ArcGIS, Jupyter Notebook, and Git

A reproducible workflow for ArcGIS spatial analysis using Jupyter Notebook and versioned with Git

workflow
ArcGIS
JupyterNotebook
Git
Author

Sarah Tanja

Published

September 24, 2026

Modified

October 8, 2026

1 Aims

Here we want to figure out a simple reproducible workflow that allows researchers to work using ArcGIS Pro’s integrated Jupyter Notebook and back up and version their work using Git. The goal is for the Git repository to become the durable record of how the analysis works, while Microsoft Teams can remain the home for source/spatial data that may prove too large for Git.

We need to:

  1. create a new Git repo inside the existing University of Washington Puget Sound Institute GitHub organization1

  2. create a .Gitignore file that will exclude anything not needed for reproducibility

  3. be mindful of GitHub’s large file size limits

    GitHub limit
    Normal Git file 100 MB
    File >50 MB GitHub warns large file
    Browser upload 25 MB max *when uploading files directly through GitHub’s website
    Single push 2 GB max
    Repository .Git size 10 GB recommended maximum
    GitLarge File Storage (LFS) file 2–5 GB/file, depending on payment plan *we are not using LFS

1 Admin for the org is currently Stefano Mazzilli (9/24/2026)

2 Background

The context for this is that I am picking up a spatial critical analysis project (Marine Protected Area Impacts on Marine Vegetation… Characterizing eelgrass and kelp conservation in Puget Sound protected areas) that was started by Sandra Dorning in September of 2025, but I only have one year in the WA Seagrant Fellowship to try and complete the analysis. I want to structure the project such that if it needs to be re-run in the future (to add more or updated data), or if I need to pass the baton before it’s finalized, that what I’ve done is reproducible.

The big-picture end-products that we want for this is a grey-literature Puget Sound Institute report guiding answers to the question “Are marine protected areas effective at protecting marine vegetation from risks in Puget Sound?”, a manuscript for submission to a peer-reviewed journal, a web-app map, and a story map.

3 Workflow environment

Important

You need a PC or Linux to use ArcGIS Pro!

  • If you don’t already have it, InstallGit

  • Sign in to OneDrive with the same account used to access PSI Microsoft Teams

  • If you don’t already have one, sign up for a GitHub account

4 Migrate existing project files

Existing files for the project live on PSI’s Microsoft Teams Critical Analysis channel at Protected Areas SLD 2026/

I started by selecting the folder’s three-dot ... More Options menu and clicking Add Shortcut to OneDrive/ My Files

I then opened my file explorer and navigated to the OneDrive folder for my University of Washington account. The UWT PSI - Protected Areas SLD 2026 folder was present in my OneDrive as a shortcut. The status symbol was a blue cloud, indicating I could only access the files when online, and they weren’t copied locally to my computer.

Warning

The entire folder is 2,652 files and 253 GB! The next step (syncing the files to keep them on my own computer) took a long time….

I right-clicked on the file and selected Always keep on this device

The GIS Analysis subdirectory contains the following subdirectories:

  • CA_DataExploration
  • CA_King_County_Parks_map
  • CA_KingCountyParks
  • CA_MPA_Round1
  • CA_RCO_Public_Lands
  • CA_WA_Bathymetry_CoNED_MLLW
  • NHDNat2Exploration

Each subdirectory contains the following files and subdirectories:

  • .aprx

    • ArcGIS Pro project file, It stores information about things such as:

      • maps

      • which layers are in those maps

      • symbology

      • labeling

      • layouts

      • connections to databases/folders

      • references to data sources

      • project configuration

      Crucially, it generally does not contain the actual spatial datasets.

      .aprx files are binary, so Git can tell that it changed but cannot show diffs. Two researchers editing the same .aprx can create merge conflicts that Git can’t resolve. The .aprx files can be included in the Git repo for version history and recovery, but not for collaborative merging.

      .aprx files are tracked in Git but are binary and cannot be merged reliably.

      Before editing an .aprx, confirm that no other active branch is modifying the same file.

      Keep .aprx changes isolated from unrelated code changes where practical.

      Pull/rebase from main immediately before beginning .aprx work.

      Complete and merge .aprx changes promptly rather than maintaining long-running branches containing .aprx modifications.

      Do not resolve .aprx merge conflicts using Git’s automatic merge tools. Determine which version is authoritative and manually reproduce the required ArcGIS Pro changes in that version.

  • .atbx
    • ArcGIS toolbox, contains geoprocessing tools
  • .gdb

    • Geodatabase file, where actual GIS datasets live, a directory containing many internal files.

      Large files

  • RasterFunctionTemplates/
    • definitions for raster processing
  • Index/
    • ArcGIS search cache
    • Not needed for reproducibility
  • ImportLog/
    • Logs of imported or converted material
    • Not needed for reproducibility
  • GpMessages/
    • ArcGIS history
    • Not needed for reproducibility
  • .backups/
    • versions for recovery after a crash
    • Not needed for reproducibility

5 Proposed .gitignore

**/.backups/