Impact of stochastic cumulus on the NOGAPS ET ensemble forecasting system. Part II: Tropical cyclone track forecast performance

The new NOGAPS Ensemble Transform (ET) ensemble forecasting system was run for July-October 2005 with and without a stochastic cumulus scheme, and the tropical cyclone (TC) track forecasts of the ensemble members and their consensus were evaluated. The TC track forecast error for the NOGAPS ET ensemble mean was reduced by the use of the stochastic cumulus scheme for all basins. The TC track forecast errors of the NOGAPS ensemble members and their mean (with stochastic physics) were compared with those from other models, the NCEP GFS ensemble members and their mean, and a multi-model consensus (CONU) for the Atlantic and eastern North Pacific basins. The NOGAPS ensemble mean TC track forecast error was significantly less than that for the GFS ensemble mean for forecast lengths greater than 48 h. For the Atlantic basin, the NOGAPS ensemble mean TC track forecast error was comparable to that for CONU for forecast lengths greater than 72 h. The relationship between ensemble spread and ensemble mean forecast error was examined for the NOGAPS, GFS, and multi-model ensembles and found to be much weaker for single-model ensemble systems than that for simple multi-model ensemble systems.