Operational water forecast ability of the HRRR-iSnobal combination: an evaluation to adapt into production environments

Operational water-resource forecasters, such as the Colorado Basin RiverForecast Center (CBRFC) in the Western United States, currently rely on historical records to calibrate the temperature-index models used forsnowmelt runoff predictions. This data dependence is increasinglychallenged, with global and regional climatological factors changing theseasonal snowpack dynamics in mountain watersheds. To evaluate and improvethe CBRFC modeling options, this work ran the physically based snow energybalance iSnobal model, forced with outputs from the High-Resolution RapidRefresh (HRRR) numerical weather prediction model across 4 years in a Colorado River Basin forecast region. Compared to in situ, remotely sensed, and the current operational CBRFC model data, the HRRR-iSnobal combinationshowed well-reconstructed snow depth patterns and magnitudes until peakaccumulation. Once snowmelt set in, HRRR-iSnobal showed slower simulatedsnowmelt relative to observations, depleting snow on average up to 34 dlater. The melting period is a critical component for water forecasting.Based on the results, there is a need for revised forcing data inputpreparation (shortwave radiation) required by iSnobal, which is arecommended future improvement to the model. Nevertheless, the presentedperformance and architecture make HRRR-iSnobal a promising combination forthe CBRFC production needs, where there is a demonstrated change to theseasonal snow in the mountain ranges around the Colorado River Basin. Thelong-term goal is to introduce the HRRR-iSnobal combination in day-to-dayCBRFC operations, and this work created the foundation to expand andevaluate larger CBRFC domains.