Development and application of an operational, relocatable, mesogamma-scale weather analysis and forecasting system

We report on the result from an operational forecast system built to predict local circulations forced by complex terrain and other variations in land-surface characteristics. The cornerstone of the prediction system in the Penn State University/National Center for Atmospheric Research (PSU/NCAR) mesoscale model, version 5 (MM5), a nonhydrostatic regional model. The specific application reported herein is for the region surrounding Dugway Proving Ground (DPG) in west-central Utah. The nature of the terrain requires a horizontal resolution of about 1 km in order to capture the important local features. This resolution is achieved by the use of grid nesting. To our knowledge, this resolution is finer than that being used in any other currently operational forecast system tasked with local and regional weather prediction. For verification purposes, forecasts are stratified according to season and mean-flow characteristics. Data for verification consist of DPG surface mesonet data. Root-mean-square errors (RMSE), time mean circulations and spatial anomaly correlation statistics are computed and composited for each hour of the day. These are compared with identical forecasts of lagged persistence, with the time lag being 24 h (MM5 forecasts are all initialized at 1200 UTC). For wind and temperature, the RMSE from MM5 is consistently lower than that of persistence. In addition, MM5 shows skill in predicting the time-mean circulations on the test range (variations of a few m/s and °Celsius). MM5 forecast errors grow slowly with time until around sunsset, after which they decrease slightly, suggesting that local nighttime 'forcing' dominates the error growth, as the surface layer decouples from the free atmosphere. Finally, spatial anomaly correlations suggest that the non-systematic, range-scale circulations exhibit low predictability.