Tailoring WRF and Noah-MP to Improve Process Representation of Sierra Nevada Runoff: Diagnostic Evaluation and Applications

Watersheds at the western margin of the Sierra Nevada mountains in California are regulated by large dams providing crucial water supply, flood control, and electricity generation. Runoff in these basins is snowmelt dominated and therefore vulnerable to alteration due to climate change. Regional climate models coupled to land surface models can be used to study the hydrologic impacts of climate change, but there is little evidence that they accurately simulate watershed-scale runoff in complex terrain. This study evaluates capabilities of the Weather Research and Forecasting (WRF) regional climate model, coupled to the Noah-multiparameterization (MP) land surface model, to simulate runoff into nine Sierra Nevada reservoirs over the period 2007–2017. Default parameterizations lead to substantial inaccuracy in results, including median bias of 61%. Errors can be traced to process representations; specifically, we modify the representation of snowflake formation in the Thompson microphysics scheme and subsurface runoff generation in the Noah-MP land surface model, including a correction representing effects of groundwater storage. The resulting parameterization improves Nash-Sutcliffe efficiency to above 0.7 across all basins and reduces median bias to 21%. To assess capabilities of the modified WRF/Noah-MP system in supporting analysis of human-altered hydrology, we use its streamflow projections to force a reservoir operations model, results of which maintain high accuracy in predicting reservoir storage and releases (mean Nash-Sutcliffe efficiency > 0.41). This diagnostic analysis indicates that coupled climate and land surface models can be used to study climate change effects on reservoir systems in mountain regions via dynamical downscaling, when adequate physical parameterizations are used