Development of a Simple, Open-Source Hurricane Wind Risk Model for Bermuda with a Sensitivity Test on Decadal Variability

Abstract

A hurricane-catastrophe model was developed for assessing risk associated with hurricane winds for Bermuda by combining observational knowledge with property value and exposure information. The sensitivity of hurricane wind risk to decadal variability of events was tested. The historical record of hurricanes passing within 185 km of Bermuda was created using IBTrACS. A representative exposure dataset of property values was developed by obtaining recent governmental Annual Rental Value data, while Miller et al. (Weather Forecast 28:159–174, 2013) provided a vulnerability relationship between increasing winds and damage. With a probabilistic approach, new events for 10,000 years were simulated for three different scenarios using (1) the complete record of annual TC counts; (2) two high-frequency periods and; (3) two low-frequency periods. Exceedance probability curves were constructed from event loss tables, focusing on aggregating annual losses from damaging events. Expected losses of low-frequency scenarios were less than losses of high-frequency scenarios or when the whole historical record was used. This framework suffers from uncertainties due to different assumptions and biases within IBTrACS. Small data sizes limit our ability to conduct a formal model validation and results should be interpreted in this context. In the future, sensitivity tests on the different components of the model will be performed.