A new statistical model for tropical cyclone track forecasting has been developed by Météo-France in La Reunion Island. This model named MOCCANA (Modèle Climatologique de Cyclones par ANAlogues) is a climatological and persistance model with an analog selection procedure. Another feature of this model is its Principal Component Analysis to calculate the regression coefficients. All of these particular procedures lead to a new approach among the common encountered statistical tropical cyclone track forecasting models. In short, the noticeable difference between MOCCANA and other statistical models is the time adaptative feature in its multivariate analysis. Such a procedure insure to have the "optimum" selection of predictors/predictands set for one forecasting attempt. A companion paper (Bessafi et al, 1999) focused on the presentation of this new methodology. This previous result has shown an equivalent performance to its counterpart statistical CLIPER model (Neumann, J.C, E.A. Randrianarison, 1976) in the absolute forecasted error distance for the Southwest Indian ocean basin. This substantiates the validity of the method and its competitivity. The application of this model to the remaining basins of the globe is presented here. This study also show the Forecasting Difficulty Level of MOCCANA for the respectively basins: North Atlantic, North-East Pacific, North-West Pacific, South-West Pacific, North Indian Ocean, South-Est Indian Ocean and the whole Southern Hemisphere. In comparison with CLIPER model, the performance of MOCCANA showed that there is not a particular hierarchy in the FDL indice for the short time forecasting. A comparison with CLIPER for the North Atlantic and North-East pacific basin show significant increase of the forecasting TC location skill by MOCCANA for all selected forecasting time.
![](images/flag.png)
![](images/blue_ribbon.png)