From tropical cyclone hazard layers to vulnerability and loss estimation: a comprehensive framework

Tropical cyclones generate hazards such as extreme winds coastal flooding from surge and rainfall driven inundation that drive severe impacts on Islands territories. Losses quantification is a key factor for the implementation of risk-reduction and adaptation strategies. In this context, the objective of this work is to evaluate a hazard-to-damage workflow to estimate losses from hazards generated by tropical cyclones using different hazard and damage simulation approaches.   

This work presents a modular workflow allowing to estimate multi-hazard damages originated by tropical cyclones focused on French island territories (Reunion, French Antilles and Mayotte). Hazard simulations generate wind speed, coastal water-level and inundation depths due to rainfall using separate components. These hazard layers are harmonised for the damage estimation using a reinsurance data base of historical losses in French territories exposed to cyclones.  

Wind speed simulation is based on three complementary approaches to reflect different data availability and use cases.  

• Dynamical way: near-surface wind from WRF (Weather Research and Forecasting) simulations, providing spatially continuous fields that can support event reconstruction and forecast oriented application  

• Observation-driven option relies on Météo-France wind stations and applies terrain-related adjustments base on land-surface roughness and topography to represent local exposure heterogeneity.  

• A parametric option based on a Holland-type wind field to generate time evolving wind footprints. These wind options deliver consistent outputs (e.g. wind speed, direction ...) on a common grid for intercomparison.  

Flooding is generated by interfacing two flooding-related components: a coastal submersion chain providing water levels from marigrams combined with inland propagation, and an inland flooding component forced by river discharge and precipitation observations.   

All generated hazard layers are then compared to the available information on risk and infrastructure in the studies territory and loss estimation can be done with two main options.