THE USE OF DECISION THEORY IN METEOROLOGY

Abstract

The concepts of decision theory are discussed, especially in the light of their application to meteorology. The use of the principles of decision-making under risk requires certain probability information to be available. The issuance of forecasts in probability terms has a firm basis in theory and has been shown to work well in practice. The best verification statistic of these forecasts is their usefulness to the user and this can be measured and compared with some standard if the utility matrix is known. A multi-dimensional contingency table technique is used to estimate the conditional probability distribution of the 5-hr. projection of ceiling height at Washington National Airport. Three predictors are screened from 164 possible predictors according to the utility criterion. Developmental and test data results are presented.