GeoZarr
February 2025 (subjective) status report
Goals
- Reinvigorate Building It Out!
- 🙅‍♂️💠All benefits from discussion are exhausted
- 🚀 GeoZarr can be successful if we act now
- 💔 GeoZarr can fail if we don’t
What is Zarr?
An community-developed, open-source format for the storage of chunked, compressed, N-dimensional arrays
- Cloud optimized (concurrent read/writes, “infinitely” scalable)
- Cross-disciplinary (originally developed for bio-imaging)
- Extremely extensible (supports any key-value store)
What’s going on today?
- Zarr is growing in popularity
- Icechunk is improving consistency and performance
- VirtualiZarr is simplifying virtualization
- Progressing towards Zarr specification V3 across languages
Zarr implementation landscape
What is GeoZarr?
A geospatial extension for the Zarr specification, formalizing metadata expectations.
Why is this an exciting time?
Multi-dimensional geospatial solutions can bring real impact across humanitarian, scientific, and industrial needs
Why is this an exciting time?
Experimental CRS handling across the Python ecosystem
Why is this an exciting time?
Generic support for analytic/functional coordinates in Xarray
Why is this an exciting time?
Draft ZEP for breaking the stalemate on building Zarr extensions
Why is this an exciting time?
We have a deadline!
What do we need?
To prove whether OGC TMS 2.0 is sufficient for performant and reliable rendering and analysis
What do we need?
To push forward Xarray functional coordinates for geospatial
What do we need?
To demonstrate how XPublish and Titiler’s Xarray extension fit together (or not)
What do we need?
To show how to get the most out of Zarr and STAC
