A Bayesian Network methodology for coastal hazard assessments on a regional scale: the BN-CRAF

Abstract

Hazard assessment is one of the key elements to be included in any coastal risk assessment framework. Characterizing storm-induced erosion and inundation involves the assessment of the coastal response under the forcing of a stochastic source (the storm), acting on a variable morphology (the beach) and inducing some damages. Hazard assessment under any present or future scenario will be affected by uncertainties either associated to the models used, the definition of climate conditions, and the characterization of the coastal morphology. In this context, Bayesian Networks (BN) can effectively address the problem as they allow accounting for these uncertainties while characterizing stochastically the system response and giving insight on the dependencies among involved variables. In this work, a BN-based methodology for storm-induced hazard assessment at regional scale is presented. The methodology is able to account for uncertainties associated with included models and forcing conditions through Monte-Carlo simulations. It produces distributions of erosion and inundation hazards under given scenarios allowing conditioned hazard assessments as a function of storm and morphological variables. Results are compared to hazards evaluated using an existing Coastal Risk Assessment Framework (CRAF), at two locations of the Catalan coast already identified as hotspots for storm-induced erosion and/or flooding.