Wind Speed Modelling and Short-Term Prediction Using Wavelets

Abstract

The mathematical method of wavelets is explained and used to predict wind conditions using short-term data collected at a site and referred to long term data from, say, a meteorological station. We model the response time-series in terms of a multiscale wavelet decomposition of the explanatory time-series. Preliminary results of this method, using hourly 10 minute averaged data from six locations in the British Isles, allow comparison with a linear regression method in terms of prediction errors over 21 days.