JupyterGIS sync meeting

🙏 Our apologies, there’s no time for introductions!

We’re excited for you to join us, but this meeting is short and task-focused. Please attend community meetings (see GeoJupyter calendar) to meet the team and add your own agenda items, and/or introduce yourself on Zulip!

JupyterGIS sync meeting (2026-10-20)¶

Please add new agenda items under the New agenda items heading!

• Join us on Google Meet

  • What time is the meeting in my time zone?

• Previous meetings

Attendees¶

Your name / GitHub ID / affiliation

• Matt / @mfisher87 / DSE

• Martin / @martinRenou / QuantStack

Agenda¶

• Review our project board

  • What items should be added?

  • Are there stale items that are no longer urgent?

  • Are there things we can change about the project board to make it more useful? Add more information? Remove steps?

• Trail

  • Monitor the many tools that practitioners use in their typical workflow and produce a provenance across tools to enable greater reproducibility, repeatability, and understanding when using multiple tools (some GUI-based).

  • Starting with prototyping forwarding JupyterGIS signals -> Jupyter Server event bus, and subscribing over websocket: https://github.com/geojupyter/prototype-jupytergis-eventfeed

  • Martin: https://github.com/notebook-link/metrics

  • How do we

• Maplibre viewer in geojupyter/jupytergis#1875

  • deck.gl? :P Gains projection support in v10! Can be used in combination with MapLibre. IMO, we should have a deck.gl + MapLibre adapter -- adapters shouldn’t be 1:1 with individual renderers.

    • COG support? https://developmentseed.org/deck.gl-raster/blog/multi-band-cog/

      • https://github.com/geomatico/maplibre-cog-protocol , https://maplibre.org/maplibre-gl-js/docs/examples/add-a-cog-raster-source/ ,

    • https://developmentseed.org/deck.gl-raster/examples/reprojection-explainer/

  • Need a way to add “warning” to layers -- we want to show e.g. “this layer won’t display with this renderer” or “this layer is ephemeral because it’s backed by an Xarray dataset in kernel memory”!

• OpenEO:

  • New example examples/99-OpenEO-NDCI_venice_lagoon-Microsoft.ipynb uses the same backend (titiler-openeo) as https://studio.mpc.ds.io/

  • auth: Arjun looking into supporting OIDC

  • CORS: JupyterLab -> use our own proxy. JupyterLite -> ??. Notebook.link -> We’ll provide our own.

  • tiff tiles -> to enable symbology

• GEE viz/analysis feedback loop (CNG)

  • Presented at CNG: https://github.com/geojupyter/presentation-cng2026/

  • Hypothesis: Users want GEE-like viz/analysis feedback loops. Test your computation quickly on a large dataset. GEE does this by using overviews efficiently, not by super-impressive hardware, i.e. if your map is zoomed out, do the computation on the fly on lowest-resolution data. The CNG/OSS ecosystem is almost able to do this, but we are missing some building blocks. Namely, Xarray datasets are not overview-aware, and can’t do overview-aware computations. If our tile server could take a dataset that contains deferred computations backed by a COG, e.g. ndvi = (ds["nir"] - ds["red"]) / (ds["nir"] + ds["red"]) and do e.g. ndvi[myslice].compute(overview_level=3), we could provide a GEE-like UX with only OSS tools.