A Dynamical-Statistical Downscaling Approach for Range-Scale Numerical Weather Prediction

At the U.S. Army Dugway Proving Ground, UT, forecasters use an ensemble NWP system known as E-4DWX (the Ensemble Four-Dimensional Weather System). E-4DWX, which was developed at the National Center for Atmospheric Research, comprises 30 WRF Model members run four times per day at a grid interval of 3.3 km, and a 4DWX-VLES (4DWX Very Large Eddy Simulation) system run eight times per day at a grid interval of 300 m. E-4DWX is based on WRF FDDA and multi-approach ensemble perturbations, whereas 4DWX-VLES runs single deterministic forecasts with an emphasis on range-scale data assimilation and high-resolution weather analysis and forecasting. 4DWX-VLES produces fine-scale gridded numerical analysis of weather variables at the range, and E-4DWX provides mesoscale probabilistic forecasts. To take advantage of E-4DWX forecasts and 4DWX-VLES analyses, a dynamic-statistical downscaling (DSD) approach is introduced to generate precision weather forecast for the range. DSD follows the idea of analog model output statistical post-processing proposed by Hamill et al. (2013, 2014) and Delle Monache et al. (2013). Here, we build and find the mesoscale weather analogs based on the E-4DWX ensemble forecasts, and make use of 4DWX-VLES analyses corresponding to the analogs to construct range-scale weather prediction and to estimate forecast uncertainties. As an initial test, we process 3 months of archived E-4DWX and 4DWX-VLES output. We start with E-4DWX ensemble means of the current forecasts, then we use self-organization maps (SOMS) as a pattern-matching technique to search for analogs. After the analogs are determined, the 4DWX-VLES analyses corresponding to the analogs are statistically integrated to generate range-scale forecasts. Preliminary results and validation will be presented.