Maximum/Minimum Temperature Forecasts Based on Model Output Statistics

Abstract

A new automated system of forecasting maximum and minimum surface temperatures in the conterminous United States has been developed. This system is based on the Model Output Statistics (MOS) approach which effectively combines numerical and statistical methods. A series of screening experiments is described, derivation and interpretation of the MOS equations are discussed, and sample statistics are presented. Verification results, comparing the MOS and perfect prog systems, are summarized. Operational aspects such as facsimile and teletype transmissions, are discussed, and recent changes in procedure are explained. The MOS system replaced the perfect prog technique in the operations of the National Weather Service in August 1973. This has resulted in increased accuracy of the autmated forecasts; at the same time we increased the station coverage from 131 to 228 cities in the conterminous United States.