Assessing Tropical Cyclone Tracks in the Bay of Bengal

Tropical cyclone (TC)-induced storm surges have been wreaking havoc on Bangladesh's coastline practically every two years since the country's independence, resulting in numerous deaths and significant economic loss. Analyses suggest it will only get worse in the future. Previously numerous modeling attempts have been made to predict the water level at the coastal regions during cyclone events forced by weather models simulating relatively coarse resolution atmospheric datasets ranging from 25 to 50 km. Recently, the Hadley Center of the UK Meteorological Office has provided high-resolution forcing data (4.5 km and 1.5 km) from the Met Office Unified Model (MOUM) [1] that can be used to simulate coastal models efficiently to improve storm surge prediction and estimate associated risks and vulnerabilities. MOUM provided nine ensembles of high-resolution wind and pressure data for 12 historic TCs in the Bay of Bengal (BoB) shown in Fig. 1. These were the most devastating of all causing unprecedented casualties and damage. By correctly simulating these storms with high-resolution atmospheric forcing, a robust coastal model can be generated that could not only provide reasonable predictions on where a Tropical Cyclone might land on the Bangladesh coastline and its corresponding surge height but also alleviate the timely and accurate dissemination of the cyclone information. This study provides a simplified way to identify which of these 9 ensembles can be efficient for developing a coastal model by analyzing the cyclone track, the cyclone landfall location, and the time for a practical storm surge forecast in the Bay of Bengal. Each ensemble is 72 hours long with the first one being initiated 60 hours from the cyclone landfall while each of the next ensembles was initiated three hours lagged from its previous one. This was done to check which lagged forecast could simulate the TC better. An intercomparison of cyclone tracks is done among the ensembles, with the best tracks from the Joint Typhoon Warning Center (JTWC) and the Indian Meteorological Department (IMD) to identify which of these datasets can better represent the TC. This operation has been done by extracting the TC path from each of the ensembles and determining the distance from the JTWC track by using RStudio. Cyclone landfall time has been analyzed for all the generated tracks to identify the most representative atmospheric data for TC. Initial results indicate that ensembles 6-9 are generally better at capturing the TC path and predicting the TC landfall but this is not consistent across all the storms. The study also shows a slight bias in track and landfall location between the best tracks of IMD and JTWC for the TCs landfalling in Bangladesh.