Dynamic storm surge prediction with one-way coupled mesoscale modeling system

The natural disasters like tropical cyclones (TCs) cause variety of damages, as it is associated with strong wind, heavy rainfall and storm surge. Among these, storm surge is one of the most hazardous elements associated with land falling TCs and became a serious concern in the coastal regions of India, Bangladesh, Myanmar, and Sri Lanka. There has been extensive loss of life and property in this region. The main factors contributing to disastrous storm surges along the east coast of India, Bangladesh and Myanmar are; 1) shallow bathymetry, 2) convergence of the bay, 3) low-lying delta region entrenched with a large number of river systems with innumerable number of inlets in head Bay region and 4) concentrated coastal population etc. About 90 % of the damage due to land falling TCs is mainly attributed to the coastal inundation with saline sea water. Thus, accurate and timely forecast and warning of storm surges are critical to mitigate threat to life and property and challenging task in both research and operational platforms. Thus, to predict the temporal and spatial variations of sea level in response to TCs over the Bay of Bengal (BoB) at least 3 days before landfall of the system, the dynamic one-way coupling of high resolution state-of-the-art mesoscale models like Weather Research Forecasting (WRF) and Hurricane WRF (HWRF) with two-dimensional shallow water storm surge model is initiated. The storm surge model used in this study needs the wind-stress forcing as the basic input parameter/forcing to generate surges associated with TCs. For this purpose, the wind-stress is computed by using the dynamical storm model of Jelesnianski and Taylor (1973). The wind-stress is calculated using maximum pressure drop, dp (hPa) value and the estimated radius of maximum wind, Rm (m). The observations/information of dp (hPa) and Rm (m) are upgraded in every 3 hr. It has been hypothesized that, the mesoscale forcing could have recognizable contribution towards the storm surge amplitude. Hence, the WRF/HWRF model simulated surface wind fields at 10 meter are used. Thus, an offline one-way coupling between WRF/HWRF and storm surge model is established. In this regard, the surface wind as simulated with WRF/HWRF is converted to wind-stress employing the relationship proposed by Garrett (1977). In order to provide the WRF/HWRF simulated wind at each grid point of storm surge model, the WRF/HWRF model is configured with same horizontal resolution and domain coverage. The WRF/HWRF model output is saved in every three hours interval and is used by storm surge model. In the present study, the severe TCs developed over BoB during the period 2007-2013 including recent very severe cyclonic storm Phailin are considered. The evolution of storm surges near the coast is very sensitive to the coastal geometry and offshore bathymetry at the location of the landfall of the cyclone. Realizing these factors, the time evolution of storm surges along the coast is estimated. In order to have a comparative analysis on the surge envelope along the coast, three important locations has been chosen for surge estimation. For this purpose, the landfall (LF) location, 25 km left of landfall (LL) and 25 km right of landfall (LR) are considered, being 25 km is the average radius of maximum wind in this region . The storm surge model developed over the BoB is able to simulate surge heights, which are in good agreement with the estimated values as provided by different meteorological/operational agencies of India, Bangladesh and Myanmar. The WRF/HWRF model also exhibits its capability towards the generation of appropriate inputs to the storm surge model. The surge prediction depends upon the improved track and intensity prediction. The data assimilation experiments with WRF/HWRF improve the track and intensity prediction, and enhance associated surge prediction for all the cases. The real-time surge prediction using high-resolution mesoscale model product is possible well in advance i.e. 48 hrs of actual landfall of the storm. However, in case of Phailin, the real time surge prediction before 75 hrs in advance helped in evacuation process. It is also noticed that, the maximum surge prediction does not depends upon the duration of the storm surge model integration. The storm surge model is successfully integrated with the mesoscale wind-stress forcing as generated with high resolution WRF/HWRF model. The spatial distribution as well as the peak surge envelope is well captured by the model. The result emphasizes the suitability of fine resolution location specific model in real-time forecast of storm surges. The peak surge envelope along the coast is well simulated by the model for all the cases. The time evolutions of surge development in the vicinity of the landfall location of each cyclone are also calculated. The maximum peak surge is observed at the right of the each cyclone with reasonable accuracy.