Estimating the Joint Probability of a Weather Event at Two Locations

Abstract

A total of 511 056 hourly observations of precipitation, sky cover, ceiling, visibility, wind speed and temperature, taken over a 13-year period at nine locations, were processed to obtain unconditional and two-location joint relative frequencies of 18 weather events, in winter and in summer. A model was developed to estimate joint weather event probabilities from unconditional probabilities and the spatial correlation coefficient. It was developed on data from the 36 pairs of locations ranging from 9 to 431 miles apart. The probability estimates given by the model were compared with the observed relative frequencies and two-thirds of the root-mean-square errors were less than 1%. The largest root-mean-square-error was 2.9%.