Climate Forecast Verifications, United States Mainland, 1974–83

Abstract

A nine year record (December 1973 to February 1983) of seasonal temperature and precipitation anomaly forecasts is examined. The four independent forecasters were J. Namias, the National Weather Service (D. Gilman and colleagues), the Analoger (T. Barnett and R. Preisendorfer), and A. Douglas. The skills of these human forecasters are compared to three benchmark forecasters, climatology, persistence and random chance, and to several simple objective forecasters. It was found that the human forecasters are all of comparable skill, and that they are generally better than climatology or random chance. However, the humans are often no better than persistence or some of the objective forecasters. In general, temperature was predicted better than precipitation. For both temperature and precipitation, winters were most-well predicted and falls were least-well predicted. Temperature was best predicted in the Southwest Desert, Pacific Coast and Northern Plains, and worst predicted on the Gulf Coast, Atlantic Coast and Southern Plains. Precipitation was best predicted in the Southwest Desert, Great Lakes, and Northern Great Basin and worst predicted along the Gulf, Atlantic and Pacific Coasts.