Precipitation Partitioning Including Multiple Variables Lead to More Accurate Models for Climate Change Scenarios

Accurate partitioning of rain and snow precipitation is fundamental to estimation of snow and streamflow by hydrology models. Hydrology models take a simplistic temperature based static approach to precipitation partitioning. Recent research suggests that switching to dynamic partitioning that considers multiple meteorological variables improves hydrological model performance related to snow magnitude and phenology metrics. This change also impacts what fraction of streamflow in a region is snowmelt driven, a key consideration in climate change assessment due to its impact on the outcome of hydrographs. We evaluated two scenarios: limiting long-term temperature rise and continuing current temperature rise trends. Our goal is to (a) quantify how dynamic partitioning changes the snowmelt contribution to runoff compared to static partitioning and (b) evaluate how this can alter transitions across watershed classification (snow-dominant, transitional, and rain dominant) in a climate change context. The VIC-CropSyst model was applied to the Yakima River Basin as a case study to address these questions. We found that the average DRS partitioning resulted in snowmelt having a higher contribution to streamflow (roughly 30% historically, 45% in the near future, and 65% in the distant future). These results helped prove our original hypothesis of if multiple variables are factored into the model’s usage, it leads to improved performance and accuracy of that model. The higher elevations show a smaller change in the amount of snowmelt that contributes to streamflow (i.e. 25% in the Naches), while this change is much larger in lower elevations (i.e. 185% in Lower Yakima) regarding the same scenario. DRS partitioning forecasts much of the Yakima River Basin go from a snow dominant to a transitional watershed, while SRS projected the same area goes from snow dominant to rain dominant.