Verification of HIGH-RESOLUTION Precipitation Forecasts For The Atlanta Olympic Games

Abstract

The Local Analysis and Prediction System (LAPS) was an integral part of the Olympic Weather Support System (OWSS) designed by the National Weather Service (NWS) to supplement the forecasting operations in the Peachtree City, GA, NWS forecasting office during the 1996 Atlanta Olympic Games. This paper presents an objective hourly verification of some of the precipitation forecasts produced by the numerical modeling component of LAPS during the summer of 1996 for the southeast United States. The scores indicated underforecasting at all thresholds when the model was initialized at 0600 UTC. A later initialization improved the bias at lower thresholds, but caused overforecasting at higher thresholds. A comparison with the precipitation forecasts by the 29-km Eta model showed that the high-resolution LAPS system was able to produce better precipitation forecasts, particularly when initialized with a high-resolution local analysis. This paper also presents a discussion of the impact of different algorithms used to collocate observed and forecasted precipitation data. Higher bias scores (BSs) were obtained when the score was computed at the model grid points instead of at the station locations. For BSs computed at the station locations, higher scores were obtained when a larger number of grid points surrounding a station was used to compute forecasted precipitation at the station.