A GPU-accelerated shallow flow model for tsunami simulations

Abstract

This work presents a numerical model for tsunami propagation and inundation, which solves two-dimensional shallow water equations using a finite-volume Godunov-type numerical scheme incorporating a Harten–Lax–van Leer–contact (HLLC) approximate Riemann solver. To improve the computational efficiency for high-resolution tsunami simulations that commonly cover large spatial domains, the model is implemented for graphics processing units using compute unified device architecture (CUDA). The accelerated tsunami model is used to simulate an idealised dam-break case involving 1 million uniform computational cells and a laboratory tsunami propagation and run-up test with complex domain topographic features. Compared with a Fortran code for the same numerical scheme running on a standard personal computer, the model is able to reduce computational time by more than 40 times.