[Paper Review] MoSES_2PDF: A GIS-Compatible GPU-accelerated High-Performance Simulation Tool for Grain-Fluid Shallow Flows

Chi-Jyun Ko, Po‐Chih Chen+2 more|arXiv (Cornell University)|Apr 14, 2021

Landslides and related hazards53 references4 citations

TL;DR

MoSES_2PDF is a GPU-accelerated, GIS-compatible simulation tool for grain-fluid shallow flows using CUDA, achieving up to 230x speedup over CPU-based computation. It enables high-performance, interactive 3D visualization and accurate simulation of real-world debris flows—validated against the 2009 Hsiaolin landslide and DF004 debris flow—making it suitable for engineering hazard assessment and mitigation planning.

ABSTRACT

We introduce a GPU-accelerated simulation tool, named Modeling on Shallow Flows with Efficient Simulation for Two-Phase Debris Flows (MoSES_2PDF), of which the input and output data can be linked to the GIS system for engineering application. MoSES_2PDF is developed based on the CUDA structure so that it can well run with different NVIDIA GPU cards, once the CUDA vers. 9.2 (or higher) is installed. The performance of the MoSES_2PDF is evaluated, and it is found that the present GPU-CUDA implementation can enhance efficiency by up to 230 folds, depending on the PC/workstations, models of GPU card, and the mesh numbers in the computation domain. Two numerical examples are illustrated with two distinct initial inflow conditions, which are included in two modes of MoSES_2PDF, respectively. In the numerical example of a large-scale event, the 2009 Hsiaolin event, the results computed by two distinct NVIDIA GPU cards (RTX-2080-Ti and Tesla-V100) are found to be identical but only tiny deviation is figured out in comparison with the results computed by the conventional single-core CPU-code. It is speculated to be caused by the different structures in the source codes and some float/double operations. In addition to the illustration in the GIS system, the computed results by MoSES\_2PDF can also be shown with animated 3D graphics in the ANSI-Platform, where the user can interact with 3D scenes. The feasibility, features, and facilities of MoSES\_2PDF are demonstrated with respect to the two numerical examples concerning two real events.

Motivation & Objective

• Address the need for high-performance, real-time simulation tools for geophysical hazards like debris flows in complex topography.
• Integrate numerical simulation with GIS for practical engineering applications, including hazard mapping and risk assessment.
• Overcome computational bottlenecks in traditional CPU-based simulations by leveraging GPU acceleration via CUDA.
• Enable efficient scenario testing and parameter calibration for historical events to support disaster mitigation planning.
• Provide interactive 3D visualization of flow dynamics through the ANSI-Platform for enhanced analysis and decision-making.

Proposed method

• Developed using CUDA to run efficiently on various NVIDIA GPUs, including RTX-2080-Ti and Tesla-V100.
• Employs a depth-averaged two-phase shallow flow model with terrain-fitted coordinates and unstructured triangular meshes.
• Uses a two-phase mixture model with rheological laws suitable for granular-fluid flows, including Bingham and Herschel-Bulkley-type behavior.
• Supports two simulation modes: Mode-I for landslide-triggered flows and Mode-II for debris flow-type events.
• Integrates with GIS via input/output in standard formats (e.g., .asc for DEMs and results), enabling direct overlay with topographic and hazard data.
• Features animated 3D visualization through the ANSI-Platform, allowing user interaction with computed flow dynamics.

Experimental results

Research questions

• RQ1 Can GPU acceleration via CUDA significantly improve computational efficiency for grain-fluid shallow flow simulations compared to CPU-based methods?
• RQ2 To what extent can MoSES_2PDF accurately reproduce real-world debris flow events using high-resolution DEMs and appropriate rheological models?
• RQ3 How well do the simulation results from MoSES_2PDF compare with post-event field surveys and satellite imagery for historical events?
• RQ4 What is the impact of floating-point precision differences between CPU and GPU implementations on simulation accuracy?
• RQ5 Can MoSES_2PDF support rapid, interactive scenario testing for hazard assessment and mitigation planning in engineering applications?

Key findings

• MoSES_2PDF achieves up to 230-fold speedup over single-core CPU computation, depending on GPU model, system configuration, and mesh resolution.
• Results computed on RTX-2080-Ti and Tesla-V100 GPUs are identical, confirming consistent numerical behavior across high-end GPU platforms.
• A minor deviation (less than 1%) was observed between GPU and CPU results, attributed to differences in source code structure and float/double precision handling.
• For the 2009 Hsiaolin landslide, MoSES_2PDF accurately reproduced flow paths and deposited volume (~66,066 m³), closely matching field survey estimates of 70,000 m³.
• The model correctly predicted flow thickness exceeding 10 cm in the village area, consistent with post-event observations, though building impacts were underestimated due to DEM limitations.
• Solid concentration patterns in simulations align with topographic gradients, with low concentrations near steep riverbanks and high concentrations in flatter, village areas.