Assessing the effects of uncertainty in windspeed and precipitation forcings on lateral snow redistribution in mountainous basins

Snow-dominated montane watersheds provide critical ecological function, water storage, and water supply for downstream population centers across the globe. Recent literature suggests that hydrologic model uncertainty in these watersheds is largely driven by meteorological forcing uncertainty. Additionally, few models simulate lateral snow transport processes such as blowing snow and avalanche, meaning that the impact of forcing uncertainty on snowpack redistribution is unknown. This pair of limitations presents a distinct challenge for modelers in both identifying accurate model structures and identifying the drivers of simulated results. In this study, we ask how uncertainty in windspeed and precipitation forcing affects modeled lateral redistribution of snow in mountain basins. We hypothesize that windspeeds and precipitation from downscaled meteorological datasets require numerical correction for effective snow redistribution, and that the magnitude of these corrections will vary across geographic regions. Analyzing the impacts of these uncertainties allows us to determine how influential windspeed and precipitation forcings are on snow transport processes and on the spatial patterns of snow accumulation and melt dynamics.

We use the Canadian Hydrologic Model (CHM), to simulate snow accumulation and melt over five water years within a set of basins in the Sierra Nevada and Rocky Mountains in the United States that have extensive airborne lidar observations from the Airborne Snow Observatory (ASO). CHM runs over a triangular mesh with a six-layer snowpack energy balance model and lateral transport through blowing snow and avalanche. We use two climate forcing datasets with different underlying resolutions to evaluate the effects of windspeed and precipitation on modeled snowpack in mountainous terrain. ERA5-Land, a 9-km resolution dataset, is selected because its global coverage is advantageous for geographic generalizability. The CONUS404 product, a 4-km resolution dynamically downscaled dataset from ERA5 over the contiguous US, is selected to test a higher resolution product over the areas of interest. In each basin, windspeed and precipitation are perturbed to assess sensitivity and the resulting snowpack distribution.

We use observed SWE, snow cover, and derived snow disappearance date from SNOTEL, snow courses, and MODSCAG to evaluate model results using a standardized benchmarking process. This enables us to decipher whether corrections to windspeed and precipitation yield similar metrics despite different underlying redistribution processes. By evaluating models across two climatically distinct regions, we can assess whether numerical precipitation and windspeed adjustments improve snow simulations, and whether they are transferable or region-specific. We present an overview of the study and results demonstrating how uncertainty in meteorologic forcing propagates into lateral snow transport processes, which can provide guidance for improving snowpack simulations across complex mountainous terrain.