🌐 GeoJupyter 🌐
@
🌥️ CNG ’25 🌥️

Exploring more approachable geospatial data workflows as a community

Matt Fisher, Schmidt DSE @ UC Berkeley

👋 Hi, I’m Matt!

💻 Research Software Engineer (RSE)

🧑‍🤝‍🧑 Community Engagement Manager

🌱 Working at Schmidt DSE @ UC Berkeley

❄️ Previously at National Snow & Ice Data Center (NSIDC)

👐 Open source maintainer: earthaccess, viscm, several conda-forge feedstocks

🐶 🌱 🎶 🎹 🥁

I’m Matt Fisher, and I’m a Research Software Engineer and Community Engagement Manager for GeoJupyter.

Before GeoJupyter, I worked in building software for research support and data analysis, visualization, management, and access at the National Snow and Ice Data Center.

I’m a maintainer of several open source projects, and maybe we’ve spoken on GitHub before! earthaccess is one of my favorite community-owned projects I help maintain. It enables access to NASA data in 3 lines of code, which is a huge improvement: before earthaccess you had to write >100 lines to authenticate and interact with the NASA earth science data API.

I also love dogs (if you see one, let me know 😁), gardening, and amateurishly playing music on drums and keyboards 😄

🤝 I want to talk to you!

🗨️ Ask me about…

🧑‍🤝‍🧑 Community engagement stuff

👐 Open source stuff

💪 Accessibility stuff (earthaccess)

📦 New Python project/environment management tools (uv, pixi)

👂 I want to ask you about…

☕ JavaScript map engines & cloud-native data support

🕳️ What gaps or friction you see in the open source geospatial ecosystem

🧩 Do you have a project that you feel fits in this community?

🛑 What’s stopping your team from trying JupyterGIS?

mfisher87.github.io/presentation-cng2025/

I have lots of stuff I want to talk to you all about! I hope we can find each other later and chat 😄

If you want to check out these slides and follow the links, they are publicly available online from this QR code or URL.

I’m not going to go through this slide now, but I’ll show it again at the end.

In this talk…

• What is GeoJupyter?
• Why GeoJupyter?
• What is JupyterGIS?
• How you can join the community

mfisher87.github.io/presentation-cng2025/

In this talk, I’m going to introduce the GeoJupyter community and discuss why it exists. I’ll then introduce and demo the JupyterGIS software. Finally, we’ll talk about how you can join the community.

Thank you

But first, I’m representing a lot of folks today. I don’t own this, our community does. I’d like the thank the many collaborators involved in this project. So many people and organizations have come together to make this happen!

Special thanks to QuantStack for leading the development of JupyterGIS, and ESA for funding them.

And personal thanks to Fernando Pérez and the Schmidt Center for Data Science & Environment for supporting me to work on this project!

I’d like to spend time individually thanking everyone, but then my time would be over 😄

One last big thank you to all the amazing role models in these communities who have inspired me to be a better community manager and developer.

Thank you all.

❓ What is GeoJupyter?

---

GeoJupyter is an open and community-owned effort to reimagine geospatial interactive computing experiences to enable more people to confidently engage with geospatial data.

---

What is GeoJupyter?

Read the slide

Jupyter is a mature community with millions of users that has already made huge strides in making programming more accessible to the world.

And it provides a powerful architecture for interactive and real-time collaborative development, which is why we’re building on top of it.

Our mission is intentionally broad. What does it look like to you to enable more people to confidently engage with geospatial data?

We want to hear from you, and I’ll share how you can do that at the end.

Our values

👐 Open source & open science - geospatial data is important to everyone!

🤸 Approachability and playfulness, like desktop GIS tools

🪶 Flexibility and reproducibility, like coding methods

🎭 Collaboration and storytelling, like Jupyter Notebooks

Open source and open science are critical values for us; we want to democratize geospatial data practice, so everything we do is free and open.

And it’s important to us to create tools that combine the strengths of many amazing existing tools.

Desktop GIS tools offer approachability and playfulness.

Coding methods offer flexibility and reproducibility.

And the Jupyter ecosystem offers collaboration and storytelling.

You got your chocolate in my peanut butter!

We believe these values belong together!

But it’s not going to happen by accident, like the Reese’s Peanut Butter Cups commercial 😄 Lots of work needs to be done!

🧑‍🤝‍🧑 It’s a communtity!

GeoJupyter is not software; it’s a community which will build many things!

GeoJupyter is a community; our interest includes but is not limited to what many think of as “GIS”.

JupyterGIS is our flagship project, led by QuantStack; a GIS platform in JupyterLab that can be used in similar ways to QGIS plus more. We’ll talk more about JupyterGIS shortly.

But our work is far from limited to JupyterGIS; not everything we build is going to be a huge project or GIS-first. It depends on what our community (that’s you!) tells us they need.

🤷 Why GeoJupyter?

🤕 Existing workflows are sometimes painfully imperfect:

• Even experts often struggle
• Linked visualizations are difficult to build
• Friction when visualizing from a programming environment

Why are we doing this?

Because working with data about our planet is hard.

What we’re bringing to the table is a holistic and human-centered approach to evaluating and improving this landscape.

So far we’ve conducted dozens of hours of user interviews to understand what folks need most, and we’ll continue to use the input of real people to guide our priorities.

The following are only a few examples of challenges we’ve heard from those practitioners.

Many of these challenges have existing solutions, but they often involve cobbling together multiple tools, introducing workflow friction.

🤷 Why GeoJupyter?

🤕 Existing workflows are sometimes painfully imperfect:

• Even experts often struggle
• Linked visualizations are difficult to build
• Friction when visualizing from a programming environment

Geospatial is hard, even for experts.

We all struggle sometimes.

Brianna’s story

When our community burned, where was the satellite information?

…we are failing the people and causes that need us the most.

– Brianna Pagán, Technical Program Lead @ Development Seed, ex-NASA Deputy DAAC Manager

Brianna, who you may have heard speak yesterday, lost her home in the recent fires in California. When she learned her home was at risk, her first thought, like a true scientist and geospatial nerd, was to look at the data. Recounting these events, she wrote:

I should be able to do this. I have a PhD in Remote Sensing, I have built satellite data processing pipelines, I help manage one of the NASA satellite data centers, I teach courses on climate resilience, adaptation and geoinformatics. […] and yet the irony is that with all my experience I could not access the information I knew existed to help our community.

End-quote. You can read Brianna’s full story by following the link at the top of the slide.

🤷 Why GeoJupyter?

🤕 Existing workflows are sometimes painfully imperfect:

• Even experts often struggle
• Linked visualizations are difficult to build
• Friction when visualizing from a programming environment

Another pain point we hear is that folks want to visualize the same data in different ways, and have those visualizations interactively linked.

Linked visualizations

For example, you might want to visualize soil samples on the map, but also scatter plot those soil samples to find relationships between those samples’ attributes.

You may find spatial correlations you wouldn’t have otherwise noticed by interacting with variable correlations, and following those correlations back to a map.

There’s a package called “glue” that can do this, and it’d be really cool to integrate that functionality with a GIS-first interface.

🤷 Why GeoJupyter?

🤕 Existing workflows are sometimes painfully imperfect:

• Even experts often struggle
• Linked visualizations are difficult to build
• Friction when visualizing from a programming environment

Another challenge that we’ve heard widely reported by users is friction when visualizing data about our planet from a programming environment, like a Jupyter Notebook.

The process users described has many repeated steps, like a dance. We heard this so often that we gave it a name: “The GIS Bounce”.

💃 The GIS Bounce 🕺 (hard mode)

🧮 Do some analysis

💾 Write out a data file

📥 Download it to local machine

🗺️ “Bounce” to QGIS, load basemap, load data file

👀 Validate, inspect, explore

🚿 “Bounce” back to Notebook and repeat

To do the GIS bounce:

First, do some analysis. Then, write out a data file to check your work so far. If you’re working in a cloud environment, download that data file. “Bounce” over to QGIS, load up a basemap, and then finally load your data. Now you can do your validation and “bounce” back to your notebook, and repeat!

That’s a lot of steps, and they are repeated every time a practitioner does a critical task: checking their work.

Checking our work shouldn’t be frustrating and time-consuming, it should be seamless, joyful, and rewarding.

💃 The GIS Bounce 🕺 (easy mode)

🧮 Do some analysis

🐍 “Bounce” to JGIS: one line of code!

(Imports don’t count 😏)

from jupytergis import explore

explore(my_geodataframe)

👀 Validate, inspect, explore

🚿 “Bounce” back to Notebook and repeat

So we developed an interface for this in JupyterGIS, a function called explore(). It’s still rough around the edges, but it’s available now! 🎉

It enables the user to pass geospatial data directly to a visualization environment with one line of code.

We’ll show more shortly!

🤷 Why GeoJupyter?

🤕 Existing workflows are sometimes painfully imperfect:

• Even experts often struggle
• Linked visualizations are difficult to build
• Friction when visualizing from a programming environment

And much, much more!!

• Visualizing features of vertical landscapes
• Repeatable GIS layer structure and symbology
• ❓

The landscape of open source geospatial tools is far from perfect.

There’s lots more to do, and we want to hear what’s important to you.

JupyterGIS

‼️ Early development ‼️

Let’s talk about some of out work in progress now!

JupyterGIS is the GeoJupyter flagship project, with development led by the amazing folks at QuantStack.

What is it? How is it different?

Read the slide

• Interactive GIS environment
• JupyterLab- and browser-native (try it in a JupyterHub!)
• Python API
• Supports QGIS project files
• Coming soon: STAC browser!
• Real-time collaboration: like Google Docs, but on a map!

🏠 All under one roof (JupyterLab)!

Demo

💡 Lite: https://jupytergis.readthedocs.io/en/latest/lite/lab/

🧪 Lab: http://34.71.209.73:8888/

Open in private window: http://34.71.209.73:8888/

First, let’s look at “The GIS Bounce” we talked about earlier.

• Open explore.jgis.
  • This notebook explores a dataset about the discrimatory US government practice of “redlining” from the 1930s-1960s. Redlining was a practice of grading neighborhoods to financially exclude people on the basis of race.
• Run the notebook cell by cell, except the final cell
  • We’ve got some data and we want to validate it before continuing our analysis..
  • Calling .plot() doesn’t get us very far; it would be challenging to verify the data is properly geolocated without a basemap.
  • We would need to write more code to inspect the features’ attributes.
  • That’s where the explore() function we discussed comes in!
  • With only one line of code, excluding imports, we can interact with this data much like we would in QGIS.
• Run final cell. Right click the “New Haven” layer, zoom to it.
  • The geolocation looks good.
  • Let’s inspect the features and be sure those look the way we expect, too.
  • We can see that this neighborhood was ranked “A” (“best”) because it was mostly white, meaning that home loans were more readily available, leading to generational harm.
  • Now that we’ve explored our data, we can go back to our example analysis and continue. We could close this panel and re-open it again later.
  • Or, because this visualizer is in a JupyterLab window, we can move it around and scale it however we like, leaving it available for more visualization later.
• Move the window to a bottom pane and scale it way down.

Let’s explore some example JupyterGIS projects.

• Open france_hiking.jGIS from JupyterLab file explorer.
  • This example JupyterGIS project file and many more are distributed with JupyterGIS, and you can check it out in your browser with JupyterLite from the “Try it now!” link in our README and docs.
  • This map shows hiking trails in France.
  • It’s a bit noisy right now, so let’s get rid of this MODIS layer.
• Right click NAGAGIBS layer, remove.
  • Let’s add a layer that shows where Strava users are running.
• Open the layer browser.
  • Our layer browser has a bunch of pre-defined layers you can add to your map
• Add “Strava.Run”.
  • The hiking layer is a bit hard to read now, though, so let’s change the layer order.
• Move “Strava.Run” underneath “Waymarked”.
  • Still a bit hard to read, let’s change the opacity.
• Set opacity of “Waymarked” to 1.
  • That’s better!

Let’s change it up and look at some earthquakes data.

• Open earthquakes.jgis.
• Activate the identify tool.
  • We have an “identify” tool that enables you to view attributes of these features.
• Select a feature.
  • We can see the earthquake’s magnitude and timestamp. It’d be nice to visualize this on the map.
• Open symbology for earthquakes.
  • Let’s set the symbology to “graduated” and select the attribute we want to symbologize on 😄
  • We want to use color to represent the magnitude, and use a quantile scale.
• Click “classify”.
  • Beautiful.
  • Now, I want to view this data over time.
• Open the time slider.
  • Let’s select the correct time attribute and set a time step of “daily”.
• Scrub with slider.
  • And let’s animate it at 10 FPS.
• Select 10FPS, press play.
  • Someone mentioned the other day that it would be cool to persist the past points when animating to see accumulation. Love that idea, and we have an open issue in GitHub where you can follow along.

Collaborative demo:

• Nick: Join collaborative session on earthquakes.jGIS
• Nick: Add WaymarkedTrails.Slopes layer
• Nick: Disable the earthquakes layer
• Nick: Add an annotation at the base of a trail “Let’s meet here at Aerial Tram on Friday”
• Matt: Looks someone’s trying to collaborate with me! Who could that be? 🤔
• Nick: Raise hand!
• Matt: That’s Nick Gondek, my co-worker. Looks like he’s having trouble focusing on my talk and looking forward to some skiing :)
• Matt: Nick is shown here as a collaborator on this JupyterGIS project.
• Matt: Follow Nick.
• Matt: I’m using the “Follow” feature to see what Nick is doing.
• Matt: I see Nick has added an “Annotation”, our name for a geospatial conversation in JGIS.
• Matt: Respond to annotation: “Sounds awesome! Meet you there.”
• Nick: Add another annotation for a run from the top of Aerial Tram, let’s take Chip’s run down to the hotel.
• Matt: Respond to annotation: Siiick!
• Matt: I think we’re done planning our post-conference recreation, I guess let’s finish up this talk 😆

⚠️ If something goes wrong ⚠️

• That’s what I get for doing a live demo.
• I knew a live demo with collaborators was playing with fire. I’ve flown too close to the sun, sorry for my hubris 😆
• Open browser debug tools. Hold on, I can fix this!! Just kidding, let’s move on 😄
• Come talk to me later and I’ll tell you or show you more 😄

Demo recording (just in case)

How you can join

🎙️ Sign up for an interview!

📆 Join a hackathon or community meeting!

🗨️ Chat with us on Zulip!

🧪 Try JupyterGIS! Where does it (not) meet your needs?

🎁 Report bugs, request features! Open Pull Requests!

😎 Share your rad vibes and leadership!

💰 Fiscal sponsorship!

💌 Contact me: Zulip, GitHub, email

I’m sure you’re wondering “how can I be a part of this rad community?”

There are lots of ways you can join our community and be a part of this mission.

We’d love to talk to you about your needs, frustrations, and opinions in an interview.

It would be amazing to see you at one of our bi-weekly hackathons or monthly community meetings! You can view our community calendar on geojupyter.org or by following the link on this slide.

We also have a community chat on Zulip, which is like an open-source Slack. You don’t even need to register an account to view our activity, but we would love it if you joined in 😄

JupyterGIS is available for install with pip, conda, or equivalents, and we’re jazzed to hear what you think 😄

We’re excited to expand our funded development team, so if your company is interested in sponsoring this work, please reach out. We are still exploring what this will look like, and while GeoJupyter will always be community owned, we’re ready to work with folks whose goals align with ours.

If you have any questions, you can always reach out to me on Zulip, GitHub, or email.

❤️ Thank you! ❤️

Thanks for your time and attention, and I hope to see you around 😄

Advance to next slide

🤝 I want to talk to you!

🗨️ Ask me about…

🧑‍🤝‍🧑 Community engagement stuff

👐 Open source stuff

💪 Accessibility stuff (earthaccess)

📦 New Python project/environment management tools (uv, pixi)

👂 I want to ask you about…

☕ JavaScript map engines & cloud-native data support

🕳️ What gaps or friction you see in the open source geospatial ecosystem

🧩 Do you have a project that you feel fits in this community?

🛑 What’s stopping your team from trying JupyterGIS?

mfisher87.github.io/presentation-cng2025/

🎉 Bonus slides

FAQ

• 😐 What about <existing tool>?
  • 🔥 We understand we’re the “new kids on the block” and we’re coming in hot.
  • We are interested in building on the amazing progress of recent decades and learning lessons from modern users.
  • Understanding geographic space and scale is centuries in the making, bigger than any company or project, including ours.
• ⏱️ Timeline?
  • ✨ JupyterGIS available to install right now!
  • 🤷 It depends! We’re listening to users to identify priorities.
  • STAC browser expected this year.