The Rank Input Method and Probability Variation Guides.
Abstract
The rank input method allows a forecaster's subjective estimate to be quantified into a probability forecast. The forecaster's estimate can be a rank input, a probability of a single category, or a categorical forecast. With the rank input the forecaster ranks the synoptic situation--very bad to very good-- in relation to the element to be forecast, e.g., surface visibility. The transnormalized regression probability model is then used to calculate the probability of the specific event. Probability of a single category can be converted to probabilities for one or more different categories. A categorical forecast can be converted to probability forecasts. A validation during REFORGER 78 concluded that the method shows promise and that forecasters were able to produce a large number of probability forecasts with a few simple rankings of the synoptic situation. Probability variation guides are tables giving forecast probability values for various inputs. Plotted on a simple graph, all values for a given skill and climatology fall along a single curve in probability space. These curves make certain decision analysis theorems much simpler in form.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1980
- Accession Number
- ADA093196
Entities
People
- Albert R. Boehm
Organizations
- Air Force Technical Applications Center