摘要
Based on the Global Regional Assimilation and Prediction System-Tropical Cyclone Model(GRAPES-TCM),an ensemble forecast experiment was performed,in which Typhoon Wipha during the period immediately prior to landfall was selected for the study and the breeding of growing mode(BGM) method was used to perturb the initial conditions of the vortex field and the environment field.The results of the experiment indicate that each member had a different initial status in BGM processing and they show a reasonable spread among members along with the forecast phase.Changes in the large-scale field,thermodynamic structure,and spread among members took place when Wipha made landfall.The steering effect of the large-scale field and the interaction between the thermodynamics and the dynamics resulted in different tracks of the members.Meanwhile,the forecast uncertainty increased.In summary,the ensemble mean did not perform as well as the control forecast,but the cluster mean provided some useful information,and performed better than the control in some instances.The position error was 34 km for 24 h forecast,153 km for 48 h forecast,and 191 km for 66 h forecast.The strike probability chart qualitatively described the forecast uncertainty.
Based on the Global Regional Assimilation and Prediction System-Tropical Cyclone Model （GRAPES-TCM）, an ensemble forecast experiment was performed, in which Typhoon Wipha during the period immediately prior to landfall was selected for the study and the breeding of growing mode （BGM） method was used to perturb the initial conditions of the vortex field and the environment field. The results of the experiment indicate that each member had a different initial status in BGM processing and they show a reasonable spread among members along with the forecast phase. Changes in the large-scale field, thermodynamic structure, and spread among members took place when Wipha made landfall. The steering effect of the large-scale field and the interaction between the thermodynamics and the dynamics resulted in different tracks of the members. Meanwhile, the forecast uncertainty increased. In summary, the ensemble mean did not perform as well as the control forecast, but the cluster mean provided some useful information, and performed better than the control in some instances. The position error was 34 km for 24 h forecast, 153 km for 48 h forecast, and 191 km for 66 h forecast. The strike probability chart qualitatively described the forecast uncertainty.
基金
National Basic Research Program of China (2009CB421500)
Shanghai Science and Technology Program (10231203700)
National Natural Science Foundation of China (40921160381)
