Use of a Bayesian Network for coastal hazards, impact and disaster risk reduction assessment at a coastal barrier (Ria Formosa, Portugal)

Abstract

Coastal communities are threatened by the impact of severe storms that may cause significant loss or damage of property and life. The main processes causing such impacts at sandy coastlines and nearby coastal communities are storm erosion, overwash and inundation. Coastal response under present conditions and under predicted climate change has been frequently assessed on the basis of numerical models, which in turn can be also used to evaluate the effectiveness of Disaster Risk Reduction (DRR) measures to mitigate the response of the coast to the imposed conditions. However, detailed morphodynamic models are computationally expensive and not commonly used by coastal managers. The present work proposes the construction of a probabilistic Bayesian Network (BN) as a surrogate for the numerical simulations. This BN is trained with a large number of morphodynamic simulations, under a variety of storm conditions and DRR measures, in order to serve as a front-end platform for visualising, analysing and evaluating combined results of the numerical modelling. The BN introduced in an early warning system will be able to serve both, as a predictive and as a working tool to determine impacts and evaluate risk reduction after measures implementation.Here, an example of the implementation and results of such a BN system is presented. The BN system was built for a coastal sector of the Ria Formosa barrier island system (South Portugal) to inform the degree of impact derived from overwash and erosion over the study area. The BN boundary conditions include variable wave height, water level, and wave period. The impact on receptors, including houses and infrastructure, was assessed. In addition, this tool can inform about the effectiveness of a particular DRR measure. The evaluated DRR measures were two primary measures (partial house removal and beach replenishment) and a non-primary measure (improve channels of communication), all measures proposed by local stakeholders. Results show that for a storm with wave characteristics of the 1 in 50 year return period and spring tide conditions, the house removal DRR reduces the overwash impact by 15% and erosion impact by 58%. The implementation of beach replenishment could reduce the erosion impact of the same event by 96% while it would have a smaller effect on the overwash impact (16%). The implementation of non-primary measures would have a much smaller effect on risk reduction. The combined effect of the above DRR measures (mainly house removal and beach nourishment) reduces storm impacts at the study area to a value near zero. The BN surrogates the model simulated onshore hazards and translates them into impacts for the current conditions, which give a high degree of confidence in the potential application of the BN as a management tool.