Mesoscale Modelling for Tropical Cyclone Forecasting over the North Indian Ocean - Monitoring and Prediction of Tropical Cyclones in the Indian Ocean and Climate Change - page 269

Mesoscale Modelling for Tropical
Cyclone Forecasting over the
North Indian Ocean
U.C. Mohanty*, Krishna K. Osuri and S. Pattanayak
Indian Institute of Technology Delhi, Hauz Khas
New Delhi - 110016, India
*e-mail: [e-mail]
1. Introduction
The coastal regions of Bay of Bengal (BoB) and Arabian Sea (AS) experience
severe damage due to landfalling tropical cyclones (TCs). The synoptic and
statistical methods have limitations in predicting the track and intensity beyond
24 hours (Mohanty and Gupta, 1997). However, the numerical forecast using
dynamical models can provide better forecast guidance for genesis, intensity
and movement of TCs up to 72 hours (Rao and Bhaskar Rao, 2003; Mandal et
al., 2004; Osuri et al., 2012a) and helps in the disaster mitigation planning.
Hence, it is necessary to evaluate the comprehensive performance of such
dynamical models in track and intensity forecasts of TCs. Davis et al. (2008)
and Osuri et al. (2012a) showed that real-time TC forecast of ARW (Advanced
Research Weather Research and Forecasting) model is generally competitive
with, and occasionally superior to, other operational forecasts for track and
intensity of landfalling TCs over Atlantic and BoB respectively. Wang et al.
(2006) demonstrated that error growth in ARW model forecasts is noticeably
slow as the forecast length increases.
Further, improved prediction of track and intensity of TCs requires accurate
representation of the initial vortex position and structure. The sparsity of
observations over the tropical oceans causes large initial position and intensity
error. The initial analysis can be improved with the assimilation of conventional
and non-conventional observations. Several past studies have demonstrated
that the assimilation of near-storm observations through 3-dimensional
variational data assimilation (3DVAR) system can substantially improve cyclone
predictions (Zhang et al., 2007; Pu et al., 2008; Singh et al., 2008; Osuri et al.,