Raster based model of inland coastal flooding propagation using linearized bottom friction and application to a real case study in Caorle, Venice (IT)

The paper presents a flood propagation model just developed by the authors, raster based and suited to very large maps (i.e. regional or national scale). The finite-difference model solves a simplified form of the shallow water equations, accounting for a linearized bottom friction and removing the convective acceleration terms. Each grid cell of the map is treated as an individual storage volume element and a simplified momentum equation is used to calculate the inter-cell fluxes. The computational efficiency is one of the key issue of inundation models applied to hyper-resolution maps. In order to increase the computational speed and the memory capacity, a TESLA K80 GPU accelerator with 4992 Nvidia CUDA cores and 12 GByte memory is used. The model is firstly tested against a well-known benchmark and then applied to a case study, i.e. the coastal flooding of Caorle area, Venice (IT) occurred in December, 2008. Flooding was caused by overtopping of waves propagating up the Livenza river mouth. The available information includes the incident wave conditions during the storm event, the water level inside the river mouth, and newspaper reports, pictures and video documenting the flooded area. The proposed inundation model has the ambition to help the coastal manager and stakeholders in producing flooding risk maps required under the EU water flood directive. In fact, using a raster based approach, the model is also suited to be integrated in a Geographical Information System (GIS) platform in a form that is readily applicable by the stakeholders with the most recent Digital Elevation Models (DEMs).