An Application of Model Output Statistics to the Development of a Local Wind Regime Forecast Procedure

Abstract

The Model Output Statistics (MOS) approach is used to develop a procedure for forecasting the occurrence of a local wind regime at Rota, Spain known as the levante. Variables derived solely from surface pressure and 500 mb height forecast fields of the Fleet Numerical Oceanography Center operational model are used as possible predictors of levante occurrence. These variables are screened by a stepwise discriminant analysis program to determine those which best discriminate between the occurrence and non-occurrence of a levante. The variables which are selected are used by the program as predictors in a levante forecast procedure. The forecast procedure is developed on a four-year data set and is then tested on a separate one-year data set. In addition to an independent data test, a ten is conducted to compare the levante forecasts produced by this procedure with the forecasts produced by the procedure used by forecasters at Rota. The MOS procedure shows skill and compares favorably with the Rota forecast procedure.