Skip to content

Latest commit

 

History

History
62 lines (42 loc) · 3.59 KB

TECHNICAL.md

File metadata and controls

62 lines (42 loc) · 3.59 KB
Raw

SWAN wave model

This document provides background on the (dummy) calculations which are done as part of the FAIR-data-example-project. The workflow consists of a few steps which include a calculation with the SWAN wave model for a lake, the Volkerak Zoommeer, in the Netherlands (see Figure below). SWAN is a model which can compute wave properties (e.g. wave height) for a sea or a lake. It requires a wind field as an input and computes to wave properties. To do this it requires a grid an output location for which the output is given.

Domain of model for Volkerak Zoommeer

FAIR-data-example-project

In the example in this project we do two calculations with different constant wind speeds (20 and 30 m/s) and compute the resulting wave properties. Please note that this is a highly simplified model. For this, the following steps are added to the snakefile:

  1. Prepare output location (rule: locations)
  2. Prepare grid (rule: grid)
  3. Setup input files for all the wind conditions (rule: create_sims)
  4. Compute calculations for all the wind conditions (rule: run)
  5. Plot the results of all the wind conditions (rule: analyse)

In every step a specific task of the workflow is executed when the previous step is successfully finished. For example, it is not possible to start a calculation when the grid files are not prepared. The snakefile includes a list of wind speeds for which the computations are performed. This means that some steps of the snakefile needs to be executed for all wind speeds. A complete overview of the workflow is shown in the Figure below.

Overview of the workflow steps in the example project

Now we can try if we can run the project and do a small exercise with it. The steps to follow can be found in EXERCISE.md.

Bonus information:

Running snakemake locally and on the h6 cluster

It is possible to run a workflow locally and on the H6 cluster. This can be achieved by specifying in the snakefile whether to run on Windows (win=True) or on Linux (win=False).

To run this demo project locally run (win=True):

snakemake -c1

Snakemake by default looks for a snakefile in the directory. It is also possible to specify the name of a snakefile to run:

snakemake --cores 1 --snakefile snakefile

To run this demo project on the h6 run (win=False):

snakemake --cores 1 --cluster "qsub -q {cluster.q} -N {cluster.N} -M {cluster.M}" --jobs 4 --cluster-config config/cluster.yaml

The flag --cluster specifies the command to sumbit a job to a cluster. With the flag --cluster-config additional wildcards for the cluster can be set. In this case the queue (q), job name (N) and email (M). The maximum number of jobs is specified with --jobs

Not all rules are submitted to the cluster. The localrules keyword in the snakefile can be used to specify the rules to be executed immediately. localrules: create_sims, grid, locations

Using a standard folder structure

In Deltares or with colleagues who are familiar, some standardized project templates may already be available. An example of of such a template is available from our Groundwater management department. This template can be created by:

cookiecutter gl:deltares/imod/cookiecutter-reproducible-project