Geology topics

Aaron Joseph Heldmyer

Publications and source records attributed to Aaron Joseph Heldmyer.

5 recordsLinked to original sources

USGS2026

Machine learning generated streamflow drought forecasts for the conterminous United States (CONUS): developing and evaluating an operational tool to enhance sub-seasonal to seasonal streamflow drought early warning for gaged locations

Forecasts of streamflow drought, when streamflow declines below typical levels, are notably less available than for floods or meteorological drought, despite widespread impacts. We apply machine learning (ML) models to forecast streamflow drought 1–13 weeks ahead at 3,219 streamgages across the conterminous United States. We applied two ML methods (Long short-term memory neural networks; Light Gradient-Boosting Machine) and two benchmark models (persistence; Autoregressive Integrated Moving Average) to predict weekly streamflow percentiles with independent models for each forecast horizon. ML models outperformed benchmarks in predicting continuous streamflow percentiles below 30%. ML models generally performed worse than persistence models for discrete classification (moderate, severe, extreme) but exceeded the benchmark models for drought onset/termination. Performance was better for less intense droughts and shorter horizons, with predictive power for 1–4 weeks for severe droughts (10% threshold). This work highlights challenges and opportunities to advance hydrological drought forecasting and supports a new experimental forecasting tool.

conterminous United States↗

USGS2026

Snow simulations predict future changes in rain-on-snow events across the upper Gallatin River watershed, a Greater Yellowstone Ecosystem headwater system

Study region: The upper Gallatin River watershed, an alpine headwater system in the Greater Yellowstone Ecosystem, in Wyoming and Montana. Study focus: As global and regional air temperatures rise, mountain headwaters across the Greater Yellowstone Ecosystem (GYE) are projected to see more precipitation falling as rain. While the hydrologic effects of this snow-to-rain transition depends on a variety of factors, it can lead to an increased occurrence of rain-on-snow (RoS) events. To investigate these changes, we used high-resolution (30 m) SnowModel simulations of the upper Gallatin River watershed. Simulations were run for 2001-2013 using two scenarios: (1) historical meteorology as control and (2) pseudo global warming (PGW) where control air temperature and precipitation conditions were perturbed to represent mean end-of-century conditions under a high-emissions scenario. New hydrological insights for the region: SnowModel outputs show that changes in PGW precipitation and snow accumulation varied with elevation. Warmer air temperatures at low elevations (< 2,500 m) led to less snow accumulation and less precipitation falling as snow. Colder baseline air temperatures for elevations above 2,500 meters (m) resulted in minor reductions in winter snowfall fraction. For PGW simulations, spring (April-June) months were rainier, and elevations above 2,500 m experienced more RoS events. Snowpacks between 2,500-3,100 m generated more snowmelt during RoS events, which was reflected in the watershed average. More high-intensity melt events can affect aquatic habitat, water quality, and the accuracy of streamflow forecasts across the region.

Montana, Wyoming↗

USGS2025

Machine learning generated streamflow drought forecasts for the Conterminous United States (CONUS): Developing and evaluating an operational tool to enhance sub-seasonal to seasonal streamflow drought early warning for gaged locations

Forecasts of streamflow drought, when streamflow declines below typical levels, are notably less available than for floods or meteorological drought, despite widespread impacts. To address this gap, we apply machine learning (ML) models to forecast streamflow drought 1-13 weeks into the future at > 3,000 streamgage locations across the conterminous United States (CONUS). We applied two ML methods (Long short-term memory (LSTM) neural networks; Light Gradient-Boosting Machine - LightGBM) and two benchmark model approaches (persistence; Autoregressive Integrated Moving Average - ARIMA) to predict weekly streamflow percentiles with independent models for each forecast horizon. To explore whether a training focus on dry weeks improved performance, both ML models were trained using all percentiles (LSTM-all, LightGBM-all) and only percentiles below 30% (LSTM<30, LightGBM<30). We evaluated model performance regionally and nationally for drought occurrence (the classification performance for a future date) and for drought onset/termination (performance identifying drought starts and ends). ML models generally performed worse than the persistence model for discrete classification (moderate, severe, extreme drought) of drought occurrence but exceeded the benchmark models for onset/termination. ML models outperformed benchmarks in predicting continuous streamflow percentiles below 30%. Occurrence performance was better for less intense droughts and shorter forecast horizons, with the ML models having predictive power at 1-4 week horizons for severe droughts (10th percentile threshold). All models struggled to forecast onset, though the best ML model was the LSTM<30 (sensitivity of 22%). Termination performance was greater, with the drought termination performance greatest for the LightGBM-all model. When estimating model uncertainty, the LSTM<30 model had the narrowest 90% percentile interval with closest to optimal capture. This work highlights the challenges and opportunities to further advance hydrological drought forecasting and supports an experimental operational streamflow drought assessment and forecast tool.

Continental United States↗

USGS2024

Streamflow timing and magnitude during snow drought depend on snow drought type and regional hydroclimate

Communities around the world rely on snowmelt to meet water demands, and periods of lower than normal snow accumulation, snow droughts, can decrease water supplies. Leveraging 172 minimally disturbed and seasonally snow-covered watersheds, we developed an approach to examine the effects of cool & dry, warm & dry, and warm & wet snow droughts on streamflow timing and magnitude by hydrologic region. Our results showed all types of snow droughts in all regions correlate with lower annual streamflow, lower maximum and minimum flows, and lower runoff ratios, with more numerous low flow days and earlier streamflow timing. However, departures from non-snow drought conditions differed substantially between drought types and regions. Consecutive snow droughts further reduced runoff ratios and increased low flow days, likely due to additional subsurface storage depletion. With warm snow drought occurrence expected to increase, we discuss impacts for water management systems whose design specifications may not reflect the changing hydroclimate.

Hydrological Sciences Journal↗

USGS2024

Evaluating distributed snow model resolution and meteorology parameterizations against streamflow observations: Finer Is not always better

Estimating snow conditions is often done using numerical snowpack evolution models at spatial resolutions of 500 m and greater; however, snow depth in complex terrain often varies on sub-meter scales. This study investigated how the spatial distribution of simulated snow conditions varied across seven model spatial resolutions from 30 to 1,000 m and over two meteorological data sets, coarser (≈12 km) and finer (4 km). Simulated snow covered area (SCA) was compared to remotely sensed SCA and simulated watershed mean peak snow water equivalent (SWE) was compared to four streamflow statistics representing different water management-relevant aspects of the hydrograph using non-parametric correlations. April 1 SWE tended to increase with model resolution, particularly below 4,000 masl. Finer meteorology simulations produced deeper April 1 SWE than coarser meteorology simulations. Finer resolution snow simulations tended to produce longer snowmelt durations and slower snowmelt rates than coarser resolution simulations. Finer resolution simulations had better agreement with SCA for both meteorology data sets, particularly at high and low elevations. However, finer resolution simulations did not generally outperform coarser simulations in snow versus streamflow statistic correlations. Snow versus streamflow correlations were most sensitive to meteorology, watershed properties, and then resolution. Watershed physiographic properties such as wetness index may increase snow versus streamflow metric correlations while elevation and slope may decrease correlations. At watershed scales, these results suggest that simulation resolution and choice of meteorology is less important than the physiographic properties of the watershed; however, if resolving snow distribution across the landscape is important, finer-resolution simulations are useful.

Colorado↗