Authors: Meraj Sohrabi, Hamed Moftakhari, Hamid Moradkhani – University of Alabama
Title: Efficient Tropical Cyclone Scenario Selection Based on Cumulative Likelihood of Impacts
Abstract: The incidence of climate-related disasters is on the rise, which makes it imperative to intensify anticipatory action. Tropical cyclones (TC) bring extreme precipitation and storm surge to coastal areas. This poses a compound flood risk to coastal communities due to the coincidence/concurrence of multiple flood drivers. In the absence of sufficient spatiotemporal coverage of historic TC data, appropriate characterization of compound flood risk mainly relies on running a large number of synthetic scenarios with the hope that it covers the wide range of potential threats posed to a coastal community. Such an approach requires huge computational resources that make it infeasible in many cases. Here we propose a dependence-informed sampling scheme that helps reduce the dimensionality of the problem and systematically select a handful of scenarios with the largest Cumulative Likelihood of Impact (CLI). The CLI is a compound flood index that ranks the candidate storms with the potential to cause compound flooding based on the regional dependencies between forcing (wind and rainfall) and coastal flooding drivers (storm surge and runoff). The analysis of historic TC records near the coast of Texas, USA shows the usefulness of CLI in improving the efficiency of hazard forecasting and providing reliable information at a lower cost. The proposed CLI successfully ranks candidate hurricane scenarios based on their potential impact and filters out the less relevant scenarios without the need for detailed hydrodynamic simulation.