Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

npj | natural hazards Article

https://doi.org/10.1038/s44304-024-00017-8

Rainfall intensification amplifies exposure


of American Southwest to conditions that
trigger postfire debris flows
Check for updates
1 2,3 3,4 1 2
Matthew A. Thomas , Allison C. Michaelis , Nina S. Oakley , Jason W. Kean , Victor A. Gensini &
Walker S. Ashley2

Short-duration, high-intensity rainfall can initiate deadly and destructive debris flows after wildfire.
Methods to estimate the conditions that can trigger debris flows exist and guidance to determine how
1234567890():,;
1234567890():,;

often those thresholds will be exceeded under the present climate are available. However, the limited
spatiotemporal resolution of climate models has hampered efforts to characterize how rainfall
intensification driven by global warming may affect debris-flow hazards. We use novel, dynamically
downscaled (3.75-km), convection-permitting simulations of short-duration (15-min) rainfall to
evaluate threshold exceedance for late 21st-century climate scenarios in the American Southwest. We
observe significant increases in the frequency and magnitude of exceedances for regions dominated
by cool- and warm-season rainfall. We also observe an increased frequency of exceedance in regions
where postfire debris flows have not been documented, and communities are unaccustomed to the
hazard. Our findings can inform planning efforts to increase resiliency to debris flows under a changing
climate.

The immediate effects of wildfire—loss of life, destruction of property, double the expected rate of intensification following the CC relation12,15–17.
interruption of critical services, and poor air quality1–3—can be devastating. Although projections of rainfall intensification vary across different dura-
Wildfire is also known to enhance susceptibility to flash floods and debris tions, spatial extents, regional climates, and storm characteristics12,18–21, there
flows triggered by short-duration, high-intensity rainfall in steep and is substantial evidence that the short-duration (<1 h) rainfall conducive to
severely burned terrain4,5. Postfire debris flows, which are fueled by the postfire debris flows will continue to intensify, and perhaps at a higher rate
entrainment of sediment on hillslopes and in channels, exhibit notably more than long-duration (≥1 h) rainfall14,15.
destructive power than flash floods owing to amplification of flow depth, Postfire debris flows have been documented throughout the American
discharge, and shear stress6 that is imparted by the most dangerous con- Southwest, including areas dominated by warm-season (June–September)
stituents (mud, boulders, and downed trees) of these gravity- and fluid- and cool-season (October–May) rainfall22. Most of the postfire debris flows
driven mass movements (Fig. 1b). They can be triggered by 15-min rainfall that have been observed in coastal areas of the Southwest (e.g., southern
intensities7 with recurrence intervals as low as one year8 in the first two years California) have occurred during the cool season. During this time, cool-
after fire9. Major postfire debris flows, which have caused substantial season precipitation in the western United States is primarily dominated by
damage or loss of life, have been initiated by relatively low rainfall intensities extratropical cyclone/frontal systems, which may host mesoscale convective
and by rainfall intensities greatly exceeding the baseline triggering features, such as narrow cold frontal rainbands that are capable of producing
thresholds10,11. Studies have projected that extreme precipitation at sub-daily intense rainfall that meets or exceeds postfire debris-flow triggering
periods will scale at or above the Clausius-Clapeyron (CC) relationship12–15, thresholds23–25. These storms may also feature atmospheric rivers, although
which states that the moisture-holding capacity in the atmosphere increases the presence, absence, or strength of an atmospheric river is not a reliable
at a rate of approximately 7% per degree Celsius. Some observational and predictor of whether short-duration, high-intensity rainfall conducive to
climate projection studies note substantial “super-CC” scaling at up to postfire debris flows will occur23,26. Most of the postfire debris flows that have

1
U.S. Geological Survey, Geologic Hazards Science Center, Landslide Hazards Program, Golden, CO, USA. 2Department of Earth, Atmosphere and Environment,
Northern Illinois University, DeKalb, IL, USA. 3University of California, San Diego, Scripps Institution of Oceanography, Center for Western Weather and Water
Extremes, La Jolla, CA, USA. 4California Geological Survey, Burned Watershed Geohazards Program, Sacramento, CA, USA.
e-mail: matthewthomas@usgs.gov

npj Natural Hazards | (2024)1:14 1


https://doi.org/10.1038/s44304-024-00017-8 Article

Fig. 1 | Annual frequency of rainfall threshold exceedances for burned areas 2014 and 2022, the U.S. Geological Survey (USGS) issued rainfall thresholds as part
across the American Southwest where postfire debris flows have been a concern in of emergency assessments of postfire debris-flow hazards. Shaded relief map pro-
the last decade, including the 2018 Montecito debris-flow event. a State-specific vided by Environmental Systems Research Institute83. State boundaries provided by
bar plots reflect the mean annual frequency of simulated 15-min rainfall intensity U.S. Census Bureau84. b Photograph shows the effects of postfire debris-flow
threshold exceedances associated with 175 burned areas across the American inundation associated with the 9 January 2018 Montecito debris-flow event, which
Southwest (Supplementary Fig. 1). Vertical black lines indicate standard error of the occurred in southern California. Photograph taken by Jeffrey A. Coe, USGS.
mean. We calculated these exceedance statistics for burned areas where, between

been documented in the interior Southwest (i.e., Arizona, Colorado, New how rainfall intensification driven by climate change may heighten our
Mexico) have been associated with warm season (June–September) con- exposure to this hazard. Despite progress in recognizing the potential effect
vection during the North American Monsoon27–29. Decaying tropical sys- of rainfall intensification on postfire debris flows35,36, we have not yet
tems, which typically arrive in the Southwest during the transition between developed an approach to quantify how the frequency, magnitude, and
the warm and cool seasons, have also triggered postfire debris flows30. seasonality of rainfall capable of triggering postfire debris flows could evolve.
As wildfire activity increases throughout the western United States31, so Here, we provide this information by tracking rainfall threshold exceedance
too has the demand for the emergency assessment of postfire debris-flow metrics across dynamically downscaled (3.75-km horizontal grid spacing),
hazards32, including areas where our understanding of postfire debris-flow convection-permitting simulations of short-duration (15-min) rainfall37 for
initiation is limited, such as in the Cascade ranges spanning northern historical (1990–2005) versus late 21st-century (2085–2100) climate sce-
California, Oregon, Washington, and British Columbia33. This demand, narios. Our approach enables a physics-based representation of the
coupled with recent observations of postfire debris flows in regions where meteorological phenomena often associated with postfire debris flows
they were previously undocumented26,34, highlights a need to understand (convective rainfall), honors the timescale of postfire debris-flow triggering

npj Natural Hazards | (2024)1:14 2


https://doi.org/10.1038/s44304-024-00017-8 Article

rainfall (low recurrence interval, sub-hourly duration), and resolves these and Colorado, but the FUT4.5 exceedance rates are generally lower than
characteristics at a spatial scale that is suitable for long-term hazard FUT8.5 (Figs. 2, 3).
planning.
Magnitude and seasonality of rainfall threshold exceedance
Results Similar to the marked rise in the mean annual frequency of threshold
We refer to our climate simulation output as WRF-BCC (Weather Research exceedance, we also observe a significant (p < 0.05) increase in the mean I15
and Forecasting Model-Bias Corrected Community Earth System Model), threshold exceedance ratio for FUT4.5 and FUT8.5 across burned areas in
the simulation epochs as HIST (historical; 1990–2005) and FUT (future; the Southwest as compared to HIST (Fig. 4a). The exceedance ratio is
2085–2100), and the future emissions scenarios as RCP (Representative calculated by dividing the simulated I15 by the I15 threshold. The exceedance
Concentration Pathway) 4.5 and 8.5. We characterize trends in 15-min ratio is equal to one when the simulated I15 is equal to the I15 threshold and it
rainfall intensity (I15 [mm per hour]) threshold exceedance for 175 burned is more than one when the simulated I15 is greater than the I15 threshold.
areas across the American Southwest (Fig. 1a), as well as for mountainous Given that the volume of debris flows scales with rainfall intensity44, this
terrain throughout California and Colorado (Figs. 2 and 3), for three WRF- result indicates that larger and potentially more destructive debris flows may
BCC simulation scenarios: (1) HIST, (2) FUT4.5, and (3) FUT8.5. We become more common. Specifically, FUT8.5 projects higher percent
provide more detailed descriptions of the climate simulations and thresh- increases in mean exceedance ratios than FUT4.5 for our burned area
olds in the “Methods” section, which follows the ”Discussion” section. analyses (FUT4.5 = +4% and FUT8.5 = +12%) and our statewide moun-
tainous terrain analyses (California FUT4.5 = +2% and FUT8.5 = +5%;
Frequency of rainfall threshold exceedance Colorado FUT4.5 = +4% and FUT8.5 = +9%). Our results also indicate
The WRF-BCC simulations project a rise in the frequency of above- that trends in the magnitude of threshold exceedance will not necessarily
threshold rainfall across the American Southwest. Specifically, we follow those associated with the frequency of threshold exceedance. For
observe a significant (p < 0.05) increase in the mean annual frequency example, Colorado may experience increases in the mean annual frequency
of I15 threshold exceedance for FUT4.5 and FUT8.5 across 175 of threshold exceedance that are less severe than California (Figs. 2, 3), but
burned areas in Arizona, California, Colorado, Nevada, New Mexico, Colorado is projected to experience greater increases in mean threshold
and Utah (Fig. 1a) where the U.S. Geological Survey (USGS) issued exceedance ratios than California (Fig. 4a).
rainfall thresholds as part of emergency assessments of postfire The overall distribution of cool- versus warm-season threshold
debris-flow hazards32. These 15-min rainfall thresholds were calcu- exceedance appears largely unaffected across mountainous terrain in Cali-
lated with an empirical debris-flow likelihood model22,38 that is rou- fornia and Colorado under a changing climate (Fig. 4b, c). WRF-BCC
tinely applied to areas of concern in the western United States where projects that California will continue to be dominated by cool-season
the control of short-duration rainfall on the hazard has been mea- (October–May) exceedances (HIST = 85%, FUT 4.5 = 93%, and FUT8.5 =
sured directly26,39,40. Across these burned areas, FUT4.5 projects a 96%) and that Colorado will continue to be dominated by warm-season
mean exceedance two times that of HIST (1.4 versus 0.7 exceedances (June–September) exceedances (HIST = 87%, FUT 4.5 = 78%, and
per year), while the mean exceedance rate for FUT8.5 is more than FUT8.5 = 76%). However, there are some shifts at the meteorological season
four times that of HIST (3.1 versus 0.7 exceedances per year). USGS- scale. In California, for example, the December–January–February (DJF)
assessed burned areas in California show the greatest increases in season experiences an increase (+24%) in the mean proportion of excee-
threshold exceedance (2.3 and 6.5 times higher than HIST for dances in FUT8.5 compared to HIST. The March–April–May (MAM) and
FUT4.5 and FUT8.5, respectively), while increases in threshold September–October–November (SON) seasons, on the other hand,
exceedance in neighboring Nevada (1.3 and 1.8 times higher than experience decreases in the mean proportion of exceedance (−8%) in
HIST for FUT4.5 and FUT8.5, respectively) and Arizona (1.5 times FUT8.5 compared to HIST. This pattern is consistent with other studies that
higher than HIST for both FUT4.5 and FUT8.5) are the lowest. have noted a “sharpening” of the precipitation season in California to the
Increases in exceedance rates for New Mexico, Colorado, and Utah core winter months45, as well as the intensification of atmospheric river
fall in-between, with mean exceedance rates 1.6–2.2 times greater storms and sub-daily intensities within them17. The Klamath Mountains
than HIST for FUT4.5 and 2.2–2.4 times greater than HIST for and Sierra Nevada are areas of California that are projected to experience
FUT8.5 (Fig. 1a). some of the greatest increases in the frequency of exceedances (Fig. 2) and
Our statewide analyses of I15 threshold exceedance in mountainous typically experience snowfall in the DJF season. The increase in the mean
terrain, for which we evaluate spatially variable thresholds equivalent to a proportion of exceedances during the DJF season may be associated with the
one-year recurrence interval41, indicate that the future frequency of I15 projected transition to more precipitation falling as rain under a warming
threshold exceedance will not be spatially uniform (Figs. 2, 3). For example, climate46. The snow-to-rain transition could also be reflected in the slight
FUT8.5 projects substantial increases (>6 exceedances per year versus ≤1.5 increase in the mean proportion of exceedances in Colorado during DJF and
in HIST) in California throughout the Sierra Nevada, Northern Coast SON (+1%) in FUT8.5 compared to HIST. We observe negligible change in
Ranges, and Klamath Mountains, with notable increases (>1.5–6 excee- Colorado for the MAM season (−0.2%) in FUT8.5 compared to HIST.
dances per year versus ≤1.5 in HIST) in the Peninsular, Transverse, and California and Colorado are projected to experience decreases in the mean
Southern Coast Ranges (Fig. 2). Rainfall intensification is less apparent proportion of exceedances during the June–July–August (JJA) season (−8%
(many cases < 1.5 exceedances per year) for the mountainous terrain of the and −2%, respectively) in FUT8.5 compared to HIST. These shifts in
Mojave and Colorado Deserts in southeastern California. This finding is threshold exceedance frequency during the JJA season and the early SON
consistent with others who have also projected greater changes in rainfall season are likely associated with projected changes in the frequency and
intensification for northern California than in southern California18,42,43. intensity of the North American Monsoon and decaying tropical systems
Increases in the frequency of threshold exceedance are not as apparent in impacting the Southwest47–49.
Colorado (Fig. 3) compared to California (Fig. 2). For example, FUT8.5
projects that limited areas in the San Juan Mountains of southwestern Discussion
Colorado will experience substantial (>6 exceedances per year) increases in We find that rainfall intensification driven by climate change is projected to
the frequency of threshold exceedance; however, there are notable (>1.5–6 amplify known triggering conditions for postfire debris flows in the
exceedances per year) increases in the frequency of threshold exceedance in American Southwest. This amplification is reflected by significant increases
the Front Range, Sangre de Cristo Mountains, and Sawatch Range of central in the frequency (Fig. 1a) and magnitude (Fig. 4a) of short-duration rainfall
and southern Colorado. Rises in above-threshold exceedance projected by intensity threshold exceedances for burned areas in the Southwest where
FUT4.5 and FUT8.5 are similar in their spatial distribution across California postfire debris flows have been a concern in the past, as well as across

npj Natural Hazards | (2024)1:14 3


https://doi.org/10.1038/s44304-024-00017-8 Article

Fig. 2 | Annual frequency of rainfall threshold exceedances for mountainous are labeled for reference, including “KLM” (Klamath Mountains), “NCR” (Northern
terrain throughout the State of California. State of California with an overlay of Coast Ranges), “Sierra Nevada” (SNV), “SCR” (Southern Coast Ranges), “TRV”
scatter plots that reflect 15-min rainfall intensity threshold (equivalent to a one-year (Transverse Ranges), and “PNR” (Peninsular Ranges). State boundaries provided by
recurrence interval under the present climate) exceedances in mountainous terrain U.S. Census Bureau84.
for the WRF-BCC HIST, FUT4.5, and FUT8.5 simulation scenarios. Several areas

Fig. 3 | Annual frequency of rainfall threshold exceedances for mountainous for the HIST, FUT4.5, and FUT8.5 simulation scenarios. Several areas are labeled for
terrain throughout the State of Colorado. State of Colorado with an overlay of reference, including “FRG” (Front Range), “SWA” (Swatch Range), “SJM” (San Juan
scatter plots that reflect 15-min rainfall intensity threshold (equivalent to a one-year Mountains), and “SDC” (Sangre de Cristo Mountains). State boundaries provided by
recurrence interval under the present climate) exceedances in mountainous terrain U.S. Census Bureau84.

mountainous terrain in California and Colorado that are potentially sus- (Fig. 1b). Although our focus is on debris flows generated by excessive
ceptible to the hazard (Figs. 2, 3). An increased frequency of exceedance surface-water runoff soon after the fire, the trends in rainfall intensification
indicates that we may expect more opportunities for sufficiently intense that we report in this study may also be problematic for steep, unburned
rainfall to initiate debris flows before postfire soil-hydraulic recovery and/or areas in the American Southwest that generally lack vegetation and are also
vegetation regrowth can reduce susceptibility to flash floods and runoff- prone to runoff-generated debris flows56. However, our results are not
generated debris flows39,50–53. Temporal changes in sediment sourcing (e.g., applicable to debris flows generated by landslides, for which the simulated
gradual entrainment from hillslopes, mass failure of the channel bed, and triggering conditions have been shown to be sensitive to the antecedent soil
shallow landsliding) have been observed to fuel multiple postfire debris saturation51.
flows over the course of a rainy season in the first year after fire40,54. Climate-enhanced postfire debris-flow likelihood and magnitude have
Therefore, rainfall intensification could result in more postfire debris flows been recognized for southern California35. Our analyses, which expand on
for settings that are not sediment supply limited. An increased magnitude of this work, have implications for hazard planning across the American
exceedance means that we may also expect enhanced mobilization of Southwest. For example, the ratio of rainfall threshold exceedance for
sediment and other dangerous debris, which would result in greater inun- FUT4.5 or FUT8.5 to HIST (Fig. 1a) could be used as a scaling factor for each
dation depths and runout distances along developed valley bottoms, state to broadly assess to what extent rainfall intensification could affect the
floodplains, and fan surfaces44,55. Therefore, rainfall intensification may frequency of threshold exceedance in areas where postfire debris flows have
result in more damaging postfire debris flows such as those associated with been a concern in the past (e.g., RCP4.5/HIST = 2.3 for California versus 1.6
the 2018 Montecito debris-flow event in Santa Barbara County, California11 for Colorado). The relatively high spatial resolution of our results (Figs. 2, 3)

npj Natural Hazards | (2024)1:14 4


https://doi.org/10.1038/s44304-024-00017-8 Article

Fig. 4 | Summary of rainfall threshold exceedance ratios and seasonality of U.S. Geological Survey issued rainfall thresholds as part of emergency assessments of
rainfall threshold exceedances for California, Colorado, and the American postfire debris-flow hazards between 2014 and 2022 (Fig. 1a, Supplementary Fig. 1).
Southwest. a Box and whisker plots that track changes in the exceedance ratio of 15- The boxes in a are bound by the first and third quartile, with the median indicated by
minute rainfall intensity thresholds for the HIST, FUT4.5, and FUT8.5 simulation a line, and the whiskers extend from the boxes to the farthest data point within 1.5
scenarios for mountainous terrain throughout California and Colorado (Figs. 2, 3, times the inter-quartile range. The bar plots illustrate differences in the seasonality of
Supplementary Fig. 1), as well as for 175 burned areas across the American South- rainfall threshold exceedance for mountainous terrain throughout b California and
west (i.e., Arizona, California, Colorado, Nevada, New Mexico, and Utah) where the c Colorado.

could also facilitate more localized planning. For example, we show that the rainfall intensification at the sub-hourly timescale, consistent with studies
intensification of short-duration rainfall could be more severe in moun- of hourly precipitation intensification that include the Southwest14,17,18,57.
tainous areas, such as northern California (Fig. 2). This includes areas where However, we recognize that representing convective processes associated
communities, emergency managers, and weather forecasters may be less with short-duration, high-intensity rainfall in climate models is
accustomed to contending with postfire debris, flows (relative to those in challenging58–60. Due to the computational constraints associated with
southern California), in part because the hazard has only recently been producing simulations at such fine spatiotemporal scales, our results are
systematically documented26,34. Therefore, this study highlights increased limited to downscaling a single general circulation model (GCM) reali-
hazard potential and the benefits of prefire planning, education, and out- zation with a single WRF configuration. Additional studies could corro-
reach for postfire hazards36. borate our findings as outputs at similarly high spatial and temporal
We use the first available projections of 15-min rainfall for 15-year resolutions become available from various modeling efforts with other
periods at 3.75-km grid spacing in the American Southwest. We observe GCM inputs, downscaling model cores, and physics parameterizations.

npj Natural Hazards | (2024)1:14 5


https://doi.org/10.1038/s44304-024-00017-8 Article

Future research could also assess changes in the meteorological char- Weather Research and Forecasting-Advanced Research WRF model
acteristics (e.g., storm types) driving over-threshold rainfall events (WRF-ARW) Version 4.1.269. Initial and lateral boundary conditions for
(including extreme mesoscale and synoptic atmospheric features) within WRF-ARW were taken from the bias-corrected70 output from Phase 5 of the
the context of projected changes to fire behavior. Such analyses would Coupled Model Intercomparison Project (CMIP5)71 version of the National
provide a more in-depth explanation of the intra-seasonal changes we Center for Atmospheric Research (NCAR) Community Earth System
observe and provide insight for forecasting high-intensity rainfall and Model (CESM)72,73. Bias correction prior to dynamic downscaling improves
postfire debris flows where rainfall intensification and increased fire regional simulation performance, particularly for temperature and
activity may geographically overlap under a changing climate. precipitation74,75.
We assess the short-duration rainfall characteristics that most com- For climate comparisons, we examined three, 15-year epochs: (1) a
monly trigger debris flows in the first two years after wildfire in the western historical period (HIST) spanning 1 October 1990–30 September 2005 and
United States. Short-duration, high-intensity rainfall produces excessive (2,3) a late-21st century period (FUT) spanning 1 October 2085–30 Sep-
surface-water runoff that can rapidly amplify stream power and mobilize tember 2100 under Representative Concentration Pathway (RCP) 4.5
debris, particularly in low-order catchments where the time-to- (FUT4.5) and 8.5 (FUT8.5). RCP 4.5 and RCP 8.5 characterize intermediate
concentration is short7. Our analyses of burned areas rely on short- and pessimistic future emissions scenarios, respectively. The simulations
duration rainfall thresholds calculated for recent wildfires using an empirical used spectral nudging every six hours, an approach where regional climate
model that was trained on debris-flow observations from southern Cali- models are driven by select spatial scales associated with the general cir-
fornia and has been tested against hundreds of debris-flow observations culation model76, to maintain consistency of synoptic-scale features between
across Arizona, Colorado, Idaho, Montana, New Mexico, and Utah22,38. Our the general circulation model (GCM) and the higher-resolution WRF-ARW
statewide analyses rely on the known recurrence interval of the short- domain at the boundaries77,78. Mesoscale phenomena are not nudged and
duration rainfall intensities associated with these same debris-flow are, therefore, allowed to dynamically evolve within the model domain. This
observations8; therefore, our analyses could be extended beyond California simulation dataset, which we refer to as WRF-BCC, is one of the only
and Colorado as part of a future study. The control of short-duration rainfall available with archived 15-min rainfall over multiple climate epochs, thus
on debris-flow triggering soon after fire has also proven valid for the highly permitting our novel analysis of short-duration rainfall in a climate change
destructive debris flows that followed recent megafires (>100,000 acres) in framework.
central and northern California26,34. However, we refrain from expanding A comparison of WRF-BCC HIST precipitation data was performed37
our analyses to the Pacific Northwest, where the effect of short-duration against the Parameter-elevation Regressions on the Independent Slopes
rainfall on debris-flow generation after wildfire is considerably more Model (PRISM)79 and a gridded precipitation observation ensemble
uncertain61 and short-duration precipitation frequency data for the present dataset80 (originates from over 12,000 unique surface stations) that accounts
climate41 is unavailable (Supplementary Fig. 1). The recent wave of very large, for observational uncertainty resulting from issues like gage undercatch,
high-severity wildfires in the Pacific Northwest62 underscores the value of complex terrain, and low station density. WRF-BCC HIST and PRISM
field-verified inventories of debris-flow occurrence to support the refine- showed 0.91 Pearson’s r correlation for average annual and monthly pre-
ment of conceptual models for the conditions that trigger postfire debris cipitation, with a bias smaller than at least one ensemble member of
flows in this area and short-duration precipitation frequency data that can be observations. A broad dry bias was noted in the southeast, east, and
used to support climate analyses like those we present here. northeast United States, whereas WRF-BCC HIST simulations were found
Topographic steepness, soil burn severity, and soil erodibility are to have a wet bias in most of the Northern Plains and Intermountain West
important factors that influence susceptibility to postfire debris flows22,38. during DJF when compared to observations. Overall, the biases are not
We implicitly account for these factors in our analyses of burned areas where statistically significant at the 90% significance level using a Mann–Whitney
postfire debris flows have been a concern in the recent past; however, our U test for the medians, and it is plausible that some of the wet bias in the high
statewide analyses of mountainous terrain do not consider future fire elevations of the Intermountain West is due to a lack of reliable observations.
behavior or changes in sediment availability. Increased wildfire activity is
projected across our study area in the mid-to-late 21st century and is gen- Rainfall thresholds
erally expected to enhance soil erosion across many watersheds in the Our analyses focus on two spatial subsets of the WRF-BCC simulation
western United States63, although there is spatial variability in these wildfire output: (1) 175 burned areas across the Southwest (i.e., Arizona, California,
projections64–66. Furthermore, changes to ecological succession/transition Colorado, Nevada, New Mexico, and Utah) where the USGS conducted
that are expected to accompany climate change67 may affect regional burn emergency assessments of postfire debris-flow hazards32 between 2014 and
severity and recovery characteristics. We prioritize a focus on changes in 2022, and (2) mountainous terrain81 throughout California and Colorado
rainfall because previous work has indicated that future debris-flow activity (Fig. 5a, Supplementary Fig. 1). Each USGS hazard assessment includes an
will be more sensitive to rainfall intensification than increases in fire fre- estimate of the 15-min rainfall intensity (I15) needed to generate postfire
quency and severity, at least for areas dominated by shrubland vegetation35. debris flows. These rainfall thresholds are based on a logistic regression22,38
We also recognize that regional variability in soil and rock properties and that predicts the statistical likelihood (i.e., a value ranging from 0 to 1) of
their connection to sediment availability is an important susceptibility postfire debris flows based on the topographic slope, soil burn severity, and
factor. Loose, unconsolidated materials lining channels and mantling hill- soil erodibility along stream segments and at drainage basin outlets. The
slopes are thought to contribute to postfire debris-flow susceptibility across a rainfall thresholds that we used were calculated for a likelihood of 0.5, a value
wide range of geologic conditions68, but these kinds of deposits have not yet that has been widely applied to estimate debris-flow triggering conditions in
been systematically identified at the spatial scales that we explore for this the western United States32. This likelihood-based rainfall threshold
study. Therefore, while our statewide rainfall threshold exceedance maps approach has been shown to generally perform as well, and in some cases
(Figs. 2, 3) identify where we may expect an amplification of rainfall con- better than, preexisting regional rainfall thresholds in Arizona, Colorado,
ditions that are conducive to triggering postfire debris flows, more work Montana, and New Mexico38, and has also performed similarly well in areas
would be helpful to identify regional factors that could affect debris-flow with no preexisting regional thresholds34.
susceptibility. To ensure consistency between our two types of threshold analyses, we
conducted a rainfall intensity-duration-frequency (IDF) analysis for each
Methods WRF-BCC HIST grid cell-of-interest (e.g., Fig. 5a) using a partial-duration
Climate data series approach82. For our statewide analyses, we assumed spatially variable
We analyzed a set of convection-permitting (3.75-km horizontal grid spa- I15 thresholds that are equivalent to the one-year rainfall recurrence interval8
cing) climate simulations that were dynamically downscaled using the for HIST. For our burned area analyses, we used the median of the

npj Natural Hazards | (2024)1:14 6


https://doi.org/10.1038/s44304-024-00017-8 Article

Fig. 5 | Conceptual overview of the methodology


for tracking rainfall threshold exceedances. a For
our analyses of U.S. Geological Survey (USGS)-asses-
sed fires, we selected the Weather Research and
Forecasting Model-Bias Corrected Community Earth
System Model (WRF-BCC) cells in the American
Southwest (i.e., Arizona, California, Colorado,
Nevada, New Mexico, and Utah) within burn peri-
meters where, between 2014 and 2022, the USGS
issued rainfall thresholds as part of an emergency
assessment of postfire debris-flow hazards (Supple-
mentary Fig. 1). For our statewide analyses, we selected
the WRF-BCC cells that intersected mountainous
terrain throughout the State of California and Color-
ado and applied a 10-km buffer. Mountainous terrain
map provided by USGS81. State boundaries provided
by U.S. Census Bureau84. b For each WRF-BCC cell
that we selected for our assessment-based analyses, we
developed log-linear fits to predict the recurrence
intervals (RIs) of local 15-minute rainfall rates. We
used these expressions to convert USGS rainfall
thresholds (e.g., dashed blue arrow) to rainfall RI
thresholds (e.g., dashed orange arrow). We assume 15-
min rainfall RI thresholds equal to one year (under the
present climate) for our statewide analyses. c We then
conducted a rainfall intensity-duration-frequency
analysis for each WRF-BCC cell and used power fits to
convert rainfall RI thresholds to WRF-BCC rainfall
thresholds (dashed gray arrow). d Finally, we tracked
WRF-BCC rainfall threshold exceedances (e.g., orange
bar) for the historical (HIST, 1990–2005) and late
21st-century RCP 4.5/8.5 (FUT, 2085–2100) scenarios.

USGS-based I15 thresholds32 that were calculated among drainage basin that anyone interested in using the WRF-BCC model output contact VG
outlets (a metric commonly distributed for hazard guidance) for each fire (vgensini@niu.edu) for additional information, including any collaboration.
and converted the I15 thresholds into rainfall recurrence intervals using The USGS emergency assessment of postfire debris-flow hazard data is
National Oceanic and Atmospheric Administration (NOAA) Atlas 14 available at https://usgs.maps.arcgis.com/apps/dashboards/c09fa874362
precipitation frequency data41 (Fig. 5b). We used NOAA Atlas 14 pre- e48a9afe79432f2efe6fe. The NOAA precipitation frequency data are avail-
cipitation data to calculate the recurrence interval of the USGS-based I15 able at https://hdsc.nws.noaa.gov/pfds/.
thresholds because we found that WRF-BCC HIST could underestimate the
frequency of short-duration, high-intensity rainfall relative to estimates Received: 27 October 2023; Accepted: 24 April 2024;
from NOAA. The median recurrence interval associated with the USGS-
based I15 thresholds (1.6 years) is similar to the return period that we used
for our statewide threshold analyses (1 year) and consistent with previous References
research8. We then used the IDF analyses to convert the recurrence intervals 1. Rooney, B. et al. Air quality impact of the Northern California Camp
into equivalent WRF-BCC I15 thresholds (Fig. 5c). We only retained WRF- Fire of November 2018. Atmos. Chem. Phys. 20, 14597–14616
BCC I15 thresholds for our analyses if the log-linear and power fit that we (2020).
used to calculate rainfall recurrence intervals (Fig. 5b) and rainfall thresholds 2. Schulze, S. S., Fischer, E. C., Hamideh, S. & Mahmoud, H. Wildfire
(Fig. 5c) resulted in a coefficient of determination (R2) ≥ 0.9. impacts on schools and hospitals following the 2018 California Camp
We identified three straightforward metrics to characterize future Fire. Nat. Hazards 104, 901–925 (2020).
trends in the frequency, magnitude, and seasonality of postfire debris-flow 3. Collar, N. M. & Earles, T. A. Unique challenges posed by fire
triggering conditions with the WRF-BCC simulation output: (1) the mean disturbance to water supply management and transfer agreements in
annual I15 threshold exceedance rate, (2) the mean I15 threshold exceedance a headwaters region. J. Environ. Manag. 339, 117956 (2023).
ratio, and (3) the mean seasonality of I15 threshold exceedance. To calculate 4. Wells, W. G. The effects of fire on the generation of debris flows in
these metrics for a given WRF-BCC grid cell with the associated HIST, Southern California. In Debris Flows/Avalanches (eds. Costa J. E. &
FUT4.5, and FUT8.5 simulation timeseries, we tallied each I15 exceedance Wieczorek G. F.) 105–114 (Geological Society of America, 1987).
(Fig. 5d), computed the ratio of the simulated I15 to the I15 threshold for each 5. Cannon, S. H. Debris-flow generation from recently burned
exceedance, and organized the exceedances into monthly bins and seasonal watersheds. Environ. Eng. Geosci. 7, 321–341 (2001).
breaks (i.e., October–May, cool season; June–September, warm season). 6. Kean, J. W., McGuire, L. A., Rengers, F. K., Smith, J. B. & Staley, D. M.
Amplification of post wildfire peak flow by debris. Geophys. Res. Lett.
Data availability 43, 8545–8553 (2016).
The WRF-BCC HIST, FUT4.5, and FUT8.5 simulation output that we used 7. Kean, J. W., Staley, D. M. & Cannon, S. H. In situ measurements of
to conduct our analyses are available via a public-facing digital repository at post-fire debris flows in southern California: comparisons of the timing
https://doi.org/10.5061/dryad.n2z34tn3c, https://doi.org/10.5061/dryad. and magnitude of 24 debris-flow events with rainfall and soil moisture
4qrfj6qgn, and https://doi.org/10.5061/dryad.nvx0k6dzm. We request conditions. J. Geophys. Res. 116, F04019 (2011).

npj Natural Hazards | (2024)1:14 7


https://doi.org/10.1038/s44304-024-00017-8 Article

8. Staley, D. M., Kean, J. W. & Rengers, F. K. The recurrence interval of 28. McGuire, L. A. et al. Characteristics of debris flow prone watersheds
post-fire debris-flow generating rainfall in the southwestern United and triggering rainstorms following the Tadpole Fire, New Mexico
States. Geomorphology 370, 107392 (2020). USA. Nat. Hazards Earth Sys. 24, 1357–1379 (2024).
9. DeGraff, J. V., Cannon, S. H. & Gartner, J. E. Timing of susceptibility to 29. Rengers, F. K. et al. Evaluating post-wildfire debris flow rainfall
post-fire debris flows in the western USA. Environ. Eng. Geosci. 21, thresholds and volume models at the 2020 Grizzly Creek Fire in
277–292 (2015). Glenwood Canyon, Colorado, USA. EGUsphere https://doi.org/10.
10. Cannon, S. H., Gartner, J. E., Wilson, R. C., Bowers, J. C. & Laber, J. L. 5194/egusphere-2023-2063 (2023).
Storm rainfall conditions for floods and debris flows from recently 30. Blake, E. The 2022 eastern Pacific hurricane season an above average
burned areas in southwestern Colorado and Southern California. season with five landfalls. Weatherwise 76, 27–35 (2023).
Geomorphology 96, 250–269 (2008). 31. Abatzoglou, J. T. & Williams, A. P. Impact of anthropogenic climate
11. Lancaster, J. T. et al. Observations and analyses of the 9 January 2018 change on wildfire across western US forests. Proc. Natl. Acad. Sci.
debris-flow disaster, Santa Barbara County, California. Environ. Eng. USA 113, 11770–11775 (2016).
Geosci. 27, 3–27 (2021). 32. United States Geological Survey (USGS). Emergency Assessment of
12. Westra, S. et al. Future changes to the intensity and frequency of Post-Fire Debris-Flow Hazards. Landslide Hazards Program https://
short-duration extreme rainfall. Rev. Geophys. 52, 522–555 (2014). usgs.maps.arcgis.com/apps/dashboards/c09fa874362e48a9afe
13. Ali, H., Fowler, H. J. & Mishra, V. Global observational evidence of 79432f2efe6fe (2023).
strong linkage between dew point temperature and precipitation 33. Wall, S. A., Roering, J. J. & Rengers, F. K. Runoff-initiated post-fire
extremes. Geophys. Res. Lett. 45, 12320–12330 (2018). debris flow Western Cascades, Oregon. Landslides 17,
14. Cannon, A. J. & Innocenti, S. Projected intensification of sub-daily and 1649–1661 (2020).
daily rainfall extremes in convection-permitting climate model 34. Thomas, M. A. et al. Postfire hydrologic response along the Central
simulations over North America: implications for future California (USA) coast: insights for the emergency assessment of
intensity–duration–frequency curves. Nat. Hazards Earth Syst. Sci. 19, postfire debris-flow hazards. Landslides 20, 2421–2436 (2023).
421–440 (2019). 35. Kean, J. W. & Staley, D. M. Forecasting the frequency and magnitude
15. Fowler, H. J. et al. Anthropogenic intensification of short-duration of postfire debris flows across Southern California. Earths Future 9,
rainfall extremes. Nat. Rev. Earth. Environ. 2, 107–122 (2021). e2020EF001735 (2021).
16. Berg, P. & Haerter, J. O. Unexpected increases in precipitation 36. Oakley, N. S. A warming climate adds complexity to post‐fire
intensity with temperature—a result of mixing precipitation types? hydrologic hazard planning. Earths Future 9, e2021EF002149 (2021).
Atmos. Res. 119, 56–61 (2013). 37. Gensini, V. A., Haberlie, A. M. & Ashley, W. S. Convection-permitting
17. Huang, X., Swain, D. L. & Hall, A. D. Future precipitation increase from simulations of historical and possible future climate over the
very high resolution ensemble downscaling of extreme atmospheric contiguous United States. Clim. Dyn. 60, 109–126 (2023).
river storms in California. Sci. Adv. 6, eaba1323 (2020). 38. Staley, D. M. et al. Prediction of spatially explicit rainfall
18. Prein, A. F. et al. The future intensification of hourly precipitation intensity–duration thresholds for post-fire debris-flow generation in
extremes. Nat. Clim. Chang. 7, 48–52 (2017). the Western United States. Geomorphology 278, 149–162 (2017).
19. Ali, H., Fowler, H. J., Lenderink, G., Lewis, E. & Pritchard, D. 39. Hoch, O. J., McGuire, L. A., Youberg, A. M. & Rengers, F. K.
Consistent large-scale response of hourly extreme rainfall to Hydrogeomorphic recovery and temporal changes in rainfall
temperature variation over land. Geophys. Res. Lett. 48, thresholds for debris flows following wildfire. J. Geophys. Res. Earth.
e2020GL090317 (2021). Surf. 126, e2021JF006374 (2021).
20. Haberlie, A. M., Ashley, W. S., Gensini, V. A. & Michaelis, A. C. The ratio 40. Tang, H. et al. Evolution of debris‐flow initiation mechanisms and
of mesoscale convective system precipitation to total precipitation sediment sources during a sequence of postwildfire rainstorms. J.
increase in future climate change scenarios. npj Clim. Atmos. Sci. 6, Geophys. Res. Earth Surf. 124, 1572–1595 (2019).
150 (2023). 41. National Oceanic and Atmospheric Administration (NOAA).
21. Touma, D. et al. Climate change increases risk of extreme rainfall Precipitation Frequency Data Server. Hydrometeorological Design
following wildfire in the western United States. Sci. Adv. 8, Studies Center (National Oceanic and Atmospheric Administration
eabm0320 (2022). (NOAA), 2023).
22. Staley, D. M. et al. Updated Logistic Regression Equations for the 42. AghaKouchak, A., Ragno, E., Love, C. & Moftakhari, H. Projected
Calculation of Post-fire Debris-flow Likelihood in the Western United Changes in California’s Precipitation Intensity-Duration-Frequency
States. U.S. Geological Survey Open-File Report 2016-1106, 1–13 Curves. California’s Fourth Climate Change Assessment CCCA4-
(U.S. Geological Survey, 2016). CEC-2018-005, 1–32 (California Energy Commission, 2018).
23. Oakley, N. S., Lancaster, J. T., Kaplan, M. L. & Ralph, F. M. Synoptic 43. Gershunov, A. et al. Precipitation regime change in western North
conditions associated with cool season post-fire debris flows in the America: the role of atmospheric rivers. Sci. Rep. 9, 9944 (2019).
Transverse Ranges of Southern California. Nat. Hazards 88, 44. Gartner, J. E., Cannon, S. H. & Santi, P. M. Empirical models for
327–354 (2017). predicting volumes of sediment deposited by debris flows and
24. Oakley, N. S. et al. Meteorological and climatological conditions sediment-laden floods in the transverse ranges of southern California.
associated with the 9 January 2018 post-fire debris flows in Montecito Eng. Geol. 176, 45–56 (2014).
and Carpinteria, California, USA. Nat. Hazards Earth Syst. Sci. 18, 45. Swain, D. L., Langenbrunner, B., Neelin, J. D. & Hall, A. Increasing
3037–3043 (2018). precipitation volatility in twenty-first century California. Nat. Clim.
25. de Orla-Barile, M., Cannon, F., Oakley, N. S. & Ralph, F. M. A Change. 8, 427–433 (2018).
climatology of narrow cold-frontal rainbands in southern California. 46. Rhoades, A. M. et al. Asymmetric emergency of low-to-no snow in the
Geophys. Res. Lett. 49, e2021GL095362 (2021). midlatitudes of the American Cordillera. Nat. Clim. Change. 12,
26. Thomas, M. A. et al. The rainfall intensity-duration control of debris 1151–1159 (2022).
flows after wildfire. Geophys. Res. Lett. 50, e2023GL103645 (2023). 47. Douville, H. K. et al. Water cycle changes. In Climate Change 2021:
27. Gorr, A. N., McGuire, L. A., Beers, R. & Hoch, O. J. Triggering The Physical Science Basis. Contribution of Working Group I to the
conditions, runout, and downstream impacts of debris flows following Sixth Assessment Report of the Intergovernmental Panel on Climate
the 2021 Flag Fire, Arizona, USA. Nat. Hazards 117, Change (eds. Masson-Delmotte, V. P. et al.) 1055–1210 (Cambridge
2473–2504 (2023). University Press, 2021).

npj Natural Hazards | (2024)1:14 8


https://doi.org/10.1038/s44304-024-00017-8 Article

48. Chand, S. S., Tory, K. J., Ye, H. & Walsh, K. J. Projected increase in El 68. Cannon, S. H. & Gartner, J. E. Wildfire-related debris flow from a
Niño-driven tropical cyclone frequency in the Pacific. Nat. Clim. hazards perspective. In Debris-Flow Hazards and Related Phenomena
Change 7, 123–127 (2017). (eds. Jakob M. & Hungr O.) 363–385 (Springer, 2005).
49. Knutson, T. et al. Tropical cyclones and climate change assessment: 69. Skamarock, W. C. et al. A Description of the Advanced Research WRF
part II: projected response to anthropogenic warming. Bull. Am. Model Version 4.3. NCAR/TN-556+STR, 1–148 (National Center for
Meteorol. Soc. 101, 303–322 (2020). Atmospheric Research, 2021).
50. Ebel, B. A. Temporal evolution of measured and simulated infiltration 70. Bruyère, C. L., Done, J. M., Holland, G. J. & Fredrick, S. Bias
following wildfire in the Colorado Front Range, USA: shifting corrections of global models for regional climate simulations of high-
thresholds of runoff generation and hydrologic hazards. J. Hydrol. impact weather. Clim. Dyn. 43, 1847–1856 (2014).
585, 124765 (2020). 71. Taylor, K. E., Stouffer, R. J. & Meehl, G. A. An overview of CMIP5
51. Thomas, M. A. et al. Postwildfire soil‐hydraulic recovery and the and the experiment design. Bull. Am. Meteorol. Soc. 93,
persistence of debris flow hazards. J. Geophys. Res. Earth Surf. 126, 485–498 (2012).
e2021JF006091 (2021). 72. Hurrell, J. W. et al. The Community Earth System Model: a framework
52. Perkins, J. P. et al. Multi‐stage soil‐hydraulic recovery and limited for collaborative research. Bull. Am. Meteorol. Soc. 94,
ravel accumulations following the 2017 Nuns and Tubbs wildfires in 1339–1360 (2013).
Northern California. J. Geophys. Res. Earth Surf. 127, 73. Monaghan, A. J., Steinhoff, D. F., Bruyère, C. L. & Yates, D. NCAR
e2022JF006591 (2022). CESM Global Bias-Corrected CMIP5 Output to Support WRF/MPAS
53. Graber, A. P., Thomas, M. A. & Kean, J. W. How long do runoff- Research https://doi.org/10.5065/D6DJ5CN4 (2014).
generated debris-flow hazards persist after wildfire? Geophys. Res. 74. Ines, A. V. M. & Hansen, J. W. Bias correction of daily GCM rainfall for
Lett. 50, e2023GL105101 (2023). crop simulation studies. Agric. For. Meteorol. 138, 44–53 (2006).
54. Guilinger, J. J. et al. Predicting postfire sediment yields of small steep 75. Christensen, J. H., Boberg, F., Christensen, O. B. & Lucas-Picher, P.
catchments using airborne lidar differencing. Geophys. Res. Lett. 50, On the need for bias correction of regional climate change projections
e2023GL104626 (2023). of temperature and precipitation. Geophys. Res. Lett. 35,
55. Barnhart et al. Multi-model comparison of computed debris-flow L20709 (2008).
runout for the 9 January 2018 Montecito, California post-wildfire 76. Omrani, H., Drobinski, P. & Dubos, T. Spectral nudging in regional
event. J. Geophys. Res. Earth Surf. 126, e2021JF006245 (2021). climate modelling: how strongly should we nudge? Q. J. R. Meteorol.
56. Coe, J. A., Kinner, D. A. & Godt, J. W. Initiation conditions for debris Soc. 138, 1808–1813 (2012).
flows generated by runoff at Chalk Cliffs, central Colorado. 77. von Storch, H., Langenberg, H. & Feser, F. A spectral nudging
Geomorphology 96, 270–297 (2008). technique for dynamical downscaling purposes. Mon. Weather Rev.
57. Kendon, E. J. et al. Do convection-permitting regional climate models 128, 3664–3673 (2000).
improve projections of future precipitation change? Bull. Am. 78. Feser, F., Rockel, B., von Storch, H., Winterfeldt, J. & Zahn, M.
Meteorol. Soc. 98, 79–93 (2017). Regional climate models add value to global model data: a review
58. Prein, A. F. et al. A review on regional convection-permitting climate and selected examples. Bull. Am. Meteor. Soc. 92,
modeling: demonstrations, prospects, and challenges. Rev. 1181–1192 (2011).
Geophys. 53, 323–361 (2015). 79. Daly, C., Neilson, R. P. & Phillips, D. L. A statistical-topographic model
59. Haarsma, R. J. High resolution model Intercomparison Project for mapping climatological precipitation over mountainous terrain. J.
(HighResMIP v1.0) for CMIP6. Geosci. Model Dev. 53, Appl. Meteorol. Climatol. 33, 140–158 (1994).
323–361 (2016). 80. Newman, A. J. et al. Gridded ensemble precipitation and temperature
60. Roberts, M. J. et al. The benefits of global high resolution for climate estimates for the contiguous U.S. J. Hydrometeorol. 16,
simulation: process understanding and enabling of stakeholder 2481–2500 (2015).
decisions at the regional scale. Bull. Am. Meteorol. Soc. 99, 81. United States Geological Survey (USGS). Global Mountain Explorer
2341–2359 (2018). 2.0. K3. https://rmgsc.cr.usgs.gov/gme/ (2023).
61. Wondzell, S. M. & King, J. G. Postfire erosional processes in the 82. Dunne, T. & Leopold, L. B. Water in Environmental Planning
Pacific Northwest and Rocky Mountain Regions. For. Ecol. Manag. (Freeman, 1978).
178, 75–87 (2003). 83. Environmental Systems Research Institute. World Shaded Relief
62. Abatzoglou, J. T., Rupp, D. E., O’Neill, L. W. & Sadegh, M. Compound https://hub.arcgis.com/maps/9c5370d0b54f4de1b48a3792d737
extremes drive the western Oregon wildfires of September 2020. 7ff2/about (2019).
Geophys. Res. Lett. 48, e2021GL092520 (2021). 84. United States Census Bureau. State Boundaries (Generalized for
63. Sankey, J. B. et al. Climate, wildfire, and erosion ensemble foretells Mapping) https://www.sciencebase.gov/catalog/item/
more sediment in western USA watersheds. Geophys. Res. Lett. 44, 52c78623e4b060b9ebca5be5 (2011).
8884–8892 (2017).
64. Stavros, E. N., Abatzoglou, J. T., McKenzie, D. & Larkin, N. Regional Acknowledgements
projections of the likelihood of very large wildland fires under a We conducted our rainfall analyses using the Comet supercomputer,
changing climate in the contiguous Western United States. Clim. which is hosted by the San Diego Supercomputer Center at the
Change 126, 455–468 (2014). University of California, San Diego. This computational resource was
65. Westerling, A. L. Wildfire Simulations for California’s Fourth Climate made available by the Atmospheric River Program (Phases 2 and 3),
Change Assessment: Projecting Changes in Extreme Wildfire Events which is supported by the California Department of Water Resources
with a Warming Climate. CCCA4-CEC-2018-014, 1–57 (California (awards 4600013361 and 4600014294, respectively) and the U.S. Army
Energy Commission, 2018). Corps of Engineers Engineer Research and Development Center (award
66. Abatzoglou, J. T. et al. Projected increases in western US forest fire USACE W912HZ-15-2-0019). We extend a special thanks to Patrick
despite growing fuel constraints. Commun. Earth Environ. 2, Mulrooney at the Center for Western Weather and Water Extremes for his
227 (2021). technical support with Comet. We also acknowledge the high-
67. Hill, A. P. & Field, C. B. Forest fires and climate-induced tree range performance computing resources provided by the Cheyenne super-
shifts in the western US. Nat. Commun. 12, 6583 (2021). computer, which is hosted by the National Center for Atmospheric

npj Natural Hazards | (2024)1:14 9


https://doi.org/10.1038/s44304-024-00017-8 Article

Research (NCAR) Computational and Information Systems Laboratory Correspondence and requests for materials should be addressed to
and sponsored by the National Science Foundation (NSF). M.A.T. and Matthew A. Thomas.
J.W.K. were supported by the USGS Landslide Hazards Program. A.C.M.
was supported by NSF grant AGS-2203516, the California Department of Reprints and permissions information is available at
Water Resources (DWR) AR Program, and the U.S. Bureau of Recla- http://www.nature.com/reprints
mation project R15AC00003. V.A.G. and W.S.A. were supported by NSF
grants 1637225 and 1800582. We appreciate the constructive feedback Publisher’s note Springer Nature remains neutral with regard to
provided by Joel B. Sankey, Joseph E. Gartner, and three anonymous jurisdictional claims in published maps and institutional affiliations.
reviewers, which improved our manuscript. Any use of trade, firm, or
product names is for descriptive purposes only and does not imply Open Access This article is licensed under a Creative Commons
endorsement by the U.S. Government. Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long
Author contributions as you give appropriate credit to the original author(s) and the source,
M.A.T. conducted the analyses and compiled the manuscript with input from provide a link to the Creative Commons licence, and indicate if changes
all authors. A.C.M. processed the climate model data. A.C.M., N.S.O., and were made. The images or other third party material in this article are
J.W.K. contributed to concept development. V.A.G. and W.S.A. provided the included in the article’s Creative Commons licence, unless indicated
climate model data and expertise. otherwise in a credit line to the material. If material is not included in the
article’s Creative Commons licence and your intended use is not permitted
Competing interests by statutory regulation or exceeds the permitted use, you will need to
The authors declare no competing interests. obtain permission directly from the copyright holder. To view a copy of this
licence, visit http://creativecommons.org/licenses/by/4.0/.
Additional information
Supplementary information The online version contains © The Author(s). Parts of this work were authored by US Federal Government
supplementary material available at authors and are not under copyright protection in the US; foreign copyright
https://doi.org/10.1038/s44304-024-00017-8. protection may apply 2024

npj Natural Hazards | (2024)1:14 10

You might also like