Flood hazard mapping by means of fully-2D and quasi-2D numerical modeling: a case study

The results of mathematical modeling of some flooding scenarios in the middle reach of the more important Italian river are presented. In 1868 and 1951 the considered region was subject to two severe floods, due to the breaking of the earthen embankment of the Po river. The inundated area was about 40km2 and water depths up to 3.5 meters were observed. In both cases the breaches were originated by the failure of sluice gates located on minor drainage channels. On the basis of the shallow water approach, four flooding scenarios were at first simulated by means of a 2D finite volume MUSCL-Hancock code. The scheme is explicit, shock capturing, high resolution and belongs to the Surface Gradient Methods class. The SGM peculiar way of treatment of the slope source term allows to rigorously satisfy the C-property and to model the potential overtopping of the man-made or natural levees, such as channel embankments, roads, railways, etc. To derive the discharge overflowing from the breach, also when the flow is backwatered due to the partial filling of the floodable region, a reach of the Po river was included in the calculations. A sensitivity analysis on the main parameters affecting the inundation dynamics, namely the breach width and its position and time of occurrence with reference to the chosen hydrological input, was performed. At the aim to derive a flood hazard mapping for the area under investigation a risk factor was also derived as a function of computed maximum water depth and arrival time of the wetting front. One of the investigated scenarios, assumed as a reference, was also simulated by means of a quasi-2D numerical model. The area susceptible to flooding was schematized through a number of storage cells, each representing a portion of the domain bounded by the same levees considered in the two dimensional model. The cells were properly joined by link channels provided with features that allow to compute the flow over user defined broad crested weirs. The flooding dynamics is not correctly described by the quasi-2D numerical model in the first hours after the breaching; afterwards the quasi-static behaviour of the flooding is well caught. However the overall comparison between the fully-2D and quasi-2D numerical results shows a fairly good agreement for what concerns the identification of flooded area, maximum water depths and discharge overflowing from the breach