Automated Guidance for Predicting Quantitative Precipitation

Abstract

An operational, automated guidance system for producing both probability of precipitation amount (PoPA) and categorical forecasts of precipitation amount for 233 cities in the conterminous United States is described. Regionalized equations giving the probabilities of≥0.25, ≥0.50, ≥1.0 and ≥2.0 inches for 6 and 24 h periods for both warm and cool seasons were developed with use of the Model Output Statistics technique. Categorical forecasts are produced by transforming the probability forecasts in such a way as to maximize the threat score. Two sets of guidance products were developed. The first uses predictors from the National Meteorological Center's (NMC) limited Area Fine Mesh (LFM) model, while the second uses predictors from NMC's Primitive Equation (PE) model and the Techniques Development Laboratory's Trajectory model. To test the PoPA system, operational forecasts at 215 cities were verified over a 10-month period. The PoPA categorical forecasts were compared by season to those produced subjectively at NMC and by the LFM and PE models. Threat scores and categorical biases were computed. The results indicate that, overall, the PoPA categorical forecasts were better than those of the PE and LFM models and were almost as good as those prepared subjectively at NMC.