Examination of the Ability of the WRF Modeling System to Predict Observed Rainfall Patterns in the San Francisco Bay Area

Our goal is to examine the skill with which the WRF model can predict the rainfall patterns observed in the San Francisco Bay Area (SFBA). In particular, we focus on the rainshadow (RS) often seen over South Bay cities such as San Jose. For both RS events and widespread rain events, such as the one on 20 Feb 2017 which caused flash-flooding and major property damage in San Jose, it is important to develop a modeling system that can provide accurate local rainfall information to water and emergency managers. Preliminary results indicate that simulations initialized with NAM analyses generally do no better than those initialized with GFS, and often perform less well. Experiments also suggest that, although we use a nested grid with the inner grid having a resolution around 1 km, more accurate results are obtained with cumulus parameterization turned on (with cu=1 or 2 giving best results). Case studies of certain rain events (RS events and the 20 Feb 2017 heavy rain event) are presented and discussed.