Accuracy of weather forecasts

Abstract

AFTER nine years of publication there has been considerable criticism of the usefulness and accuracy of long range weather forecasts. Gordon1 has suggested that the forecasts have not yet attained a practical or economical degree of usefulness, and that advance computer technology and statistical power will not improve their accuracy unless a breakthrough in ideas is achieved. Existing methods are broadly based on an analogue or matching technique in which overall weather circulations of a particular month for many decades in the past are compared with those of the month preceding that for which the forecast is to be made.