Abstract: A flood prediction model is usually an integration of a hydrologic module, a one-dimensional (1D) module, a two-dimensional (2D) hydrodynamic module, and others for surface water flow simulation. This paper compares the related models updated in the domestic and international literature, and develops a dynamic two-way coupling of hydrologic and 2D hydrodynamic models (DBCM and DCM2D), to achieve its performances－such as computational efficiency and accuracy－higher than the previous models. DBCM divides the watershed into non-inundation and inundation zones, spatially nonoverlapping, with the former calculated on coarse grids for hydrologic modeling, while the latter on fine grids for 2D hydrodynamic modeling. These two sets of zones are joined and coupled through movable interfaces, by applying the conservation rules of water volume and momentum. DBCM overcomes the deficiencies of the previous models that couple the two sets but breaking the conservation rules, and thus achieves better accuracy. DCM2D is an expanded DBCM by adding a 1D hydrodynamic module, and therefore achieves an internally bidirectional coupling of the distributed hydrologic module and the 1D and 2D hydrodynamic modules. These new models are aimed to better describe the real flood process while considering both computational accuracy and efficiency.

Key words: watershed flood prediction, DCM2D, DBCM, two-way coupling, computational accuracy and efficiency