Evaluation of surface sensible weather forecasts by the WRF and the eta models over the western United States

An evaluation of the surface sensible weather forecasts using high-density observations provided by the MesoWest cooperative networks illustrates the performance characteristics of the Cooperative Institute for Regional Prediction (CIRP) Weather Research and Forecast (WRF) and the Eta Models over the western United States during the 2003 warm season (June–August). In general, CIRP WRF produced larger 2-m temperature and dewpoint mean absolute and bias errors (MAEs and BEs, respectively) than the Eta. CIRP WRF overpredicted the 10-m wind speed, whereas the Eta exhibited an underprediction with a comparable error magnitude to CIRP WRF. Tests using the Oregon State University (OSU) Land Surface Model (LSM) in CIRP WRF, instead of a simpler slab-soil model, suggest that using a more sophisticated LSM offers no overall advantage in reducing WRF BEs and MAEs for the aforementioned surface variables. Improvements in the initialization of soil temperature in the slab-soil model, however, did reduce the temperature bias in CIRP WRF. These results suggest that improvements in LSM initialization may be as or more important than improvements in LSM physics. A concerted effort must be undertaken to improve both the LSM initialization and parameterization of coupled land surface–boundary layer processes to produce more accurate surface sensible weather forecasts.