Comparison of Microphysics Schemes for Simulation of Snow Cover Fraction in the Sierra Nevada

Abstract:

Mountain snow cover is in important component of regional climate due to its high albedo and feedback properties. Models are often used to examine snow covered area (SCA) of large geographical areas and to understand snow cover variability. We utilize the Weather Research and Forecasting (WRF) regional climate model, combined with the Noah land surface model with multiparameterization options (Noah-MP), to simulate snow cover fraction (SCF) in a 3 km domain over the central Sierra Nevada, which includes the Stanislaus, Tuolumne, Merced, and San Joaquin watersheds. The impact of several microphysics schemes on SCF simulation is compared. The Thompson microphysics scheme is capable of identifying total domain-wide SCA with an error of 2144.2 km² (25%). The Goddard scheme has a RMSE of 2393.6 km² (28%) for simulating SCA, and the Community Atmospheric Model 5.1 scheme has a RMSE of 2384.0 km² (27%). While WRF correctly simulates the presence or absence of snow in an average snow year, the model consistently simulates too much snow on the western side of the Sierra Nevada and too little on the eastern side, regardless of microphysics scheme. For the Thompson microphysics option, WRF simulates SCF on a mid-winter day with an overall bias of ~17% compared to satellite observations. However, the bias is ~31% on the western side of the range and ~-23% on the eastern side. Further work will aim to understand the differences in SCF simulation skill on either side of the Sierra Nevada range. Wet and dry years will also be examined to determine whether the east/west bias prevails in years with different accumulation patterns.