Caelan Simeone

Drought is a highly consequential natural disaster that may likely increase in both severity and extent across the conterminous United States (CONUS). The mechanisms affecting the propagation of drought from the atmosphere to streamflow are complex and interactive, making the prediction of streamflow drought difficult with current modeling approaches. Machine learning is an emerging tool in the field of hydrology that may be well-suited to prediction of streamflow drought across large and topographically diverse areas. Here, we train and analyze 3198 random forest models at U.S. Geological Survey streamgages to understand common meteorological drivers of streamflow drought and to define physiographic characteristics of basins sensitive to these drivers. We also develop a novel dynamic regionalization approach using donor gages to predict daily streamflow drought at pseudo-ungaged locations. Our results show that teleconnections, temperature, evaporative demand, and snow–water equivalent are important drivers of streamflow drought in the West, Southwest, and Northern Rocky Mountains (Northern Rockies) regions of the United States, and precipitation and soil moisture are primary drivers of streamflow drought in the Northeast, Southeast, and the Northwest regions. Prediction using dynamic regionalization shows comparable performance to at-site models. This method may be applicable to other hydrologic prediction problems requiring transferability to ungaged locations.

Drought is a highly consequential natural disaster that may likely increase in both severity and extent across the conterminous United States (CONUS). The mechanisms affecting the propagation of drought from the atmosphere to streamflow are complex and interactive, making the prediction of streamflow drought difficult with current modeling approaches. Machine learning is an emerging tool in the field of hydrology that may be well-suited to prediction of streamflow drought across large and topographically diverse areas. Here, we train and analyze 3,198 random forest models at U.S. Geological Survey streamgages to understand common meteorological drivers of streamflow drought and to define physiographic characteristics of basins sensitive to these drivers. We also develop a novel dynamic regionalization approach using donor gages to predict daily streamflow drought at pseudo-ungaged locations. For this study, CONUS was divided into nine regions for ease of reference in describing results. Our results show that teleconnections, temperature, evaporative demand, and snow-water equivalent are important drivers of streamflow drought in the West, Southwest, and Northern Rocky Mountains (Northern Rockies) regions of the United States, and precipitation and soil moisture are primary drivers of streamflow drought in the Northeast, Southeast, and the Northwest regions. Prediction using dynamic regionalization shows comparable performance to at-site models.

Drought is a highly consequential natural disaster that may likely increase in both severity and extent across the conterminous United States (CONUS). The mechanisms affecting the propagation of drought from the atmosphere to streamflow are complex and interactive, making the prediction of streamflow drought difficult with current modeling approaches. Machine learning is an emerging tool in the field of hydrology that may be well-suited to prediction of streamflow drought across large and topographically diverse areas. Here, we train and analyze 3,198 random forest models at U.S. Geological Survey streamgages to understand common meteorological drivers of streamflow drought and to define physiographic characteristics of basins sensitive to these drivers. We also develop a novel dynamic regionalization approach using donor gages to predict daily streamflow drought at pseudo-ungaged locations. For this study, CONUS was divided into nine regions for ease of reference in describing results. Our results show that teleconnections, temperature, evaporative demand, and snow-water equivalent are important drivers of streamflow drought in the West, Southwest, and Northern Rocky Mountains (Northern Rockies) regions of the United States, and precipitation and soil moisture are primary drivers of streamflow drought in the Northeast, Southeast, and the Northwest regions. Prediction using dynamic regionalization shows comparable performance to at-site models.

Drought is a highly consequential natural disaster that may likely increase in both severity and extent across the conterminous United States (CONUS). The mechanisms affecting the propagation of drought from the atmosphere to streamflow are complex and interactive, making the prediction of streamflow drought difficult with current modeling approaches. Machine learning is an emerging tool in the field of hydrology that may be well-suited to prediction of streamflow drought across large and topographically diverse areas. Here, we train and analyze 3,198 random forest models at U.S. Geological Survey streamgages to understand common meteorological drivers of streamflow drought and to define physiographic characteristics of basins sensitive to these drivers. We also develop a novel dynamic regionalization approach using donor gages to predict daily streamflow drought at pseudo-ungaged locations. For this study, CONUS was divided into nine regions for ease of reference in describing results. Our results show that teleconnections, temperature, evaporative demand, and snow-water equivalent are important drivers of streamflow drought in the West, Southwest, and Northern Rocky Mountains (Northern Rockies) regions of the United States, and precipitation and soil moisture are primary drivers of streamflow drought in the Northeast, Southeast, and the Northwest regions. Prediction using dynamic regionalization shows comparable performance to at-site models.

Drought is a highly consequential natural disaster that may likely increase in both severity and extent across the conterminous United States (CONUS). The mechanisms affecting the propagation of drought from the atmosphere to streamflow are complex and interactive, making the prediction of streamflow drought difficult with current modeling approaches. Machine learning is an emerging tool in the field of hydrology that may be well-suited to prediction of streamflow drought across large and topographically diverse areas. Here, we train and analyze 3,198 random forest models at U.S. Geological Survey streamgages to understand common meteorological drivers of streamflow drought and to define physiographic characteristics of basins sensitive to these drivers. We also develop a novel dynamic regionalization approach using donor gages to predict daily streamflow drought at pseudo-ungaged locations. For this study, CONUS was divided into nine regions for ease of reference in describing results. Our results show that teleconnections, temperature, evaporative demand, and snow-water equivalent are important drivers of streamflow drought in the West, Southwest, and Northern Rocky Mountains (Northern Rockies) regions of the United States, and precipitation and soil moisture are primary drivers of streamflow drought in the Northeast, Southeast, and the Northwest regions. Prediction using dynamic regionalization shows comparable performance to at-site models.

Rapid shifts between droughts and floods, termed hydrological whiplash, challenge water management, yet their timing and drivers remain poorly understood at continental scales. While drought-to-flood (DtF) transitions have received growing attention, flood-to-drought (FtD) transitions — though rarer — pose distinct operational challenges that are less well characterized. These wet-to-dry shifts can disrupt post-flood recovery, strain warm-season water demands, and create compounding risks for infrastructure and water quality. We analyzed daily streamflow records from 3,219 USGS streamgages (1981-2024) to characterize both DtF and FtD transitions across CONUS, with particular emphasis on understanding why these transitions are not symmetric inverses of each other. We test a wide variety of hydrological extreme transition definitions to examine the sensitivity of the number of transitions identified and their rapidity. Additionally, we identify a subset of transitions that may be impactful based on the maximum change in percentile magnitude during a transition. DtF transitions are faster than FtD transitions, and short-term (<= 30-days) transitions in both directions are concentrated in the Northeast, Northwest, and Rocky Mountains regions. Short-term DtF transitions are additionally concentrated in southern California and along the line from North Dakota down to Texas where precipitation approximately equals potential evapotranspiration. We find direction-specific controls: snow-dominated, urban, regulated, and minimally disturbed basins show the most frequent impactful DtF transitions, while regulated basins are most prone to impactful FtD transitions. Rapid and impactful transitions make up a substantially larger fraction of DtF transitions than FtD transitions across nearly all basin types. A key finding is that hydrological and meteorological whiplash rarely coincide: only 19-24% of hydrological extreme transitions co-occur with hydroclimate whiplash, revealing that basin storage, regulation, and routing processes create a fundamental decoupling between climate forcing and streamflow response. Our findings highlight the need to better understand quick hydrological transitions under increasing hydroclimatic volatility, particularly the understudied FtD direction, and the mechanisms by which anthropogenic modifications reshape the hydrological whiplash risk.

Rapid shifts between droughts and floods, termed hydrological whiplash, challenge water management, yet their timing and drivers remain poorly understood at continental scales. While drought-to-flood (DtF) transitions have received growing attention, flood-to-drought (FtD) transitions — though rarer — pose distinct operational challenges that are less well characterized. These wet-to-dry shifts can disrupt post-flood recovery, strain warm-season water demands, and create compounding risks for infrastructure and water quality. We analyzed daily streamflow records from 3,219 USGS streamgages (1981-2024) to characterize both DtF and FtD transitions across CONUS, with particular emphasis on understanding why these transitions are not symmetric inverses of each other. We test a wide variety of hydrological extreme transition definitions to examine the sensitivity of the number of transitions identified and their rapidity. Additionally, we identify a subset of transitions that may be impactful based on the maximum change in percentile magnitude during a transition. DtF transitions are faster than FtD transitions, and short-term (<= 30-days) transitions in both directions are concentrated in the Northeast, Northwest, and Rocky Mountains regions. Short-term DtF transitions are additionally concentrated in southern California and along the line from North Dakota down to Texas where precipitation approximately equals potential evapotranspiration. We find direction-specific controls: snow-dominated, urban, regulated, and minimally disturbed basins show the most frequent impactful DtF transitions, while regulated basins are most prone to impactful FtD transitions. Rapid and impactful transitions make up a substantially larger fraction of DtF transitions than FtD transitions across nearly all basin types. A key finding is that hydrological and meteorological whiplash rarely coincide: only 19-24% of hydrological extreme transitions co-occur with hydroclimate whiplash, revealing that basin storage, regulation, and routing processes create a fundamental decoupling between climate forcing and streamflow response. Our findings highlight the need to better understand quick hydrological transitions under increasing hydroclimatic volatility, particularly the understudied FtD direction, and the mechanisms by which anthropogenic modifications reshape the hydrological whiplash risk.