Simulation of Bay of Bengal cyclone tracks: a multimodel superensemble approach

In recent years, significant progress has been made in the prediction of track of tropical cyclones by using high resolution forecast model and high resolution temporal and spatial data. Further improvements been made by the use of ensemble prediction system (EPS). The perturbation of initial conditions in such cases gives rise to a number of tracks as followed by different members. Ensemble mean of all members in some cases was found to be better as compared to the individual members. Mylne et al (1999) suggested that ensemble systems using perturbations of initial conditions display insufficient spread in the ensemble, characterised by an excessive portion of verifying observations falling outside the range of the spread. Another approach to the EPS was suggested by Krishnamurti et al (1999) which uses a multi model superensemble appraoch and has been tested for the Atlantic hurricanes of 1998. It is showed that the cyclone tracks as obtained from the superensemble forecasts are superior to the forecasts made by the member models.

In the present study, we made an attempt to predict the track of tropical cyclones which originate in the Bay of Bengal using the Multi model superensemble approach. We have used RAMS, MM5, FSU (Global and Regional) models and their various versions, generated by plugging-in different available physical parameterization schemes in the models. The basic concept behind the multi model superensemble approach is to correct the prediction errors of different models using the past behaviour of these models. The superensemble prediction is then made from a combination of multimodel predictions, which has the least forecast error. Krishnamurti et al (1999) showed that the coefficients of correction show significant resilience which can be used to correct the predictions. Preliminary results suggest that the superensemble approach gives a better track of the cyclones as compared to the member models.