Authors: Hongping Gu, Wei Zhang, Sherly R. Disanayakalage, SnehadEEP Ballav – Utah State University
Title: Quantifying Sensitivity of 2025 Texas Flood in WRF-Hydro to Forcing Uncertainties
Presentation Type: Poster Presentation
Abstract: Flash-flood prediction remains highly sensitive to uncertainties in hydrometeorological forcing, particularly for short-duration, high-intensity convective events. In this study, we investigate the sensitivity of WRF-Hydro simulated streamflow to different precipitation forcing datasets during the July 4, 2025 flash flood event in the Guadalupe River Basin, Texas. Four widely used forcing products—NLDAS, AORC, HRRR, and MRMS—are evaluated within a consistent WRF-Hydro modeling framework at 1 km resolution. We focus on two key USGS streamflow stations, 08165500 (Guadalupe River at Hunt, TX) and 08167000 (near Kerrville, TX), which exhibit the largest observed discharge peaks during the event. Model simulations are compared against USGS observations to assess differences in peak magnitude, timing, and cumulative runoff response across forcing datasets. Results show that precipitation forcing uncertainty leads to substantial variability in simulated streamflow, particularly in peak discharge magnitude and timing. MRMS forcing generally produces the highest peak flows and best matches observed peak magnitudes, while NLDAS tends to underestimate peak intensity. HRRR captures the temporal evolution of precipitation reasonably well but shows mixed performance in peak magnitude. AORC exhibits intermediate behavior between NLDAS and MRMS. Differences in accumulated precipitation directly translate into differences in runoff generation and flood peak response, highlighting the importance of accurately representing convective rainfall. Overall, this study demonstrates that high-resolution, observation-constrained precipitation datasets such as MRMS are critical for improving flash-flood prediction in WRF-Hydro.