied to the 6-hourly model precipitation totals, MAP is found using these pixels (as shown in Fig. 2) for each of the topographic regions. From these resultant model MAP values, an event exceeding the 90th percentile of nonzero precipitation is considered to be an EPE. The 90th percentile thresholds and sample sizes of EPEs and all precipitation events are shown for each region (for the available record) in Table 2.

Fig. 2.
Fig. 2.

Resultant domains after applying the slope threshold criteria, showing the composite mean precipitation of 90th percentile EPEs for each region (i.e., the sample of EPEs in Table 2). The black lines indicate contours of model terrain and are only shown to generally illustrate the location of topographic features in the region.

Citation: Journal of Hydrometeorology 24, 5; 10.1175/JHM-D-22-0105.1

Table 2.

Sample sizes of model precipitation events, model EPEs, and 90th percentile 6-hourly precipitation amounts.

Table 2.

The PR region has a larger sample size of events but a significantly lower 90th percentile intensity when compared to the CTR region. Relative histograms of the proportion of frozen precipitation to total precipitation for the EPEs in each region (Fig. 3) confirm that frozen precipitation does not contribute significantly to a large majority of these model EPEs. Therefore, no additional analysis was performed for the frozen precipitation portion of the model EPEs.

Fig. 3.
Fig. 3.

Relative histogram of the proportional contribution of frozen precipitation to the total accumulated precipitation for the model EPEs in each topographic region.

Citation: Journal of Hydrometeorology 24, 5; 10.1175/JHM-D-22-0105.1

Due to the limiting constraints of the model domain, which extends to about 120°–121°W, this analysis focuses on near coastal mesoscale precursors only. These are CAPE; potential instability as diagnosed by the sign of the vertical gradient of equivalent potential temperature, from 915 to 750 hPa; and (projected) low-level moisture flux magnitude at the 850-hPa pressure level. The levels used for the equivalent potential temperature analysis were determined based on the model layers and to give the vertical gradient calculations similar step sizes. Upstream domains within which the instability forcings are considered are shown in Fig. 4 for each topographic region, and the low-level moisture flux domains are similar but slightly expanded further into the North Pacific. As with all modeling approaches, the base scale at which the upstream precursors are inferred is set by the horizontal resolution of the model (i.e., 5 km), which likely includes some convective processes. Of course, there may be some smaller-scale topographic variability and additional convective processes contributing to EPEs in these mountainous regions that are not captured by this WRF Model. However, the fact that the climatological precipitation distribution shows clear orographic influences and is well represented (see Fig. S3) illustrates the suitability of the model in capturing precipitation processes at the barrier scale. Within these upstream domains, the spatial mean and maximum were taken of the temporal mean and maximum fields (i.e., the 6-hourly mean or maximum values of the input model hourly data) to result in four combinations (aside from e/dz, where only the spatiotemporal mean is considered). These four cases are analyzed with the hypothesis that one might show a more significant relationship to the model EPEs.

Fig. 4.
Fig. 4.

Upstream domains from (a) the CTR and (b) the PR, showing the composite mean of 6-hourly (temporal mean) lagged CAPE associated with EPE time steps only. The same domains are used in defining upstream potential instability and were slightly expanded southward and/or westward for the (projected) low-level moisture flux. The black lines indicate contours of terrain as represented by the model and are only shown to generally illustrate the location of topographic features in each region.

Citation: Journal of Hydrometeorology 24, 5; 10.1175/JHM-D-22-0105.1

From the output WRF variables, specific humidity (q) is estimated using the water vapor mixing ratio (w), and the low-level (850 hPa) moisture flux is calculated as the product of specific humidity with the horizontal wind components (v = u, υ), as shown in Eq. (1):
qv=(w1+w)v.
This low-level moisture flux vector was projected onto a vector with a direction perpendicular to the mean aspect of the windward-facing slopes (as indicated by the values given in the bottom left corner of Fig. 1b). The magnitude of this projected vector (QV) is used as the moisture flux quantity in the Monte Carlo analysis, with large and positive magnitudes indicating increased favorability for orographic interactions. As the primary interest of this work is to identify potential candidate physics-based sources of predictability, these forcings are considered at 6- and 12-h lag times with respect to the precipitation events (both extreme and nonextreme). The remainder of this paper will focus on results from the 6-h lag, as the findings were the most significant, but results for the 12-h lag are qualitatively similar. The relationship of these lagged variables to the model EPEs are assessed via relative histograms and statistical metrics of the samples. Sampling errors on each of the statistical metrics were found via a log-normalization of the variables.

The Monte Carlo methodology was designed to test whether the relationship of model precipitation to the upstream precursor variables is significantly different for an EPE compared to the same relationship for any precipitation event in general. Random values are chosen from this sample of all precipitation events [i.e., any time step where the MAP value is greater than zero after the 5 mm (6 h)−1 threshold is applied] to create a sample of the same size as the EPEs for the topographic region in question. This process is repeated 3000 times, and at that level the results were found to be insensitive to increases in the number of simulations used. To ensure the (linear) independence of the samples being drawn, an appropriate sampling interval was determined for each region by the correlation distance. The presence of any diurnal cycle in CAPE was also determined by a comparison of the spatial autocovariance functions for the mean field of each 6-hourly time step within the upstream domains (as shown in Fig. 4). Both the serial autocorrelation and spatial autocovariance functions are conditional on positive precipitation occurrence only.

This process was completed for CAPE and low-level moisture flux individually, followed by a coupled simulation that pulled the paired values of both variables. Statistical metrics and relative histograms are found for each of the 3000 simulations and the results are shown for the means (statistical values and distributions) of these simulations. For the single variable simulations, these mean statistical metrics and distributions are then compared to the corresponding EPE-associated statistics. From the Monte Carlo results for the coupled analysis, the sample of low-level moisture flux values between the population median and 75th percentile was compared to the same sample but with the additional criteria that the related CAPE value must also be greater than its population median. The precipitation events associated with each of these subsamples (i.e., at the 6-h lag) were then examined.

4. Results and discussion

a. Climatological and event-scale validation of modeled WRF precipitation

1) Monthly prism climatological validation

The first step in the validation process was to compare the mean and median precipitation climatologies between WRF (aggregated to monthly totals) and monthly PRISM for the full wet season, first half of the wet season and second half of the wet season. After remapping WRF precipitation to the PRISM grid, the monthly and synoptic climatologies were assessed at the original PRISM grid scale of 4 km, as well as for two spatially averaged grids (8 and 12 km) to ensure that differences introduced by edge effects or slight spatial variations in precipitation distributions are mitigated. The monthly mean precipitation climatologies at 12 km for the full wet season are shown in Fig. 5 for WRF and PRISM. The climatologies for the first and second half of the wet season, as well as the median climatologies, displayed similar characteristics as the full wet season.

Fig. 5.
Fig. 5.

Mean monthly precipitation climatologies, at the 12-km resolution, for the full wet season for (a) WRF, (b) PRISM, and (c) the relative differences [i.e., (WRF − PRISM)/PRISM at each grid point] between the two products.

Citation: Journal of Hydrometeorology 24, 5; 10.1175/JHM-D-22-0105.1

Overall, the spatial distributions of precipitation are similar between WRF and PRISM for the CTR and PR at all spatial resolutions (i.e., 4, 8, and 12 km), with similar structures in the orographic signal of precipitation. At the original 4-km resolution (Fig. S3), the orographic precipitation signal in PRISM is more smoothed out than in WRF, which is likely a result of the interpolation method in the former. Within each of these regions of orographic precipitation, the locations of localized maxima are similar. The results illustrated in Fig. 5 illuminate the previously mentioned issues in the WTR region, likely stemming from proximity of the upstream region to the western and northern boundaries of the model domain. The most significant precipitation in PRISM is located in the WTR, while in WRF the intensities in this region are much lower and comparable to those seen in the southern half of the PR. In WRF, there are also large relative differences in the frequency of occurrence of zero precipitation in the northwest corner of the domain (not shown), indicating the likelihood of boundary effects. As previously noted, the WTR region is excluded from the remainder of the analysis based on these (and other) findings.

The relative differences are defined such that a positive relative difference indicates a larger climatological value in WRF, i.e., (WRF − PRISM)/PRISM at each grid point in the domain. WRF tends to have higher climatological mean precipitation than PRISM in regions of high elevation. Spatially averaging from 4 to 12 km reduces the maximum relative difference of the pixel values from about 310%–200%, indicating that some of these differences at the original PRISM resolution were a result of small variations in the precipitation distributions between the two products. Though these maximum relative differences remain high at the 12-km resolution, it is important to note that the highest relative differences cover a small area coincident with the most significant topographic features of the region. This trend of higher climatological mean precipitation intensities in the mountainous regions is generally consistent with validation efforts of past studies (Prein et al. 2015). It is probable that some of this wet bias is due to overestimation of precipitation intensity and frequency by the WRF Model in these regions but is likely also partially due to biases common in gridded observational products (such as both PRISM and CNRFC). These biases, which were discussed in section 3a, are most prevalent in regions of complex topography.

These climatologies were also compared for a range of low precipitation thresholds [ranging from 2 to 10 mm (6 h)−1] applied to WRF in order to determine an effective choice for the analysis. This is to account for the increased simulation of low precipitation frequency in mesoscale models, which has been found to be a result of both the convective and microphysics parameterizations (Chang et al. 2020; Lucas-Picher et al. 2021; Tsintikidis and Georgakakos 1999). To produce realistic results in the “model world,” model precipitation must match observations as much as possible so that the inference of precursors is within a range of precipitation for which the model and observations match more closely. Chosen was the threshold which appropriately modified the WRF climatological values to better match PRISM, while keeping orographic trends in precipitation biases in mind. Based on this analysis, a threshold of 5 mm (6 h)−1 was determined to be an appropriate value. Comparing PRISM to WRF with this threshold applied, the spatial distributions of climatological precipitation (Fig. 5) and number of zero monthly precipitation occurrences (not shown) are similar, as indicated by the low relative differences across most of the region.

The PRISM rain gauge density was then examined to determine the extent to which these differences in climatological magnitudes may be driven by the reliance on interpolation in areas of sparse gauge coverage, particularly in topographically complex regions. Figure 6 shows only those PRISM gauges with 60% (or greater) nonmissing values throughout the study period (2002–10), along with those areas with a relative difference between the full wet season mean precipitation climatologies [i.e., (WRF − PRISM)/PRISM] that are greater than 1 at the 4-km grid scale. The number of pixels with a relative difference greater than 1 are relatively few, indicating that WRF and PRISM are similar across most of the Southern California domain.

Fig. 6.
Fig. 6.

Relative differences between the full wet season mean climatologies of WRF and PRISM at the native 4-km resolution (pixels), alongside the PRISM rain gauge network (points). Only pixels with a relative difference greater than 1 and gauges with 60% or greater nonmissing values throughout the record are shown.

Citation: Journal of Hydrometeorology 24, 5; 10.1175/JHM-D-22-0105.1

Aside from the San Jacinto Mountain area (around 33.75°N, 117.25°W), the majority of these pixels with significant relative differences remain below about 1.6. The distribution of the rain gauges with >60% nonmissing values reveals a pattern of low gauge density coincident with, and upstream from, many of the regions of highest relative differences. Across the Southern California domain, there are large gaps in rain gauge coverage across the most significant topographic features, with most of the gauges being located at low elevations on the windward side and with the next closest gauges oftentimes on the lee side of the mountain range. The few pixels with very significant overestimation of climatological mean precipitation (>200%) are located in mountainous regions with significant slope and where ground truth data are lacking (meaning that the pixel values are a result of interpolation in the observational product).

Contrasting these mountainous regions with the highest relative differences and low rain gauge densities, an area of relatively high gauge density (for those gauges with valid measurements for 60% of the record or greater) was located within the San Gabriel Mountains. Across these topographic features, the highest relative differences between WRF and PRISM reach a maximum of about 0.6 for the full wet season mean climatologies, which reflects markedly more similar precipitation estimates than nearby mountainous regions with high relative differences. This serves as reinforcing evidence that the PRISM precipitation estimates in regions of significant topography are likely influenced by the sparse gauge densities and reliance on interpolation. This may beg the question of whether the gridded CNRFC product would be a better option for use in this section of the validation. Though PRISM’s rain gauge coverage is sparse in the mountainous regions, the gridded CNRFC gauge density is significantly lower than PRISM for a majority of the study period across the entirety of the domain and does not meaningfully improve until 2008 (see section 1a in the online supplemental material information for further discussion). No observational precipitation product is without its own biases, so the validation process is conducted using the available information and keeping these biases in mind.

2) Event-scale precipitation validation

Validation of the modeled WRF precipitation is warranted at the event scale. As the principal interest of this study is in determining the statistical relationship of upstream precursors to the population of extreme precipitation events, it is important to establish whether the model adequately represents the extreme event distributions. To do this, the relative histograms of 90th percentile WRF (aggregated to daily) and daily PRISM precipitation, as determined by the MAP within the two different areal supports outlined in section 2a, are compared for each topographic region. Figures 7a–d show the results for the mountain range scale (MRS) and USGS slope-aspect (SA) areal supports (shown in Fig. 8), respectively. Similarly, the distributions of 6-hourly (90th percentile) EPEs from WRF and CNRFC were compared and are shown in Fig. S5. The distributions are similar for both topographic regions, for the 6-hourly and daily temporal resolutions, and for the two areal supports, indicating that the distribution of both daily and 6-hourly EPEs in WRF is consistent with observations. For both the daily (WRF versus PRISM) and 6-hourly (WRF versus CNRFC) comparison, there is a tendency for the WRF precipitation distribution to be slightly shifted toward higher precipitation magnitudes in comparison to the observational product. Comparing the daily and 6-hourly results, the 6-hourly WRF distributions tend to match CNRFC more closely than the daily WRF distributions match PRISM. This is consistent with the notion that the most significant improvements to the representation of precipitation characteristics in high-resolution models with explicit convection occur at higher spatiotemporal scales, as well as for extreme events.

Fig. 7.
Fig. 7.

Relative histogram of daily 90th percentile precipitation for WRF and PRISM for (a) the CTR and (b) the PR using the “MRS” areal supports as shown in Fig. 1b, and (d) the CTR and (e) the PR using the USGS “SA” areal supports as shown in Fig. 8.

Citation: Journal of Hydrometeorology 24, 5; 10.1175/JHM-D-22-0105.1

Fig. 8.
Fig. 8.

Precipitation domains defined based on the slope and aspect criteria shown in Table 1, showing the composite mean of (a) the top nine daily PRISM precipitation events and (b) the top nine daily WRF precipitation events in the CTR and PR.

Citation: Journal of Hydrometeorology 24, 5; 10.1175/JHM-D-22-0105.1

The most notable discrepancy is in the CTR at the daily temporal resolution, particularly for the SA areal support, with the center of mass of the 90th percentile WRF distribution shifted toward higher precipitation intensities. This could be reflecting the ability of WRF to model the additional lifting mechanism supplied by topography, leading to higher precipitation totals in the mountainous regions (where the most significant features are located in the CTR). Considering that observational precipitation products tend to have some precipitation undercatch, particularly for extreme events and in mountainous regions (Lucas-Picher et al. 2021; Vergara-Temprado et al. 2020), the WRF and observational EPE distributions are similar to one another. Therefore, the conclusion can be made that the WRF Model sufficiently represents the 90th percentile daily and 6-hourly EPE distributions.

To assess the agreement between the two products for particular significant historical events (rather than the overall distribution of extreme events), the top nine daily precipitation events from WRF were then compared to the corresponding daily event in PRISM (and vice versa with the nine most significant daily events from PRISM). The top nine events from each topographic region were combined to produce a singular relative histogram with a larger sample size (n = 18). For both the top WRF events (Figs. S4a,c) and top PRISM events (Figs. S4b,d), the distribution of the corresponding events from the alternate dataset is shifted to the left (i.e., toward lower precipitation amounts). The fact that the events from the other product in both cases have a portion of their distributions with little precipitation indicates some differences between the two datasets at this daily event scale. Examination of the dates of the top events from each product shows that several have a top event in the corresponding precipitation product that is separated by 1–3 days, with the WRF events tending to occur earlier than PRISM. Additionally, there are several events where there is significant precipitation in the corresponding product but located in a different topographic region. This could indicate potential errors in the WRF Model (or in its boundary forcing) in correctly capturing the low-level wind direction for some events. Together, these findings signal that WRF can be relied upon to get the distribution of “real-world EPEs” correctly, even if the timing or location of the most significant precipitation differs from observations for a particular event.

b. Upstream candidate precursor analysis

1) Convective instability

Because it is well known that Southern California is impacted by ARs, which feature a concentrated region of strong moisture flux below the planetary boundary layer (PBL) and are typically associated with neutral stability (Ralph et al. 2005), the convective stability associated with the EPEs is first assessed. The vertical gradient in equivalent potential temperature (θe) can be used to diagnose the convective stability of an atmospheric layer, with a decrease in equivalent potential temperature with height, i.e., Δθez < 0, indicative of convectively unstable conditions (American Meteorological Society 2022). When such a layer of air is forced to ascend, explosive convection can occur due to the differing cooling rates for the moist bottom and comparatively dry top of the layer (the latter of which will follow the dry adiabatic lapse rate for much longer before reaching saturation, hence cooling more rapidly). This reduces the environmental lapse rate and results in the transition of a layer of air to become unstable, dependent upon the layer being sufficiently lifted from its initial position. The lifting must be initiated by external forcings (as the parcel is not yet positively buoyant), which, for example, can be supplied in the form of orographically forced ascent (Wallace and Hobbs 2006). While the presence of positive CAPE indicates preexisting instability, a convectively unstable layer may initially present as stable and does not destabilize unless it is lifted externally. This stability metric is therefore differentiated from the presence of CAPE. The composite mean vertical profile of equivalent potential temperature with altitude for the top nine events in each topographic region indicate such conditions in the bottommost levels of the atmosphere, with an abrupt change to positive θe increases around 2–3 km in altitude.

The model 6-hourly EPE-associated distributions of the vertical θe gradient (at the 6-h lag time) for the 800-, 850-, and 890-hPa levels are shown in Fig. 9. The highest relative frequencies in the bottom two layers (850 and 890 hPa) occur at negative values, indicating convectively unstable conditions. Peaks in the relative histograms occur at the most negative values in the bottommost layer (890 hPa), become slightly less negative in the middle layer (850 hPa) and reach positive values, i.e., convectively stable conditions, at the topmost layer (800 hPa). The decrease in convective instability with altitude mirrors the sharp decrease in moisture near the PBL, suggesting that this “explosion” of convection is a possibility for some of the events. These results indicate that convective instability is a potential precursor in its own right and motivates a consideration of boundary layer moisture flux in conjunction with the primary stability metric of interest (i.e., CAPE).

Fig. 9.
Fig. 9.

Model EPE-associated relative histograms of the vertical gradient in equivalent potential temperature at the 6-h lag to each event for (a) 890, (b) 850, and (c) 800 hPa. The blue and gray bars indicate the two topographic regions (CTR and PR, respectively).

Citation: Journal of Hydrometeorology 24, 5; 10.1175/JHM-D-22-0105.1

2) Monte Carlo simulation: CAPE

The Monte Carlo analysis was first done for CAPE individually in order to determine the prevalence of atmospheric instability preceding the EPEs, irrespective of additional background conditions. Prior to the initiation of this simulation, the serial autocorrelation function [Eq. (2)] was used to determine an appropriate sampling interval for each spatiotemporal mean and maximum combination of CAPE for each topographic region:
ρ=i=0n(xixi¯)(xlag,ixlag,i¯)/Nlagσxσx,lag.
In the conditional autocorrelation function (ρ), Nlag is the number of values at a particular lag and is found using the mean (xi¯ and xlag,i¯) and standard deviations (σx and σx,lag) of each pair of values (i.e., the input sample of points and their lagged counterparts), rather than the stationary statistical metrics of the entire sample. The inputs at each lag consider only those points with positive precipitation and a corresponding positive precipitation at the lag of interest [which is considered up to n = 20 (6-hourly) time steps]. An example is shown in Fig. 10a for the temporal mean/spatial mean case of CAPE, with the point at which the dashed tangent lines intersect the ρ = 0 line indicating the correlation distance. The correlation distances range from 6 to 7 (6-hourly) time steps for both topographic domains and spatiotemporal mean and maximum cases, which corresponds to between 36 and 42 h. Very roughly, this is consistent with the typical timespan of wintertime synoptic disturbances moving through the region. The spatial autocovariance functions of mean CAPE for each 6-hourly time step (i.e., 0000, 0600, 1200, and 1800 UTC) were then compared to determine whether the sampling interval should consider any diurnal dependencies. Figure 10b shows an example of the autocovariance function for temporal mean CAPE in the CTR (for the 0000 UTC time step). The patterns are largely consistent throughout the day, with a northwest-to-southeast diagonally oriented signal in all regions. Because of the strong resemblance of the autocovariance signals throughout the day, there is no apparent diurnal cycle of CAPE for this particular set of precipitation events. This is somewhat expected because this analysis excludes any summertime precipitation, so instability driven by differential heating throughout the day is not a significant factor.
Fig. 10.
Fig. 10.

(a) Conditional serial autocorrelation functions of CAPE for the CTR and PR regions for the spatiotemporal mean case and (b) conditional spatial autocovariance function for the 0000 UTC time step in the CTR, for temporal mean CAPE.

Citation: Journal of Hydrometeorology 24, 5; 10.1175/JHM-D-22-0105.1

Composite mean CAPE for the top nine precipitation events in each region (not shown) indicate a center of mass of CAPE that shifts southward along the coast with the topographic region. This is apparent at both the 6- and 12-h lag, but with the CAPE feature being located further offshore at the latter time step. Peak magnitudes of CAPE are slightly higher for the PR region than the CTR (about 200 J kg−1 compared to 180 J kg−1, respectively). Statistical metrics of the EPE-associated CAPE (at the 6-h lag) are shown in Table 3 for the temporal mean/spatial mean and temporal maximum/spatial maximum cases, as these cases capture the range of values from the four combinations. For both cases, the mean and median are significantly above 0 J kg−1, indicating that a majority of events are associated with some level of instability. The statistical metrics of the temporal maximum/spatial maximum case are noteworthy in their magnitudes—for example, the 75th percentile in the PR region is roughly 500 J kg−1, which is significant for this geographic region. Even the mean and median values are relatively large, indicating that within the 6-hourly time step preceding the event and somewhere within the upstream domain, CAPE reaches significant levels when compared to the typical climatology of the region.

Table 3.

Statistical metrics (and sampling errors) of the EPE-associated CAPE (J kg−1) sample at the 6-h lag. The top (bottom) entry in each cell indicates the statistical metrics for the spatiotemporal mean (maximum) case.

Table 3.

Overall, the statistical metrics suggest slightly increased importance of unstable conditions in the PR, which might be due to its more southern location, as well as the lower elevations and discontinuities in the topographic features of the region. While the CTR has extensive and significant barriers, the PR region tends to have more isolated peaks with lower elevations so that additional lifting mechanisms might be required to force a parcel to rise above its LFC. Overall, the statistical metrics in CTR and PR are relatively similar, which is likely a result of some overlap of the upstream domains. The median values are much lower than the mean values for all topographic regions and spatiotemporal mean/maximum cases of CAPE, illustrating the skewedness of the EPE-associated distributions.

The EPE-associated and Monte Carlo mean relative histograms are shown in Fig. 11. Both the EPE-associated and Monte Carlo relative histograms resemble an exponential distribution, with the latter being more “smoothed out” due to the averaging of the individual simulation distributions. For both topographic regions, there is a significant area in the middle portion of the CAPE distribution with increased relative frequency for the EPE-associated sample (about 30–150 J kg−1 for the spatiotemporal mean case). For all regions and spatiotemporal mean/maximum cases, this area of increased relative frequency lies well above the upper 95% confidence limit of the Monte Carlo distributions. In contrast, the Monte Carlo distributions have a majority of events associated with much lower (or zero) CAPE. There are also subsequent “secondary peaks” in the histogram tails where the EPE-associated distributions have increased relative frequencies compared to the Monte Carlo distributions (above the upper 95% confidence limit).

Fig. 11.
Fig. 11.

EPE-associated and mean Monte Carlo relative histograms for each topographic region for (a) temporal mean/spatial mean CAPE, (b) temporal mean/spatial maximum CAPE, (c) temporal maximum/spatial mean CAPE, and (d) temporal maximum/spatial maximum CAPE. Solid lines indicate the EPE-associated distributions and the dashed lines in the same color indicate the corresponding Monte Carlo distributions, with the shaded region representing 95% confidence intervals for each. To enhance the distribution characteristics at lower values, CAPE is plotted on a logarithmic scale for the x axis.

Citation: Journal of Hydrometeorology 24, 5; 10.1175/JHM-D-22-0105.1

These trends are apparent in all spatiotemporal mean/maximum cases, particularly for the temporal mean/spatial mean case (Fig. 11a). This is reaffirmed by the mean statistical metrics, where the temporal mean/spatial mean case had the largest relative differences from the EPE-associated value of all the spatiotemporal mean/maximum combinations (Table 4). The relative differences of the statistical metrics are defined as (EPE − MC)/MC, so that a positive relative difference indicates a larger EPE-associated value. Though the results from the spatial and temporal maximum cases are notable due to the large magnitudes of CAPE associated with the EPEs across all topographic regions, the coupled analysis later revealed significant variability and erratic tendencies of these cases. Therefore, the remainder of the results presented in this paper will focus on the spatiotemporal mean case, as averaging increases the confidence in the metrics and distributions.

Table 4.

Mean statistical metrics of the Monte Carlo simulation samples (top entry) and differences (relative differences) between the EPE-associated and these Monte Carlo mean statistical metrics (bottom entry). All table entries have units of J kg−1, aside from the relative differences (which are unitless).

Table 4.

3) Monte Carlo simulation: Projected low-level moisture flux

The single variable Monte Carlo simulation was repeated for the projected low-level moisture flux in order to establish a baseline relationship between this precursor and the EPEs. The correlation distances found from the serial autocorrelation functions were between 6 and 8 timesteps, corresponding to a period between 36 and 48 h. This is similar to the correlation distances for CAPE, hinting at the relationship between these variables. As was done for CAPE, composite mean maps of projected low-level moisture flux were produced for the 6- and 12-h lags to the top nine precipitation events in each region (not shown). A clear band of increased magnitudes upstream from each topographic region is present at both lag times (though more apparent at the 6-h lag), indicating the importance of favorably oriented low-level moisture flux in generating the most extreme precipitation events in each region. Table 5 shows the EPE-associated statistical metrics of the projected moisture flux magnitude for the temporal mean/spatial mean case. For all of the topographic regions, the mean and median values are both significant in magnitude and relatively similar to each other, indicative of relatively low skewness in the distributions. The positive magnitudes of these metrics indicate that the EPEs are associated with favorably oriented low-level moisture flux for all of the topographic domains. The statistical metrics for the CTR have larger magnitudes than the PR, suggestive of increased importance of orographic forcing for this region. This is consistent when considering the fact that the CTR contains the tallest and most continuous barrier in the Southern California domain.

Table 5.

As in Table 3, but for EPE-associated projected low-level (850 hPa) moisture flux (g kg−1).

Table 5.

Comparing the EPE-associated and Monte Carlo distributions (Fig. 12), the evident trend is for the former to be shifted to the right, signifying that the EPEs are associated with considerably higher and more favorably oriented low-level moisture flux compared to any precipitation event in general. The EPE-associated and Monte Carlo distributions of the CTR and PR have a similar shape, generally resembling a gamma distribution. The large and positive relative differences between the statistical metrics of the two samples (Table 6) reaffirm that the EPE-associated samples have considerably larger magnitudes of projected low-level moisture flux.

Fig. 12.
Fig. 12.

As in Fig. 11, but for EPE-associated and mean Monte Carlo temporal mean/spatial mean projected low-level moisture flux.

Citation: Journal of Hydrometeorology 24, 5; 10.1175/JHM-D-22-0105.1

Table 6.

As in Table 4, but for the Monte Carlo simulation results for projected low-level (850 hPa) moisture flux (g kg−1).

Table 6.

4) Monte Carlo simulation: Coupled cape and low-level moisture flux

For the coupled simulation, the precursor variables are pulled as a pair using the maximum correlation distance between the two as the sampling interval to ensure that both samples are independent. From the 3000 simulations, the population quantiles of CAPE and projected moisture flux magnitude were determined, and these values are used to subset the simulation results into two subsamples. As a first pass, low-level moisture flux was subset including only those values greater than the median. This is compared to the same subsample with additional criteria applied, that is, that the associated CAPE must also be greater than its respective median value. This initial comparison revealed a statistically significant difference between the two samples in the upper-middle portion of the distribution. Because of this, the subsamples were further subset to consider only the portion of the distribution in the upper-middle quartile of projected low-level moisture flux (i.e., greater than the median but less than the 75th percentile).

Within the projected low-level moisture flux distributions (Fig. S11), there are several bins in each topographic domain with a statistically significant increase in relative frequency for the subsample with the additional CAPE criteria (at the 95% confidence level). The distributions of the associated precipitation events for each of these samples were then compared (Fig. 13). Overall, the center of mass of the precipitation distributions for the sample with the additional criteria of significant CAPE are shifted to the right, i.e., toward larger 6-hourly precipitation magnitudes. This trend is more apparent in the PR region, where there are a large number of events with increased precipitation intensities, which is compensated for by a reduced relative frequency of about 15% at the lowest precipitation intensities.

Fig. 13.
Fig. 13.

Relative histograms of the precipitation events associated with each sample from Fig. S11, for (a) the CTR and (b) the PR.

Citation: Journal of Hydrometeorology 24, 5; 10.1175/JHM-D-22-0105.1

Differences in the 6-hourly precipitation rates related to the presence of significant CAPE are further isolated by considering only those precipitation events contributing to the bins in the moisture flux distribution with statistically significantly (at the 95% confidence limit) increases in relative frequency compared to the sample with moisture flux criteria only (X = 50% < QV < 75%, Y = 50% < QV < 75% and CAPE > 50%). The mean precipitation intensities for the two samples are shown in Table 7, along with the probability of a positive difference between the two sample means in the rightmost column. The difference is taken such that a positive difference denotes higher precipitation intensities for the sample with the additional CAPE criteria. For both topographic regions, the probability of a positive difference is greater than 50%, indicating that it is more likely that the sample with significant CAPE will have higher precipitation intensities. Mean precipitation intensities are relatively low, which would be expected when excluding the top quantile of low-level moisture flux. The projected low-level moisture flux captures both magnitude and direction, so the exclusion of this top quantile removes those events with the most significant potential for orographically enhanced rainfall. The inclusion of this upper quantile increases the mean precipitation intensities significantly (not shown), with similar trends in the probability of a positive difference between the two samples. Together with the results from the single variable Monte Carlo analysis for CAPE, these results suggest that the presence of upstream CAPE increases precipitation intensities and is a likely precursor to EPEs in the mountainous regions of Southern California. Numerous studies have linked the presence of AR conditions to extreme precipitation in the Southern California region (Lamjiri et al. 2017; Oakley et al. 2018a), but not all ARs result in such an event. Additionally, ARs act over relatively large spatial and temporal scales where the precipitation response is not always consistent from one mountain barrier to another within the Southern California region. The presence of this signal of upstream instability may provide an additional mesoscale forcing in the region that could be used to discriminate between the occurrence of a shorter duration extreme event from a nonextreme event within a particular mountainous region.

Table 7.

Mean precipitation intensities [mm (6 h)−1] for the precipitation events in X (50% < QV < 75%) and Y (50% < QV < 75% and CAPE > 50%), with the standard errors indicated for each sample. The rightmost column shows the difference and probability of a positive difference between the mean precipitation values X¯andY¯.

Table 7.

5. Conclusions and recommendations

Upstream instability as a potential candidate forcing of EPEs in the mountainous regions of Southern California was further explored in a modeling framework, using CAPE, projected low-level moisture flux and convective instability as metrics. Upstream stability metrics associated with EPEs, particularly CAPE, have yet to be investigated in a thorough manner for the Southern California region. This study represents a comprehensive “model world” analysis of such precursor conditions to EPEs in this region, illustrating the likely role of CAPE in increasing precipitation intensities in the Southern California mountainous regions. While past studies have employed coarse resolution GCMs, the present study uses a high-resolution model capable of resolving mesoscale and topographic response forcings. This work seeks to identify candidate precursors to short duration and high spatial resolution EPEs, occurring at time scales and in locations that are more relevant to the adverse land surface impacts most common to the region (i.e., landslides and flash floods).

The model validation process shows that WRF adequately models the climatological features and extreme precipitation event distributions (at both the daily and 6-hourly temporal resolutions), albeit with some differences in maximum intensities and event timing, allowing for a fully model-based analysis. A comparison with the PRISM dataset demonstrates that WRF is able to capture the regional and climatological signal of precipitation well, but with an overestimation at higher elevations. Sparse rain gauge densities are a known source of uncertainty in gridded precipitation products, particularly in regions of complex topography, and the analyses in this study have highlighted this for the PRISM and CNRFC precipitation datasets in the Southern California region. These findings are generally consistent with past validation efforts of simulated precipitation by high-resolution models, with a wet bias over regions of complex topography (Lucas-Picher et al. 2021; Li et al. 2021). Though there may be overestimation of precipitation in these models, the lack of ground truth and tendency of precipitation undercatch for rain gauges in such regions has been pointed to as important considerations in assessing this discrepancy (Vergara-Temprado et al. 2020; Lundquist et al. 2019).

The upstream precursor analysis revealed statistically significant relationships between the EPEs and atmospheric stability metrics considered. Convectively unstable conditions were found to be frequently associated with the model EPEs at atmospheric levels most consistent with the typical PBL height. A decrease in convective instability with height signifies that explosive convection possibly contributed to some of these EPEs. The Monte Carlo analysis of lagged upstream CAPE shows statistically significant increases in the magnitude of EPE-associated CAPE when compared to CAPE associated with any precipitation event. Relative histograms indicate considerably increased relative frequency in the middle portion of the distribution for the EPE-associated CAPE distribution (from about 30–150 J kg−1 in the temporal mean/spatial mean case) compared to the Monte Carlo samples of nonextreme events. These mean CAPE values are relatively low and would not trigger convection alone, but may provide an additional component of the lifting mechanisms at play in generating extreme precipitation. Considering the significant slopes and proximity to the coast of these mountain ranges, this amount of instability may be all that is required to facilitate the extreme (short duration) precipitation intensities that are some of the highest observed rates in California (Lamjiri et al. 2018).

The larger positive values of projected low-level moisture flux of the EPE-associated distribution reaffirm the importance of favorably oriented low-level winds with sufficient moisture in orographic precipitation. In the coupled analysis, increases in low-level moisture flux occurred with concurrent significant CAPE in the upper-middle quartile of the distribution. Considering the precipitation events associated with each subsample, the additional criteria of significant CAPE resulted in increased precipitation intensities. This serves as further evidence that the presence of upstream atmospheric instability can modulate the orographic precipitation response and facilitate regions of enhanced precipitation within a larger storm. These findings are consistent with past studies in other mountainous regions in the western United States, which have found that decreases in upstream stability result in increased precipitation enhancement at higher elevations (Dettinger et al. 2004; James and Houze 2005). Additionally, it has been shown that instability of an AR event can result in different precipitation distributions and intensities than would have occurred under neutral or stable conditions (Leung and Qian 2009).

These results suggest that upstream instability, as indicated by CAPE, is an important precursor candidate for the genesis of short-duration EPEs in the mountainous regions of Southern California. The shorter duration and higher intensity of EPEs in the Southern California region, in comparison to regions to the north along the USWC, alludes to the likelihood of increased importance of precipitation processes occurring at higher spatiotemporal scales. By assessing the atmospheric forcings of short-duration EPEs at similar spatiotemporal scales as the events themselves, mesoscale forcings contributing to such events are identified. The presence of upstream instability preceding these mountainous EPEs indicates a likely role of this forcing, along with synoptic and topographically forced ascent, in lifting an air parcel to result in enhanced precipitation and orographic convection along these mountain slopes.

As these relationships are identified at the 6-h lag (with similar findings at the 12-h lag), this could be useful for predictability purposes when incorporated into a framework that considers additional background atmospheric conditions. Developing such a framework with “real world” observational data constitutes an important next step to affirm the implications of these findings in enhancing predictability of EPEs. Much of the work on predictability of extreme precipitation in the region focuses on processes acting at larger spatiotemporal scales, such as ENSO or seasonal to subseasonal predictability (such as L’Heureux et al. 2021; Chen et al. 2018; Gershunov 1998, to name a few). Additional atmospheric indicators of EPEs occurring at spatiotemporal scales associated with the most common land surface impacts in Southern California (i.e., flash floods and landslides) would therefore be useful for the purposes of disaster mitigation and hazard warnings. In addition, surface conditions relevant to land surface hazards could be incorporated in order to directly connect these findings to vulnerability impacts of these EPEs. For this type of impacts study, remotely sensed precipitation estimates could be utilized in conjunction with rain gauge estimates to develop a bias adjusted product with high spatiotemporal resolution and accuracy. Such a framework could be used to inform on hazard mitigation and planning in the present, as well as projections of how these hazards, and hence how mitigation strategies, might differ in the future under climatic change.

Acknowledgments.

The authors thank the Hydrologic Research Center for ongoing funding of this research; Christopher Castro, Hsin-I Chang, and Mike Eklund for their assistance in data acquisition of the WRF Model; and Jason Sperfslage for IT assistance in data transfer. We also thank the two anonymous reviewers for their constructive feedback, which has undoubtedly helped to improve the paper.

Data availability statement.

The WRF mesoscale model data are available and may be requested from Hsin-I Chang at the University of Arizona’s Department of Hydrology and Atmospheric Sciences. For the precipitation products used in this study, the PRISM data are openly available from the PRISM Climate Group (https://prism.oregonstate.edu/). The 6-hourly CNRFC observational precipitation (QPE) data used in this study are publicly available through the California Nevada River Forecasting Center (https://www.cnrfc.noaa.gov/arc_search.php). In defining the areal support domains for the observational precipitation validation, the 1-arcsecond Digital Elevation Model data used were obtained from the U.S. Geologic Survey and can be downloaded at https://apps.nationalmap.gov/datasets/.

APPENDIX

Comparison of PRISM and CNRFC Precipitation Products

The drawbacks of using either of the observational products as the basis for validating the WRF Model were further explored by a comparison with the gridded 6-hourly CNRFC precipitation product. The results for the WTR region are presented in this comparison, as the lateral boundary proximity issue does not apply to these observational products. Scatterplots of daily and monthly MAP values from the two products, found within the bounding boxes of the topographic regions in Fig. 1b, are shown in Fig. A1. These scatterplots include only those pixel values greater than 2 mm day−1 for both products. The monthly precipitation plots (Figs. A1a–c) all have very little scatter and are close to the 45° reference line (high degree of similarity in the datasets). The reduced slopes of these linear fits indicate that CNRFC tends to have slightly lower precipitation estimates than PRISM for a given month. The scatterplots of daily precipitation (Figs. A1d–f), on the other hand, show some scatter in all three topographic regions and a more significant departure from the 45° reference line. The fact that PRISM and CNRFC precipitation have a poorer correspondence at the daily temporal resolution serves as an indication for observational errors due to differing rain gauge networks, temporal resolutions, interpolation methodologies, or differing station measurement accuracies.

Fig. A1.
Fig. A1.

Scatterplots of precipitation between the CNRFC and PRISM products for (top) monthly precipitation in the (a) WTR, (b) CTR, (c) PR, and (bottom) daily precipitation in the (d) WTR, (e) CTR, and (f) PR. The 45° reference line, indicating a 1:1 relationship of precipitation values between the two products, is denoted by the black dashed line and the gray shaded region indicates the 95% confidence limits.

Citation: Journal of Hydrometeorology 24, 5; 10.1175/JHM-D-22-0105.1

To further illustrate the impacts of gauge density on the precipitation estimates of PRISM and CNRFC, a case study of a significant precipitation event that impacted the Southern California region from 16 to 23 December 2010 was considered. Because this is the most recent month in the record of analysis for this study, this event had high input rain gauge densities in both products. The comparison of precipitation estimates found using the actual density during the time of the event were compared to estimates found using only those gauges with the historical CNRFC density from 2003, when significantly fewer rain gauges were available. To do this, control points (see Fig. A2) were chosen in PRISM for each topographic region, and from these points the five closest CNRFC rain gauges were used to find a corresponding precipitation value using the IDW methodology previously used for the remapping process (see section 2a).

Fig. A2.
Fig. A2.

Storm total precipitation for the sample event from 16 to 23 Dec 2010, for (a) regridded WRF, (b) PRISM, and (c) regridded CNRFC. The magenta points indicate the PRISM control points and the smaller black points indicate the closest pixel locations in each product.

Citation: Journal of Hydrometeorology 24, 5; 10.1175/JHM-D-22-0105.1

For all topographic regions, the resultant CNRFC precipitation estimates are markedly less similar to the PRISM control point values when using the rain gauge density from 2003 than they are with gauge density from 2010. This suggests that input rain gauge density is a significant factor in the resultant precipitation distributions and intensities when being interpolated to a continuous field in this region of complex topography. The CNRFC gauge with the highest storm total precipitation in each topographic region was also compared to the mean of the four nearest PRISM pixels to this gauge. The mean PRISM values (as well as the maximum of the four pixels used in this mean) were significantly reduced compared to the CNRFC rain gauge estimate, likely due to differences in where each product places the most intense precipitation features for this study event.

The modeled WRF event total precipitation was then contrasted with these observational datasets. All of the precipitation products were first regridded to the coordinate system of PRISM using the same IDW methodology, the results of which are shown in Fig. A2. Using the same control points, a precipitation estimate was found for each product via the IDW method using the four closest pixel values to the control points. These estimates (Table A1) illustrate differences between the modeled WRF precipitation and the two observational products, with WRF having lower event total precipitation at the control points. WRF has the highest precipitation intensities of any of the three products, but the locations of these maxima differ from the maxima in the two observational products. PRISM and CNRFC have significant precipitation occurring along the entirety of the CTR, while the most significant features in WRF are less expansive and occur over slopes with a more westerly orientation (i.e., over the eastern half of the CTR and San Jacinto Mountains in the PR). In other words, WRF correctly identifies that a significant precipitation event occurred, but the distribution of precipitation along the topographic features is not exactly aligned with the observational products. The influence of the WRF boundary conditions of the magnitude and direction of the moisture flux is important for the precipitation distribution in this mountainous terrain.

Table A1.

Storm total (16–23 Dec 2010) mean (minimum/maximum) precipitation (mm) of the four nearest pixels to the control point in each topographic region for WRF, PRISM, and CNRFC.

Table A1.

REFERENCES

  • American Meteorological Society, 2022: Potential instability. Glossary of Meteorology, https://glossary.ametsoc.org/wiki/Potential_instability.

  • Aragon, C. M., P. C. Loikith, N. McCullar, and A. Mandilag, 2020: Connecting local-scale heavy precipitation to large-scale meteorological patterns over Portland, Oregon. Int. J. Climatol., 40, 47634780, https://doi.org/10.1002/joc.6487.

    • Search Google Scholar
    • Export Citation
  • Bracken, C., B. Rajagopalan, M. Alexander, and S. Gangopadhyay, 2015: Spatial variability of seasonal extreme precipitation in the western United States. J. Geophys. Res. Atmos., 120, 45224533, https://doi.org/10.1002/2015JD023205.

    • Search Google Scholar
    • Export Citation
  • Brown, D. P., and A. C. Comrie, 2004: A winter precipitation ‘dipole’ in the western United States associated with multidecadal ENSO variability. Geophys. Res. Lett., 31, L09203, https://doi.org/10.1029/2003GL018726.

    • Search Google Scholar
    • Export Citation
  • Calianno, M., I. Ruin, and J. J. Gourley, 2013: Supplementing flash flood reports with impact classifications. J. Hydrol., 477, 116, https://doi.org/10.1016/j.jhydrol.2012.09.036.

    • Search Google Scholar
    • Export Citation
  • California Department of Water Resources, 2021: Water year 2021: An extreme year. Tech. Rep., 12 pp., https://water.ca.gov/-/media/DWR-Website/Web-Pages/Water-Basics/Drought/Files/Publications-And-Reports/091521-Water-Year-2021-broch_v2.pdf.

  • Cannon, F., C. W. Hecht, J. M. Cordeira, and F. M. Ralph, 2018: Synoptic and mesoscale forcing of Southern California extreme precipitation. J. Geophys. Res. Atmos., 123, 13 71413 730, https://doi.org/10.1029/2018JD029045.

    • Search Google Scholar
    • Export Citation
  • Carpenter, T. M., J. Wang, S. V. Taylor, E. Shamir, J. S. Sperfslage, and K. P. Georgakakos, 2007: Surveying flash flood response in mountain streams. Eos, 88, 6972, https://doi.org/10.1029/2007EO060001.

    • Search Google Scholar
    • Export Citation
  • Cayan, D. R., K. T. Redmond, and L. G. Riddle, 1999: ENSO and hydrologic extremes in the western United States. J. Climate, 12, 28812893, https://doi.org/10.1175/1520-0442(1999)012<2881:EAHEIT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Champion, A. J., S. Blenkinsop, X.-F. Li, and H. J. Fowler, 2019: Synoptic-scale precursors of extreme U.K. summer 3-hourly rainfall. J. Geophys. Res. Atmos., 124, 44774489, https://doi.org/10.1029/2018JD029664.

    • Search Google Scholar
    • Export Citation
  • Chan, S. C., E. J. Kendon, N. Roberts, S. Blenkinsop, and H. J. Fowler, 2018: Large-scale predictors for extreme hourly precipitation events in convection-permitting climate simulations. J. Climate, 31, 21152131, https://doi.org/10.1175/JCLI-D-17-0404.1.

    • Search Google Scholar
    • Export Citation
  • Chang, W., J. Wang, J. Marohnic, V. R. Kotamarthi, and E. J. Moyer, 2020: Diagnosing added value of convection-permitting regional models using precipitation event identification and tracking. Climate Dyn., 55, 175192, https://doi.org/10.1007/s00382-018-4294-0.

    • Search Google Scholar
    • Export Citation
  • Chatzimanolis, S., and M. S. Caterino, 2007: Toward a better understanding of the “transverse range break”: Lineage diversification in Southern California. Evolution, 61, 21272141, https://doi.org/10.1111/j.1558-5646.2007.00186.x.

    • Search Google Scholar
    • Export Citation
  • Chen, X., L. R. Leung, Y. Gao, Y. Liu, M. Wigmosta, and M. Richmond, 2018: Predictability of extreme precipitation in western U.S. watersheds based on atmospheric river occurrence, intensity, and duration. Geophys. Res. Lett., 45, 11 69311 701, https://doi.org/10.1029/2018GL079831.

    • Search Google Scholar
    • Export Citation
  • CNRFC, 2003: QPE (6-hour Observed Precipitation). NOAA, accessed 1 August 2021, https://www.cnrfc.noaa.gov/arc_search.php.

  • de Orla-Barile, M., F. Cannon, N. S. Oakley, and F. M. Ralph, 2022: A climatology of narrow cold-frontal rainbands in Southern California. Geophys. Res. Lett., 49, e2021GL095362, https://doi.org/10.1029/2021GL095362.

    • Search Google Scholar
    • Export Citation
  • Dettinger, M. D., 2016: Historical and future relations between large storms and droughts in California. San Francisco Estuary Watershed Sci., 14, 1, https://doi.org/10.15447/sfews.2016v14iss2art1.

    • Search Google Scholar
    • Export Citation
  • Dettinger, M. D., K. Redmond, and D. Cayan, 2004: Winter orographic precipitation ratios in the Sierra Nevada—Large-scale atmospheric circulations and hydrologic consequences. J. Hydrometeor., 5, 11021116, https://doi.org/10.1175/JHM-390.1.

    • Search Google Scholar
    • Export Citation
  • Dettinger, M. D., F. M. Ralph, T. Das, P. J. Neiman, and D. R. Cayan, 2011: Atmospheric rivers, floods and the water resources of California. Water, 3, 445478, https://doi.org/10.3390/w3020445.

    • Search Google Scholar
    • Export Citation
  • Froude, M. J., and D. N. Petley, 2018: Global fatal landslide occurrence from 2004 to 2016. Nat. Hazards Earth Syst. Sci., 18, 21612181, https://doi.org/10.5194/nhess-18-2161-2018.

    • Search Google Scholar
    • Export Citation
  • Georgakakos, K. P., N. E. Graham, T. M. Modrick, M. J. Murphy Jr., E. Shamir, C. R. Spencer, and J. A. Sperfslage, 2014: Evaluation of real-time hydrometeorological ensemble prediction on hydrological scales in Northern California. J. Hydrol., 519, 29783000, https://doi.org/10.1016/j.jhydrol.2014.05.032.

    • Search Google Scholar
    • Export Citation
  • Gershunov, A., 1998: ENSO influence on intraseasonal extreme rainfall and temperature frequencies in the contiguous United States: Implications for long-range predictability. J. Climate, 11, 31923203, https://doi.org/10.1175/1520-0442(1998)011<3192:EIOIER>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Gimeno, L., and Coauthors, 2016: Major mechanisms of atmospheric moisture transport and their role in extreme precipitation events. Annu. Rev. Environ. Resour., 41, 117141, https://doi.org/10.1146/annurev-environ-110615-085558.

    • Search Google Scholar
    • Export Citation
  • Guirguis, K., A. Gershunov, R. E. S. Clemesha, T. Shulgina, A. C. Subramanian, and F. M. Ralph, 2018: Circulation drivers of atmospheric rivers at the North American West Coast. Geophys. Res. Lett., 45, 12 57612 584, https://doi.org/10.1029/2018GL079249.

    • Search Google Scholar
    • Export Citation
  • Guirguis, K., A. Gershunov, T. Shulgina, R. E. S. Clemsha, and F. M. Ralph, 2019: Atmospheric rivers impacting Northern California and their modulation by a variable climate. Climate Dyn., 52, 65696583, https://doi.org/10.1007/s00382-018-4532-5.

    • Search Google Scholar
    • Export Citation
  • Guirguis, K., A. Gershunov, M. J. DeFlorio, T. Shulgina, L. Delle Monache, A. C. Subramanian, T. W. Corringham, and F. M. Ralph, 2020: Four atmospheric circulation regimes over the North Pacific and their relationship to California precipitation on daily to seasonal timescales. Geophys. Res. Lett., 47, e2020GL087609, https://doi.org/10.1029/2020GL087609.

    • Search Google Scholar
    • Export Citation
  • Han, Z., and H. O. Sharif, 2021: Analysis of flood fatalities in the United States, 1959–2019. Water, 13, 1871, https://doi.org/10.3390/w13131871.

    • Search Google Scholar
    • Export Citation
  • Harris, S. M., and L. M. V. Carvalho, 2018: Characteristics of Southern California atmospheric rivers. Theor. Appl. Climatol., 132, 965981, https://doi.org/10.1007/s00704-017-2138-1.

    • Search Google Scholar
    • Export Citation
  • Hecht, C. W., and J. M. Cordeira, 2017: Characterizing the influence of atmospheric river orientation and intensity on precipitation distributions over north coastal California. Geophys. Res. Lett., 44, 90489058, https://doi.org/10.1002/2017GL074179.

    • Search Google Scholar
    • Export Citation
  • Hoell, A., M. Hoerling, J. Eischeid, K. Wolter, R. Dole, J. Perlwitz, T. Xu, and L. Cheng, 2016: Does El Niño intensity matter for California precipitation? Geophys. Res. Lett., 43, 819825, https://doi.org/10.1002/2015GL067102.

    • Search Google Scholar
    • Export Citation
  • Houze, R. A., Jr., 2012: Orographic effects on precipitating clouds. Rev. Geophys., 50, RG1001, https://doi.org/10.1029/2011RG000365.

  • James, C. N., and R. A. Houze Jr., 2005: Modification of precipitation by coastal orography in storms crossing Northern California. Mon. Wea. Rev., 133, 31103131, https://doi.org/10.1175/MWR3019.1.

    • Search Google Scholar
    • Export Citation
  • Jong, B.-T., M. Ting, and R. Seager, 2016: El Niño’s impact on California precipitation: Seasonality, regionality and El Niño intensity. Environ. Res. Lett., 11, 054021, https://doi.org/10.1088/1748-9326/11/5/054021.

    • Search Google Scholar
    • Export Citation
  • Kam, J., K. Stowers, and S. Kim, 2019: Monitoring of drought awareness from Google trends: A case study of the 2011–17 California drought. Wea. Climate Soc., 11, 419429, https://doi.org/10.1175/WCAS-D-18-0085.1.

    • Search Google Scholar
    • Export Citation
  • Kean, J. W., and Coauthors, 2019: Inundation, flow dynamics, and damage in the 9 January 2018 Montecito debris-flow event, California, USA: Opportunities and challenges for post-wildfire risk assessment. Geosphere, 15, 11401163, https://doi.org/10.1130/GES02048.1.

    • Search Google Scholar
    • Export Citation
  • Khodayar, S., N. Kalthoff, and C. Kottmeier, 2018: Atmospheric conditions associated with heavy precipitation events in comparison to seasonal means in the western Mediterranean region. Climate Dyn., 51, 951967, https://doi.org/10.1007/s00382-016-3058-y.

    • Search Google Scholar
    • Export Citation
  • Kim, H.-M., Y. Zhou, and M. A. Alexander, 2019: Changes in atmospheric rivers and moisture transport over the northeast Pacific and western North America in response to ENSO diversity. Climate Dyn., 52, 73757388, https://doi.org/10.1007/s00382-017-3598-9.

    • Search Google Scholar
    • Export Citation
  • Kim, J., D. E. Waliser, P. J. Neiman, B. Guan, J.-M. Ryoo, and G. A. Wick, 2013: Effects of atmospheric river landfalls on the cold season precipitation in California. Climate Dyn., 40, 465474, https://doi.org/10.1007/s00382-012-1322-3.

    • Search Google Scholar
    • Export Citation
  • Kirshbaum, D. J., B. Adler, N. Kalthoff, C. Barthlott, and S. Serafin, 2018: Moist orographic convection: Physical mechanisms and links to surface-exchange processes. Atmosphere, 9, 80, https://doi.org/10.3390/atmos9030080.

    • Search Google Scholar
    • Export Citation
  • Lamjiri, M. A., M. D. Dettinger, F. M. Ralph, and B. Guan, 2017: Hourly storm characteristics along the U.S. West Coast: Role of atmospheric rivers in extreme precipitation. Geophys. Res. Lett., 44, 70207028, https://doi.org/10.1002/2017GL074193.

    • Search Google Scholar
    • Export Citation
  • Lamjiri, M. A., M. D. Dettinger, F. M. Ralph, N. S. Oakley, and J. J. Rutz, 2018: Hourly analyses of the large storms and atmospheric rivers that provide most of California’s precipitation in only 10 to 100 hours per year. San Francisco Estuary Watershed Sci., 16, 1, https://doi.org/10.15447/sfews.2018v16iss4art1.

    • Search Google Scholar
    • Export Citation
  • Lee, S.-K., H. Lopez, E.-S. Chung, P. DiNezio, S.-W. Yeh, and A. T. Wittenberg, 2018: On the fragile relationship between El Niño and California rainfall. Geophys. Res. Lett., 45, 907915, https://doi.org/10.1002/2017GL076197.

    • Search Google Scholar
    • Export Citation
  • Leung, L. R., and Y. Qian, 2009: Atmospheric Rivers induced heavy precipitation and flooding in the western U.S. simulated by the WRF regional climate model. Geophys. Res. Lett., 36, L03820, https://doi.org/10.1029/2008GL036445.

    • Search Google Scholar
    • Export Citation
  • L’Heureux, M. L., M. K. Tippett, and E. J. Becker, 2021: Sources of subseasonal skill and predictability in wintertime California precipitation forecasts. Wea. Forecasting, 36, 18151826, https://doi.org/10.1175/WAF-D-21-0061.1.

    • Search Google Scholar
    • Export Citation
  • Li, P., K. Furtado, T. Zhou, H. Chen, and J. Li, 2021: Convection-permitting modelling improves simulated precipitation over the central and eastern Tibetan Plateau. Quart. J. Roy. Meteor. Soc., 147, 341362, https://doi.org/10.1002/qj.3921.

    • Search Google Scholar
    • Export Citation
  • Lin, Y.-L., S. Chiao, T.-A. Wang, M. L. Kaplan, and R. P. Weglarz, 2001: Some common ingredients for heavy orographic rainfall. Wea. Forecasting, 16, 633660, https://doi.org/10.1175/1520-0434(2001)016<0633:SCIFHO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Lucas-Picher, P., D. Argüeso, E. Brisson, Y. Tramblay, P. Berg, A. Lemonsu, S. Kotlarski, and C. Caillaud, 2021: Convection-permitting modeling with regional climate models: Latest developments and next steps. Wiley Interdiscip. Rev.: Climate Change, 12, e731, https://doi.org/10.1002/wcc.731.

    • Search Google Scholar
    • Export Citation
  • Lundquist, J., M. Hughes, E. Gutmann, and S. Kapnick, 2019: Our skill in modeling mountain rain and snow is bypassing the skill of our observational networks. Bull. Amer. Meteor. Soc., 100, 24732490, https://doi.org/10.1175/BAMS-D-19-0001.1.

    • Search Google Scholar
    • Export Citation
  • Luong, T. M., C. L. Castro, T. M. Nguyen, W. W. Cassell, and H.-I. Chang, 2018: Improvement in the modeled representation of North American monsoon precipitation using a modified Kain–Fritsch convective parameterization scheme. Atmosphere, 9, 31, https://doi.org/10.3390/atmos9010031.

    • Search Google Scholar
    • Export Citation
  • Matti, J. C., and D. M. Morton, 2000: Geology of the San Bernardino National Forest. U.S. Geological Survey, 31 pp., https://www.fs.usda.gov/Internet/FSE_DOCUMENTS/stelprdb5207093.pdf.

  • Miller, M., 2017: A history of significant weather events in Southern California organized by weather type. Weather.gov, 152 pp., https://www.weather.gov/media/sgx/documents/weatherhistory.pdf.

  • Mock, C. J., 1996: Climatic controls and spatial variations of precipitation in the western United States. J. Climate, 9, 11111125, https://doi.org/10.1175/1520-0442(1996)009<1111:CCASVO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Modrick, T. M., and K. P. Georgakakos, 2015: The character and causes of flash flood occurrence changes in mountainous small basins of Southern California under projected climatic change. J. Hydrol. Reg. Stud., 3, 312336, https://doi.org/10.1016/j.ejrh.2015.02.003.

    • Search Google Scholar
    • Export Citation
  • Oakley, N. S., J. T. Lancaster, M. L. Kaplan, and F. M. Ralph, 2017: Synoptic conditions associated with cool season post-fire debris flows in the transverse ranges of Southern California. Nat. Hazards, 88, 327354, https://doi.org/10.1007/s11069-017-2867-6.

    • Search Google Scholar
    • Export Citation
  • Oakley, N. S., J. T. Lancaster, B. J. Hatchett, J. Stock, F. M. Ralph, S. Roj, and S. Lukashov, 2018a: A 22-year climatology of cool season hourly precipitation thresholds conducive to shallow landslides in California. Earth Interact., 22, https://doi.org/10.1175/EI-D-17-0029.1.

    • Search Google Scholar
    • Export Citation
  • Oakley, N. S., F. Cannon, E. Boldt, J. Dumas, and F. M. Ralph, 2018b: Origins and variability of extreme precipitation in the Santa Ynez River Basin of Southern California. J. Hydrol.: Reg. Stud., 19, 164176, https://doi.org/10.1016/j.ejrh.2018.09.001.

    • Search Google Scholar
    • Export Citation
  • NCAR Command Language, 2019: UCAR/NCAR/CISL/TDD, https://doi.org/10.5065/D6WD3XH5.

  • Pandey, G. R., D. R. Cayan, and K. P. Georgakakos, 1999: Precipitation structure in the Sierra Nevada of California during winter. J. Geophys. Res., 104, 12 01912 030, https://doi.org/10.1029/1999JD900103.

    • Search Google Scholar
    • Export Citation
  • Patricola, C. M., J. P. O’Brien, M. D. Risser, A. M. Rhoades, T. A. O’Brien, P. A. Ullrich, D. A. Stone, and W. D. Collins, 2020: Maximizing ENSO as a source of western US hydroclimate predictability. Climate Dyn., 54, 351372, https://doi.org/10.1007/s00382-019-05004-8.

    • Search Google Scholar
    • Export Citation
  • Prein, A. F., and Coauthors, 2015: A review on regional convection-permitting climate modeling: Demonstrations, prospects, and challenges. Rev. Geophys., 53, 323361, https://doi.org/10.1002/2014RG000475.

    • Search Google Scholar
    • Export Citation
  • PRISM Climate Group, 2014: PRISM climate data. PRISM Climate Group, accessed 12 July 2021, https://prism.oregonstate.edu.

  • QGIS, 2022: QGIS Geographic Information System. QGIS association, https://www.qgis.org/en/site/forusers/download.html.

  • Ralph, F. M., and M. D. Dettinger, 2011: Storms, floods, and the science of atmospheric rivers. Eos, 92, 265266, https://doi.org/10.1029/2011EO320001.

    • Search Google Scholar
    • Export Citation
  • Ralph, F. M., P. J. Neiman, and R. Rotunno, 2005: Dropsonde observations in low-level jets over the northeastern Pacific Ocean from CALJET-1998 and PACJET-2001: Mean vertical-profile and Atmospheric-River characteristics. Mon. Wea. Rev., 133, 889910, https://doi.org/10.1175/MWR2896.1.

    • Search Google Scholar
    • Export Citation
  • Ralph, F. M., P. J. Neiman, G. A. Wick, S. I. Gutman, M. D. Dettinger, D. R. Cayan, and A. B. White, 2006: Flooding on California’s Russian River: Role of atmospheric rivers. Geophys. Res. Lett., 33, L13801, https://doi.org/10.1029/2006GL026689.

    • Search Google Scholar
    • Export Citation
  • Ramezani Ziarani, M., B. Bookhagen, T. Schmidt, J. Wickert, A. de la Torre, and R. Hierro, 2019: Using convective available potential energy (CAPE) and dew-point temperature to characterize rainfall-extreme events in the south-central Andes. Atmosphere, 10, 379, https://doi.org/10.3390/atmos10070379.

    • Search Google Scholar
    • Export Citation
  • Ricard, D., V. Ducrocq, and L. Auger, 2012: A climatology of the mesoscale environment associated with heavily precipitating events over a northwestern Mediterranean area. J. Appl. Meteor. Climatol., 51, 468488, https://doi.org/10.1175/JAMC-D-11-017.1.

    • Search Google Scholar
    • Export Citation
  • Risanto, C. B., H.-I. Chang, T. M. Luong, C. L. Castro, H. P. Dasari, and I. Hoteit, 2023: Retrospective sub-seasonal forecasts of extreme precipitation events in the Arabian Peninsula using convective-permitting modeling. Climate Dyn., https://doi.org/10.1007/s00382-022-06336-8, in press.

    • Search Google Scholar
    • Export Citation
  • Roe, G. H., 2005: Orographic precipitation. Annu. Rev. Earth Planet. Sci., 33, 645671, https://doi.org/10.1146/annurev.earth.33.092203.122541.

    • Search Google Scholar
    • Export Citation
  • Rutz, J. J., W. J. Steenburgh, and F. M. Ralph, 2014: Climatological characteristics of atmospheric rivers and their inland penetration over the western United States. Mon. Wea. Rev., 142, 905921, https://doi.org/10.1175/MWR-D-13-00168.1.

    • Search Google Scholar
    • Export Citation
  • Ryoo, J.-M., Y. Kaspi, D. W. Waugh, G. N. Kiladis, D. E. Waliser, E. J. Fetzer, and J. Kim, 2013: Impact of Rossby wave breaking on U.S. West Coast winter precipitation during ENSO events. J. Climate, 26, 63606382, https://doi.org/10.1175/JCLI-D-12-00297.1.

    • Search Google Scholar
    • Export Citation
  • Seager, R., M. Hoerling, S. Schubert, H. Wang, B. Lyon, A. Kumar, J. Nakamura, and N. Henderson, 2015: Causes of the 2011–14 California drought. J. Climate, 28, 69977024, https://doi.org/10.1175/JCLI-D-14-00860.1.

    • Search Google Scholar
    • Export Citation
  • Sierks, M. D., J. Kalansky, F. Cannon, and F. M. Ralph, 2020: Characteristics, origins, and impacts of summertime extreme precipitation in the Lake Mead Watershed. J. Climate, 33, 26632680, https://doi.org/10.1175/JCLI-D-19-0387.1.

    • Search Google Scholar
    • Export Citation
  • Skamarock, W. C., and Coauthors, 2008: A description of the Advanced Research WRF version 3. NCAR Tech. Note NCAR/TN-475+STR, 113 pp., https://doi.org/10.5065/D68S4MVH.

  • Tsintikidis, D., and K. P. Georgakakos, 1999: Microphysical and large-scale dependencies of temporal rainfall variability over a tropical ocean. J. Atmos. Sci., 56, 724748, https://doi.org/10.1175/1520-0469(1999)056<0724:MALSDO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • U.S. Geological Survey, 2020: Delevation program 1-arcsecond resolution digital elevation model. U.S. Geological Survey, accessed 15 October 2021, https://www.usgs.gov/3d-elevation-program/about-3dep-products-services.

  • Vergara-Temprado, J., N. Ban, D. Panosetti, L. Schlemmer, and C. Schär, 2020: Climate models permit convection at much coarser resolutions than previously considered. J. Climate, 33, 19151933, https://doi.org/10.1175/JCLI-D-19-0286.1.

    • Search Google Scholar
    • Export Citation
  • Viale, M., and M. N. Nuñez, 2011: Climatology of winter orographic precipitation over the subtropical central Andes and associated synoptic and regional characteristics. J. Hydrometeor., 12, 481507, https://doi.org/10.1175/2010JHM1284.1.

    • Search Google Scholar
    • Export Citation
  • Wallace, J. M., and P. V. Hobbs, 2006: Atmospheric Science: An Introductory Survey. 2nd ed. Elsevier, 350 pp.

  • Warner, M. D., C. F. Mass, and E. P. Salathé Jr., 2012: Wintertime extreme precipitation events along the Pacific Northwest coast: Climatology and synoptic evolution. Mon. Wea. Rev., 140, 20212043, https://doi.org/10.1175/MWR-D-11-00197.1.

    • Search Google Scholar
    • Export Citation
  • Western Regional Climate Center, 2022: Cooperative climatological data summaries. Western Regional Climate Center, accessed 10 May 2022, https://wrcc.dri.edu/Climate/west_coop_summaries.php.

  • Xiong, Y., and X. Ren, 2021: Influences of atmospheric rivers on north pacific winter precipitation: Climatology and dependence on ENSO condition. J. Climate, 34, 277292, https://doi.org/10.1175/JCLI-D-20-0301.1.

    • Search Google Scholar
    • Export Citation

Supplementary Materials

Save