LISFLOOD-FP 8.2: GPU-accelerated multiwavelet discontinuous Galerkin solver with dynamic resolution adaptivity for rapid, multiscale flood simulation
Abstract. The second-order discontinuous Galerkin (DG2) solver of the shallow water equations in LISFLOOD-FP 8.0 is well-suited for predicting small-scale transients that emerge in rapid, multiscale floods caused by impact events like tsunamis. However, this DG2 solver can only be used for simulations on a uniform grid where it may yield inefficient runtimes even when using its graphics processing unit (GPU) parallelised version (GPU-DG2). To maximise runtime reduction, the LISFLOOD-FP 8.2 version integrates GPU parallelised dynamic (in time) grid resolution adaptivity of multiwavelets (MW) with the DG2 solver (GPU-MWDG2). The GPU-MWDG2 solver requires selecting a maximum refinement level, L, based on size and resolution of the Digital Elevation Model (DEM) and an error threshold, ε ≤ 10-3, to preserve similar accuracy as a GPU-DG2 simulation on a uniform grid. The accuracy and efficiency of dynamic GPU-MWDG2 adaptivity is assessed for four tsunami-induced flooding test cases involving increasingly complex tsunamis: from single-wave impact events to wave trains. At ε = 10-3, the GPU-MWDG2 simulations yield predictions similar to the GPU-DG2 simulations but using ε = 10-4 can improve the accuracy in velocity-related predictions. In terms of efficiency, the GPU-MWDG2 simulations show progressively larger speedups over the GPU-DG2 simulations from L ≥ 10, which become significant (≥ 3.3- and 4.5-fold at ε = 10-4 and 10-3 , respectively) for simulating a single-wave impact event. The LISFLOOD-FP 8.2 code is open source, DOI: 10.5281/zenodo.4073010, as well as the simulation data and the input files and scripts to reproduce them, DOI: 10.5281/zenodo.13909072, with additional documentation at https://www.seamlesswave.com/Adaptive (last accessed: 9 October 2024).