Cloud-atmosphere source profile

cloud-atmosphere-profile.json is the authoritative Stage 4.1a source, freshness, aggregation, and display configuration.

Normal generation uses a locally prepared snapshot at .prepared/cloud-atmosphere-latest.geojson. Raw COG and NetCDF inputs and the prepared snapshot are intentionally not checked in because they are large, mutable refresh products. P-Tree requires a user account; credentials remain in the operator’s .netrc and are never copied into this tree. Publication must retain the JAXA/EORC and applicable upstream credit required by the current source terms.

The checked fixture under fixtures/ exercises the schema and renderer. It is synthetic, is identified as a fixture in its metadata, and must not be presented as an observation.

For first-time production setup, follow the credentialed P-Tree download quick start. It covers registration, safe .netrc configuration, a read-only login test, the complete refresh sequence, expected outputs, and troubleshooting.

Refresh workflow:

Configure a .netrc entry for ftp.ptree.jaxa.jp.
Run make fetch-cloud-atmosphere-data.
Run make prepare-cloud-atmosphere-data; it verifies every staged raster digest before aggregation.
Review .prepared/cloud-atmosphere-latest.manifest.json and the reported layer ages and coverage.
Run make generate-cloud-atmosphere.

The process-start UTC instant controls calculated solar geometry. Each source observation retains its own interval. Every enabled layer uses the latest published observation ending no later than process start. Its exact end UTC and age are shown, and latest available is added when publication exceeds the layer’s preferred freshness target. This makes a verified, authorized offline snapshot reproducible without silently turning its age target into an expiration date. A missing cell or property means unobserved, not clear sky and not zero.