This Dataset repo is the lightweight index and documentation surface for the
Himawari AHI ARCO collection. The large Zarr payloads live in the public
Hugging Face Storage Bucket:
text
hf://buckets/Gary0404/himawari-arco/arco_v1/
The repo keeps small metadata files such as *.manifest.json and
*.coverage.json. New formal ARCO payloads should be published to the bucket,
not to this Dataset repo.
Available Bucket Payloads
text
B03:
bucket prefix: hf://buckets/Gary0404/himawari-arco/arco_v1/B03/
months: 2023-01 through 2025-11
chunks: time/y/x = 6 x 1024 x 1024
B08:
bucket prefix: hf://buckets/Gary0404/himawari-arco/arco_v1/B08/
months: 2023-01 through 2025-12
chunks: time/y/x = 6 x 1024 x 1024
B13/B09/B10 formal products are planned as bucket-first products after canary
validation. Pilot payloads may still appear in this repo until they are either
migrated or retired.
{band}(time, y, x): Himawari AHI band values
source_present(time): 1 if a source observation exists for the target time
source_time_index(time): compact source time index, or -1 for missing source time
time/y/x: coordinates
spatial_ref: geostationary projection metadata
Missing source times are kept on the regular 10-minute UTC time grid. For those
times, the band array reads as NaN, source_present is 0, and
source_time_index is -1.
Repository Role
This repo is intentionally small:
text
keep here:
README.md
.gitattributes
*.manifest.json
*.coverage.json
small validation or example files
do not keep here as the steady state:
full monthly *.zarr payload directories
The large B03/B08 Zarr payloads have been copied and validated in the bucket.
They are removed from this repo so the Dataset repo remains a lightweight
catalog instead of a second large storage copy.