Combining Monte Carlo and Ensemble Probabilities in Tropical Cyclone Forecasts near Landfall

The Monte Carlo probability (MCP) model, which has been used for official tropical cyclone (TC) warnings to the public by the United States’ National Hurricane Center (NHC), can estimate the probability of wind speed in the vicinity of a TC during the forecast period. It has been successful in the operational environment for many years. However, due to its strong dependence on a given forecast track (e.g., forecast from the NCEP Global Forecast System), the MCP model may generate a poor probability map for TCs near landfall. In this study, we proposed and tested a modified MCP method for TC forecasts near landfall. We first adjusted the MCP model by adding limits to the direction angle and motion distance to deal with the substantial change in TC moving direction and the low wind speeds during landfall. Then, we combined ensemble probability maps generated from ECMWF, United Kingdom Met Office (UKMO), and NCEP ensemble forecasts, obtained from The International Grand Global Ensemble (TIGGE), into the MCP model to configure a modified MCP model. Wind speed probability maps for the 0–120-h forecast from both the original and modified MCP models are compared. It is found that the modified MCP model can provide a better wind speed probability map during landfall, especially at wind speeds of 20–64 kt near TC landfall. The results from this study prove the benefits of combining the MCP model with ensemble forecasting in potential applications for improved TC forecasts.