Extended-range ensemble forecasting at ECMWF

摘要: The performance of a number time-lagged extended-range ensembles forecasts with ECMWF operational model at T63 resolution was evaluated. Ensemble averaging reduces the RMS error. There is positive correlation between spread and skill, however, much this could be explained by annual cycle. During winter skill tend to anticorrelated, which may indicate systematic error in model.