A comparison of the NOGAPS and GFDN dynamical track prediction models during the 1997 western North Pacific typhoon season

Abstract

The performance of both the U.S. Navy (NOGAPS) and regional (GFDN) dynamical track prediction models during the 1997 western North Pacific typhoon season is documented. In the context of the Systematic Approach of Carr and Elsberry, a knowledge base of six conceptual models (summary in Table 8.1) is proposed that associates recurring tropical cyclone (TC) forecast track errors with various types of TC and environmental structures. Twenty-one storms of the 27 analyzed have periods in which at least one significant track error source was identified (summary in Table 8.3). More situations (23) were identified in the NOGAPS forecasts than in the GFDN forecasts (14). Individual case studies are presented to illustrate recurring scenarios with poor performance in either the NOGAPS model, GFDN model, or both. Use of these conceptual models and their supporting case studies may allow the JTWC forecaster to better understand how the NOGAPS model and GFDN model may perform in specified synoptic environments. It is hoped that the JTWC forecaster can use the information in this study to provide more accurate TC tracks by rejecting inappropriate model guidance during future typhoon seasons in the western North Pacific. In addition, this study may provide feedback to dynamical model producer as to situations in which large track errors have occurred, in hopes that the model might be improved in the future