LAUSR

Predicting the phase of precipitation is fundamental to water supply and hazard forecasting.Phaseprediction methods (PPMs) are used to predict snow fraction, or the ratio of snowfall to total precipitation. Common temperature-based regression (Dai method) and threshold at freezing (0 C)PPMs had comparable accuracyto a humidity-based PPM (TRHmethod)using 6- and 24-hour observations. Using a daily climate dataset from 1980-2015,the TRHmethodestimates14% and 6% greater precipitation-weighted snow fraction than the 0Cand Dai methods, respectively. The TRH method predicts four times less area with declining snow fraction than the Dai method (2.1% and 8.1% of the study domain, respectively) from 1980-2015, with the largest differences in the Cascade and Sierra Nevada mountains and Southwestern U.S. Future RCP8.5 projections suggest warming temperatures of 4.2°C and declining relative humidity of 1% over the 21st century. The TRHmethod predicts a smaller reduction in snow fraction than temperature-only PPMs by 2100, consistent withlower humidity bufferingdeclines insnow fraction caused byregional warming.