Short term power forecasts for large offshore wind turbine arrays

Methods are presented for precise prediction of windspeeds at a wind- farm using a combination of numerical weather predictionmodels and on-site wind speedmeasurements. Simple techniques are used to inves- tigate the properties of the data. More advanced techniques are tested for the predictive value in forecasting including techniques from autore- gressive models, data assimilation, artificial neural networks and kernel dressing. The performance of the resulting forecasts are compared with the measured wind speedsusing a range of forecast windows.