The WRF model is the next generation community mesoscale model designed to enhance collaboration between the research and operational sectors. The WRF model has two dynamical cores, the Advanced Research WRF (ARW) and the Non-hydrostatic Mesoscale Model (NMM). There are also two options for a “hot-start” initialization of the WRF model – the Local Analysis and Prediction System (LAPS) and the Advanced Regional Prediction System (ARPS) Data Analysis System (ADAS). Both LAPS and ADAS are three-dimensional weather analysis systems that integrate multiple meteorological data sources into one analysis over the domain of interest. These analysis systems allow mesoscale models to benefit from the addition of high-resolution data sources in their initial conditions. Having a series of initialization options and WRF cores, as well as a number of different model parameterizations within each core, provides SMG with needed flexibility. It also creates challenges, such as determining which configuration options are best to address specific forecast concerns. The goal of this project is to assess the different configurations available and determine which configuration will best predict surface wind speed and direction at EAFB.
Results from different combinations of WRF initializations/model parameterizations will be presented (ADAS-ARW, ADAS-NMM, LAPS-ARW, and LAPS-NMM, and variations of the WRF model physics) at a 1-km resolution over EAFB and adjacent areas. At least six wind cycling cases will be included, as well as two null cases (non-wind cycling days). Each model run is integrated 12 hours, each run beginning prior to the wind cycling event. To quantify model performance, a standard objective evaluation which consists of point forecast error statistics is employed. The forecasts of wind speed and direction will be compared with an independent sample of EAFB wind tower data not used in the initialization of the WRF model. Standard verification statistics are calculated such as root mean square (RMS) error, bias, and standard deviation. In addition, a subjective evaluation of the wind cycling cases will be presented in which the forecast fields are manually examined and verified for this meteorological phenomena. A summary of the relative skill of the various WRF configurations and how each configuration behaves relative to the others is given, as well as a determination of the best model configuration for predicting wind cycling at EAFB.