Optimizing CVI-based heuristic approach for regional coastal vulnerability assessment in Sicily and 2000-2019 shoreline change calibration

Coastal vulnerability can be defined as its proneness to be damaged by changing climatic and/or geomorphic conditions. In this study, coastal vulnerability focuses entirely on the physical impact of erosion, distinguishing the coastal behaviour between negative (susceptibility) and positive (resilience) responses to climate change-induced conditions, without taking into account any socioeconomic variables. The Coastal Vulnerability Index (CVI) approach has become one of the most used and simple assessing method for coastal regions around the world. CVI approach requires the selection of a set of critical variables which are expected to significantly drive system processes, which are then reclassified assigning to each obtained class a score expressing its expected incidence, following to an expert based analysis. In this study we propose a modified version of the CVI approach, for application to the entire coastline of Sicily as a first-pass regional scale coastal-vulnerability assessment. In particular, we modified the set of considered variables, maintaining coast classification and slope, but excluding the shoreline change, which was rather used for validating the obtained model. At the same time, long term or negligible variables (sea level change and tidal range, respectively) were not considered and substituted by three covariates much more directly related to the short-term sea wave energy (number of storm surges; mean wave power; dominant wave direction). Finally, a new variable expressing the sediment yield from the fluvial system was also included, exploiting the estimation offered by the Watem-Sedem model. To apply the CVI at a regional scale, centroids along the coastline were generated and the vulnerability class for each factors captured, so to compute the composite vulnerability CVI, which was reclassified into its four quartiles. By crossing the obtained CVI classes and the observed 2000-2019 shoreline variation, a largely satisfactory validation of the heuristic model was obtained, with the advancing and retreating coastal sectors being strongly concentrated in the I and IV CVI classes, respectively. A general negative trend between advancement and CVI score prediction arose from class I to class IV, with a less marked discrimination between class II and class III. At the same time, it was observed how the sediment yield factor allowed to correctly classify some advancing coastal sector, which otherwise would have been predicted as subjected to retreatment. The obtained results produced a reliable picture of the regional coastline status, posing also the basis for potential steps toward a stochastic modelling of coastal erosion processes in Sicily.