3. Open Data, Tools, and Workflows

Motivation

The goal for this week to build a foundation that allows us to jump into data analysis in the coming weeks.

Specific learning goals

After completing this unit, you will be able to

  • Define open reproducible science and explain its importance.
  • Describe how reproducibility can benefit yourself and others.
  • Evaluate data and workflows according to open-science principles including FAIR.
  • List tools that can help you implement open reproducible science workflows.
  • Explain why version control is useful in a scientific workflow.
  • Explain how a GitHub repository stores and tracks changes to files.

Skills:

  • Use GitHub-Desktop to track changes in files using git commit and git push.
  • Create a new git repository and publish the repository on GitHub.com.
  • Create a copy of (i.e. fork) other users’ files on GitHub.com.
  • Use the git clone command to download a copy of a GitHub repository to your computer.

Recap: Data Analysis Workflow

We can describe the data analysis workflow as a linear sequence of steps (Figure 1):

Flowchart for environmental data analysis as a linear process with steps: data collection, data cleaning, data analysis, data interpretation, and decision making
Figure 1: Environmental Data Analysis Flowchart; Credit: NumberAnalytics

However, this workflow is embedded into a larger context of doing open-science.

Recap: Open Science and Workflows

Open and reproducible science is a collection of practices (Figure 2) that allow us to easily share, work and collaborate with others1.

An open science workflow highlighting the roles of data, code, and workflows as well as stakeholder engagement to find solutions for environments problems
Figure 2: An open science workflow highlighting the roles of data, code, and workflows. Source: Max Joseph, Earth Lab at University of Colorado, Boulder.

Benefits of Open Reproducible Science

Benefits of openness and reproducibility in science include:

  • Transparency in the scientific process, as anyone including the general public can access the data, methods, and results.
  • Ease of replication and extension of your work by others, which further supports peer review and collaborative learning in the scientific community.
  • It supports you! You can easily understand and re-run your own analyses as often as needed and after time has passed.

What does this mean in practice?

  • Use scientific programming to process data
    • Allows you to automate tasks
    • Provides a written ‘recipe’ that is always followed the same way
  • Organize your work using expressive names for files and directories
    • See the paper I shared on organization
  • Use FAIR data to enhance the reproducibility of projects
  • Protect your raw data
    • Make sure your raw data is not modified
    • Keep outputs separate from inputs
    • Organize your work in directories
  • Use version control
    • Allows you to track and manage any changes to your code
    • Allows you to publish your code
  • Document your workflows
    • Use tools that document workflows
    • Describe and add comments to your code
    • (Can be many other things as well)

A case study

Hey ISAT 420 Students,
I found this cool dataset online and I want you to check this out. The data is attached to this email.

Best wishes, TG

data.csv

TipReflection Questions:
  • Look at data.csv
  • What are the issues with this dataset?
  • What do you think the dataset represents
  • How would you go on to analyze the data?

FAIR and Open Data

It is impossible to know from the csv that I shared with you, what the data represents.

The dataset I shared is the global carbon budget for the year 2023 (Figure 3), which is compiled by the Global Carbon Project:

Figure 3: Schematic representation of the overall perturbation of the global carbon cycle caused by anthropogenic activities, averaged globally for the decade 2013-2022. See legends for the corresponding arrows and units. The uncertainty in the atmospheric CO2 growth rate is very small (±0.02 Gt C yr-1) and is neglected for the figure. Source: GCP 2023

This dataset is actually fairly well described and documented. This includes the website and an article in the open-access journal: Earth System Science Data2

Data Analysis Process

Ok, we now know where the data is coming from, what is represents, and that it is likely trustworthy.

So let’s think about what we can do with this. Considering our data analysis workflow (Figure 4):

flowchart LR
  A("<s>Data Collection</s>") --> B("<s>Data Cleaning</s>")
  B --> C(Data Analysis)
  C --> D(Data Visualization)
  D --> E(Data Communication)

Figure 4: Schematic representation of what we might want to do with the global carbon project dataset

How can we do this in a transparent and reproducible way?

Using the tools that we have

We can use Jupyter Notebook to read, process, and visualize our data. The most simple code for our analysis might be something like:

Step 1: Read the dataset
Step 2: Make some calculations to process our data
Step 3: Make a visualization of our data
Step 4: Save the plot as a file

Demonstration: Jupyter Notebook

The code is found on GitHub: https://github.com/ISAT-DrG/ISAT_420_S26_Shared

Checking what we have done for reproducibility

Remember, being reproducible is not all or nothing. Let’s see which criteria we are currently doing well with …

… and what we could do to improve.

Git and GitHub for Transparency and Reproducibility

The code I just demonstrated is part of a git repository that is hosted on GitHub.

This means that any changes to the code are tracked within the git repository and the code is also quasi-archived.3

We use git and GitHub as a version control system (Figure 5):

Figure 5: A version control system saves changes to a document, sequentially as you add and commit them to the system. Source: Software Carpentry.

This also means that different users can make independent changes to the same document (Figure 6):

Figure 6: Different versions of the same document can be saved within a version control system. Source: Software Carpentry.

If there are conflicts between the users’changes, these conflicts can be resolved by choosing which changes you want to keep (Figure 7):

Figure 7: Two sets of changes to the same base document can be merged together within a version control system if there are no conflicts (areas where both users modified the same part of the same document in different ways). If there are conflicts, they can resolved by choosing which change you want to keep. After conflicts are resolved, all other changes submitted by both users can then be merged together. Source: Software Carpentry.

Demonstration: Updating our analysis for 2025

Let’s say we want to make a change in our analysis. This change should be transparent and reproducible.

We can use git and GitHub for this.

Step 1: Make and save changes
Step 2: Commit change
Step 3: Push our changes to GitHub

We can now examine the changes in the Github Repository (or also using GitHub-Desktop).

Practice: Git workflow: Commit and Synching Changes with GitHub

TipActivity: Create a repository using GitHub Desktop and make a change

Goal: Practice what we just learned

  1. Create a new repository using GitHub Desktop
    File \(\rightarrow\) NewRepository
  2. View the files: Show in Explorer (Windows) or Show in Finder (Mac)
  3. Add a new text document with some text: test.txt
  4. Commit your new file with a message: Added a new test file
  5. Modify your text file by adding some additional text.
  6. Commit your changed file with a message: Changed file content
  7. Push your change to GitHub
  8. View your changes on GitHub.com

Git and GitHub for sharing and collaboration

Getting Code from GitHub - Git workflow: Fork and Clone

Let’s say we are doing open-science and have found some freely-available code on GitHub that we want to build on. What can you do to get started?

Create a Copy of Other Users’ Files on GitHub.com (Fork a Repo)

Using GitHub.com, you can make a copy of a GitHub repository (also known as a repo) owned by another user or organization (a task referred to as forking a repository).

To fork a repo:

  1. Navigate to the repo that you wish to fork. Example:
    https://github.com/ISAT-DrG/ISAT_420_S26_Shared

  2. On the upper right corner you will see a button that says Fork (Figure 8)

    Figure 8: To fork a repo, first navigate to the repo you want to fork. Then click the fork button in the upper right hand corner of your screen. You can then create a copy of of this repo in your account.

  3. Click on the Fork button and select your user account when it asks you where you want to fork the repo.

  4. You can now use GitHub-Desktop to create a local copy of the repository on your computer.

    • File \(\rightarrow\) Clone Repository
    • Select the forked repository from the list
    • Decide on where you want to save the repository: Local path
    • Hit the button: Clone

Collaborating with Shared Code

Often times, there is a central repository with code that many people are contributing to. This includes many of the Python packages that are part of the Pangeo ecosystem. To manage this collaboration, individual contributors are working on their own forked copies of the project. When they are ready to share (or suggest) a change to the overall project, they can use a Pull Request (Figure 9).

Figure 9: LEFT: To sync changes made and committed locally on your computer to your GitHub account, you push the changes from your computer to your fork on GitHub. RIGHT: To suggest changes to another repository, you submit a Pull Request to update the central repository. Source: Colin Williams, NEON.

A Pull Request is a workflow that allows

  • Your collaborators to see exactly what items where changed line by line
  • A group of people to view, review, and comment on the code line by line
  • Approve or reject any changes that will then be merged into the base repository.

GitHub Workflow: Creating a Pull Request

For now, we don’t need to know this workflow in detail. However, it is described in Chapter 8, Lesson 2 of the CU-Boulder Intro to Earth Data Science Textbook (2025)

Wrap-up

TipSkills/ Tool Check

Can you

  • use Github-Desktop to create a new repository?
  • publish your repository to GitHub (set to private)?
  • add a file to your repository, commit the change, and push the change to GitHub?
  • fork and clone a shared repository so that you have the contents on your computer?
  • use Anaconda to run a jupyter notebook on your computer?
    • Open jupyter notebook and run the first couple of cells of this notebook: w3_basic_python_test.ipynb
    • I recommend to run the rest at home to get a refresher on how python works.

Acknowledgements

The content and activities on this page are in a large part based on the CU-Boulder EarthLab Intro to Earth Data Science Textbook (2025). Relevant chapters are:

Footnotes

  1. CU Boulder’s Intro to Earth Data Science (2020) textbook , Section 1: Introduction to Open Reproducible Workflows↩︎

  2. A bit of shameless bragging here: I am a topic editor for this journal. This means, I help govern the policies of this journal as well as administer the review process for articles.↩︎

  3. Technically GitHub should not be relied on to be the solve code archive because there is no guarantee that GitHub or the company that owns it will always exist.↩︎