Forecasting Tropical Cyclone Tracks using MM5

The Fifth-generation PSU/NCAR Mesoscale Model (MM5) was used to predict tropical cyclone tracks that developed near the Philippines in 2003. The model’s sensitivity to the parameterization schemes for the cumulus, planetary boundary layer and explicit moisture was investigated to determine the optimal model configuration for predicting the track of a tropical cyclone. The 6-hour forecast skill of the model in locating the eye of the storm (i.e. location of minimum sea level pressure) indicated a greater sensitivity to the cumulus parameterization scheme (CPS) than to the planetary boundary layer or explicit moisture parameterization schemes. Among the three CPSs used in this study, the Betts-Miller CPS consistently resulted in the most accurate and smoothest tracks upon validation with observations. Cases under Betts-Miller CPS also gave the most well defined eye for all cyclone events. CLIPER (CLImatology/PERsistence technique) outperforms all schemes before the first 24 hours, but some MM5 schemes emerge as more accurate after 2 days. Results suggest that the optimal configuration of MM5 for forecasting the tracks of tropical cyclones over the Philippine domain uses the Betts-Miller CPS, Medium Range Forecast PBL and Schultz moisture parameterization.

[Presented at the Samahang Pisika ng Pilipinas (SPP) Congress, 2004.]