Statistical prediction of monthly mean temperature anomalies in the United States during winter months

WILLIAM H. KLEIN, RUNHUA YANG

Abstract

Anomalies of mean monthly air temperature at 50 surface stations in the contiguous United States during the winter months of January, February and March from 1951 to 1980 are statistically screened as functions of earlier, centered or future time means of different length. Potential predictions include fields of 700 mb height, air and sea surface temperature, snow cover and ENSO indices. Future height anomalies are damped in accordance with the accuracy of daily numerical prognoses out to 10 days produced operationally at the National Meteorological Center. All statistical significance is evaluated by means of Monte Carlo simulations. The principal result is that judicious use of 1-10 day numerical height predictions offers the greatest promise for immediate improvement of monthly mean temperature forecasts.

Full Text:

PDF