Dynamical Downscaling Of Global Reanalysis Data For High-Resolution Spatial Modeling Of Snow Accumulation/Melting At The Central/Southern Sierra Nevada Watersheds

Data scarcity is one of the major constraints for hydrological modeling at a mountainous region with complex terrains. In this study, WRF (Weather Research and Forecasting model) model was configured to produce highresolution atmospheric data for spatial modeling of snow at seven watersheds in central/southern Sierra Nevada. The configured WRF model downscaled a global reanalysis dataset into hourly time intervals and 3-km spatial grid resolution. Performance of the configured WRF model was validated by comparing basin averages of daily precipitation and daily mean temperature between the WRF simulation and data from Parameter-elevation Relationships on Independent Slopes Model (PRISM). The WRF model demonstrated satisfactory performance in simulating peaks and variations of daily precipitation and daily mean temperature. The atmospheric data which were obtained by the dynamical downscaling of Climate Forecast System Reanalysis dataset (CFSR) were then used for calibration and validation of the snow model for the seven watersheds. The dynamical downscaling and the subsequent simulation of snow were performed through the application of The Watershed Environmental Hydrology Hydro-Climate Model (WEHY-HCM) into the central/southern Sierra Nevada. The daily scale validation of the calibrated snow model showed that the simulated snow water equivalent (SWE) was mostly in agreement with observed SWE, however, tended to somewhat underestimate SWE. The snow model simulation with the precipitation input from PRISM indicated that the results using the dynamical downscaling had comparable performance with the results using PRISM precipitation, while some observation points showed improvements of the simulated SWE by using PRISM precipitation. Spatial comparison of the snow cover between the snow model simulations and Moderate Resolution Imaging Spectroradiometer (MODIS) observations showed that the snow model was capable of detecting the snow cover at high elevation bands while the snow cover detection at low elevation bands, which correspond to the boundary between snow cover and no snow cover area, was less accurate.