The Application of Signal Detection Theory to Weather Forecasting Behavior

Abstract

A variety of measures are used to judge the skill and accuracy with which forecasters predict the weather and to verify forecasts: Such measures can confound accuracy with decision strategy and sometimes give conflicting indications of performance. Signal detection theory (SDT) provides a theoretical framework for describing forecasting behavior and minimizing these problems. We illustrate the utility of signal detection theory in this context, show how it can be used to understand the effects of time pressure created by frequent weather activity on forecasting judgments, and illustrate how to achieve a specific social policy.