#52: NetCDF Was Built for Supercomputers. Zarr Is Built for the Cloud.

Share

NetCDF/HDF5 treat data as a single file with complex internal pointers. Updating one time-step often requires rewriting the whole file or managing distributed locks. Zarr inverts this: it treats data as a directory of independent chunks on S3.

NetCDF is monolithic. You have a single file containing temperature measurements for every grid point, every altitude, and every time-step. Modify time-step 100? Traditional workflows either rewrite the entire file or use complex journaling. Parallel writes collide. Locking becomes the bottleneck.

Zarr explodes the array. The same dataset becomes a directory structure: temperature/0/0/0.bin, temperature/0/0/1.bin… each chunk is a separate object on S3. Metadata (dimensions, data types, compression) lives in lightweight JSON files. Data lives in independent binary chunks.

This changes parallelism fundamentally. A hundred machines can write a hundred different time-steps simultaneously. No coordination. No locking. Each chunk is a separate object—S3 handles the atomicity.

The query pattern determines chunking. Climate models chunk by time (fetch all space, one time-step). Satellite data cubes chunk by space (fetch all bands, one region). The same data, organized for different access patterns.

Zarr handles N-dimensional arrays natively: Time × Elevation × Latitude × Longitude. NetCDF forces you to manage those relationships implicitly. Zarr’s chunking strategy is explicit and optimizable.

The rule: If it has a time axis, it probably belongs in Zarr. Use COG for 2D imagery. Use Zarr for everything else—weather forecasts, climate models, hyperspectral time-series. NetCDF still works, but Zarr scales on the cloud.

Whenever you’re ready, here are 4 ways I can help you grow in GIS & spatial data:

​Spatial Lab​ – My private community where GIS professionals, data engineers, and analysts connect, swap workflows, and build repeatable systems together.

​Modern GIS Accelerator​ – A guided program to help you break out of legacy GIS habits and learn modern, scalable workflows.

​Career Compass​ – A career-focused program designed to help GIS pros navigate the job market, sharpen their pitch, and find roles beyond traditional GIS paths.

​Sponsorship​: Interested in sponsoring this newsletter (or other content)? ​Learn more here​ and fill out the form to get in touch!

#51: You Don’t Choose Between Vector Tiles and PMTiles

Prev

#53: There Is No Correct Chunk Shape, Only the Shape That Matches Your Query

Next
Comments
Add a comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Get every update, in your inbox.
Get every update, in your inbox.
Get every update, in your inbox.
One tip, every day
Get every update, in your inbox.
Subscribe below and join 11,000+ others learning modern GIS.