Supported formats

Zarr

Browzarr reads Zarr stores through zarrita.js:

• Zarr v2 and Zarr v3 metadata formats
• Remote stores over HTTP(S) (S3, GCS, static hosting, ...) — see Remote data
• Local .zarr directories via the offline apps
• Consolidated metadata (.zmetadata) speeds up the initial parse when present

Codecs

Common compressors and filters are supported (e.g. blosc, gzip, zlib, shuffle, astype). If a chunk fails to decode, the codec is likely missing — open an issue with the store URL.

Icechunk

Icechunk repositories can be opened remotely from the catalog or by URL. Version selection is available when a store exposes multiple versions.

NetCDF

.nc files are decoded in-browser with netcdf4-wasm (WebAssembly):

• NetCDF-3 (classic) and NetCDF-4 (HDF5-based)
• Chunked (HDF5) files perform best; contiguous large files are slower

Chunking guidance

• Aim for 1–10 MB per chunk
• Align chunks with the slicing pattern you expect (e.g. one time step per chunk for animation)
• Avoid many tiny chunks — per-request overhead dominates