The Impact of Various Forecasting System Components on Short-Term Wind Forecasts

Today's state of the art forecasting systems consist of three major components: data assimilation (e.g. 3-DVAR, Kalman Filtering etc), numerical modeling (available models from national centers, in-house customized numerical model runs) and statistical post-processing of the obtained forecasts. In this study we will evaluate how each of these components, as well as their synergy, contribute to the skill of short-term wind forecasts. To measure of the wind forecast accuracy, Mean Absolute Error (MAE) will be calculated in addition to wind ramp event statics such as equitable threat score (ETS) and False Alarm Ratio (FAR). The quantitative impact of each component will be assessed by applying the factor separation method on these forecast accuracy measures.