Authors: Wei Zhang – Utah State University
Title: When Bias Correction Helps and When It Hurts: WRF-Hydro Streamflow Sensitivity to CONUS404-Based Forcing in Snow-Dominated and Convective Flood Regimes
Presentation Type:
Abstract: High-resolution hydroclimate datasets such as CONUS404 and its bias-adjusted successor CONUS404-BA are increasingly used to force hydrologic models, yet their event-scale impacts on streamflow remain uncertain. We assess WRF-Hydro sensitivity to raw CONUS404 and Daymet-based CONUS404-BA in two contrasting regimes: snow-dominated in water year 2011 in the Upper Colorado River Basin (UCRB) and the 30 June-7 July, 2002 convective flood in the Guadalupe River Basin, Texas. For the Guadalupe River Basin, we also conduct a hybrid experiment (CONUS404_BA_StageIV) that replaces only the event-period precipitation in CONUS404-BA with Stage IV radar-gauge analyses while fixing all other inputs. We analyze domain-mean forcing differences, and resulting snow, water-balance and hydrograph responses, and assess streamflow performance using observed U.S. Geological Survey (USGS) streamgage records. In the UCRB, CONUS404-BA reduces wet and cold biases relative to NLDAS, improves snow accumulation and melt timing, and substantially enhances streamflow and water-balance performance versus raw CONUS404. In Guadalupe, however, bias adjustment dampens storm intensity and spatial gradients, smoothing the peak-rainfall maximum and yielding a muted flood hydrograph with degraded streamflow skill at USGS streamgages. Raw CONUS404 preserves recognizable storm structure but underestimates total rainfall depth, whereas CONUS404_BA_StageIV recovers a physically plausible hydrograph and improves skill at most USGS streamgages. These results show that event-agnostic bias adjustment can benefit snow regimes yet degrade convective extremes, underscoring the need to preserve event-scale precipitation structure for flood applications.