Evaluation of dynamical track predictions for tropical cyclones in the Atlantic during 1997-98

Abstract

Carr and Elsberry (1999; NPS Tech Report) have described eight conceptual models that explain most cases of large (> 300 n mi at 72 h) western North Pacific tropical cyclone (TC) track errors by the Navy Operational Global Atmospheric Prediction System (NOGAPS) and the Geophysical Fluid Dynamics Lab (Navy version - GFDN) models. This study is for TCs in the Atlantic basin and includes the European Centre for Medium-range Weather Forecasting (ECMWF) and the United Kingdom Meteorological Office global models, whereas the GFDL model is eliminated. A detailed examination is made of large (> 250 n mi at 72 h) errors made by the three dynamical models for two seasons of Atlantic TC tracks (1997-98). The percentages of > 250 n mi 72-h errors for the NOGAPS, UKMO, and ECMWF models were 23%, 26%, and 19%, respectively. The same error mechanisms found to apply in other basins also affect the dynamical models in the Atlantic. The NOGAPS and UKMO models have a tendency to over-represent TCs and other circulations, which leads to a cyclonic rotation, or even merger, via the Excessive Direct Cyclone Interaction (E-DCI) process, just as was found in the western North Pacific. The primary ECMWF error source was Excessive Midlatitude CycloGenesis (MCG).