Emergency Response Transport Forecasting Using Historical Wind Field Pattern Matching

Abstract

Historical pattern matching, or analog forecasting, is used to generate short-term mesoscale transport forecasts for emergency response at the Idaho National Engineering and Environmental Laboratory. A simple historical pattern-matching algorithm operating on a database from the spatially and temporally dense Eastern Idaho Mesonet is used to generate a wind field forecast, which then is input to an existing puff diffusion model. The forecasts are rated both by a team of meteorologists and by a computer scoring method. Over 60% of the forecasts are rated as acceptable. The forecasts also are compared with a persistence method, using both a subjective human evaluation and root-mean-square error calculations.