An Application of Discriminant Analysis to a Tropical Cyclone Intensification Forecasting Algorithm
Abstract
A stepwise screening discriminant analysis is used to develop an algorithm for forecasting tropical cyclone intensification. Some potential predictors used in the analysis are derived from FNOC (Fleet Numerical Oceanography Center) zonal wind analyses. Other potential predictors considered are simple climatological parameters and persistence of intensification. The potential predictors are screened by a stepwise discriminant analysis program to determine which significantly discriminate between three (intensifying, filling, and neither) intensification categories. The best discriminators are then used as predictors in an intensification forecasting algorithm. The algorithm is developed and tested on Western Pacific tropical cyclone data from the years 1974-82. In addition to the independent test, stability of the algorithm is investigated and a test is conducted to compare forecasts produced by the algorithm with forecasts made by JTWC (Joint Typhoon Warning Center) and with forecasts of persistence of intensification. The algorithm shows skill relative to persistence and compares favorably with the official forecasts, although most of the differences between the scores are not significant. For this data set, JTWC did not forecast tropical cyclone intensification significantly better than the derived algorithm or persistence.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1985
- Accession Number
- ADA159399
Entities
People
- B. J. Cook