Performance of TC track ensemble prediction system in NMC/Beijing

The TC track ensemble prediction system was developed based on the perturbation technique of the medium range ensemble prediction system and the BOGUS vortex scheme used in the TC track prediction system in NMC/Beijing. All the ensemble members have different background fields and the same BOGUS vortex in the same position-no vortex and position perturbation added yet. This system consists of 14 perturbed run and one control run, runs twice a day in typhoon season and provides 120h track forecasts.

The results from case study shows that the adding of the BOGUS vortex could improve the TC track ensemble prediction ability of the medium range ensemble prediction system now running in operation in NMC through the followings two aspects:

1) Removing the large initial position spread when TC could be described in the initial fields of the ensemble members

2) Improving the ability of TC track prediction when there are no vortexes in the initial fields of the ensemble members.

The TC track ensemble prediction system was put into real time run in TC season 2007. The statistic analyses show that the mean track errors of No.0704-No.0716 from all the ensemble members are less 5-10km than those from the control run. Compared with the mean track of all the ensemble members, the cluster mean is much better in some TC when there are obviously different tracks, for example, some members give the westward movement and some members give the northward movement. Also the strike probability that TC passes a position within 120km could provide more useful information than those from the ensemble mean track.

More research will be carried out in the near future: the TC vortex preturbation and improvement of the perturbation in tropical region. At the same time more useful products from the TC track ensemble prediction system will be developed.