A Self-Contained and Automated Method for Flood Hazard Maps Prediction in Urban Areas

Abstract

Water depths and velocities predicted inside urban areas during severe storms are traditionally the final result of a chain of hydrologic and hydraulic models. The use of a single model embedding all the components of the rainfall–runoff transformation, including the flux concentration in the river network, can reduce the subjectivity and, as a consequence, the final uncertainty of the computed water depths and velocities. In the model construction, a crucial issue is the management of the topographic data. The information given by a Digital Elevation Model (DEM) available on a regular grid, as well as all the other elevation data provided by single points or contour lines, allow the creation of a Triangulated Irregular Network (TIN) based unstructured digital terrain model, which provides the spatial discretization for both the hydraulic and the hydrologic models. The procedure is split into four steps: (1) correction of the elevation z* measured in the nodes of a preliminary network connecting the edges with all the DEM cell centers; (2) the selection of a suitable hydrographic network where at least one edge of each node has a strictly descending elevation, (3) the generation of the computational mesh, whose edges include all the edges of the hydrographic network and also other lines following internal boundaries provided by roads or other infrastructures, and (4) the estimation of the elevation of the nodes of the computational mesh. A suitable rainfall–runoff transformation model is finally applied to each cell of the identified computational mesh. The proposed methodology is applied to the Sovara stream basin, in central Italy, for two flood events—one is used for parameter calibration and the other one for validation purpose. The comparison between the simulated and the observed flooded areas for the validation flood event shows a good reconstruction of the urban flooding.