Development of an Open-source Hurricane Risk Model for Bermuda

Local insurers have advised that proxy exposure databases are often used for Bermuda, based on the distribution of other jurisdictions in vendor models’ databases. We use recent Annual Rental Value data from the Bermuda Government to develop a representative dataset of property value for each of the 36 electoral constituencies in Bermuda. Elevation data for each representative property is captured using the Bermuda Government’s Digital Terrain Model in ArcGIS. Recent published work resulting from analysis of damage from Hurricane Fabian (Miller et al 2013) has allowed the estimation of damage functions that incorporate the effects of topographic upslope enhancement, (not captured by HURDAT data).
Given Bermuda’s location and small size (57 km2) it constitutes a ‘small target’ for TCs in the Atlantic. Therefore, despite being in a region through which hurricanes frequently move (Figure 1), the record of hurricane direct hits is sparsely populated with large-impact events. It is evident sample size is compromising the assessment of hurricane risk – the relatively short historical (instrumented) record of hurricanes hitting Bermuda is likely to not reflect the true damage probability distribution.