FORECASTING WIND POWER IN HIGH WIND PENETRATION MARKETS, USING MULTI-SCHEME ENSEMBLE PREDICTION METHODS

Investigations from operational forecasting with a 75-member multi-scheme ensemble prediction system (MSEPS) have been carried out in order to find the various unknown error sources in wind power predictions. Various methods of predicting wind power and an error decomposition was used to identify the model components to further improve. The best of the wind power predictions was based on 300 input weather parameters. Additionally, the ensemble's ability to predict the forecasting error is shown, which can be used to estimate the reserve requirements and thereby lower the costs of balancing wind power, making this energy source more compatible in the liberalised markets. Results are presented for Germany, Denmark , Ireland and Australia.