Model precipitation skill evaluated with radar data

A large scale, frontal precipitation system is simulated using version 3.2 of the Weather Research and Forecasting – Advanced Weather Research (WRF-ARW) model to test the impact of horizontal resolution (3km, 1km and 0.33km) and bulk microphysical parameterization (BMP) scheme on model performance. The forecasts are then evaluated as a function of horizontal scale against different types of radar observations (composite rainfall estimates, single S-band observations and vertically-pointing X-band profiles). It is found that for stratiform precipitation, the forecasts are more sensitive to changes in the BMP than to changes in horizontal resolution. However, any differences between model configurations do not generate significant improvements in performance measured with respect to radar observations. This is further illustrated through the evaluation of different forecasting systems for 24 rainfall events over the continental United States, which shows that precipitation forecasts from the Canadian model GEM (global model with variable resolution, convection parameterized) and from WRF-ARW run at 4 km grid spacing (limited area model, convection explicit) compare better to each other than to radar rainfall estimates.