Further Evaluation of the Quasi-Lagrangian Model's Forecast Track Guidance

Abstract

A comparison is presented of the forecast track errors in the 1990 North Atlantic hurricane season cases from three operational models. The Quasi-Lagrangian Model (QLM) is a baroclinic model. The National Hurricane Center Model 1990 (NHC90) is an advanced statistical-dynamical model, and CLIPER (climatology and persistence) is a simple statistical model that uses only climatology and persistence. Mean errors over the season were smaller in the QLM than in the NHC90 or CLIPER. The forecasts were also divided into several 5°-10°latitude zones based on the initial storm latitude. The mean errors in the QLM were the smallest in many zones after the first 12 h. Mean forecast track errors in some storms are presented to show the variation in the performance of the models from one storm to another. Individual cases in these terms are also discussed to gain further insight into the relative accuracy of the models. The QLM generally performed better than the NHC90 and CLIPER for storms that moved northward or had unusual tracks. The analysis and forecasts of an operational global model are used to derive the QLM initial state and for prediction over QLM lateral boundary points, respectively. Both the global analysis and forecast programs were modified in 1991. It is shown that the performance of the QLM relative to the CLIPER and NHC90 was not altered despite the changes in the global model.