Extended Abstract (856K)7D.5
Beta test of a prototype dynamical model track prediction evaluation system model for the Atlantic
Mark A. Boothe, NPS, Monterey, CA; and R. L. Elsberry and L. L. Salvador
The U. S. Weather Research Program Hurricane Landfall has sponsored the Joint Hurricane Testbed program to facilitate a transition of research toward operations at the Tropical Prediction Center/National Hurricane Center (NHC). This paper describes the first-year effort and future plans for producing a dynamical model expert system module for evaluating tropical cyclone track predictions in the Atlantic. This project is to adapt a similar expert system for the western North Pacific that has been used successfully at the Joint Typhoon Warning Center, Hawaii, for use in the Atlantic. The five dynamical model tracks and predicted fields utilized in the Atlantic system were: NCEP Aviation and Geophysical Fluid Dynamics Lab (GFDL) models, U. S. Navy Operational Global Atmospheric Prediction System (NOGAPS) and Navy version of the GFDL model (called GFDN), and the UK Met office global model. Only the tracks were available for the barotropic model LBAR. After conversion by the Computer Science Corporation (Dan Martinez and Jim Peak), a beta test was carried out at the Naval Postgraduate School (NPS) for a two-month period. Simulated real-time model evaluations were made for 00 UTC and 12 UTC track forecasts from mid-August to early October. Model fields received at NPS were processed and transferred to NHC where the Navy liaison at NHC (LCDR Laura Salvador) independently exercised the prototype expert system. Performance of both analysts will be presented, especially for difficult cases in which the dynamical model consensus spread was large at 72 h. Performance of the dynamical model track error characteristics based on several seasons will be presented. Based on these error characteristics, and the beta test, several changes are proposed to improve the performance of the prototype expert system, e. g., modification of consensus spread thresholds and the assignment of objective error mechanism options, which had been based upon performance in the western North Pacific. Plans for the display of other dynamical models and a more streamlined data distribution to NHC for the 2002 season will be proposed.
Session 7D, Tropical Cyclone Prediction V (Parallel with Session 7B and 7C)
Tuesday, 30 April 2002, 4:00 PM-5:30 PM
Previous paper Next paper