Rain-On-Grid Local-Inertial Formulation to Model Within Grid Topography

Flood inundation can be simulated using externally coupled hydrology and hydraulic models (decoupled models), large-scale hydrological models and rain-on-grid or direct run off models. External coupling can only simulate the fluvial (river) flooding in the hydraulic domain. The large-scale hydrological models transfer the overland flow component to the downstream point based on simplified hydrological routing methods that are not accurate especially when backwater, flood-plain reach interaction processes are involved, as in the case of low-lying terrains of deltas. To overcome the aforementioned issues, the rain-on-grid models incorporate the effective rainfall as a source term in the continuity equation of the two-dimensional (2D) shallow water equations (SWEs). This represents the pluvial flooding that is typical of urban areas. Owing to the improved computational efficiency and similar accuracy in comparison with the full 2D SWEs, local-inertial approximations have been widely used for flood forecasting, especially in large basins. However, the availability of very high-resolution data from LiDAR that can resolve smaller urban landscape features demands more computational cost. Therefore, to further increase computational efficiency without compromising on the accuracy, a bathymetry-based sub-grid approach is introduced into the local-inertial rain-on-grid model. Though sub-grid formulation operates at a coarser grid, the fluxes through the cell interfaces are estimated based on pre-stored hydraulic properties, thus capturing the within grid topographic features. To illustrate the advantage of the sub-grid-based local-inertial rain-on-grid formulation, the Adyar basin comprising the Chennai city in Tamil Nadu, India, is chosen as a case study. The study shows that the proposed sub-grid-based local-inertial model can simulate flood flow through the urban features with improved accuracy.