Evaluating the use of a nonlinear two-dimensional model in downslope windstorm forecasts

Severe downslope windstorms are a mesoscale, primarily wintertime, phenomen on that affect regions in the lee of large mountain ranges. The resolution of current weather prediction models is too coarse to explicitly predict do wnslope windstorms. Hence, additional operational tools are needed for maki ng downslope windstorm forecasts. Current windstorm forecast techniques com monly utilize a tool referred to as a " decision tree. '' Although decision trees provide valuable guidance, operational forecasters have not found th is type of tool to be highly reliable. With recent advances in computer tec hnology, a new type of operational tool is available for forecasting downsl ope windstorms: two- dimensional, nonlinear, mesoscale numerical models. Th is study investigates whether this type of model, initialized with upstream profiles taken from operational Eta Model forecasts, can produce accurate downslope windstorm forecasts. Numerical simulations for high- wind events that affected seven regions in the United States between January 1993 and April 1997 indicate this tool is able to produce lee- slope wind speeds that meet the local peak gust thres hold for a High Wind Warning for a majority of those cases where observed w inds met this threshold. These simulations were initialized with upstream s oundings taken from the 12- and 18- h Eta forecasts valid at the time of ea ch high- wind event. A comparison for one region between the number of even ts for which High Wind Watches were posted and the number of events for whi ch the two- dimensional model prediction met the peak gust threshold sugges ts this new tool would be a definite improvement over the current forecast technique. On the other hand, a preliminary test of the model's ability to differentiate between windstorm and nonwindstorm events suggests the false warning rate for this tool may be high. Further testing of this tool is ong oing and will continue through the winter months of 2000/ 01.