Development of Statistical Post-Processing Techniques for Improved Low-level Wind Speed Forecasts in West Texas

The wind energy industry needs accurate forecasts of wind speeds at hub height and in the rotor layer to accurately project expected power output from a wind farm. These forecasts aid the industry in deciding if there will be enough power to meet the energy need or if back up power sources will be necessary, and they are also used to site future wind farms. Current numerical weather prediction (NWP) models struggle to accurately predict low level winds such as those at turbine heights, partially because of systematic biases within the models. These systematic errors are addressed through this study with statistical post-processing techniques such as model output statistics (MOS) and the analog ensemble (AnEn) approach. Additionally, the analog ensemble approach is used to take advantage of the spread of solutions produced by the ensemble members in creating a reliable forecast distribution. This study uses reforecasts from the Weather Research and Forecasting (WRF-ARW) model version 3.5.1 and observations from SODAR instruments in West Texas over periods of up to two years to examine the skill added to forecasts when applying both MOS and AnEn techniques. Different aspects of the techniques are tested such as model horizontal and vertical resolution, number of predictors, and training set length. Further, experiments are composited over different times to address the time-dependent nature of systematic wind biases. Both MOS and the AnEn are applied to several levels representing heights in the turbine rotor layer (40, 60, 80, 100, 120 m). Ultimately the results of this study present the degree of improvement each technique provides to raw WRF forecasts, and the translation of adjusted low-level wind forecasts into the real-time Texas Tech University (TTU) weather prediction system is discussed.