Caroline S. Juang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
Caroline S. Juang has a total of 51 co-authors. Here are the top co-authors by number of shared publications:
- A. Park Williams
Department of Geography University of California Los Angeles CA USA
8 shared publications E. Eva Borbas
2 shared publications- Ewa Agata Czyz
University of Zürich
2 shared publications Glynn Hully
2 shared publications- Jason E. Smerdon
Lamont-Doherty Earth Observatory, Columbia University, Palisades, USA
2 shared publications - Jatan Buch
Department of Earth and Environmental Engineering; Columbia University; New York, NY; USA
2 shared publications
Latest Publications
- Open Access
- 1 November 2026
- Published by Wiley
Preconditioning Drought and Fire Weather Jointly Elevate Megafire Risk in the Western United States
Antecedent drought and fire weather operate across different timescales and jointly shape conditions conducive to large-fire growth. Nevertheless, their joint association with megafire risk is understudied. Here we quantify joint rarity for 19,761 wildfires across the western United States (WUS) during 1984--2024. Fires exposed to jointly rare antecedent drought and fire-weather environments had a megafire rate of 7.58%, whereas 4.88% of all fires were megafires. Among fires exposed to these environments, 46.7% would not have been tagged as extremes-driven under a univariate extreme threshold perspective. Additionally, the annual share of fires exposed to jointly rare antecedent drought and fire-weather environments increased over time and was positively correlated with annual megafire rates, while exposure trends and the climate configurations associated with jointly rare exposure varied regionally across the WUS, with the strongest exposure trend observed in the Southwest.
- 1 November 2026
- Published by Wiley
Open AccessPreconditioning Drought and Fire Weather Jointly Elevate Megafire Risk in the Western United States
Antecedent drought and fire weather operate across different timescales and jointly shape conditions conducive to large-fire growth. Nevertheless, their joint association with megafire risk is understudied. Here we quantify joint rarity for 19,761 wildfires across the western United States (WUS) during 1984--2024. Fires exposed to jointly rare antecedent drought and fire-weather environments had a megafire rate of 7.58%, whereas 4.88% of all fires were megafires. Among fires exposed to these environments, 46.7% would not have been tagged as extremes-driven under a univariate extreme threshold perspective. Additionally, the annual share of fires exposed to jointly rare antecedent drought and fire-weather environments increased over time and was positively correlated with annual megafire rates, while exposure trends and the climate configurations associated with jointly rare exposure varied regionally across the WUS, with the strongest exposure trend observed in the Southwest.
- Open Access
- 12 September 2026
- Published by American Geophysical Union (AGU) in Earth's Future
- Vol. 14 (9)
Contribution of La‐Niña‐Like Pacific Ocean Trend in Sea‐Surface Temperatures to Western United States Wildfire Area, 1984–2022
In the western United States (US), the annual wildfire extent increased by 306% over 1984–2022, largely due to a 1,320% increase in forest‐fire area. During this time, associated with the Pacific Decadal Oscillation (PDO), the tropical Pacific Ocean exhibited a “La‐Niña‐like” trend with an enhanced zonal sea‐surface temperature gradient (gSST) between its warm western and cool eastern sides. In contrast, most atmosphere‐ocean models simulate human‐caused climate change should force an “El‐Niño‐like” gSST trend, a discrepancy from observations that is the focus of vigorous investigation. To understand the relevance of uncertainties in projections of gSST to western US wildfire activity, we statistically estimate the contribution of the observed increase in gSST of the past four decades. We compare the observed burned area outcome against a counterfactual scenario in which the observed La Niña‐like gSST trend did not occur. We estimate that the strengthened gSST gradient accounted for 19% [11%, 33%] of the increase in western US wildfire area. In forest areas alone, we estimate the gSST trend accounted for 32% [21%, 45%] of the observed increase, due mostly to increased spring and summer vapor‐pressure deficit (VPD). Thus, the strengthened gSST was not the dominant driver of the observed increase in wildfire area, but it appeared to play an important secondary role. Should the positive gSST trends not reverse, western US climate may remain more conducive to forest fire than would be expected based on current climate projections.
- 12 September 2026
- Published by American Geophysical Union (AGU) in Earth's Future
- Vol. 14 (9)
Open AccessContribution of La‐Niña‐Like Pacific Ocean Trend in Sea‐Surface Temperatures to Western United States Wildfire Area, 1984–2022
In the western United States (US), the annual wildfire extent increased by 306% over 1984–2022, largely due to a 1,320% increase in forest‐fire area. During this time, associated with the Pacific Decadal Oscillation (PDO), the tropical Pacific Ocean exhibited a “La‐Niña‐like” trend with an enhanced zonal sea‐surface temperature gradient (gSST) between its warm western and cool eastern sides. In contrast, most atmosphere‐ocean models simulate human‐caused climate change should force an “El‐Niño‐like” gSST trend, a discrepancy from observations that is the focus of vigorous investigation. To understand the relevance of uncertainties in projections of gSST to western US wildfire activity, we statistically estimate the contribution of the observed increase in gSST of the past four decades. We compare the observed burned area outcome against a counterfactual scenario in which the observed La Niña‐like gSST trend did not occur. We estimate that the strengthened gSST gradient accounted for 19% [11%, 33%] of the increase in western US wildfire area. In forest areas alone, we estimate the gSST trend accounted for 32% [21%, 45%] of the observed increase, due mostly to increased spring and summer vapor‐pressure deficit (VPD). Thus, the strengthened gSST was not the dominant driver of the observed increase in wildfire area, but it appeared to play an important secondary role. Should the positive gSST trends not reverse, western US climate may remain more conducive to forest fire than would be expected based on current climate projections.
- Open Access
- 27 August 2026
- Published by Elsevier in SSRN Electronic Journal
A quarter-century record of hourly Land Surface Temperature over the United States: Long term trends
Diurnal land surface temperature (LST) extremes directly impact society, but their long-term trends are difficult to evaluate at regional scales. Here, we develop and utilize a unified GOES-derived LST product (UGEOLST) to quantify daytime and nighttime surface thermal conditions and their diurnal variability over nearly a quarter century (2000–2023). Representative daytime and nighttime conditions are derived using the 90th percentile of daytime values and the median of nighttime values within diurnal windows, respectively, designed to reduce cloud contamination and spurious extremes. Seasonal trends are evaluated using Mann–Kendall tests and Sen’s slope estimates. Results show statistically significant and spatially heterogeneous changes across much of the contiguous United States (CONUS), with the most coherent increases in daytime extremes occurring during fall and winter. These patterns are consistent with hydroclimatic controls on surface energy partitioning, including snow cover variability, cloud modulation of shortwave radiation, and soil moisture limitation in transition-zone climates. Nighttime LST trends are more spatially heterogeneous, with enhanced warming in localized zones. The diurnal temperature range (DTR) is increasing, particularly throughout the great plains and parts of the Pacific west coast. Sen’s slope estimates indicate local warming rates exceeding 0.2 °C per year in some regions, particularly across the Great Plains during fall for daytime extremes. The hottest surface temperatures occur in arid desert environments such as Death Valley. These results demonstrate the value of geostationary satellite observations for resolving diurnal and seasonal variations in LST extremes and for distinguishing between surface-driven and atmosphere-driven modes of thermal change. Our UGEOLST product provides a continuous record for characterizing evolving surface thermal behavior, with implications for environmental monitoring, heat risk assessment, and climate resilience.
- 27 August 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessA quarter-century record of hourly Land Surface Temperature over the United States: Long term trends
Diurnal land surface temperature (LST) extremes directly impact society, but their long-term trends are difficult to evaluate at regional scales. Here, we develop and utilize a unified GOES-derived LST product (UGEOLST) to quantify daytime and nighttime surface thermal conditions and their diurnal variability over nearly a quarter century (2000–2023). Representative daytime and nighttime conditions are derived using the 90th percentile of daytime values and the median of nighttime values within diurnal windows, respectively, designed to reduce cloud contamination and spurious extremes. Seasonal trends are evaluated using Mann–Kendall tests and Sen’s slope estimates. Results show statistically significant and spatially heterogeneous changes across much of the contiguous United States (CONUS), with the most coherent increases in daytime extremes occurring during fall and winter. These patterns are consistent with hydroclimatic controls on surface energy partitioning, including snow cover variability, cloud modulation of shortwave radiation, and soil moisture limitation in transition-zone climates. Nighttime LST trends are more spatially heterogeneous, with enhanced warming in localized zones. The diurnal temperature range (DTR) is increasing, particularly throughout the great plains and parts of the Pacific west coast. Sen’s slope estimates indicate local warming rates exceeding 0.2 °C per year in some regions, particularly across the Great Plains during fall for daytime extremes. The hottest surface temperatures occur in arid desert environments such as Death Valley. These results demonstrate the value of geostationary satellite observations for resolving diurnal and seasonal variations in LST extremes and for distinguishing between surface-driven and atmosphere-driven modes of thermal change. Our UGEOLST product provides a continuous record for characterizing evolving surface thermal behavior, with implications for environmental monitoring, heat risk assessment, and climate resilience.
- Open Access
- 18 June 2026
- Published by Elsevier in SSRN Electronic Journal
A quarter-century record of hourly Land Surface Temperature over the United States: Long term trends
Diurnal land surface temperature (LST) extremes directly impact society, but their long-term trends are difficult to evaluate at regional scales. Here, we develop and utilize a unified GOES-derived LST product (UGEOLST) to quantify daytime and nighttime surface thermal conditions and their diurnal variability over nearly a quarter century (2000–2023). Representative daytime and nighttime conditions are derived using the 90th percentile of daytime values and the median of nighttime values within diurnal windows, respectively, designed to reduce cloud contamination and spurious extremes. Seasonal trends are evaluated using Mann–Kendall tests and Sen’s slope estimates. Results show statistically significant and spatially heterogeneous changes across much of the contiguous United States (CONUS), with the most coherent increases in daytime extremes occurring during fall and winter. These patterns are consistent with hydroclimatic controls on surface energy partitioning, including snow cover variability, cloud modulation of shortwave radiation, and soil moisture limitation in transition-zone climates. Nighttime LST trends are more spatially heterogeneous, with enhanced warming in localized zones. The diurnal temperature range (DTR) is increasing, particularly throughout the great plains and parts of the Pacific west coast. Sen’s slope estimates indicate local warming rates exceeding 0.2 °C per year in some regions, particularly across the Great Plains during fall for daytime extremes. The hottest surface temperatures occur in arid desert environments such as Death Valley. These results demonstrate the value of geostationary satellite observations for resolving diurnal and seasonal variations in LST extremes and for distinguishing between surface-driven and atmosphere-driven modes of thermal change. Our UGEOLST product provides a continuous record for characterizing evolving surface thermal behavior, with implications for environmental monitoring, heat risk assessment, and climate resilience.
- 18 June 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessA quarter-century record of hourly Land Surface Temperature over the United States: Long term trends
Diurnal land surface temperature (LST) extremes directly impact society, but their long-term trends are difficult to evaluate at regional scales. Here, we develop and utilize a unified GOES-derived LST product (UGEOLST) to quantify daytime and nighttime surface thermal conditions and their diurnal variability over nearly a quarter century (2000–2023). Representative daytime and nighttime conditions are derived using the 90th percentile of daytime values and the median of nighttime values within diurnal windows, respectively, designed to reduce cloud contamination and spurious extremes. Seasonal trends are evaluated using Mann–Kendall tests and Sen’s slope estimates. Results show statistically significant and spatially heterogeneous changes across much of the contiguous United States (CONUS), with the most coherent increases in daytime extremes occurring during fall and winter. These patterns are consistent with hydroclimatic controls on surface energy partitioning, including snow cover variability, cloud modulation of shortwave radiation, and soil moisture limitation in transition-zone climates. Nighttime LST trends are more spatially heterogeneous, with enhanced warming in localized zones. The diurnal temperature range (DTR) is increasing, particularly throughout the great plains and parts of the Pacific west coast. Sen’s slope estimates indicate local warming rates exceeding 0.2 °C per year in some regions, particularly across the Great Plains during fall for daytime extremes. The hottest surface temperatures occur in arid desert environments such as Death Valley. These results demonstrate the value of geostationary satellite observations for resolving diurnal and seasonal variations in LST extremes and for distinguishing between surface-driven and atmosphere-driven modes of thermal change. Our UGEOLST product provides a continuous record for characterizing evolving surface thermal behavior, with implications for environmental monitoring, heat risk assessment, and climate resilience.
- Open Access
- 1 May 2026
- Published by IOP Publishing in Environmental Research Letters
- Vol. 21 (9)
Opposing contributions of the tropical Pacific to recent burned forest area trends in Eastern Australia and the Southwestern United States
The Southwestern U.S. and Eastern Australia are two of the most fire-prone regions of the world. These two regions also have significant and opposite-signed hydroclimate teleconnections to sea-surface temperature anomalies in the tropical Pacific, such that an El-Niño-like west-to-east tropical Pacific SST gradient is associated with hot, dry conditions in Eastern Australia and cool, wet conditions in the Southwestern U.S. Additionally, the tropical Pacific has experienced a pronounced La-Niña-like cooling over recent decades. Here, we investigate the roles of the tropical Pacific and anthropogenic climate change as drivers of observed trends in burned forest area and vapor pressure deficit (VPD) in the Southwestern U.S. and Eastern Australia, using observations and CMIP6 model data. We characterize the strength and spatial structure of relationships between fire-favorable climate variables in these regions and SST anomalies in the Pacific. We show that in both regions, VPD has increased with a large contribution from anthropogenic climate change. In the Southwestern U.S., the recent trend in the tropical Pacific west-east sea surface temperature gradient positively contributed to the VPD-induced increase in burned forest area since 1984. However, in Eastern Australia, the tropical Pacific SST gradient trend has likely offset the anthropogenic climate change-induced tendency to increase burned forest area.
- 1 May 2026
- Published by IOP Publishing in Environmental Research Letters
- Vol. 21 (9)
Open AccessOpposing contributions of the tropical Pacific to recent burned forest area trends in Eastern Australia and the Southwestern United States
The Southwestern U.S. and Eastern Australia are two of the most fire-prone regions of the world. These two regions also have significant and opposite-signed hydroclimate teleconnections to sea-surface temperature anomalies in the tropical Pacific, such that an El-Niño-like west-to-east tropical Pacific SST gradient is associated with hot, dry conditions in Eastern Australia and cool, wet conditions in the Southwestern U.S. Additionally, the tropical Pacific has experienced a pronounced La-Niña-like cooling over recent decades. Here, we investigate the roles of the tropical Pacific and anthropogenic climate change as drivers of observed trends in burned forest area and vapor pressure deficit (VPD) in the Southwestern U.S. and Eastern Australia, using observations and CMIP6 model data. We characterize the strength and spatial structure of relationships between fire-favorable climate variables in these regions and SST anomalies in the Pacific. We show that in both regions, VPD has increased with a large contribution from anthropogenic climate change. In the Southwestern U.S., the recent trend in the tropical Pacific west-east sea surface temperature gradient positively contributed to the VPD-induced increase in burned forest area since 1984. However, in Eastern Australia, the tropical Pacific SST gradient trend has likely offset the anthropogenic climate change-induced tendency to increase burned forest area.
- Open Access
- 30 April 2026
- Published by American Geophysical Union (AGU) in Earth's Future
- Vol. 14 (5)
Evolving Fire Frequency in the Western United States and Its Links to Human Influence
Burned area and wildfire damages in the western United States (WUS) have increased dramatically in recent decades. Wildfire frequency, however, has shown trends that vary by data set, region, and fire size. Using a comprehensive fire occurrence data set screened for reporting artifacts from 1992 to 2020, we show a significant downward trend in annual fire frequency in WUS, with a significant contribution from decreasing human ignitions. Human‐caused fires significantly declined in California and Arizona, but increased in Wyoming. The relationship between human‐caused fires and population suggests a pyric transition, where fire frequency correlates positively to population in areas with very low population such as Wyoming and negatively where population is denser, such as in California and Arizona. Understanding the human influence on fire frequency trends beyond anthropogenic warming has strong implications for future projections of wildfire activity.
- 30 April 2026
- Published by American Geophysical Union (AGU) in Earth's Future
- Vol. 14 (5)
Open AccessEvolving Fire Frequency in the Western United States and Its Links to Human Influence
Burned area and wildfire damages in the western United States (WUS) have increased dramatically in recent decades. Wildfire frequency, however, has shown trends that vary by data set, region, and fire size. Using a comprehensive fire occurrence data set screened for reporting artifacts from 1992 to 2020, we show a significant downward trend in annual fire frequency in WUS, with a significant contribution from decreasing human ignitions. Human‐caused fires significantly declined in California and Arizona, but increased in Wyoming. The relationship between human‐caused fires and population suggests a pyric transition, where fire frequency correlates positively to population in areas with very low population such as Wyoming and negatively where population is denser, such as in California and Arizona. Understanding the human influence on fire frequency trends beyond anthropogenic warming has strong implications for future projections of wildfire activity.
- Open Access
- 29 March 2025
- Published by American Geophysical Union (AGU) in Geophysical Research Letters
- Vol. 52 (7)
Influence of Time‐Averaging of Climate Data on Estimates of Atmospheric Vapor Pressure Deficit and Inferred Relationships With Wildfire Area in the Western United States
Vapor pressure deficit (VPD) is a driver of evaporative demand and correlates strongly with wildfire extent in the western United States (WUS). Vapor pressure deficit is the difference between saturation vapor pressure ( e s ) and actual vapor pressure ( e a ). Because e s increases nonlinearly with temperature, calculations of time‐averaged VPD vary depending on the frequency of temperature measurements and how e a is calculated, potentially limiting our understanding of fire‐climate relationships. We calculate eight versions of monthly VPD across the WUS and assess their differences. Monthly VPDs calculated from daily data are 2%–6% higher, and more accurate, than when calculated from monthly data. Using daily maximum and minimum temperature, instead of mean, increases VPD by ∼20%, but can overestimate true values depending on how e a is calculated. These differences do not meaningfully impact correlations with annual wildfire area, however, suggesting our understanding of historical fire‐VPD relations is not very sensitive to how VPD is calculated.
- 29 March 2025
- Published by American Geophysical Union (AGU) in Geophysical Research Letters
- Vol. 52 (7)
Open AccessInfluence of Time‐Averaging of Climate Data on Estimates of Atmospheric Vapor Pressure Deficit and Inferred Relationships With Wildfire Area in the Western United States
Vapor pressure deficit (VPD) is a driver of evaporative demand and correlates strongly with wildfire extent in the western United States (WUS). Vapor pressure deficit is the difference between saturation vapor pressure ( e s ) and actual vapor pressure ( e a ). Because e s increases nonlinearly with temperature, calculations of time‐averaged VPD vary depending on the frequency of temperature measurements and how e a is calculated, potentially limiting our understanding of fire‐climate relationships. We calculate eight versions of monthly VPD across the WUS and assess their differences. Monthly VPDs calculated from daily data are 2%–6% higher, and more accurate, than when calculated from monthly data. Using daily maximum and minimum temperature, instead of mean, increases VPD by ∼20%, but can overestimate true values depending on how e a is calculated. These differences do not meaningfully impact correlations with annual wildfire area, however, suggesting our understanding of historical fire‐VPD relations is not very sensitive to how VPD is calculated.
- Open Access
- 11 December 2022
- Published by Wiley in ESS Open Archive
Modeling wildfire activity in the western United States with machine learning
The annual area burned due to wildfires in the western United States (WUS) increased by more than 300% between 1984 and 2020. However, accounting for the nonlinear, spatially heterogeneous interactions between climate, vegetation, and human predictors driving the trends in fire frequency and sizes at different spatial scales remains a challenging problem for statistical fire models. Here we introduce a novel stochastic machine learning (ML) framework to model observed fire frequencies and sizes in 12 km x 12 km grid cells across the WUS. This framework is implemented using Mixture Density Networks trained on a wide suite of input predictors. The modeled WUS fire frequency corresponds well with observations at both monthly (r= 0.94) and annual (r= 0.85) timescales, as do the monthly (r= 0.90) and annual (r= 0.88) area burned. Moreover, the annual time series of both fire variables exhibit strong correlations (r >= 0.6) in 16 out of 18 ecoregions. Our ML model captures the interannual variability and the distinct multidecade increases in annual area burned for both forested and non-forested ecoregions. Evaluating predictor importance with Shapley additive explanations, we find that fire month vapor pressure deficit (VPD) is the dominant driver of fire frequencies and sizes across the WUS, followed by 1000-hour dead fuel moisture (FM1000), total monthly precipitation (Prec), mean daily maximum temperature (Tmax), and fraction of grassland cover in a grid cell. Our findings serve as a promising use case of ML techniques for wildfire prediction in particular and extreme event modeling more broadly.
- 11 December 2022
- Published by Wiley in ESS Open Archive
Open AccessModeling wildfire activity in the western United States with machine learning
The annual area burned due to wildfires in the western United States (WUS) increased by more than 300% between 1984 and 2020. However, accounting for the nonlinear, spatially heterogeneous interactions between climate, vegetation, and human predictors driving the trends in fire frequency and sizes at different spatial scales remains a challenging problem for statistical fire models. Here we introduce a novel stochastic machine learning (ML) framework to model observed fire frequencies and sizes in 12 km x 12 km grid cells across the WUS. This framework is implemented using Mixture Density Networks trained on a wide suite of input predictors. The modeled WUS fire frequency corresponds well with observations at both monthly (r= 0.94) and annual (r= 0.85) timescales, as do the monthly (r= 0.90) and annual (r= 0.88) area burned. Moreover, the annual time series of both fire variables exhibit strong correlations (r >= 0.6) in 16 out of 18 ecoregions. Our ML model captures the interannual variability and the distinct multidecade increases in annual area burned for both forested and non-forested ecoregions. Evaluating predictor importance with Shapley additive explanations, we find that fire month vapor pressure deficit (VPD) is the dominant driver of fire frequencies and sizes across the WUS, followed by 1000-hour dead fuel moisture (FM1000), total monthly precipitation (Prec), mean daily maximum temperature (Tmax), and fraction of grassland cover in a grid cell. Our findings serve as a promising use case of ML techniques for wildfire prediction in particular and extreme event modeling more broadly.
- Open Access
- 4 October 2022
- Published by Wiley in ESS Open Archive
Modeling wildfire activity in the western United States with machine learning
The annual area burned due to wildfires in the western United States (WUS) increased by more than 300% between 1984 and 2020. However, accounting for the nonlinear, spatially heterogeneous interactions between climate, vegetation, and human predictors driving the trends in fire frequency and sizes at different spatial scales remains a challenging problem for statistical fire models. Here we introduce a novel stochastic machine learning (ML) framework to model observed fire frequencies and sizes in 12 km x 12 km grid cells across the WUS. This framework is implemented using Mixture Density Networks trained on a wide suite of input predictors. The modeled WUS fire frequency corresponds well with observations at both monthly (r= 0.94) and annual (r= 0.85) timescales, as do the monthly (r= 0.90) and annual (r= 0.88) area burned. Moreover, the annual time series of both fire variables exhibit strong correlations (r >= 0.6) in 16 out of 18 ecoregions. Our ML model captures the interannual variability and the distinct multidecade increases in annual area burned for both forested and non-forested ecoregions. Evaluating predictor importance with Shapley additive explanations, we find that fire month vapor pressure deficit (VPD) is the dominant driver of fire frequencies and sizes across the WUS, followed by 1000-hour dead fuel moisture (FM1000), total monthly precipitation (Prec), mean daily maximum temperature (Tmax), and fraction of grassland cover in a grid cell. Our findings serve as a promising use case of ML techniques for wildfire prediction in particular and extreme event modeling more broadly.
- 4 October 2022
- Published by Wiley in ESS Open Archive
Open AccessModeling wildfire activity in the western United States with machine learning
The annual area burned due to wildfires in the western United States (WUS) increased by more than 300% between 1984 and 2020. However, accounting for the nonlinear, spatially heterogeneous interactions between climate, vegetation, and human predictors driving the trends in fire frequency and sizes at different spatial scales remains a challenging problem for statistical fire models. Here we introduce a novel stochastic machine learning (ML) framework to model observed fire frequencies and sizes in 12 km x 12 km grid cells across the WUS. This framework is implemented using Mixture Density Networks trained on a wide suite of input predictors. The modeled WUS fire frequency corresponds well with observations at both monthly (r= 0.94) and annual (r= 0.85) timescales, as do the monthly (r= 0.90) and annual (r= 0.88) area burned. Moreover, the annual time series of both fire variables exhibit strong correlations (r >= 0.6) in 16 out of 18 ecoregions. Our ML model captures the interannual variability and the distinct multidecade increases in annual area burned for both forested and non-forested ecoregions. Evaluating predictor importance with Shapley additive explanations, we find that fire month vapor pressure deficit (VPD) is the dominant driver of fire frequencies and sizes across the WUS, followed by 1000-hour dead fuel moisture (FM1000), total monthly precipitation (Prec), mean daily maximum temperature (Tmax), and fraction of grassland cover in a grid cell. Our findings serve as a promising use case of ML techniques for wildfire prediction in particular and extreme event modeling more broadly.
- Top Cited PapersOpen Access
- 8 March 2022
- Published by American Geophysical Union (AGU) in Geophysical Research Letters
- Vol. 49 (5)
Rapid Growth of Large Forest Fires Drives the Exponential Response of Annual Forest‐Fire Area to Aridity in the Western United States
Annual forest area burned (AFAB) in the western United States (US) has increased as a positive exponential function of rising aridity in recent decades. This non‐linear response has important implications for AFAB in a changing climate, yet the cause of the exponential AFAB‐aridity relationship has not been given rigorous attention. We investigated the exponential AFAB‐aridity relationship in western US forests using a new 1984–2019 database of fire events and 2001–2020 satellite‐based records of daily fire growth. While forest‐fire frequency and duration grow linearly with aridity, the exponential AFAB‐aridity relationship results from the exponential growth rates of individual fires. Larger fires generally have more potential for growth due to more extensive firelines. Thus, forces that promote fire growth, such as aridification, have more potent effects on larger fires. As aridity increases linearly, the potential for growth of large fires accelerates, leading to exponential increases in AFAB.
- 8 March 2022
- Published by American Geophysical Union (AGU) in Geophysical Research Letters
- Vol. 49 (5)
Top Cited PapersOpen AccessRapid Growth of Large Forest Fires Drives the Exponential Response of Annual Forest‐Fire Area to Aridity in the Western United States
Annual forest area burned (AFAB) in the western United States (US) has increased as a positive exponential function of rising aridity in recent decades. This non‐linear response has important implications for AFAB in a changing climate, yet the cause of the exponential AFAB‐aridity relationship has not been given rigorous attention. We investigated the exponential AFAB‐aridity relationship in western US forests using a new 1984–2019 database of fire events and 2001–2020 satellite‐based records of daily fire growth. While forest‐fire frequency and duration grow linearly with aridity, the exponential AFAB‐aridity relationship results from the exponential growth rates of individual fires. Larger fires generally have more potential for growth due to more extensive firelines. Thus, forces that promote fire growth, such as aridification, have more potent effects on larger fires. As aridity increases linearly, the potential for growth of large fires accelerates, leading to exponential increases in AFAB.