What is reproducible modeling

Within and outside Deltares, there is currently a push towards reproducible model projects. The goal is to ensure proper quality assurance of projects to leave more time for the actual analysis. In recent yearss, Deltares has embraced Open-Source software with the motto ‘dare to share’, which has been a crucial step towards achieving project reproducibility. In this tutorial, we will explore the essential steps and best practices to ensure that your numerical models are reproducible and easily shareable.

Why Reproducibility Matters

Reproducibility is a cornerstone of research. It allows others to validate and build upon your work, promotes transparency, and fosters collaboration within the scientific community. By following reproducible practices in your numerical modeling projects, you can:

  • Increase the reliability and credibility of your research.
  • Facilitate collaboration and knowledge exchange with colleagues and clients.
  • Enable easier troubleshooting and debugging of your models.
  • Ensure long-term accessibility and usability of your work.

Definition of a reproducible project

To establish a clear understanding, it is important to define what we mean by reproducible projects. A reproducible numerical modeling project is a project which meets the following requirements:

  1. Can reproduce the results of the project from source data with preferably 1 command.
  2. Take the minimal effort to set up on a new machine, regardless of the operating system.
  3. Is easy to understand how the project is set up (e.g., standardized folder structure, readable scripts, documentation)
  4. Keep a log of changes made to the project (scripts and data version control)
  5. Allow easy sharing of the project within Deltares, so projects are easily findable.
  6. Allow easy sharing of data (e.g., generic data formats, meta-data)

Checklist

To determine how FAIR your project is the FAIR-data-checklist can be used. This checklist scores your project based on multiple categories.

Example projects

An overview of FAIR example projects is given here.



  • No labels

Disclaimer