SEAMLESS-WAVE

Logo

SEAMLESS-WAVE is a developing “SoftwarE infrAstructure for Multi-purpose fLood modElling at variouS scaleS” based on "WAVElets" and their versatile properties. The vision behind SEAMLESS-WAVE is to produce an intelligent and holistic modelling framework, which can drastically reduce iterations in building and testing for an optimal model setting, and in controlling the propagation of model-error due to scaling effects and of uncertainty due statistical inputs.

View the Project on GitHub ci1xgk/Fellowship_Webpage

LISFLOOD-FP8.0

Initially developed by the University of Bristol, the LISFLOOD-FP hydrodynamic modelling framework has undergone many developments. It include solvers of the 2D shallow water equations with various mathematical and/or numerical complexities for raster-formatted hydraulic simulations.

The local inertia solver, known as the ACC solver, is widely used to simulate floods with gradually-varying, subcritical flow over sufficiently rough surfaces with Manning’s coefficient of at least 0.03. It has a version with CPU-specific optimisations and enhanced with a subgrid channel model (Neal et al., 2012, 2018).

LISFLOOD-FP8.0 includes second-order discontinuous Galerkin (DG2) and first-order finite volume (FV1) solvers of the 2D shallow water equations for modelling a wide range of flows, including rapidly-propagating, supercritical flows, shock waves, or flows over very smooth surfaces. The DG2/FV1 solvers are parallelised for the multi-core CPU architecture, but do not integrate with the subgrid channel model nor with the CPU-specific optimisations.

To decrease the computational costs, the ACC, FV1 and DG2 solvers have been accelerated using Graphics Processing Units (GPUs) (Shaw et al., 2021; Sharifian et al., 2023). The suitable applications for each of these solvers of are summarised in the table below, along with their potential limitations.

Solver Suitable applications Limitations
ACC Fluvial flooding; Pluvial flooding on catchment-scale resolutions Not recommended for supercritical flows, e.g., thin flows in pluvial flooding simulations at fine resolution
FV1 Fluvial and Pluvial flooding; Dam-breaks Might fail in capturing small-scale transients of flows
DG2 Fluvial flooding; Dam-breaks; tsunamis; Flows around hydraulic structures High computational cost for large-scale applications; Restrictive time-step for applications that involve thin flows

ACC solver adapted on a static, non-uniform grid, generated by multiwavelets Galerkin projection of the DEM

LISFLOOD-FP8.1 includes a new GPU-accelerated solver, known as the non-uniform ACC solver. The non-uniform grid is generated by the multiresolution analyses (MRA) of multiwavelets (MWs) to a Galerkin projecion of the digital elevation model (DEM). The ACC solver runs on this non-uniform grid without any grading across resolution levels. The technical background of the non-uniform ACC solver can be found in Sharifian et a;, 2023 and the new parameters required for the non-uniform ACC solver with video tutorials for reproducing real-world catchment- and urban-scale case studies are detailed in this section: “Using the non-uniform ACC solver”.

Adaptive FV1/DG2 solvers on dynamic non-uniform grid generated by Haar wavelets (HW)/multiwavelets (MW)

LISFLOOD-FP8.2 include a new GPU-parallelised adaptive FV1/DG2 solvers that run on an adaptive grid generated by the MRA of the HW/MW, respectively, whereby adaptive grid refers to a dynamic-in-time non-uniform grid that is generated every timestep. The new solvers, namely the GPU-parallelised MW adaptive DG2 solver, called GPU-MWDG2, is overviewed in a technical paper and the guidance for running GPU-MWDG2 simualtions is available in this section: “Using GPU-MWDG2 in LISFLOOD-FP”.

DG2-RANS solver with GPU acceletation to simulate viscous turbulent shallow vortical flows

A turbulent shallow vortical flow simulator has also been added to LISLFOOD-FP. It adapts the DG2 numerical method to solve Reynolds-Average Navier-Stokes (RANS) Equations based on the k-ε turbulence model and incorporates further robustness treatments to preserve the positivity of the turbulence fluctuations. It is accelerated on the multicore CPU and the GPU. This new solver is overviewed in a technical paper and guidance for running it on LISFLOOD-FP is available in the following section: “Using DG2-RANS in LISFLOOD-FP”.

The rest of this page includes instructions on how to download, install and run the code of LISFLOOD-FP 8.0 for using the uniform grid DG2 and FV1 solvers.

Guidance is provided on producing the Galerkin polynomial projection of the DEM and initial condition files (with slope-coefficients), and on post-processing the coarse Galerkin polynomial output files to downscale them into a floodplain map at finer resolutions.

The documented instructions are also covered in video tutorials and through realistic case studies.


Download

Zenodo for external users

University of Sheffield users


Compilation

LISFLOOD-FP 8.0 is cross-platform and can be compiled on Windows or Linux using CMake version 3.13 or above. University of Sheffield users can also compile and run it on the HPC Computing Facilities. The compilation process on each of the platforms is explained below.


Case studies

Case studies are presented to instruct users on how to set up and run the DG2/FV1/ACC solvers of LISFLOOD-FP. These studies are also useful to identify the most optimal DG2 configuration for the targeted application.

1. Merewether urban flooding

This case study (Smith et al., 2017) is used to explain how to set up and run LISFLOOD-FP for beginners using any of the DG2, FV1 and ACC two-dimensional hydrodynamic solvers on the CPU or the GPU.

2. Dam-break wave over an urban area

This case study (Soares-Frazao and Zech, 2008) is aimed to demonstrate when local slope limiting with DG2 should be applied.

3. Valley flooding following a dam failure

This case study (Néelz and Pender, 2013) is aimed to demonstrate how to handle an inflow boundary condition located inside the computational area, at source point(s).

4. Carlisle 2005 urban flooding

This case study (Horritt et al., 2010) is particularly aimed to demonstrate how to handle multiple inflow boundary conditions from different source point(s). It is also used to demonstrate a new postprocessing toolkit for producing metrics to evaluate the floodplain extent predictions when comparing the simulation outcomes of two different hydrodynamic solvers.

5. Eden 2015 fluvial flooding

This case study (Xia et al., 2019) involves overland flow driven by spatially- and temporally-varying rainfall data over a 2500 kilometre square catchment. It is aimed to demonstrate how to use spatially-varying Manning coefficients provided in an input file, along with the new rain-on-grid option of LISFLOOD-FP 8.0. It is also used to demonstrate a new postprocessing toolkit for downscaling coarse-resolution planar DG2 flood maps.


Video tutorials

  1. Downloading Lisflood

  2. Compiling on Windows

  3. How to set up and run a simulation in Lisflood

  4. Running a dam break test with slope limiting

  5. Running a test with a point-source inflow

  6. Running a test with multiple inflows

  7. Running a test with spatially and/or temporally varying Manning coefficient rainfall


License

The source code of LISFLOOD-FP8.1 is available under a GNU General Public License v3.0 for any non-commercial use and can be downloaded at 10.5281/zenodo.4073011.


Key references

G. Kesserwani, J.L. Ayog and D. Bau (2018). Discontinuous Galerkin formulation for 2D hydrodynamic modelling: trade-offs between theoretical complexity and practical convenience. Computer Methods in Applied Mechanics and Engineering, 342, 710-741.

J. Ayog, G. Kesserwani, J. Shaw, M.K. Sharifian, and D Bau (2021). Second-order discontinuous Galerkin flood model: comparison with industry-standard finite volume models. Journal of Hydrology, 594: 125924.

J. Shaw, G. Kesserwani, J. Neal, P. Bates, and M. K. Sharifian (2021). LISFLOOD-FP 8.0: the new discontinuous Galerkin shallow water solver for multi-core CPUs and GPUs. Development and technical paper, Geoscientific Model Development, 14, 3577–3602.

M. K. Sharifian, G. Kesserwani, A. Chowdhury, J. Neal, and P. Bates (2023). LISFLOOD-FP 8.1: New GPU accelerated solvers for faster fluvial/pluvial flood simulations, Geosci. Model Dev. Discuss..


About the Developers

While the original version of LISFLOOD-FP were set in motion by University of Bristol, the recent versions 8.0 and 8.1 were mainly developed as part of the SEAMLESS-WAVE project lead by Georges Kesserwani in University of Sheffield, who is main contact about these versions.

Contact Georges

The main GPU kernels for the FV1 and DG2 solvers were initiated by James Shaw during his postdoctoral research, before joining The Floow as a senior software engineer. Further developments were mainly carried out by Mohammad Kazem Sharifian, including the GPU kernels of the ACC solver for the uniform and non-uniform grids. Mohammad is now a Flood Modeller at RMS (Moody’s Analytics Company) and is happy to receive enquiries about the software, bug reports, feature requests, or any relevant matter that needs the community’s attention.

Contact Mohammad

back