Variability in river discharge, particularly very high flow and low flow conditions, has far-reaching environmental and economic consequences. The retention of water in surface storage, concentrated in lakes, ponds, wetlands, floodplains, and temporary water in flood prone areas, can potentially contribute to flow generation and flood regulation. However, the impact of surface water storage on river discharge can be challenging to isolate and quantify. A suite of hydrologic signatures were generated for 72 gages across the conterminous United States. The hydrologic signatures were selected to characterize all flows as well as isolating high and low flows, and machine learning models were developed to explain watershed variability in signature values. Wetland related variables, including multi-sensor-based surface water extent and hydroperiod, were compared with other drivers, including climate, topography, and land cover. An improved understanding of how surface water dynamics influence river discharge can be used to improve the resilience of river systems to climate extremes.

Variability in river discharge, particularly very high flow and low flow conditions, has far-reaching environmental and economic consequences. The retention of water in surface storage, concentrated in lakes, ponds, wetlands, floodplains, and temporary water in flood prone areas, can potentially contribute to flow generation and flood regulation. However, the impact of surface water storage on river discharge can be challenging to isolate and quantify. A suite of hydrologic signatures were generated for 72 gages across the conterminous United States. The hydrologic signatures were selected to characterize all flows as well as isolating high and low flows, and machine learning models were developed to explain watershed variability in signature values. Wetland related variables, including multi-sensor-based surface water extent and hydroperiod, were compared with other drivers, including climate, topography, and land cover. An improved understanding of how surface water dynamics influence river discharge can be used to improve the resilience of river systems to climate extremes.

Understanding and projecting the downstream benefits of terrestrial surface water storage (volumetric water stored in lakes and wetlands, SWstorage) requires watershed hydrologic models. Use of external datasets to calibrate and validate modeled SWstorage dynamics remains uncommon, particularly across major river basins. Here, we: (1) develop and assess the utility of a novel remote sensing-based (RS) SWstorage approach for verifying watershed-model SWstorage estimates, (2) compare average modeled and RS SWstorage volume across the landscape, and (3) compare variability in modeled and RS SWstorage through time. We used SWstorage informed by Sentinel-1 and -2 (RS SWstorage), with Soil and Water Assessment Tool (SWAT) model simulations (SWAT SWstorage) across the ~450,000 km2 Upper Mississippi River Basin. We found that RS SWstorage was, on average, lower than SWAT SWstorage in tile-drained agricultural regions where static Digital Elevation Model (DEM)-generated depressions used in the SWAT model often did not contain RS surface water. Conversely, RS SWstorage was higher than SWAT SWstorage in wetland-rich regions where surface water was shallower than DEM vertical accuracy. In modeled subbasins where DEM-generated maximum SWstorage capacity was low relative to SWAT SWstorage volumes, SWAT SWstorage was effectively capped and unable to vary through time, whereas RS SWstorage in the same subbasins continued to vary. Thus, RS SWstorage allows for a more accurate representation of where, when, and how much water is on the landscape. This finding is useful for informing watershed model initial conditions and highlights the potential for RS to be used in SWstorage calibration or data assimilation.

Understanding and projecting the downstream benefits of terrestrial surface water storage (volumetric water stored in lakes and wetlands, SWstorage) requires watershed hydrologic models. Use of external datasets to calibrate and validate modeled SWstorage dynamics remains uncommon, particularly across major river basins. Here, we: (1) develop and assess the utility of a novel remote sensing-based (RS) SWstorage approach for verifying watershed-model SWstorage estimates, (2) compare average modeled and RS SWstorage volume across the landscape, and (3) compare variability in modeled and RS SWstorage through time. We used SWstorage informed by Sentinel-1 and -2 (RS SWstorage), with Soil and Water Assessment Tool (SWAT) model simulations (SWAT SWstorage) across the ~450,000 km2 Upper Mississippi River Basin. We found that RS SWstorage was, on average, lower than SWAT SWstorage in tile-drained agricultural regions where static Digital Elevation Model (DEM)-generated depressions used in the SWAT model often did not contain RS surface water. Conversely, RS SWstorage was higher than SWAT SWstorage in wetland-rich regions where surface water was shallower than DEM vertical accuracy. In modeled subbasins where DEM-generated maximum SWstorage capacity was low relative to SWAT SWstorage volumes, SWAT SWstorage was effectively capped and unable to vary through time, whereas RS SWstorage in the same subbasins continued to vary. Thus, RS SWstorage allows for a more accurate representation of where, when, and how much water is on the landscape. This finding is useful for informing watershed model initial conditions and highlights the potential for RS to be used in SWstorage calibration or data assimilation.

High-frequency observations of surface water at fine spatial scales are critical to effectively manage aquatic habitat, flood risk and water quality. We developed inundation algorithms for Sentinel-1 and Sentinel-2 across 12 sites within the conterminous United States (CONUS) covering >536,000 km2 and representing diverse hydrologic and vegetation landscapes. These algorithms were trained on data from 13,412 points spread throughout the 12 sites. Each scene in the 5-year (2017-2021) time series was classified into open water, vegetated water, and non-water at 20 m resolution using variables not only from Sentinel-1 and Sentinel-2, but also variables derived from topographic and weather datasets. The Sentinel-1 model was developed distinct from the Sentinel-2 model to enable the two time series to be integrated into a single high-frequency time series, while open water and vegetated water were both mapped to retain mixed pixel inundation. Results were validated against 7,200 visually inspected points derived from WorldView and PlanetScope imagery. Classification accuracy for open water was high across the 5-year period, with an omission and commission error of only 3.1% and 0.9% for Sentinel-1 and 3.1% and 0.5% for Sentinel-2, respectively. Vegetated water accuracy was lower, as expected given that the class represents mixed pixels. Sentinel-2 showed higher accuracy (10.7% omission and 7.9% commission error) relative to Sentinel-1 (28.4% omission and 16.0% commission error). Our results demonstrated that Sentinel-1 and Sentinel-2 time series can be integrated to improve the temporal resolution when mapping open and vegetated waters, although sensor-specific differences, such as sensitivity to vegetation structure versus pixel color, complicate the data integration for subpixel, vegetated water compared with open water.

High-frequency observations of surface water at fine spatial scales are critical to effectively manage aquatic habitat, flood risk and water quality. We developed inundation algorithms for Sentinel-1 and Sentinel-2 across 12 sites within the conterminous United States (CONUS) covering >536,000 km2 and representing diverse hydrologic and vegetation landscapes. These algorithms were trained on data from 13,412 points spread throughout the 12 sites. Each scene in the 5-year (2017-2021) time series was classified into open water, vegetated water, and non-water at 20 m resolution using variables not only from Sentinel-1 and Sentinel-2, but also variables derived from topographic and weather datasets. The Sentinel-1 model was developed distinct from the Sentinel-2 model to enable the two time series to be integrated into a single high-frequency time series, while open water and vegetated water were both mapped to retain mixed pixel inundation. Results were validated against 7,200 visually inspected points derived from WorldView and PlanetScope imagery. Classification accuracy for open water was high across the 5-year period, with an omission and commission error of only 3.1% and 0.9% for Sentinel-1 and 3.1% and 0.5% for Sentinel-2, respectively. Vegetated water accuracy was lower, as expected given that the class represents mixed pixels. Sentinel-2 showed higher accuracy (10.7% omission and 7.9% commission error) relative to Sentinel-1 (28.4% omission and 16.0% commission error). Our results demonstrated that Sentinel-1 and Sentinel-2 time series can be integrated to improve the temporal resolution when mapping open and vegetated waters, although sensor-specific differences, such as sensitivity to vegetation structure versus pixel color, complicate the data integration for subpixel, vegetated water compared with open water.

This data release presents truth data and benchmark results describing simulation of hydrologic drought events in the conterminous United States. This data release supports a publication (Simeone and others, 2024) which documents drought benchmarking methods and their application to the results of the National Hydrologic Model Precipitation-Runoff Modeling System v1.0 (NHM-PRMS). Truth data used were observations at U.S. Geological Survey streamgages across the conterminous United States. These data include 4662 U.S. Geological Survey streamgages with a historical period from 1984-2016. The following files are included in this data release: 1) kappa_long_nhm.csv: Benchmark results for the Cohen's kappa evaluation metrics in long table format. 2) spear_bias_dist_long_nhm.csv: Benchmark results for the Spearman's, bias, and distributional evaluation metrics in long table format. 3) ann_eval_long_nhm.csv: Benchmark results for the annual drought evaluation metrics in long table format. 4) streamflow_percentiles_nhm.zip: A zip file containing individual streamflow percentile data files used in this analysis as truth data. 5) input_data_nhm.zip: A zip file with input data for individual streamgages used for our data analysis pipeline as truth data. 6) streamflow_gages_in_study.csv: Metadata information for the 4662 U.S. Geological Survey streamgages contained in the above datasets.

This data release presents truth data and benchmark results describing simulation of hydrologic drought events in the conterminous United States. This data release supports a publication (Simeone and others, 2024) which documents drought benchmarking methods and their application to the results of the National Water Model (NWM) version 2.1. Truth data used were observations at U.S. Geological Survey streamgages across the conterminous United States. These data include 4662 U.S. Geological Survey streamgages with a historical period from 1984-2016. The following files are included in this data release: 1) kappa_long_nwm.csv: Benchmark results for the Cohen's kappa evaluation metrics in long table format. 2) spear_bias_dist_long_nwm.csv: Benchmark results for the Spearman's, bias, and distributional evaluation metrics in long table format. 3) ann_eval_long_nwm.csv: Benchmark results for the annual drought evaluation metrics in long table format. 4) streamflow_percentiles_nwm.zip: A zip file containing individual streamflow percentile data files used in this analysis as truth data. 5) input_data_nwm.zip: A zip file with input data for individual streamgages used for our data analysis pipeline as truth data. 6) streamflow_gages_in_study.csv: Metadata information for the 4662 U.S. Geological Survey streamgages contained in the above datasets.

This data release presents truth data and benchmark results describing simulation of hydrologic drought events in the conterminous United States. This data release supports a publication (Simeone and others, 2024) which documents drought benchmarking methods and their application to the results of the National Water Model (NWM) version 2.1. Truth data used were observations at U.S. Geological Survey streamgages across the conterminous United States. These data include 4662 U.S. Geological Survey streamgages with a historical period from 1984-2016. The following files are included in this data release: 1) kappa_long_nwm.csv: Benchmark results for the Cohen's kappa evaluation metrics in long table format. 2) spear_bias_dist_long_nwm.csv: Benchmark results for the Spearman's, bias, and distributional evaluation metrics in long table format. 3) ann_eval_long_nwm.csv: Benchmark results for the annual drought evaluation metrics in long table format. 4) streamflow_percentiles_nwm.zip: A zip file containing individual streamflow percentile data files used in this analysis as truth data. 5) input_data_nwm.zip: A zip file with input data for individual streamgages used for our data analysis pipeline as truth data. 6) streamflow_gages_in_study.csv: Metadata information for the 4662 U.S. Geological Survey streamgages contained in the above datasets.

Effective monitoring and prediction of flood and drought events requires an improved understanding of how and why surface-water expansion and contraction in response to climate varies across space. This paper sought to (1) quantify how interannual patterns of surface-water expansion and contraction vary spatially across the Prairie Pothole Region (PPR) and adjacent Northern Prairie (NP) in the United States, and (2) explore how landscape characteristics influence the relationship between climate inputs and surface-water dynamics. Due to differences in glacial history, the PPR and NP show distinct patterns in regards to drainage development and wetland density, together providing a diversity of conditions to examine surface-water dynamics. We used Landsat imagery to characterize variability in surface-water extent across 11 Landsat path/rows representing the PPR and NP (images spanned 1985-2015). The PPR not only experienced a 2.6-fold greater surface-water extent under median conditions relative to the NP, but also showed a 3.4-fold greater change in surface-water extent between drought and deluge conditions. The relationship between surface-water extent and accumulated water availability (precipitation minus potential evapotranspiration) was quantified per watershed and statistically related to variables representing hydrology-related landscape characteristics (e.g., infiltration capacity, surface storage capacity, stream density). To investigate the influence stream connectivity has on the rate at which surface water leaves a given location, we modeled stream-connected and stream-disconnected surface water separately. Stream-connected surface water showed a greater expansion with wetter climatic conditions in landscapes with greater total wetland area, but lower total wetland density. Disconnected surface water showed a greater expansion with wetter climatic conditions in landscapes with higher wetland density, lower infiltration and less anthropogenic drainage. From these findings, we can expect that shifts in precipitation and evaporative demand will have uneven effects on surface-water quantity. Accurate predictions regarding the effect of climate change on surface-water quantity will require consideration of hydrology-related landscape characteristics including wetland storage and arrangement.