Improving GLAMEPS wind speed forecasts by statistical postprocessing

The Grand Limited Area Model Ensemble Prediction System (GLAMEPS) is a short range multimodel ensemble system, a combination of the subensembles from the ALADIN, HIRLAM STRACO and Kain-Fritsch and ECMWF models. In this report it is shown that the (gridbox average) GLAMEPS 10 meter wind speed forecasts can be improved significantly by statistical postprocessing when verified against station observations. Results show that both reliability and resolution can be improved for wind speeds up to 16 m/s. Special attention has been paid to the improvement for higher wind speeds. Furthermore, experiments indicated that use of the mean of the control runs of the four models instead of the ensemble mean do not result in losses in Brier skill scores for forecasts 18, 30 and 42 hours ahead. Postprocessing of the complete ensemble performed better than postprocessing each of the four subensembles separately. These conclusions were derived from data from November and December 2011 for stations all over Europe and need to be confirmed for datasets of longer periods and outputs of the current GLAMEPS version.