Forecast Accuracy
The chart shows the range of accuracy achieved by live GH Forecaster services in the UK, Spain, the USA, Ireland and Canada. The results are presented as Mean Absolute Error normalised by the wind farm capacity. A range of results are achieved largely due to the variety of wind regimes and topography. However, the adaptive GH Forecaster modelling methods have proved to be accurate in all tested conditions.
Click the image for a larger version. 
The chart shows a 25 day time history of GH Forecaster wind farm power predictions made 3 hours in advance, compared with actual production. The combination of physical models and adaptive “learning” algorithms enables accurate predictions of the timing and magnitude of power production at short forecast horizons.
Click the image for a larger version. 
When forecasting out to 12 hours in the future the trends of wind energy output are predicted well. There are some instances when the timing of weather systems and therefore wind energy is not predicted perfectly. However, it can be seen that the forecasts contain valuable information concerning the likely power production.
Click the image for a larger version. |
