Run SFINCS Directly from Deltares’ Repository

The SFINCS (Super-Fast INundation of CoastS) model is a reduced-complexity simulation engine developed by Deltares. It is designed to simulate compound coastal flooding with high computational efficiency while maintaining reliable accuracy.

This guide shows you how to run SFINCS through the Inductiva API using Deltares' publicly available container image directly from their repository.

The use case covered in this guide is featured in the following publication: Camila Gaido Lasserre, Kees Nederhoff, Curt D. Storlazzi, Borja G. Reguero, Michael W. Beck (2024). "Improved efficient physics-based computational modeling of regional wave-driven coastal flooding for reef-lined coastlines." Ocean Modelling.

Prerequisites

Download the use case here and copy the sfincs_netbndbzsbzifile.nc file from Models/HighReliefArea/SFINCS/BoundaryConditions/rp_000_SLR000 to the Models/HighReliefArea/SFINCS/InputFiles folder.

The InputFiles folder should contain the following files:

sfincs.dep          sfincs.man
sfincs.gms          sfincs.msk
sfincs.ind          sfincs.obs
sfincs.inp          sfincs.obs.bak
sfincs.inp.bak      sfincs_netbndbzsbzifile.nc

Run Simulations using SFINCS Docker Container

You can reference the Deltares’ Docker image directly in your simulation script using the inductiva.simulators.CustomImage class, as shown below:

import inductiva

cloud_machine = inductiva.resources.MachineGroup(
    "c2d-highcpu-8",
    data_disk_gb=100)

custom_simulator = inductiva.simulators.CustomImage(
    container_image="deltares/sfincs-cpu:latest")

input_dir = "/Path/to/Models/HighReliefArea/SFINCS/InputFiles"

task = custom_simulator.run(
    input_dir=input_dir,
    commands=["sfincs"],
    on=cloud_machine)

The CustomImage simulator enables you to run simulations using any Docker image of your choice by specifying the container_image parameter.

When the simulation is complete, we terminate the machine, download the results and print a summary of the simulation as shown below.

Task status: Success

Timeline:
    Waiting for Input         at 07/05, 15:59:41      2.927 s
    In Queue                  at 07/05, 15:59:43      34.238 s
    Preparing to Compute      at 07/05, 16:00:18      22.302 s
    In Progress               at 07/05, 16:00:40      1443.74 s
        └> 1443.575 s      sfincs
    Finalizing                at 07/05, 16:24:44      2.737 s
    Success                   at 07/05, 16:24:47      

Data:
    Size of zipped output:    74.44 MB
    Size of unzipped output:  88.29 MB
    Number of output files:   7

Estimated computation cost (US$): 0.028 US$

As you can see in the “In Progress” line, the part of the timeline that represents the actual execution of the simulation, the core computation time of this simulation was approximately 1443.7 seconds (24 minutes and 44 seconds).

Scaling Up Your Simulation

Scaling up your simulation is as simple as updating the machine_type parameter to a 16 vCPU machine (c2d-highcpu-16).

By increasing the number of virtual CPUs, we’ve reduced the processing time from 25 minutes and 31 seconds to 18 minutes and 21 seconds.

Here are the results of running the same simulation on a few machines: