Bay of Bengal cyclone extreme water level estimate uncertainty

Abstract

Accurate estimates of storm surge magnitude and frequency are essential to coastal flood risk studies. Much research has focused on tide–surge interaction and joint probability techniques to combine multiple cyclone characteristics. In the Bay of Bengal, extreme water levels are derived from numerical storm surge models based on an idealised cyclone event; however, uncertainty within such calculations for this region is poorly understood, especially when propagated through to the flood hazard. We use the IBTrACs data set to estimate natural variability in four key parameters used to describe an idealised cyclone and create a set of idealised but equally likely “1 in 50 year” recurrence interval cyclone events. Each idealised cyclone is then used to force a storm surge model to give predicted peak water levels along the northern Bay of Bengal coast. Finally, extreme water level uncertainty is propagated through an inundation model to predict flood extent and depth over inland coastal floodplains. The descriptive parameters of 18 cyclone events (between 1990 and 2008) appear to show no statistically significant variation (at the 5 % level) due to landfall location, which allows us to pool characteristics for the entire Bay of Bengal. We find that the natural variability of cyclone parameters translates into large uncertainty both for storm surge height (of the order of metres) and for coastal inundation (hundreds of km2). Using the variability estimates for a 1-in-50-year cyclone event making landfall at the 2007 Sidr location, cyclone central pressure drop uncertainty had the greatest effect upon simulated storm surge magnitude. However, uncertainty within cyclone track characteristics (track speed, landfall and genesis location) has greater influence on subsequent inundation extent. Storm surge hazard uncertainty due to cyclone parameter variability was found to be comparable to the inundation difference simulated when the peak surge coincided with either a mean spring high or low water. Our research indicates the importance of improving extreme water level estimates along the Bay of Bengal coastline for robust flood hazard management decisions in the Bay of Bengal.