The role of spatial dependence in global-scale coastal flood risk assessment

Coastal flooding is among the world’s deadliest and costliest natural hazards. The impacts caused by coastal flooding can be particularly high when an event affects a large spatial area, as witnessed during Hurricane Katrina and Cyclone Xaver. Current large-scale flood risk studies assume that the probabilities of water levels during such events do not vary in space. This failure to capture flood spatial dependence can lead to large misestimates of the hazard and risk at large spatial scales, and therefore potentially misinform the risk management community. In this contribution, we assess the effects of spatial dependence on coastal flood risk estimation at the global scale. To this end, we compare the assessments using two spatial dependence scenarios: i) complete dependence and ii) modelled dependence of water level return periods. For the complete dependence scenario, we use the existing risk information calculated by the GLOFRIS global risk modelling framework. To estimate the spatially-dependent risks, we use an event-based multivariate statistical approach and consider 10,000-year extreme coastal flood events derived from the global synthetic dataset of spatially-dependent extreme sea levels. The associated spatially coherent return periods of each event are then combined with the GLOFRIS spatially-constant inundation layers to create the spatially-dependent inundation map. These hazard maps, overlaid with exposure layers and vulnerability information, are further used to assess the coastal flood impacts. The flood risk is estimated using Weibull’s plotting formula and presented in terms of expected annual population and expected annual damage. This study will improve our understanding of flood spatial dependence and will provide improved risk estimation at the global scale. Such reliable estimates could lead to improved large-scale flood risk management through better wide-area planning decisions, more accurate insurance coverage, and better emergency response.