Verification of Vortex '94 Forecasts.
Abstract
Forecast verification involves the assessing the quality of a set of forecasts. There are many different methods of measuring the quality of a set of forecasts, and this study examines two; the measures-oriented approach and the distributions-oriented approach. It is argued that the distributions-oriented approach offers a more comprehensive look at the set of forecasts and observations and the relationships between them. This approach also highlights the strengths and weaknesses of the forecasting system so that improvements can be made in specific areas. This study evaluates the forecasts made during VORTEX '94. The forecasts included experimental probabilistic forecasts for the VORTEX forecast area as well as probability contour forecasts. These forecasts were evaluated by looking at the conditional distribution of the forecast given the event, p(f x), the conditional distribution of the event given the forecast, p(x f), as well as the marginal distributions of forecasts, p(f), and observations, p(x). It is only through looking at these distributions of forecasts and observations that the true quality of a set of forecasts can be evaluated.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1996
- Accession Number
- ADA311886
Entities
People
- Robert D. Duncomb Jr
Organizations
- Air Force Institute of Technology