Probabilistic wind speed forecast for wind power prediction using pseudo ensemble approach

Accurate wind forecast at a wind farm is an essential process in wind energy industry and marketing. Numerical Weather Prediction (NWP) models can be used but they provide wind forecast as a single value for a given time horizon. Therefore, forecasting wind speed as a deterministic value doesn't represent the uncertainty of the wind speed forecast. Ensemble NWP forecast can be used to calculate the probability of occurrence of different wind speeds classes. The main disadvantage of this approach is the extensive computational resources required to run multiple copies of the NWP model. This paper, explores the possibility of using pseudo ensemble method for generating probabilistic wind forecast for wind farm applications. The proposed method utilizes the spatial and temporal neighborhoods of the forecast point to generate forecast dataset and then calculate the required probabilities. A case study using the proposed method is tested and validated using wind data from NWP model and measurements from three ground weather stations in Oman.