Model Uncertainty in Flood Modelling. Case Study at Vu Gia Thu Bon Catchment - Vietnam

Abstract

Flood map nowadays is seen as an indispensable tool in urbanism, flood prevention and mitigation. For this reason, establishing flood map is mighty necessary for developing the socio economy of a river catchment. In recent years, creating this kind of map based on hydraulic models has been applied and proved good efficiencies in mitigating the consequences of flood catastrophes to human at many regions on the world. However, because of the lack of observed data in large catchments as well as developing countries, applying this work for these regions becomes a huge challenge for hydrologists. The insufficiency of meteo hydrological data, coarse resolution of topography, land cover data …brings many difficulties to choose a suitable model or decide a reasonable model structure for flood modelling in these catchments. This study via the flood modelling process at downstream of Vu Gia Thu Bon catchment, a coastal region in Viet Nam central will compare the differences between 1D model, 2D model, Quasi 2D model and 1D/2D coupling model for flood simulation. The study also presents the uncertainties of input data such as topography, land use, rainfall, and boundary condition when modelling flood events. These simulations are carried out on the modules of Mike by DHI software: Mike 11, Mike 21, Mike Flood. The results might show strong and weak points of each model. These could help modelers to get several judgments when selecting the model to build the flood map. This study is expected to give some usefulnesses for flood modeling in the coastal part of a big catchment.