Climate Forecast Verification via Multinomial Stochasters,
Abstract
The problem of attaching some quantitative measure of skill to forecasts of temperature, precipitation and other physical fields over extensive regions of the atmosphere and hydrosphere is examined. It is suggested that to each forecaster we may assign a competitive stochaster, a device or person that performs the same forecast over the same regions of space and time as the forecaster, but using a specially designed random procedure. This notion is illustrated for the case of a multinomial stochaster, by means of numerical studies of actual temperature and precipitation forecasts over the U.S. mainland for various seasons over the past three years. Specially designed tables and charts show how quantitative judgments of forecaster skills can be made in a variety of ways. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1977
- Accession Number
- ADA056920
Entities
People
- Rudolph W. Preisendorfer
Organizations
- Scripps Institution of Oceanography