Wind power prediction based on numerical and statistical models

The issue of wind power prediction is approached in this work by means of numerical and statistical prediction models. Two high resolution regional atmospheric systems are employed in order to provide accurate local wind forecasts while a combination of statistical post processes is utilized targeting to the local adaptation of the results and the reduction of possible systematic biases. A variety of power estimation models are employed for the prediction of the wind power potential in real time applications over two areas of Greece: the islands of Crete and Kefalonia. The results obtained prove that accurate power prediction can be reached if the local environmental conditions are credibly estimated while the use of the power output in previous time steps do not contribute significantly to the improvement of the final forecast.