Zephyr - the next generation prediction.

Two of the most successful short-term prediction models (and the only ones in operational use at utilities) are going to be merged into one: the Risø model, developed by Landberg and the Wind Power Prediction Tool WPPT, developed at the Department of Mathematical Modelling IMM of the Danish Technical University. This paper will describe a new project funded by the Danish Ministry of Energy where the largest Danish utilities (Elkraft, Elsam, Eltra and SEAS) are participating. Two advantages can be achieved by combining the effort: The software architecture will be state-of-the-art, using the Java2TM platform and Enterprise Java Beans technology, and it will ensure that the best forecasts are given on all prediction horizons from the short range (0-9 hours) to the long range (36-48 hours). This is because the IMM approach uses online data and advanced statistical methods, which is advantageous in the short horizon, while the use of meteorological models like the HIRLAM model of the Danish Meteorological Institute is advantageous in the long term. The Risø approach is not based on data input from the wind farms, and can therefore deliver a starting guess, as well as a back-up for the case of missing online data.