Short-Term Wind Speed Forecasting by Spectral Analysis

Abstract

In this paper, we apply a wind speed forecasting framework developed earlier by the authors to the two-dimensional wind velocity measurements collected from a meteorological station in the Marmara region of Turkey over a short-term. In the application of the framework, first the measurements are de-trended for the diurnal and the weekly patterns. Other stages of the framework are covariance-factorization via a recently developed subspace method and one-step-ahead and/or multi-step-ahead Kalman filter predictions.