Track forecasting of tropical cyclones in the North Indian Ocean with 3DVAR and 4DVAR Data Assimilation Techniques

The North Indian Ocean particularly on the East coast of India is vulnerable to tropical cyclones that cause heavy wind, torrential rain and enormous damage to the life and property when they hit the coastal regions year after year. Most of the studies on the North Indian Ocean focus on the Indian summer monsoon. Few studies consider the North-East monsoon that occurs during the period October-December and brings in about 70 % of the annual rainfall to the coastal regions of Tamil Nadu, Andhra Pradesh and Orissa. Scarce are exhaustive studies on track forecasting of tropical cyclone, originating on the east coast and scarcer still are studies that incorporate state of the art data assimilation for these events. Hence, in this study the main focus is on the cyclonic activity in the North Indian ocean where in three cyclones Phyan in (2009) Laila in (2010) and Jal in (2010) are analyzed in detail with and without data assimilation techniques. For this purpose, the above mentioned cyclones have been simulated by the community meso-scale numerical weather prediction model-Weather Research and Forecasting (WRF) on a desktop mini super computer CRAY CX1. The weather prediction is very sensitive to initial conditions and the physics parameterization schemes used in the models. A 0.5 degree resolution Global Forecast System (GFS) data from the National Centers for Environmental Prediction (NCEP) has been used as initial and boundary conditions. The sensitivity of the WRF model to tropical cyclone Jal with available physics parameterization schemes conducted in-house led to the best set of schemes and this has been used for analyzing the three cyclones under consideration. The first part of this study focuses on the analysis of all the above mentioned cyclones and the simulated cyclone tracks are compared with Joint Typhoon Warning Center (JTWC) observation cyclone track data. The results clearly indicate that the error in the initial condition is 32.7 km for the case of cyclone Jal, 390.9 km for the case of cyclone Phyan and 247 km for the case of cyclone Laila. Figure.1 shows the results of the numerical simulations for the three cyclones, from which one can infer that the model gives better result in the case of cyclone Jal compared to the other two cases. This is because of better initial conditions in the GFS data for the simulation of the cyclone Jal. Nevertheless, good initial conditions of GFS data are not available for the other two cyclone cases. Hence, it is imperative to improve the initial conditions for simulations instead of relying on low resolution GFS data alone. The main objective of this work is to study the effect of improved initial conditions for cyclones under consideration by using the Advanced Research WRF (ARW) data assimilation package. The NCEP ADP Global Upper Air observations data will be used for assimilation. Track results are compared with and without assimilation. Furthermore, the structure and evolution of all cyclones are analyzed with assimilated data and results thereof are presented. This study also present a comparison of the results with 3DVAR and 4DVAR assimilation techniques for tropical cyclones in the North Indian Ocean region.