A New Way to Predict Forecast Skill

Forecast skill (Anomaly Correlated Coefficient, ACC) is a quantity to show the forecast quality ofthe products of numerical weather forecasting models. Predicting forecast skill, which is the foundationof ensemble forecasting, means submitting products to predict their forecast quality before they are used.Checking the reason is to understand the predictability for the real cases. This kind of forecasting servicehas been put into operational use by statistical methods previously at the National Meteorological Center(NMC), USA (now called the National Center for Environmental Prediction (NCEP)) and European Centerfor Medium-range Weather Forecast (ECMWF). However, this kind of service is far from satisfactorybecause only a single variable is used with the statistical method. In this paper, a new way based onthe Grey Control Theory with multiple predictors to predict forecast skill of forecast products of theT42L9 of the NMC, China Meteorological Administration (CMA) is introduced. The results show: (1)The correlation coefficients between forecasted and real forecast skill range from 0.56 to 0.7 at differentseasons during the two-year period. (2) The grey forecasting model GM(1,8) forecasts successfully thehigh peaks, the increasing or decreasing tendency, and the turning points of the change of forecast skill ofcases from 5 January 1990 to 29 February 1992.