Author(s): Zhi Li; Na Zheng; Mario Morales-Hernandez; Daniel Caviedes-Voullieme
Linked Author(s):
Keywords: Shallow flow; Surface-subsurface exchange; Parallel computing; GPU
Abstract: Surface-subsurface (SS) flow exchange is a critical process in hydrological modeling. However, the SS models could be computationally demanding, especially when using 2D SWE and 3D RE instead of their simplified forms. This is due to both the small time steps to keep SWE solvers stable, and the nonlinear soil-water retention function that requires iterative subsurface flow solvers. Advances in High-Performance Computing (HPC) offer solutions for accelerating large-scale hydrological simulations. Here, we integrate the SERGHEI-SWE shallow water solver (Caviedes-Voullieme et al. 2023) and the SERGHEI-RE subsurface solver (Li et al. 2024) into a unified model, named SERGHEI-SWE-RE, to simulate SS exchange. Key features of the model include: (1) the use of complete forms of governing equations, (2) modularity with asynchronous coupling for efficiency, and (3) performance portability across diverse computational platforms via the Kokkos framework. These features will be demonstrated through testings with classical benchmark problems and real-world hydrological applications.
Year: 2025