The need for qualification of weather forecasts has long been recognized. The diverse uses of predictions and variations in the forecaster's ability to prognosticate weather events correctly have suggested to previous writers that expressions of confidence or other means of weighting can serve the public to advantage. A particular method of expressing and scoring 5-day temperature forecasts in terms of probability statements has been tried experimentally by the U. S. Weather Bureau's Extended Forecast Section and found to provide valuable information in addition to that normally furnished. Furthermore, it is demonstrated that the participants in the experiment possess definite skill in selecting probability statements, and that relative verification results are comparable to those obtained by scoring the conventional forecasts.




Present affiliation: Operations Evaluation Group, Navy Department, Washington, D. C.