1. Introduction
Shallow landslides, especially those that mobilize as debris flows, have long posed a threat to life and property situated among California’s complex terrain (Campbell 1975; Taylor and Brabb 1987; Wieczorek 2002; Coe et al. 2004; Jibson 2005; Corbett and Collins 2023). Persistent and, at times, intense rainfall has typically been associated with many impactful landslide events in California, with most landslides occurring during the peak of the cool/wet season from December through March. This was exemplified recently when a persistent sequence of storm events began in late December 2022 and continued through March 2023, causing widespread shallow landsliding that resulted in damage to homes, roads, and other infrastructure in many areas of the state (Moody’s RMS 2023; CGS 2024).
To mitigate the loss of life and damage to homes and infrastructure from precipitation-induced landslides, researchers and emergency managers often turn to rainfall thresholds to identify when potentially impactful landslides might occur. In California, these have varied from antecedent rainfall and intensity thresholds (Campbell 1975) to rainfall intensity–duration curves (Keefer et al. 1987; Ellen and Wieczorek 1988), process-based metrics (Thomas et al. 2018), and early warning systems (NOAA-USGS Debris Flow Task Force 2005). Elsewhere in the world, metrics linking rainfall parameters (season or storm total, storm duration, and storm intensity) are also prevalent and well integrated to early warning systems (Guzzetti et al. 2020). These metrics can provide guidance for decision-makers in landslide-prone communities preceding and during storm events. However, all approaches to rainfall thresholds do not address the atmospheric characteristics that lead to these rainfall conditions. This raises the question, are there common atmospheric conditions associated with landslide-producing storms that may inform early warning of landslide hazards?
We focus our efforts on shallow landslide events (Fig. 1) as these are typically triggered during rainfall rather than at some time afterward. This temporal connection is dependent on the time required for rainfall to infiltrate into the shallow subsurface. Thus, shallow landslides are generally more well connected to temporally coincident rainfall conditions compared to deeper-seated landslides (i.e., those triggered by longer-term groundwater seepage such as the 2014 Oso, Washington, landslide; Wartman et al. 2016; Collins and Reid 2020). Given the prevalence of widespread shallow landslide events documented in California over the past 50 years (e.g., Taylor and Brabb 1972; Campbell 1975; Ellen and Wieczorek 1988; DeGraff 1994; Jibson 2005; Collins and Corbett 2019; Corbett and Collins 2023), examining the atmospheric characteristics associated with historic landslide events is a natural line of inquiry.
Examples of shallow landslides in California that are assessed in this study. (a) Landslide runout into homes from the 2005 La Conchita event (from Jibson 2005), (b) landslide impacting a home in the 2017 East Bay event, (c) widespread landsliding in the 2018 Tuolumne event, and (d) hillslope failure from the 2020 San Diego event (photo credit: Douglas Alden, UCSD/SIO).
Citation: Earth Interactions 28, 1; 10.1175/EI-D-24-0003.1
Previous work has demonstrated a relationship between atmospheric rivers (ARs; narrow corridors of enhanced water vapor transport in the atmosphere) and landslide activity along the U.S. West Coast (Biasutti et al. 2016; Young et al. 2017; Oakley et al. 2018a; Cordeira et al. 2019; Collins et al. 2020). However, these studies have explored this connection either through limited atmospheric variables or through compositing such that individual event characteristics are not represented. Herein, we build on this research by exploring variables related to ARs in greater depth as well as other atmospheric variables relevant to precipitation. We identify 18 impactful shallow landslide events associated with 13 storms in California and evaluate key meteorological characteristics of these storms to determine whether there are common features that can provide situational awareness of landslide potential. The goal is to assess whether, in the absence of well-constrained rainfall intensity–duration thresholds that could be applied to precipitation forecasts, there are storm characteristics that can be related to landslide potential. As analog events are often used in weather forecasting applications (e.g., Chattopadhyay et al. 2020), developing a description of past storm events that triggered landslides can provide insights into landslide potential for incoming storms and aid in communication efforts. This work also serves to bridge the gap between the disciplines of meteorology and geomorphology by providing detailed meteorological analysis of landslide-triggering events and explanations of the relevance of these variables to increase understanding of landslide drivers across disciplines.
2. Methods
a. Landslide catalog development
This study focuses on widespread, impactful shallow landslide events in California. The term “shallow landsliding” is used herein to describe landslides with failure surfaces that are typically less than 3 m deep but can be up to 10 m deep. The focus on this depth range explicitly tries to ensure that the landslide events under consideration were triggered in close temporal proximity to storm rainfall events through a direct hydrological connection from surface infiltration to subsurface seepage and soil water pore pressure generation (e.g., Iverson 2000; Lu and Godt 2013). We aim to represent multiple geomorphic provinces (Jenkins 1938) of California where both soils and terrain are susceptible to landslide initiation. The number of landslides associated with each storm event ranges from a few to several thousand and generally consists of debris slide and debris flow landslide types (Varnes 1978). The landslides selected caused damage to life, property, or infrastructure (Table 1).
Landslide events assessed in this analysis. The “Event name” column is the name assigned to each event. The “Event county” column lists the California counties in which the landslide event occurred. The “Event window” column refers to the time window during which the landslide occurred in which atmospheric characteristics are assessed, described in section 2a. If a more precise landslide initiation time is known, it is provided in the “Landslide initiation time” column; a — indicates no precise time, only a window is available. The remaining columns describe the number of landslides, type, and impact. References for each of the events are provided in Table S1.
Two major constraints guided the landslide catalog development. First, we only consider post-1995 events such that widespread NEXRAD weather radar data were available to assess the presence of any high-intensity rainfall features. Second, we require that landslide timing can be constrained to a 6-h window, so we can assess the atmospheric conditions coincident with landslide initiation. This duration strikes a balance between providing a reasonable number of events to study while not creating a time window so wide that the atmospheric conditions in that window may not be relevant to landslide initiation.
Prospective landslide events were gathered from a variety of resources including peer-reviewed literature, reports from the California Geological Survey, U.S. Geological Survey, National Weather Service, print and social media, and anecdotal reports. All landslide events were assigned a 6-h start and end time window, as well as a “landslide initiation time” where applicable. Where event timing is better constrained, a minimum 3-h window is assigned to the event. This accounts for uncertainty in event reporting as well as uncertainty in the atmospheric reanalysis data used to assess each event. Our analysis includes 18 landslide events that occurred during 13 unique storms (Table 1). We consider 37°N latitude as a dividing line between Northern and Southern California, which places seven events in each of these regions. We consider four events in the Sierra Nevada separately. We assign rectangular areas encompassing the extent of landslide activity to each event (Fig. 2). Given that atmospheric reanalysis data have a coarse (31 km) resolution compared to the extent of landslide events (see section 2b), the landslide event polygons are liberally sized (i.e., long dimensions ranging from 3 to 67 km).
Map of 18 landslide areas analyzed in this study. Red rectangles indicate areas of landslide activity. Labels correspond to event names in Table 1. The dashed horizontal gray line is at 37°N, which we use as the dividing line between Northern and Southern California. The Sierra Nevada mountain range spanning the eastern boundary of California is noted as well. Coordinates for each rectangle are provided in Table S2. Shaded relief map base from Tozer et al. (2019).
Citation: Earth Interactions 28, 1; 10.1175/EI-D-24-0003.1
b. Meteorological data analysis
For each landslide event, we examine meteorological variables relevant to precipitation-producing storms in California. We use the fifth major global reanalysis produced by ECMWF (ERA5; Hersbach et al. 2018), an hourly, 31-km horizontally spaced grid atmospheric reanalysis dataset. The variables we examine are commonly considered in the assessment of characteristics of hydrologically critical or impactful storms in California (e.g., Weaver 1962; Haynes 2001; Young et al. 2017; Hecht and Cordeira 2017; Cannon et al. 2018; Oakley et al. 2017, 2018b; California-Nevada River Forecast Center 2023). Although there are numerous other meteorological variables that could be considered for analysis, these are first-order variables relevant to precipitation that are available in and meaningful at the spatial scale of ERA5. Variables analyzed and discussed herein are as follows:
-
Geopotential height of 300 and 500 hPa, used for identifying areas of high and low pressure at mid-to-upper levels of the atmosphere
-
Wind speed and direction at 300 hPa, 850 hPa, and 10 m, used to provide information on upper-level jet stream characteristics, potential for large-scale lift in the atmosphere, and orographic precipitation potential at low levels
-
Convective available potential energy (CAPE), an indicator of instability in the atmosphere, indicator of potential for high-intensity rainfall
-
Mean sea level pressure (MSLP), used here for identifying surface cyclones, the minimum MSLP during a storm event serves as an indicator of the storm’s intensity
-
Integrated water vapor (IWV), vertically integrated water vapor, used for evaluating the presence and strength of ARs, and indicative of moisture available for precipitation
-
Integrated water vapor transport (IVT), vertically integrated water vapor, wind speed, and wind direction, used for evaluating the presence and strength of ARs, also indicative of moisture available for precipitation
We plot each variable over the 6-h window of each landslide event. For the MSLP variable, we identify the lowest value in the center of the barometric surface low pressure associated with a storm for each landslide event time window. We determine a general location of the low using the following criteria: Pacific Northwest (PNW) if it is poleward of 42°N, Northern California (N CA) if it is 37°–42°N, and Southern California (S CA) if the low center is equatorward of 37°N. All surface low pressure centers were offshore and east of 140°W longitude, and most were also east of 130°W.
c. Atmospheric river analysis
There are numerous algorithms available to objectively identify atmospheric rivers in forecast models, satellite observations, reanalysis, and climate projection datasets that have varying geometry, threshold, and duration requirements, dependent on the authors’ goals and region studied (Shields et al. 2018). While earlier methods relied on satellite retrievals of IWV (e.g., Ralph et al. 2004; Neiman et al. 2008), the IVT variable is now more commonly used, with various thresholds considered (e.g., Rutz et al. 2014; Guan and Waliser 2019).
Given the variations in AR definitions and detection algorithms, it is valuable to consider uncertainty in the detection of AR features. We apply the Toolkit for Extreme Climate Analysis Bayesian AR Detector (TECA-BARD) AR detection method (O’Brien et al. 2020), which applies a statistical method to account for detection uncertainty, to the ERA5 reanalysis IVT. We first examine the TECA-BARD probability of AR detection for the 6-h period preceding the end of the landslide window (Table 1). In cases where probabilities are <10% and/or the authors’ subjective interpretation of meteorological information and other published literature suggest an event should feature an AR, we additionally examine a 24-h period preceding the end of the landslide window in case the landslide event occurred when an AR had begun to dissipate and was no longer likely to be detected. We generate images of the probability of AR detection over California and report the highest probability of detection in the vicinity of the landslide event being assessed. Figures S1–S3 in the online supplemental material provide examples of the TECA-BARD AR detection results.
The atmospheric river scale (AR scale; Ralph et al. 2019) was developed to provide situational awareness on the presence of AR conditions (IVT > 250 kg m−1 s−1). The AR scale considers the magnitude and duration of IVT at a given point and prescribes a ranking from 1 to 5 for each event, with AR1 as the weakest and AR5 as the strongest. We use IVT from ERA5 to calculate the AR scale at landfall points along the California coast as well as at the ERA5 grid cell associated with the center of each landslide polygon. We use a 96-h window prior to landslide occurrence to calculate the AR scale across the state for each storm event in our catalog. Additionally, we assess the maximum IVT and IWV value at the landslide polygon center point.
The AR scale calculation does not consider the length and width of the moisture plume. Thus, it is possible to have an event that ranks on the AR scale but does not meet the criteria of one or more AR algorithms, which often include length and/or width criteria. For example, the storm in Figs. 3b and 3c has a well-defined narrow plume of IVT and IWV extending from the tropics to California and an AR3–4 classification. In contrast, the storm in Figs. 3e and 3f reaches AR1, though its IVT plume length only extends several hundred kilometers. Indeed, TECA-BARD does not capture this event as an AR (Table 2). Thus, whether this moisture plume could be considered an AR is debatable and dependent on the criteria used despite its classification of AR1 on the AR scale. The AR scale provides a method to categorize and rank moisture transport; however, a storm ranking on the AR scale may not always have a synoptic-scale AR feature.
Examples of visualizations of meteorological variables used for analysis in this study. (a)–(c) Meteorological conditions for the 2017 Redding landslide event at the time of 0100 UTC. (d)–(f) Meteorological conditions for the 2001 Santa Paula landslide event at 0900 UTC. Variables shown in the top row are (a) 300-hPa height (decameters) as contours and 300-hPa wind barbs and wind speed shown in filled contours (m s−1); jet streak is labeled and referred to in a later section; (b) 500-hPa height (decameters) and IVT, shown as vectors (smallest vectors are equal to 250 kg m s−1) and filled contours; (c) MSLP (hPa) as contours, IWV (mm) as filled contours, and 10-m wind (m s−1). (d)–(f) As in (a)–(c), but for the 2001 Santa Paula event. Tables 2 and 3 describe features seen in the figures. Yellow stars indicate the landslide event locations.
Citation: Earth Interactions 28, 1; 10.1175/EI-D-24-0003.1
Rainfall and moisture variable characteristics for landslide events. The “Event” column provides the event name (see Table 1). Superscripts in the “Event” column indicate landslide events that happened in different locations in the same storm. The “Percent avg WYTD precip” column provides the percent of WYTD average precipitation prior to the landslide event. The “TECA-BARD AR” column provides the percent likelihood of detection of an AR object based on the TECA-BARD detection algorithm for a 6-h period (24-h period). Where no (24-h) value is provided, the 24-h value was equal to 6 h. The AR scale (coast) column provides the AR scale value at the coastal landfall location west of the landslide. A range is provided where the coastal scale is ambiguous. The “AR scale (loc.)” column provides the AR scale value at the landslide location. For both AR scale columns, — indicates the location did not meet AR scale criteria. The “IVT” and “IVT pctile” columns provide the IVT value and percentile rank of the IVT value at the landslide location. The “IWV” and “IWV pctile” columns provide the IWV value and the percentile rank of the IWV value at the landslide location.
d. Precipitation data analysis
Though this study focuses on atmospheric conditions associated with landslide-triggering storms, we recognize that antecedent rainfall plays an important role in landslide activity. We evaluate water year (beginning 1 October) to date precipitation, using PRISM precipitation estimates (PRISM Climate Group 2014) at daily temporal resolution and 4-km spatial resolution. Using the PRISM grid cell most central to each landslide polygon, we first calculate the 1991–2020 daily average precipitation from PRISM. We then calculate the 1 October accumulation to the date that most closely corresponds to the period preceding each landslide event and then calculate water-year-to-date percentage of average precipitation for each event. We sought hourly rain gauge–based precipitation data for each landslide event in the catalog, but such data were not available for all events at locations representative of the landslides. Thus, we do not include analysis of hourly precipitation data in this study. High-resolution (<4 km, 1 h) precipitation reanalysis products became available for California toward the completion of this study (e.g., Rasmussen et al. 2023; Rahimi et al. 2022); exploration of these datasets for landslides warrants its own independent study. We provide suggestions for the use of these reanalyses with respect to landslide-triggering rainfall in section 4.
To examine the spatial patterns and intensity of rainfall for each event, we use NEXRAD radar observations from the National Centers for Environmental Information (NCEI) radar archive (https://www.ncei.noaa.gov/maps/radar/). We identify the presence or absence of features such as thunderstorms and convective bands within each storm event.
e. Atmospheric variable anomaly analysis
We calculate anomalies for 300-hPa wind speed, IVT, IWV, and CAPE to assess how these characteristics of each event compare to climatology. For each landslide event polygon, we use the values from the ERA5 grid cell central to that polygon. As the horizontal resolution of the ERA5 grid is 31 km, in some cases, the grid cells are larger than the landslide event polygons. For each landslide event window, we determine the maximum value for each variable at each 1-h time step and report the highest value among time steps. We use the 1991–2020 (30 year) period for the December through April season (i.e., the wet season in California) at the same grid cell as a baseline for calculating percentile rankings. To account for spatial uncertainties in the reanalysis dataset, we also examined the eight adjacent grid points for each event but did not find substantial differences in the values and thus chose to report the ERA5 gridcell values central to the landslide polygon.
Additionally, we determine the El Niño–Southern Oscillation (ENSO) phase (El Niño, La Niña, or neutral) associated with the wet season within which each landslide event occurred using the oceanic Niño index (ONI) analysis presented by NOAA CPC (2024).
3. Results
a. Season-to-date rainfall
Considering cumulative rainfall totals for the California water year (beginning 1 October) to landslide date (hereafter WYTD), all but one (Tuolumne 2018) of the 18 landslide events were associated with above-average precipitation. The breakdown of rainfall anomalies across events is as follows and described in Table 2:
-
100%–150% of average: five events
-
151%–200% of average: seven events
-
201%–250% of average: three events
-
>250% of average: two events
The 2018 Tuolumne event observed only 74% of average WYTD precipitation at the time of the event but still had widespread landsliding. This was due to the landslide area experiencing multiple hours of persistent high-intensity rainfall associated with a mesoscale convective band (Collins et al. 2020). In addition, all storms except the 2018 Tuolumne event reached >100% of the long-term average WYTD precipitation at least 7–14 days prior to the landslide event, with a majority (11 of 18 events) reaching and maintaining above-average precipitation more than 30 days prior to the event.
b. Atmospheric moisture variables
1) Atmospheric river detection
We consider AR probability of detection among the 13 storm events rather than 18 landslide events as AR detection periods overlap for landslide events in the same storm event. Summarizing results presented in Table 2, the likelihood of TECA-BARD-based AR detection across the 13 storm events was
-
≥90%: six storms, four of which resulted in more than one widespread landsliding event in the state
-
70%–80%: one storm
-
40%–50%: one storm
-
10%–20%: one storm in Southern California
-
≤10%: four storms, all in Southern California
These probabilities of detection generally agree with other published works that address these storms (e.g., Hatchett et al. 2020; Moore et al. 2020; Rhoades et al. 2023) and author interpretation of ARs in meteorological information. However, two storm/landslide events, the 2005 La Conchita and 2018 Tuolumne events, had probability of AR detection much lower than anticipated. The storm associated with the 2005 La Conchita event features a narrow plume of IWV and IVT exceeding typical AR thresholds extending from the subtropics into Southern California (Fig. S2) and is described as an AR in previous literature (Dettinger et al. 2011). Previous research and media identified an AR immediately preceding the 2018 Tuolumne event [Center for Western Weather and Water Extremes (CW3E) 2018; Weather Underground 2018; Collins et al. 2020], and meteorological analyses show a synoptic feature characteristic of AR events (Fig. S3). It is difficult to determine without substantial analysis why the TECA-BARD algorithm did not recognize these events as having a high likelihood of AR but nevertheless provides an example of the potential for disagreement among algorithms and meteorologist interpretation in AR detection.
2) Atmospheric river scale
We first assess the AR scale classification at a coastal landfall location upstream of each of the 18 landslide events (Fig. 4):
-
AR5: 3
-
AR4: 3
-
AR3: 7
-
AR2: 2
-
AR1: 1
-
No AR conditions: 2
Examples of the AR scale based on Ralph et al. (2019) calculated for a 96-h period prior to the landslide event for (a) the Redding 2017 event and (b) the Santa Paula 2001 event. Colored circles along the coastline represent the AR scale value at each coastal landfall location. The color-filled star provides the AR scale value at the landslide location. Panel (a) corresponds to the event shown in Figs. 3a–c, and (b) corresponds to the event shown in Figs. 3d–f. Shaded relief from Amante and Eakins (2009).
Citation: Earth Interactions 28, 1; 10.1175/EI-D-24-0003.1
The three AR1 and AR2 events occurred in Southern California, as did the two events that did not register on the AR scale. All Northern California and Sierra Nevada events were AR3 or greater at landfall. In Southern California, only two of seven events reached AR3 or greater at landfall.
As an AR moves inland and into elevated terrain, there is typically less water vapor in the vertical column. Additionally, wind speeds are retarded by friction with the land surface. This generally results in lower values of IWV and IVT over land than at the coast (Rutz et al. 2014); thus, a lower AR scale value can be expected. The AR scale summary at landslide location, which varies in distance from coastal landfall location (Fig. 2), is as follows:
-
AR5: 0
-
AR4: 2
-
AR3: 5
-
AR2: 4
-
AR1: 2
-
No AR conditions: 5
Four of the non-AR condition landslide locations were in Southern California, and the fifth was in the Sierra Nevada (2019 Yosemite event). As expected, the AR scale ranking was lower at the landslide location than the coastal location. The four exceptions were for landslide events with coastal locations, where the landfall and landslide location AR scale values were equal (Table 2).
3) Integrated water vapor transport
Whereas the AR scale considers both IVT magnitude and duration, we can also consider the intensity of AR conditions by IVT magnitude alone at each landslide location using categories proposed by Ralph et al. (2019), as noted in Table 2:
-
Exceptional (≥1250 kg m−1 s−1): 0 events
-
Extreme (1000 kg m−1 s−1): 1 event
-
Strong (751–1000 kg m−1 s−1): 2 events
-
Moderate (501–750 kg m−1 s−1): 4 events
-
Weak (250–500 kg m−1 s−1): 10 events
-
None (<250 kg m−1 s−1): 1 event
In some cases, the storm event IVT exceeded the AR threshold (250 kg m−1 s−1) at the landslide location but not for a sufficient duration for the storm to reach a ranking on the AR scale. This typically occurred at inland locations. None of the Southern California landslide event locations were associated with IVT conditions in the moderate or higher range.
Assessing peak IVT anomalies at each landslide location, we find
-
16 of 18 events exceed the 95th percentile IVT
-
12 of 18 events exceed the 99th percentile IVT
The 2006 Klamath event and 2020 San Diego landslide event were below 95th IVT percentile. However, one should be cautious in making the statement that most events feature extreme IVT. Most locations in California typically receive all their precipitation in less than 100 days yr−1 (Pierce et al. 2013), with most precipitation occurring in just a handful of storms per year (e.g., Lamjiri et al. 2018; Oakley et al. 2018b). IVT is generally elevated only during these storm events. For the events and climatological baseline considered here, reaching IVT of 250 kg m−1 s−1 is at least the 88th percentile for all 18 event locations and the 95th percentile for 11 of 18 events. In other words, just by reaching the general IVT threshold for ARs, a storm already exhibits anomalous IVT.
In addition to the magnitude of IVT, its orientation with respect to large-scale terrain is also important for orographic precipitation efficiency. A perpendicular orientation of incoming IVT to terrain is the most favorable (e.g., Lin et al. 2001; Ricciotti and Cordeira 2022). As California mountain ranges trend northwest to southeast or run east–west in the case of Southern California’s Transverse Ranges, an angle of approach from the southwest to south is most effective for orographic precipitation (Hecht and Cordeira 2017; Oakley et al. 2018a,b). Nearly all (17 of 18) events experienced IVT from a south-to-southwest direction (e.g., Fig. 3b). The one exception was the 2001 Santa Paula event (Fig. 3e), where IVT orientation was from the south-southeast; this is still a somewhat favorable orientation for orographic rainfall in this area.
4) Integrated Water Vapor
It is helpful to consider both IVT and IWV when evaluating moisture available for precipitation. IVT is driven by both wind and moisture, thus either may dominate the resultant IVT value (Gonzales et al. 2020). On the other hand, IWV considers only water vapor. For IWV, a value of 20 mm is considered to be a minimum threshold for achieving AR conditions (Ralph et al. 2004; Wick et al. 2013). Figures 3c and 3f provide examples of IWV for two storm events. IWV among the 18 landslide events as reported in Table 2 can be summarized as follows:
-
IWV > 25 mm: 5 events
-
IWV = 20–25 mm: 8 events
-
IWV < 20 mm: 5 events
Some events with IWV less than 20 mm at the landslide location still register on the AR scale, which is based on IVT. Like IVT, IWV is low for much of the wet season. Reaching a threshold of 20 mm for IWV is at least the 85th percentile for all 18 landslide event locations. IWV anomalies can be summarized as follows:
-
>90th percentile IWV: all 18 events
-
>99th percentile IWV: 7 events
For the 2017 Yosemite event, 20 mm is not reached in the historic record analyzed though the 16 mm observed exceeds the 99th percentile.
c. Storm characteristics
1) Mid- to upper-level disturbance type and orientation
We evaluate the synoptic-scale disturbances associated with each of the 13 storm events by placing them into broad categories based on 300- and 500-hPa geopotential height patterns (Table 3):
-
Trough: two storm events (1998 Laguna Beach and 2017 Redding). Troughs are elongated areas of relatively low atmospheric pressure (American Meteorological Society 2012).
-
Trough with embedded short-wave disturbance: four storm events (1997 Klamath/Sourgrass, 2017 East Bay/Yosemite, 2018 Tuolumne, and 2019 Mokelumne/Sausalito/Riverside). Short waves are disturbances whose wavelength is much smaller than the long-wave trough in which they are embedded and can enhance upper-level divergence, resulting in focused areas of precipitation (Aguado and Burt 2007).
-
Closed low: two storm events (1998 East Bay and 2005 La Conchita). A closed low is a trough that has developed one or more closed height contours. As the low closes off, it will tend to propagate downstream more slowly than an open-wave trough potentially producing persistent precipitation over an area (Oakley and Redmond 2014).
-
Closed low with embedded short-wave disturbance: one storm event (1995 Ventura).
-
Cutoff low: three storm events (2001 Santa Paula, 2005 Chino Hills, and 2020 San Diego). A cutoff low is a trough that becomes closed and eventually completely removed from the mean westerly flow and may propagate downstream more slowly than an open-wave trough (Nieto et al. 2008), creating potential for greater precipitation totals. Cutoff lows are climatologically more prevalent in Southern California than Northern California (Abatzoglou 2016; Barbero et al. 2019). Figures 3d and 3e show an example of the cutoff low in the 2001 Santa Paula event.
-
Upper-level jet with embedded short-wave disturbance: one storm event (2006 Marin/Klamath).
Disturbance and jet characteristics for landslide events. Superscripts in the “Landslide event” column indicate events that happened in different locations in the same storm. The “500–300-hPa disturbance characteristics” column characterizes each event [see section 3c(1)]. “Broad” describes troughs that span more than 15° of longitude at their widest point. This distinction is made as narrower troughs tend to have increased upper-level divergence on their downstream side and enhanced precipitation. The “Orientation” provides the orientation of the upper-level disturbance as positive (POS), negative (NEG), or neutral (NEU) orientation, or a dash where not applicable. The “300-hPa jet speed” column provides the wind speed at 300 hPa at the ERA5 grid cell representing the landslide location. The “300-hPa jet speed pctile” column provides the percentile value of the wind speed in the previous column. The “Landslide location WRT jet” provides the location of the landslide event with respect to the upper-level (300 hPa) jet. The “L” and “R” are left and right, and “S” indicates south. The “Surface low location and central MSLP” column describes the location of the center of the MSLP surface low pressure system and provides its lowest central low pressure in the event window. Details on these location categories are provided in section 2b.
The 500-hPa disturbance orientation at the time of the event may provide insight into a storm’s potential for high-intensity rainfall. A negatively tilted trough has an axis-oriented northwest to southeast (relative to a line of longitude) and creates conditions favorable for instability and convection. In this situation, cold air advection occurs at mid-to-upper levels above relatively warm air at lower levels (Macdonald 1976; Oakley et al. 2017). However, instability can still occur within neutral and positively tilted troughs (Figs. 3a,b shows a positively tilted trough), as well as in cutoff low pressure systems (e.g., Figs. 3d,e). For this analysis, we examine all 18 landslide events as trough orientation can change throughout an event. Seven landslide events feature negatively tilted troughs, and three feature positively tilted troughs. Five have a neutral orientation, and the three cutoff lows do not have an orientation (Table 3). Thus, the negative orientation is slightly more prevalent across events.
2) Upper-level jet characteristics
We examine the speed and position of the upper-level (300 hPa) jet stream and associated jet streaks relative to the locations of the landslide events. Stronger upper-level wind speeds may be associated with enhanced upper-level divergence, upward vertical motions, and precipitation. All but two of the landslide events (both cutoff lows) had the jet stream positioned overhead, and all but two landslide events were associated with 300-hPa wind speeds exceeding the 80th percentile (Table 3). One of the two with <80th percentile 300-hPa winds was the 2020 San Diego event, a cutoff low. The landslide impacts occurred under the low center, an area of weak upper-level winds. The other was the 2006 Klamath event, where the landslide location was situated under the periphery of the upper-level jet. Among the 16 landslide events that exceeded the 80th percentile for 300-hPa wind speed, 12 had 300-hPa wind speeds exceeding 90th percentile and five exceeded 99th percentile. Three of these five landslide events occurred during a single storm on 14–15 February 2019. The maximum 300-hPa jet speed among events was 77 m s−1 in the 2019 Sausalito event, which can be considered extreme (Lin 2009).
Jet streaks are areas of local wind maxima embedded within the jet stream. In an idealized scenario, the right entrance and left exit regions of the streak are areas where upper-level divergence occurs, facilitating upward vertical motions and precipitation (Lin 2009). Figure 3a shows an example of a jet streak over Northern California. Additionally, the downstream side of a curved jet is also a favorable region for upper-level divergence, upward vertical motions, and precipitation (Lin 2009).
Though there are a range of scenarios, nearly all investigated landslide events occur in the vicinity of a jet streak or curved jet exit. The position of the landslide location relative to the upper-level jet for each event is described in Table 3. Thus, it is likely that jet dynamics play a role in facilitating upward vertical motions and heavy precipitation in many of these events.
3) Surface low characteristics
The location of a storm’s surface low pressure system influences wind strength and direction as well as moisture transport at a location of interest. Precipitation can occur at a distance from the low center (e.g., ahead of or along the cold front) as well as in the vicinity of the surface low. In our study, we found a mix of surface lows near the location of the landslide events as well as at more distant locations. Whereas all Northern California and Sierra Nevada landslide events had surface lows in the vicinity of either Northern California or the Pacific Northwest (which includes areas offshore of British Columbia), landslide events in Southern California had a broader mix, with four lows situated near Southern California, two in the Pacific Northwest, and one in Northern California (Table 3).
A storm’s intensity is often described by its central MSLP. Storms impacting California with lower MSLP are often associated with stronger winds or higher storm total precipitation (Raphael and Mills 1996). Over land along the U.S. West Coast, historic record low MSLP varies from around 963 hPa in western Washington to 988 hPa near San Diego, California (Weather Prediction Center 2023). The unofficial lowest MSLP recorded in California was 973.4 hPa in Crescent City (NWS Eureka 2019). Many of the events in our catalog have their event minimum MSLP over the ocean; these may be lower than the records over land.
We examine the MSLP minima for each of the 18 landslide events by surface low location:
-
Surface low in Pacific Northwest: six of seven events had MSLP < 990 hPa, two of which (1997 Klamath and 1995 Ventura events) had MSLP < 975 hPa.
-
Surface low in Northern California: six of seven events had MSLP < 1000 hPa, the most extreme of which were the 1998 East Bay (978 hPa) and the 2006 Marin and Klamath events (987 hPa).
-
Surface low in Southern California: All four events had MSLP ≥ 1000 hPa.
Most of the Southern California events are associated with closed or cutoff lows, which are primarily mid- to upper-level features in the atmosphere and are often associated with weak or nonexistent surface lows (Nieto et al. 2008).
4) Low-level winds
The characteristics of low-level winds provide insight into favorable conditions for orographic lift and enhanced rainfall. We evaluate wind speed and direction at 10 m above ground level and 850 hPa (∼1500 m above sea level) for each landslide event (Table 4). There are a variety of factors that affect orographic precipitation outcomes such as the shape of the terrain barrier and stability profile of the atmosphere; however, the amount of precipitation that occurs along the windward slope of the terrain barrier is related to the magnitude of upslope flow (Neiman et al. 2002; Colle 2004). To help contextualize our results, we consider the general classifications used by Colle (2004) to describe flow upstream of a terrain barrier. The 850-hPa-level wind speeds among events can be summarized as follows:
-
High, >30 m s−1: one event
-
Moderate to high, 20–30 m s−1: three events
-
Moderate, 15–20 m s−1: three events
-
Light to moderate, 10–15 m s−1: six events
-
Light, 5–10 m s−1: five events
Low-level wind, CAPE, and radar observations for landslide events. Superscripts in the “Landslide event” column indicate landslide events that occurred in different locations in the same storm. The “850-hPa wind speed and direction” column provides wind speed estimates for the landslide location as well as the wind direction during the event time window at the 850-hPa level, which is approximately 1500 m above sea level. Letters indicate the cardinal, ordinal, and intercardinal directions; VAR indicates variable direction. The “10-m wind speed and direction” column provides wind speed estimates for the landslide location as well as the wind direction during the event time window at 10 m above ground level. The “CAPE” column provides the maximum CAPE value in the landslide time window. The “CAPE percentile” column provides the percentile value of the previous column relative to the December–April climatology at that location. The “Maximum dBZ, radar feature” column provides the value or range of the maximum dBZ values observed in radar imagery in the landslide area during the event window. It also provides a brief description of the characteristics of the radar observations and notes any identifiable features.
One characteristic of ARs is a low-level jet (LLJ), a peak in low-level wind speeds below 1500 m, typically at 1000 m, with speeds in the ∼20–40+ m s−1 range (Ralph et al. 2005; Demirdjian et al. 2020). With the exception of the 2005 Chino Hills event, all events with 850-hPa winds > 20 m s−1 were associated with ARs; thus, the high wind speeds may reflect the passage of the LLJ in the core of the AR during the event window. At the 10-m level, wind speeds among events can be summarized as
-
Light, 5–10 m s−1: 16 events
-
Light to moderate, 10–15 m s−1: 2 events
The most favorable direction for orographic forcing for precipitation depends on the local terrain. At the relatively coarse scale of the ERA5 data with respect to terrain (Fig. 1), low-to-mid-level winds from southerly to a west-southwesterly direction will tend to be favorable for the areas of interest, with southerly flow most favorable for the east–west-oriented Transverse Ranges. At 850 hPa, 15 of 18 the landslide events have winds in the S-SW direction range. Of the three that do not fall in that range, two feature a west-to-southwest direction and one features a south-southeast direction, which are still reasonably favorable directions for orographic lift in their respective locations (Table 4).
5) Convective available potential energy
CAPE is a measure (J kg−1) of the amount of energy in the atmosphere that can be released to create strong upward vertical motions that could support high-intensity rainfall. In the winter season that is the focus of our study, CAPE averages less than approximately 50–100 J kg−1 (Riemann-Campe et al. 2009). Relatively small amounts of CAPE (e.g., ∼200 J kg−1) have been associated with convection and high-intensity rainfall in California (e.g., Jorgensen et al. 2003; Oakley et al. 2017; Cannon et al. 2020) compared to the high values (e.g., >1000 J kg−1) associated with thunderstorm activity in other parts of the United States.
CAPE among the 18 landslide events ranged from 67 to 1094 J kg−1, with a mean of 272 J kg−1. All 18 events had greater than 80th percentile CAPE based on December–April 30-yr climatology. However, 12 of the landslide events had greater than or equal to 95th percentile CAPE values, indicating a relationship between anomalous CAPE and landslide-producing storms. The 2018 Tuolumne event had more than twice the CAPE of any other event in our study (1094 J kg−1) and was associated with a very intense band of rainfall. Interestingly, it was the only event in our landslide catalog that had less than 100% WYTD average precipitation, indicating that extreme CAPE (instability) and associated intense rainfall may be able to overcome antecedent rainfall deficits and still produce landslides. Overall, the wide range of CAPE values suggests that exceedance of a particular threshold is not a critical factor in landslide production though over half of the events were associated with moderate-to-high CAPE values. These results are similar to results of Oakley et al. (2017) who found a range of CAPE values associated with postfire debris flow events in Southern California.
6) Radar reflectivity and mesoscale features
We examine radar observations to assess rainfall intensity and whether any convective bands, thunderstorms, or other identifiable mesoscale features were present at the time of landslide triggering in our event catalog. All 18 events featured areas of moderate intensity rainfall (at least 40 dBZ) within the event window, and seven events had areas of high intensity rainfall (>50 dBZ). Though the 2005 La Conchita event did not have radar observations available at the landslide initiation time, there was a ∼20-km diameter area of 40–55 dBZ at the event location approximately 1.5 h prior to the event time. We assume the presence of similar conditions at the initiation time. Figure 5 provides an example of various patterns of radar reflectivity for periods of high-intensity rainfall during four of the landslide events.
Radar reflectivity from four landslide events. See Table 4 for descriptions. Event and image times are (a) 2006 Marin, 1740 UTC 31 Dec 2005, (b) 2017 Redding, 0140 UTC 10 Feb 2017, (c) 2019 Mokelumne, 1635 UTC 14 Feb 2019, and (d) 2020 San Diego, 2345 UTC 10 Apr 2020. Images sourced from NCEI (https://www.ncei.noaa.gov/maps/radar/).
Citation: Earth Interactions 28, 1; 10.1175/EI-D-24-0003.1
We found a range of mesoscale features present among events with reflectivity values greater than 50 dBZ. The San Diego 2020 event (Fig. 5d) featured isolated intense thunderstorms producing high-intensity rainfall. The 1998 East Bay (Coe and Godt 2001), 2018 Tuolumne (Collins et al. 2020), and 2019 Mokelumne (Fig. 5c) events all featured squall lines, which are associated with high-intensity precipitation. The 2017 Redding event (Fig. 5b) was associated with a Shasta County convergence zone (SCCZ) band, a localized terrain-forced convergence zone resulting in quasi-stationary convective precipitation bands (Roberts 2019). Though some landslide events were associated with very high-intensity rainfall (>50 dBZ), not all events occurred under these conditions. Very high-intensity rainfall is not a critical factor for shallow landslide initiation; published intensity–duration thresholds often cite moderate intensities at 1 h or more durations (e.g., Oakley et al. 2018a and references therein). This contrasts with postfire debris flows, where triggering generally requires high-intensity rainfall to mobilize surficial sediment into channels resulting in >50 dBZ values in most events (e.g., Oakley et al. 2017).
d. El Niño–Southern Oscillation
ENSO has long been looked to as a large-scale predictor of seasonal precipitation outcomes in California and the western United States. Historically, El Niño events, especially very strong episodes, were associated with a greater likelihood of above-normal precipitation in California (e.g., Redmond and Koch 1991; Cayan et al. 1999; Mo and Higgins 1998; Hoell et al. 2016). The presence of El Niño conditions increases the likelihood of wetter than normal conditions in far Southern California; however, for central and northern portions of the state, especially during weak to moderate El Niño conditions, the relationship is weaker, more variable, or not present at all (Hoell et al. 2016; Lee et al. 2018; Western Regional Climate Center 2022). While ENSO can influence the strength and position of the upper-level jet and seasonal precipitation in California, connecting ENSO to mesoscale precipitation processes at shorter time scales relevant to landslide triggering to is more tenuous (e.g., Li et al. 2020). Thus, we document the ENSO conditions associated with each event to evaluate its usefulness as a predictor of landslide activity (Table 5). Among the 13 individual storm events statewide, the ENSO breakdown is as follows:
-
El Niño: 6 (46%)
-
Neutral: 2 (15%)
-
La Niña: 5 (38%)
The El Niño–Southern Oscillation phase associated with each landslide event based on the ONI. Superscripts in the “Event” column indicate landslide events that occurred in different locations in the same storm event.
For the seven unique Northern California/Sierra Nevada storm events (i.e., counting multiple landslide events during a single storm only once), the ENSO breakdown is as follows:
-
El Niño: 2
-
Neutral: 1
-
La Niña: 4
For the seven Southern California storm events, the ENSO breakdown is as follows:
-
El Niño: 5
-
Neutral: 1
-
La Niña: 1
These results suggest that shallow landslides in California can occur in any ENSO phase but may have a slightly greater likelihood of occurring in Southern California during El Niño events.
4. Discussion
Our results are similar to other studies that examine atmospheric conditions associated with rainfall-driven landslides movements in California though previous studies examined fewer atmospheric variables. Studies focused on Northern California found most of these events were associated with AR conditions (Young et al. 2017; Cordeira et al. 2019). In contrast, Biasutti et al. (2016) indicates that, for a few events in the Bay Area, landslide triggering occurred without a well-defined AR and rather with orographic precipitation associated with a frontal passage following record precipitation. These studies and Pike and Sobieszczyk (2008) note the importance of orographic precipitation and found that south-to-southwest orientation of low-level winds and/or IVT was prominent in their impacted focus areas of Northern California. Through compositing, studies also identify a surface low along the Pacific Northwest coast as a common feature during landslide events (Young et al. 2017; Cordeira et al. 2019), similar to our findings.
In Southern California, previous studies (Biasutti et al. 2016; Young et al. 2017; Oakley et al. 2018a) also identified landslide-related events as featuring ARs. These studies tended to have a lower fraction of events meeting AR criteria than seen for Northern California events, as in this study. Predominantly southerly low-level winds and/or IVT favorable for orographic precipitation in the east–west-oriented Transverse Ranges are also prevalent in these analyses. As in this study, surface low locations for impactful Southern California storms vary from offshore of Southern California to the Pacific Northwest (Young et al. 2017).
The attributes of landslide-triggering storms found in this study and others are also characteristic of moderate-to-significant rainfall-producing storms in California that may or may not have triggered landslides. For example, heavy rainfall and flood events throughout California are often associated with ARs of varying strength (e.g., Ralph et al. 2006; Konrad and Dettinger 2017; Cannon et al. 2018; Corringham et al. 2019). Closed and cutoff lows, as well as short-wave disturbances, have also been found to feature prominently in Southern California heavy precipitation events (Haynes 2001; Cannon et al. 2018; Oakley et al. 2018b). Many studies indicate the importance of south-to-southwesterly IVT and low-level winds as favorable for orographic precipitation and as playing a key role in major California precipitation events (e.g., Ralph et al. 2005; Hecht and Cordeira 2017; Oakley et al. 2018b). Studies of California extreme precipitation also note the importance of upper-level jet characteristics (e.g., Haynes 2001; Hecht and Cordeira 2017; Oakley et al. 2018b). The considerable overlap between our catalog of landslide storms and other storms of hydrologic interest limits our ability to draw substantial conclusions that might aid in significantly increasing warning times for potential landslide hazards based on atmospheric variables alone.
These results underscore the continued need for maintaining and expanding antecedent soil moisture and precipitation-based monitoring in California (e.g., Collins et al. 2012; Thomas et al. 2018) and elsewhere, especially at the hourly to subhourly temporal resolution, as many landslide-triggering rainfall intensity–duration thresholds consider hourly data (e.g., Guzzetti et al. 2020; Oakley et al. 2018a). Future work could evaluate newly available high-resolution (≤4 km, 1 h) gridded precipitation reanalyses (Rasmussen et al. 2023; Rahimi et al. 2022) to assess time series of hourly precipitation at various durations leading up to landslide events to determine whether common hyetographs exist and how they relate to soil moisture conditions. Improvements in atmospheric modeling, such as additional high spatial (≤3 km) and temporal (≤1 h) resolution forecast models, ensemble forecasts, and evaluation of high temporal resolution forecasts could also inform the potential for exceedance of rainfall-based landslide-triggering thresholds.
As California covers nearly 10° of latitude with significant climate variations, we could not meaningfully composite atmospheric and geomorphic characteristics across all events, thus limiting conclusions about common storm characteristics. Future work on this topic may achieve greater success by focusing on smaller, more climatologically and geomorphically homogenous regions. A locally or regionally focused study could also create an opportunity to evaluate null cases, where identified favorable atmospheric conditions were present and landslides did not occur.
Our limited sample size of 18 landslide and 13 storm events prevented us from performing additional detailed geomorphic analyses such as examining the relationship among vegetation, geology, and storm characteristics. This challenge of small sample size results from the difficulty in constraining landslide timing. We advocate for additional studies that provide landslide inventories as well as the timing associated with triggering.
5. Conclusions
We examined the atmospheric characteristics of 18 impactful shallow landslide events across California associated with 13 unique storm events. Some storm characteristics were nearly ubiquitous across landslide events, such as extreme IVT relative to climatology at the landslide location, low-level winds favorable for orographically forced precipitation, or positioning of the upper-level jet relative to the landslide location in an area favorable for upward vertical motions. Other characteristics were variable across events, such as atmospheric river (AR) strength, synoptic patterns, or presence of mesoscale convective features. The common atmospheric characteristics observed across landslide events share many characteristics of other storm events of hydrologic interest in California. Therefore, from the findings of this study, we cannot definitively say whether incoming storms with these characteristics will or will not trigger landslides. However, given the prevalence of certain characteristics across events, our results suggest that landslides may be less likely to occur when these atmospheric conditions are not present. For example, all Northern California events featured moderate to strong AR conditions, while Southern California events occurred primarily under weak or non-AR conditions. Thus, it may be less likely for landslides to occur in Northern California during weak or non-AR conditions, while landslides are possible in Southern California under a broader range of AR conditions. Other key findings of similarities and differences across events include
-
Antecedent rainfall is important; all but one event had reached above average WYTD precipitation.
-
The one below-average WYTD precipitation event uniquely experienced several hours of intense rainfall associated with a slow-moving convective band and extreme CAPE, suggesting that the mesoscale characteristics of some storms can overcome below-average antecedent rainfall conditions and still trigger landslides.
-
ARs were more prominent in landslide-producing storms in Northern California than in Southern California. Six of thirteen storm events had a high (>90%) probability of AR detection, and two had moderate probability of detection (40%–80%). The remaining five with a low (≤10%) probability of AR detection were all located in Southern California.
-
The strength of AR conditions in landslide-producing storms trended higher in Northern California than in Southern California. Considering the AR scale (Ralph et al. 2019), all Northern California and Sierra Nevada events were associated with AR3 or higher storms, with the majority as AR3. In Southern California, non-AR and AR 1–2 events were primarily associated with landslide triggering.
-
Extreme IWV and IVT values were present in many events; 12 of 18 landslide events featured IVT values > 99th percentile and 7 of 18 had IWV > 99th percentile considering a wet season (December–April) climatology.
-
Flow direction favorable for orographic forcing is important among landslide events; 17 of 18 landslide events featured IVT from a southerly-to-southwesterly orientation relative to event location. Low-level wind directions were also favorable for orographic forcing in nearly all events as well.
-
The presence, strength, and position of the upper-level jet were relevant in most landslide events; 16 of 18 events had a 300-hPa jet overhead and 12 of 18 had >90th 300-hPa wind speeds. Nearly all events occurred in the vicinity of an upper-level jet streak or curved jet exit, areas that are generally favorable for enhanced upward vertical motions and precipitation.
-
Landslide-producing storms exhibited a variety of mid-to-upper-level synoptic features. Embedded short-wave disturbances were most prevalent for Northern California, whereas closed and cutoff lows were most prevalent among Southern California events.
-
High-intensity rainfall features (>50 dBZ in radar observations) played a role in 7 of 18 landslide events, typically in the form of convective bands or isolated convection.
-
Landslide-producing storms in California occurred in all ENSO phases: six during El Niño, five in La Niña, and two in neutral conditions. El Niño appears to have a larger effect in Southern California compared to Northern California; only two of seven landslide events in Northern California/Sierra Nevada were associated with El Niño conditions, whereas five of seven in Southern California occurred during El Niño conditions.
Our landslide catalog and analysis results can be used as a resource to compare historic landslide events to future or forecast scenarios, which may assist in evaluating and preparing for hazards in operational settings. Documenting and describing these storm characteristics can also help bridge the gap between the disciplines of meteorology and geomorphology. This assessment of landslide-triggering storms demonstrates the complex relationship among antecedent precipitation, storm characteristics, and landslide occurrence and illuminates additional research and monitoring efforts needed to improve landslide early warning.
Acknowledgments.
Funding for this study was provided by the U.S. Geological Survey Landslide Hazards Program and the California Department of Water Resources Atmospheric River Program, Contract 4600014294. We thank Brian Kawzenuk at CW3E/UCSD/SIO for visualizing the Atmospheric River Scale and Benjamin Hatchett at CSU/CIRA for support in visualizing the TECA-BARD atmospheric river detection algorithm for this study. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. government.
Data availability statement.
ERA5 reanalysis data used in this study are available from Copernicus Climate Services at https://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-complete?tab=form. PRISM data used in this study are available from the PRISM Climate Group at Oregon State University at https://www.prism.oregonstate.edu/. Radar data used in this study are available from the National Centers for Environmental Information at https://www.ncei.noaa.gov/maps/radar/. Oceanic Niño index data used in this study are available from the NOAA Climate Prediction Center at https://origin.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ONI_v5.php.
REFERENCES
Abatzoglou, J. T., 2016: Contribution of cutoff lows to precipitation across the United States. J. Appl. Meteor. Climatol., 55, 893–899, https://doi.org/10.1175/JAMC-D-15-0255.1.
Aguado, E., and J. E. Burt, 2007: Understanding Weather and Climate. Pearson Prentice Hall, 562 pp.
Amante, C., and B. W. Eakins, 2009: ETOPO1 1 arc-minute global relief model: Procedures, data sources and analysis. NOAA Tech. Memo. NESDIS NGDC-24, 25 pp., https://doi.org/10.7289/V5C8276M.
American Meteorological Society, 2012: Trough. Glossary of Meteorology, https://glossary.ametsoc.org/wiki/Trough.
Barbero, R., J. T. Abatzoglou, and H. J. Fowler, 2019: Contribution of large-scale midlatitude disturbances to hourly precipitation extremes in the United States. Climate Dyn., 52, 197–208, https://doi.org/10.1007/s00382-018-4123-5.
Biasutti, M., R. Seager, and D. B. Kirschbaum, 2016: Landslides in West Coast metropolitan areas: The role of extreme weather events. Wea. Climate Extremes, 14, 67–79, https://doi.org/10.1016/j.wace.2016.11.004.
California-Nevada River Forecast Center, 2023: Storm summaries. Accessed 22 April 2023, https://www.cnrfc.noaa.gov/storm_summaries.php.
Campbell, R. H., 1975: Soil slips, debris flows, and rainstorms in the Santa Monica Mountains and vicinity, southern California. USGS Professional Paper 851, 51 pp., https://doi.org/10.3133/pp851.
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, 13 714–13 730, https://doi.org/10.1029/2018JD029045.
Cannon, F., and Coauthors, 2020: Observations and predictability of a high‐impact narrow cold‐frontal rainband over Southern California on 2 February 2019. Wea. Forecasting, 35, 2083–2097, https://doi.org/10.1175/WAF-D-20-0012.1.
Cayan, D. R., K. T. Redmond, and L. G. Riddle, 1999: ENSO and hydrologic extremes in the western United States. J. Climate, 12, 2881–2893, https://doi.org/10.1175/1520-0442(1999)012<2881:EAHEIT>2.0.CO;2.
CGS, 2024: Reported California landslides database. Accessed 25 February 2024, https://cadoc.maps.arcgis.com/apps/webappviewer/index.html?id=bc48ad40e3504134a1fc8f3909659041.
Chattopadhyay, A., E. Nabizadeh, and P. Hassanzadeh, 2020: Analog forecasting of extreme-causing weather patterns using deep learning. J. Adv. Model. Earth Syst., 12, e2019MS001958, https://doi.org/10.1029/2019MS001958.
Coe, J. A., and J. W. Godt, 2001: Debris flows triggered by the El Niño rainstorm of February 2–3, 1998, Walpert Ridge and Vicinity, Alameda County, California. U.S. Geological Survey Miscellaneous Field Studies Map MF-2384, 22 pp., https://pubs.usgs.gov/mf/2002/mf-2384/.
Coe, J. A., J. W. Godt, and P. Tachker, 2004: Map showing recent (1997–98 El Niño) and historical landslides, Crow Creek and vicinity, Alameda and Contra Costa Counties, California. U.S. Geological Survey Scientific Investigations Map, SIM 2859, 16 pp., https://pubs.usgs.gov/sim/2004/2859/.
Colle, B. A., 2004: Sensitivity of orographic precipitation to changing ambient conditions and terrain geometries: An idealized modeling perspective. J. Atmos. Sci., 61, 588–606, https://doi.org/10.1175/1520-0469(2004)061<0588:SOOPTC>2.0.CO;2.
Collins, B. D., and S. C. Corbett, 2019: Terrestrial lidar data of the February 14, 2019, Sausalito Boulevard landslide, Sausalito, California. U.S. Geological Survey Data Series 1112, 12 pp., https://doi.org/10.3133/ds1112.
Collins, B. D., and M. E. Reid, 2020: Enhanced landslide mobility by basal liquefaction: The 2014 State Route 530 (Oso), Washington, landslide. Geol. Soc. Amer. Bull., 132, 451–476, https://doi.org/10.1130/B35146.1.
Collins, B. D., J. D. Stock, K. A. Foster, M. P. W. Whitman, and N. E. Knepprath, 2012: Monitoring the subsurface hydrologic response for precipitation induced shallow landsliding in the San Francisco Bay area, California, U.S.A. Landslides and Engineering Slopes: Protecting Society through Improved Understanding, E. Eberhardt et al., Eds., Taylor and Francis Group, 1249–1255.
Collins, B. D., N. S. Oakley, J. P. Perkins, A. E. East, S. C. Corbett, and B. J. Hatchett, 2020: Linking mesoscale meteorology with extreme landscape response: Effects of Narrow Cold Frontal Rainbands (NCFRs). J. Geophys. Res. Earth Surf., 125, e2020JF005675, https://doi.org/10.1029/2020JF005675.
Corbett, S. C., and B. D. Collins, 2023: Landslides triggered by the 2016–2017 storm season, San Francisco East Bay region, California. U.S. Geological Survey Science Investigations Map 3503, 1 pp., https://doi.org/10.3133/sim3503.
Cordeira, J. M., J. Stock, M. D. Dettinger, A. M. Young, J. F. Kalansky, and F. M. Ralph, 2019: A 142-year climatology of northern California landslides and atmospheric rivers. Bull. Amer. Meteor. Soc., 100, 1499–1509, https://doi.org/10.1175/BAMS-D-18-0158.1.
Corringham, T. W., F. M. Ralph, A. Gershunov, D. R. Cayan, and C. A. Talbot, 2019: Atmospheric rivers drive flood damages in the western United States. Sci. Adv., 5, eaax4631, https://doi.org/10.1126/sciadv.aax4631.
CW3E, 2018: AR Update: 22 March outlook. Center for Western Weather and Water Extremes, accessed 25 February 2024, https://cw3e.ucsd.edu/cw3e-ar-update-22-march-2018-outlook/.
DeGraff, J. V., 1994: The geomorphology of some debris flows in the southern Sierra Nevada, California. Geomorphology, 10, 231–252, https://doi.org/10.1016/0169-555X(94)90019-1.
Demirdjian, R., J. R. Norris, A. Martin, and F. M. Ralph, 2020: Dropsonde observations of the ageostrophy within the pre-cold-frontal low-level jet associated with atmospheric rivers. Mon. Wea. Rev., 148, 1389–1406, https://doi.org/10.1175/MWR-D-19-0248.1.
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, 445–478, https://doi.org/10.3390/w3020445.
Ellen, S. D., and G. F. Wieczorek, 1988: Landslides, floods, and marine effects of the storm of January 3–5, 1982, in the San Francisco Bay region, California. USGS Professional Paper 1434, 311 pp. , https://doi.org/10.3133/pp1434.
Gonzales, K. R., D. L. Swain, E. A. Barnes, and N. S. Diffenbaugh, 2020: Moisture- versus wind-dominated flavors of atmospheric rivers. Geophys. Res. Lett., 47, e2020GL090042, https://doi.org/10.1029/2020GL090042.
Guan, B., and D. E. Waliser, 2019: Tracking atmospheric rivers globally: Spatial distributions and temporal evolution of life cycle characteristics. J. Geophys. Res. Atmos, 124, 12 523–12 552, https://doi.org/10.1029/2019JD031205.
Guzzetti, F., S. L. Gariano, S. Peruccacci, M. T. Brunetti, I. Marchesini, M. Rossi, and M. Melillo, 2020: Geographical landslide early warning systems. Earth-Sci. Rev., 200, 102973, https://doi.org/10.1016/j.earscirev.2019.102973.
Hatchett, B. J., and Coauthors, 2020: Observations of an extreme atmospheric river storm with a diverse sensor network. Earth Space Sci., 7, e2020EA001129, https://doi.org/10.1029/2020EA001129.
Haynes, A., 2001: Synoptic pattern typing for heavy precipitation events in southern California. Western Region Tech. Attachment NWS 01-15, 8 pp., https://www.weather.gov/media/wrh/online_publications/TAs/ta0115.pdf.
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, 9048–9058, https://doi.org/10.1002/2017GL074179.
Hersbach, H., and Coauthors, 2018: ERA5 hourly data on single levels from 1940 to present. Copernicus Climate Change Service (C3S) Climate Data Store (CDS), accessed 10 May 2021, https://doi.org/10.24381/cds.adbb2d47.
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, 819–825, https://doi.org/10.1002/2015GL067102.
Iverson, R. M., 2000: Landslide triggering by rain infiltration. Water Resour. Res., 36, 1897–1910, https://doi.org/10.1029/2000WR900090.
Jenkins, O. P., 1938: Geomorphic provinces of California as outlined on the New State Geologic Map: ABSTRACT. AAPG Bull., 22, 1717.
Jibson, R. W., 2005: Landslide hazards at La Conchita, California. USGS Open-File Rep. 2005-1067, 12 pp., https://pubs.usgs.gov/of/2005/1067/508of05-1067.html.
Jorgensen, D. P., Z. Pu, P. O. G. Persson, and W. Tao, 2003: Variations associated with cores and gaps of a Pacific narrow cold frontal rainband. Mon. Wea. Rev., 131, 2705–2729, https://doi.org/10.1175/1520-0493(2003)131<2705:VAWCAG>2.0.CO;2.
Keefer, D. K., and Coauthors, 1987: Real-time landslide warning during heavy rainfall. Science, 238, 921–925, https://doi.org/10.1126/science.238.4829.921.
Konrad, C. P., and M. D. Dettinger, 2017: Flood runoff in relation to water vapor transport by atmospheric rivers over the western United States, 1949–2015. Geophys. Res. Lett., 44, 11 456–11 462, https://doi.org/10.1002/2017GL075399.
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, 17, https://doi.org/10.15447/sfews.2018v16iss4art1.
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, 907–915, https://doi.org/10.1002/2017GL076197.
Li, X.-F., and Coauthors, 2020: Global distribution of the intensity and frequency of hourly precipitation and their responses to ENSO. Climate Dyn., 54, 4823–4839, https://doi.org/10.1007/s00382-020-05258-7.
Lin, Y.-L., 2009: Mesoscale Dynamics. Cambridge University Press, 630 pp., https://doi.org/10.1017/CBO9780511619649.
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, 633–660, https://doi.org/10.1175/1520-0434(2001)016%3C0633:SCIFHO%3E2.0.CO;2.
Lu, N., and J. W. Godt, 2013: Hillslope Hydrology. Cambridge University Press, 437 pp.
Macdonald, N. J., 1976: On the apparent relationship between convective activity and the shape of 500 mb troughs. Mon. Wea. Rev., 104, 1618–1622, https://doi.org/10.1175/1520-0493(1976)104<1618:OTARBC>2.0.CO;2.
Mo, K. C., and R. W. Higgins, 1998: Tropical influences on California precipitation. J. Climate, 11, 412–430, https://doi.org/10.1175/1520-0442(1998)011<0412:TIOCP>2.0.CO;2.
Moody’s RMS, 2023: Moody’s RMS estimates US$5-7 billion in total U.S. economic losses from California Flooding. Moody’s RMS, accessed 25 February 2024, www.rms.com/newsroom/press-releases/press-detail/2023-01-25/moodys-rms-estimates-us5-7-billion-in-total-us-economic-losses-from-california-flooding.
Moore, B. J., A. B. White, D. J. Gottas, and P. J. Neiman, 2020: Extreme precipitation events in Northern California during winter 2016–17: Multiscale analysis and climatological perspective. Mon. Wea. Rev., 148, 1049–1074, https://doi.org/10.1175/MWR-D-19-0242.1.
Neiman, P. J., F. M. Ralph, A. B. White, D. E. Kingsmill, and P. O. G. Persson, 2002: The statistical relationship between upslope flow and rainfall in California’s coastal mountains: Observations during CALJET. Mon. Wea. Rev., 130, 1468–1492, https://doi.org/10.1175/1520-0493(2002)130<1468:TSRBUF>2.0.CO;2.
Neiman, P. J., F. M. Ralph, G. A. Wick, J. D. Lundquist, and M. D. Dettinger, 2008: Meteorological characteristics and overland precipitation impacts of atmospheric rivers affecting the west coast of North America based on eight years of SSM/I satellite observations. J. Hydrometeor., 9, 22–47, https://doi.org/10.1175/2007JHM855.1.
Nieto, R., M. Sprenger, H. Wernli, R. M. Trigo, and L. Gimeno, 2008: Identification and climatology of cut-off lows near the tropopause. Ann. N. Y. Acad. Sci., 1146, 256–290, https://doi.org/10.1196/annals.1446.016.
NOAA CPC, 2024: Cold and warm episodes by season. NOAA Climate Prediction Center, accessed 3 March 2024, https://origin.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ONI_v5.php.
NOAA-USGS Debris Flow Task Force, 2005: NOAA-USGS debris-flow warning system-Final report. U.S. Geological Survey Circular 1283, 60 pp., http://pubs.usgs.gov/circ/2005/1283/.
NWS Eureka, 2019: Unofficial record low pressure at Crescent City. Twitter, accessed 29 June 2023, https://twitter.com/NWSEureka/status/1199547731708006402.
Oakley, N. S., and K. T. Redmond, 2014: A climatology of 500-hPa closed lows in the northeastern Pacific Ocean, 1948–2011. J. Appl. Meteor. Climatol., 53, 1578–1592, https://doi.org/10.1175/JAMC-D-13-0223.1.
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, 327–354, https://doi.org/10.1007/s11069-017-2867-6.
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.
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, 164–176, https://doi.org/10.1016/j.ejrh.2018.09.001.
O’Brien, T. A., and Coauthors, 2020: Detection of atmospheric rivers with inline uncertainty quantification: TECA-BARD v1.0.1. Geosci. Model Dev., 13, 6131–6148, https://doi.org/10.5194/gmd-13-6131-2020.
Pierce, D. W., and Coauthors, 2013: The key role of heavy precipitation events in climate model disagreements of future annual precipitation changes in California. J. Climate, 26, 5879–5896, https://doi.org/10.1175/JCLI-D-12-00766.1.
Pike, R. J., and S. Sobieszczyk, 2008: Soil slip/debris flow localized by site attributes and wind-driven rain in the San Francisco Bay region storm of January 1982. Geomorphology, 94, 290–313, https://doi.org/10.1016/j.geomorph.2006.09.024.
PRISM Climate Group, 2014: Prism Climate Data. Oregon State University, accessed 16 December 2020, https://prism.oregonstate.edu.
Rahimi, S., W. Krantz, Y.-H. Lin, B. Bass, N. Goldenson, A. Hall, Z. J. Lebo, and J. Norris, 2022: Evaluation of a reanalysis-driven configuration of WRF4 over the western United States from 1980 to 2020. J. Geophys. Res. Atmos., 127, e2021JD035699, https://doi.org/10.1029/2021JD035699.
Ralph, F. M., P. J. Neiman, and G. A. Wick, 2004: Satellite and CALJET Aircraft observations of atmospheric rivers over the eastern North Pacific Ocean during the winter of 1997/98. Mon. Wea. Rev., 132, 1721–1745, https://doi.org/10.1175/1520-0493(2004)132<1721:SACAOO>2.0.CO;2.
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, 889–910, https://doi.org/10.1175/MWR2896.1.
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.
Ralph, F. M., J. J. Rutz, J. M. Cordeira, M. Dettinger, M. Anderson, D. Reynolds, L. J. Schick, and C. Smallcomb, 2019: A scale to characterize the strength and impacts of atmospheric rivers. Bull. Amer. Meteor. Soc., 100, 269–289, https://doi.org/10.1175/BAMS-D-18-0023.1.
Raphael, M., and G. Mills, 1996: The role of mid-latitude Pacific cyclones in the winter precipitation of California. Prof. Geogr., 48, 251–262, https://doi.org/10.1111/j.0033-0124.1996.00251.x.
Rasmussen, R. M., and Coauthors, 2023: CONUS404: The NCAR–USGS 4-km long-term regional hydroclimate reanalysis over the CONUS. Bull. Amer. Meteor. Soc., 104, E1382–E1408, https://doi.org/10.1175/BAMS-D-21-0326.1.
Redmond, K. T., and R. W. Koch, 1991: Surface climate and streamflow variability in the western United States and their relationship to large-scale circulation indices. Water Resour. Res., 27, 2381–2399, https://doi.org/10.1029/91WR00690.
Rhoades, A. M., and Coauthors, 2023: Recreating the California New Year’s flood event of 1997 in a regionally refined Earth System Model. J. Adv. Model. Earth Syst., 15, e2023MS003793, https://doi.org/10.1029/2023MS003793.
Ricciotti, J. A., and J. M. Cordeira, 2022: Summarizing relationships among landfalling atmospheric rivers, integrated water vapor transport, and California watershed precipitation 1982–2019. J. Hydrometeor., 23, 1439–1454, https://doi.org/10.1175/JHM-D-21-0119.1.
Riemann-Campe, K., K. Fraedrich, and F. Lunkeit, 2009: Global Climatology of Convective Available Potential Energy (CAPE) and Convective Inhibition (CIN) in ERA-40 reanalysis. Atmos. Res., 93, 534–545, https://doi.org/10.1016/j.atmosres.2008.09.037.
Roberts, M. T., 2019: Mesoscale controls on the Shasta County convergence zone. M.S. thesis, Dept. of Meteorology and Climate Science, San José State University, 105 pp., https://doi.org/10.31979/etd.sq5w-e3a9.
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, 905–921, https://doi.org/10.1175/MWR-D-13-00168.1.
Shields, C. A., and Coauthors, 2018: Atmospheric River Tracking Method Intercomparison Project (ARTMIP): Project goals and experimental design. Geosci. Model Dev., 11, 2455–2474, https://doi.org/10.5194/gmd-11-2455-2018.
Taylor, F. A., and E. E. Brabb, 1972: Maps showing distribution and cost by counties of structurally damaging landslides in the San Francisco Bay Region, California, Winter of 1968–69. U.S. Geological Survey Miscellaneous Field Studies Map MF-327, 1 pp., https://pubs.usgs.gov/mf/0327/plate-1.pdf.
Taylor, F. A., and E. E. Brabb, 1987: Map showing landslides in California that have caused fatalities or at least $1,000,000 in damages from 1906 to 1984. U.S. Geological Survey Miscellaneous Field Studies Map 1867, 1 pp., https://www.usgs.gov/maps/map-showing-landslides-california-have-caused-fatalities-or-least-1000000-damages-1906-1984.
Thomas, M. A., B. B. Mirus, and B. D. Collins, 2018: Identifying physics-based thresholds for rainfall-induced landsliding. Geophys. Res. Lett., 45, 9651–9661, https://doi.org/10.1029/2018GL079662.
Tozer, B., D. T. Sandwell, W. H. F. Smith, C. Olson, J. R. Beale, and P. Wessel, 2019: Global bathymetry and topography at 15 arc sec: SRTM15+. Earth Space Sci., 6, 1847–1864, https://doi.org/10.1029/2019EA000658.
Varnes, D. J., 1978: Slope movement types and processes. Special Report 176: Landslides: Analysis and Control, R. L. Schuster and R. J. Krizek, Transportation and Road Research Board, National Academy of Science, 11–33, http://onlinepubs.trb.org/Onlinepubs/sr/sr176/176-002.pdf.
Wartman, J., D. R. Montgomery, S. A. Anderson, J. R. Keaton, J. Benoît, J. dela Chapelle, and R. Gilbert, 2016: The 22 March 2014 Oso landslide, Washington, USA. Geomorphology, 253, 275–288, https://doi.org/10.1016/j.geomorph.2015.10.022.
Weather Prediction Center, 2023: All-time record low sea level pressures. Accessed 22 April 2023, https://www.wpc.ncep.noaa.gov/research/roth/SLPrecords.html.
Weather Underground, 2018: Atmospheric River Slams California with coastal flooding, severe thunderstorm threat. Accessed 25 February 2024, https://www.wunderground.com/cat6/atmospheric-river-slams-california-coastal-flooding-severe-thunderstorm-threat.
Weaver, R. L., 1962: Meteorology of hydrologically critical storms in California. Hydrometeorlogical Rep. 37, 227 pp., https://www.weather.gov/media/owp/hdsc_documents/PMP/HMR37.pdf.
Western Regional Climate Center, 2022: SOI-precipitation relationships. Accessed 29 April 2023, https://wrcc.dri.edu/Climate/soi_precip.php.
Wick, G. A., P. J. Neiman, and F. M. Ralph, 2013: Description and validation of an automated objective technique for identification and characterization of the integrated water vapor signature of atmospheric rivers. IEEE Trans. Geosci. Remote Sens., 51, 2166–2176, https://doi.org/10.1109/TGRS.2012.2211024.
Wieczorek, G. F., 2002: Catastrophic rockfalls and rockslides in the Sierra Nevada, USA. Catastrophic Landslides: Effects, Occurrence, and Mechanisms, S. G. Evans and J. V. Degraff, Eds., Reviews in Engineering Geology, Vol. 15, Geological Society of America, 165–190, https://doi.org/10.1130/REG15-p165.
Young, A. M., K. T. Skelly, and J. M. Cordeira, 2017: High-impact hydrologic events and atmospheric rivers in California: An investigation using the NCEI Storm Events Database. Geophys. Res. Lett., 44, 3393–3401, https://doi.org/10.1002/2017GL073077.
