Large-Scale GIS-Based Urban Flood Modelling: A Case Study on the City of Ouagadougou

Abstract

African cities are prone to recurrent flooding due to unfavourable rainfall conditions, often insufficient drainage infrastructure and fast and poorly controlled urban development. It is thus important to put forward efficient tools to characterize flooding and its consequences over large urban areas, e.g. the entire agglomeration. However, scarce data and long computation time limit the use of classic hydraulic models in such cases and require to propose alternative models. Therefore, a GIS-based urban flood model is proposed as a case study in the city of Ouagadougou (Burkina Faso), in order to produce flood mapping at small spatial resolution over the entire conglomeration. Spatial discretization is done using a Digital Elevation Model (DEM) forced by various obstacles (urban blocks) or drainage axis (roads or water collectors). The mesh size is set to 10 m to get a good representation of the urban objects and drainage directions. Runoff at the cell scale is calculated using an SCS model, which parameters are calibrated from rainfall–runoff measurements in small urban catchments. Runoff is first routed with a Lag and Route model over 1057 catchments which drain nearly 10 ha. The velocity of the model derives from a geomorphological formula involving the slope and the upstream area of each cell. Runoff is then routed in the main hydrological network (i.e. streets and channels) with a Kinematic Wave model. The dimensions of the cross sections derive from a geomorphological formula, involving slope and upstream areas of the cells. An example of flooded areas is shown, as an application of the models over the entire conglomeration. The model strength resides in its quick implementation at large scale using easily accessible data, and in a fast computation time. The model appears as a valuable tool for decision-makers, real-time forecasting and infrastructures management. This modelling approach may be complemented by finer, local scale models which may use the computed fluxes as boundary conditions.