Author(s): Na Zheng; Zhi Li; Junbo Wang

Linked Author(s):

Keywords: Urban flooding; High-performance computing; Kokkos; Bidirectional coupled model; SWMM

Abstract: In recent years, rapid urbanization and the increasing frequency of extreme climate events have led to a significant rise in urban flood disasters. Over the past few decades, hydrodynamic simulation models have become a widely adopted approach for studying urban flooding. These models typically integrate two-dimensional (2D) surface runoff models with one-dimensional (1D) pipeline models for drainage systems. However, when simulating large-scale flood events, computational efficiency often becomes a critical bottleneck. High-performance computing (HPC) has emerged as a powerful solution to accelerate urban flood simulations by utilizing advanced computational algorithms and parallel processing techniques. Despite its advantages, the wide variety of HPC hardware platforms available today presents challenges for the portability of model deployment. To address this issue, we propose a high-performance bidirectional coupled model, named SERGHEI-SWMM. Built upon the SERGHEI-SWE and SWMM frameworks, this model captures hydrodynamic processes by solving shallow water equations and can accurately simulate urban flood dynamics. Furthermore, the model leverages the Kokkos framework, ensuring excellent portability and enabling seamless operation across different HPC platforms without requiring modifications to the source code. We evaluate the model’s accuracy and acceleration performance using both an idealized test case and a real-world application. SERGHEI-SWMM serves a robust tool for simulating urban floods and investigating the complex hydrological and hydrodynamic factors influencing the dynamic evolution of urban flooding.

DOI: https://doi.org/10.64697/978-90-835589-7-4_41WC-P1692-cd

Year: 2025