PACE OCI Icechunk Stores

This Icechunk repository provides virtual access to NASA PACE OCI Level-3 products for in-region AWS access. The data remain in the original NASA S3 bucket and are accessed through virtual references, avoiding duplication while providing a Zarr-compatible interface.

There are the following Icechunk stores:

• PACE_OCI_L3M_CHL (note has been changed to _BGS)
• PACE_OCI_L3M_RRS
• PACE_OCI_L3M_KD

Each L3 product has daily, monthly, and 8D at 2 different grids 0p1deg and 4km. The group names are:

• daily/0p1deg daily/4km
• monthly/0p1deg monthly/4km
• 8Day/0p1deg 8Day/4km

The PACE_OCI_L3M_CHL daily groups are separated into subgroups based on differences in chunking before 2026-02 and afterwards.

• daily/0p1deg/chunks_512 + daily/0p1deg/chunks_16 and daily/4km/chunks_512 + daily/4km/chunks_16

• chunks_512: Files with chlorophyll chunked as (time=1, lat=512, lon=1024).
• chunks_16: Files with chlorophyll chunked as (time=1, lat=16, lon=1024).

Each dataset includes a time coordinate derived from the start date in the filename, allowing it to be opened as a standard xarray dataset with xr.open_zarr().

Read in a dataset (requires you are in AWS us-west-2)

See pace-icechunk-examples.ipynb for more examples and plot examples. See environment.yml for the basic environment needed. NASA Earthdata requires authentication. Use earthaccess to get the s3 credentials.

ds = create_ds("PACE_OCI_L3M_RRS", "daily/0p1deg")
ds

Using the function below:

icechunk 1.x will not work. icechunk.http_storage does not exist in icechunk 1.x (it arrived in 2.0). Every icechunk 2.x release needs Python 3.12 or newer, so on an older Python pip install icechunk quietly installs 1.1.x, and the code below fails with AttributeError: module 'icechunk' has no attribute 'http_storage'. Check what you have: python -c "import icechunk; print(icechunk.__version__)".

import earthaccess
import icechunk as ic
import xarray as xr

def create_ds(
    product="PACE_OCI_L3M_RRS",
    group="daily/0p1deg",
):
    url = f"https://data.source.coop/fish-pace/pace-oci/inregion/{product}"
    
    storage = ic.http_storage(url)
    
    auth = earthaccess.login()
    s3_credentials = auth.get_s3_credentials(daac="OBDAAC")
    
    url_prefix = "s3://ob-cumulus-prod-public/"
    
    virtual_creds = ic.credentials.containers_credentials({
        url_prefix: ic.credentials.s3_credentials(
            access_key_id=s3_credentials["accessKeyId"],
            secret_access_key=s3_credentials["secretAccessKey"],
            session_token=s3_credentials["sessionToken"],
        )
    })
    
    store = repo = ic.Repository.open(
        storage,
        authorize_virtual_chunk_access=virtual_creds,
    ).readonly_session("main").store

    ds = xr.open_zarr(
        store,
        consolidated=False,
        group=group,
    )

    return ds