Track prediction of Indian Ocean cyclones using Lagrangian advection model

A new model has been developed for track prediction of Indian Ocean cyclones. The model utilizes environmental steering flow using the forecasts from a high-resolution global model and the effect due to earth’s rotation (the beta-effect) to determine the future movement of cyclone. A new approach based on vertical profile of potential vorticity is used to determine weights for different vertical levels for computation of mean steering current. Despite the fact that the model is based on the dynamical framework, the operational cost and time for running the model is only a fraction of what is needed by a normal numerical weather prediction model. This new approach will enhance flexibility in defining the initial position of the cyclone in the model, and also, it is possible to create a large ensemble of predicted tracks to assess the impact of the uncertainty of initial cyclone position on the predicted tracks. The performance of the model for ten cyclones, viz. GONU (02–08 Jun, 2007), SIDR (11–16 November, 2007), NARGIS (27 Apr–04 May, 2008), RASHMI (25–27 October, 2008) KHAI-MUK (14–16 November, 2008), NISHA (25–27 November, 2008), SEVEN (04–08 December, 2008), BIJLI (14–18 April, 2009), AILA (23–26 May, 2009), and PHYAN (09–11 November, 2009), have been tested in the present study. The forecast errors of the present model have been computed with respect to the Joint Typhoon Warning Center best track analysis positions. The forecast skill improvement (mean of ten cyclones) of the model with respect to the Climatology and Persistence (CLIPER) statistical model varies from 7 to 67 % between 12 and 72 h. Copyright Springer Science+Business Media B.V. 2012