A balanced watershed decomposition method for rain-on-grid simulations in HEC-RAS

Abstract

Abstract Rain-on-grid simulations for the modeling of 2D unsteady flows in response to precipitation input are implemented in Hydrologic Engineering Center's River Analysis System (HEC-RAS) software by means of a finite volume method and a sparse parallel linear solver running on a unique processor. Obviously, this may fail due to memory shortage when the discretization yields too large linear systems. Such simulations are good candidates for the design of partition and domain decomposition methods for the modeling of very large catchments by means of hydrological sub-units. Thinking in load-balanced computations, the proposed area-balanced partition method is based on the flow accumulation information. Implemented in HEC-RAS, the domain decomposition method comprises 2D–1D models combining 2D sub-unit meshes to 1D channels for extracting outflow data from upstream units and passing the information to downstream units. The workflow is automated using the HEC-RAS Controller. As a case study, we consider the French Saar catchment (1,747 km2). The new partition methods demonstrate a better equilibrium for sub-basin areas and the computational load for the sub-problems. The total computational time is substantially reduced. Domain decomposition is required for carrying rain-on-grid simulations over the Saar watershed (2,800,000 elements). The resulting discharge and flood maps show very good agreements with available data.