Sonora 2024 Data Processing

Sonora 2024 spectra are staged as source data under data/atmospheres/sonora/. The converter reads the staged spectra.zip archive and writes local Zarr/Parquet artifacts without deleting the source zip.

Use this page when you need to rebuild or validate the local Sonora cache. Use Atmosphere capabilities when you want the broader “which atmosphere libraries are available?” map.

Local layout¶

data/atmospheres/sonora/2024/
  provenance/
  raw/
    spectra.zip
  processed/
    catalog.parquet
    catalog_fragments/
    validation/
    sonora_2024.zarr/

Filename gravity is preserved as g_m_s2. The converter also records cgs logg as log10(g_m_s2 * 100) for compatibility with AtmosphereParams. Wavelengths remain in microns and fluxes remain in the released W/m2/m column.

That gravity conversion is easy to miss. Sonora filenames encode acceleration in m s^-2; stellar-atmosphere logg conventionally means log10(cm s^-2). The factor of 100 is the unit conversion from meters to centimeters.

Source-member policy¶

The local spectra.zip may contain macOS resource-fork sidecars:

__MACOSX/spectra/._*.spec

Those entries look like spectra by suffix, but they are not model spectra. The converter skips __MACOSX/ and ._* entries, records the skipped member names in validation JSON, and reports both valid and skipped counts during dry-run.

The current staged archive reports:

valid_count=1440
skipped_count=4

The processed catalog count is therefore 1,440, not the raw archive’s 1,444 .spec-like entries.

Conversion¶

Dry-run first:

env -u VIRTUAL_ENV uv run --no-sync --extra data python \
  scripts/convert_sonora_2024.py --dry-run

Convert the staged archive:

env -u VIRTUAL_ENV uv run --no-sync --extra data python \
  scripts/convert_sonora_2024.py --overwrite

The source zip is not deleted. Validation JSON records readback status, float32 roundoff, the source zip hash, and member provenance.

Validation¶

After conversion, run the local artifact validation:

env -u VIRTUAL_ENV uv run --no-sync --extra data pytest -q \
  tests/validation/test_atmospheres_local_artifacts.py

The Sonora portion asserts:

catalog.parquet exists and has 1,440 rows.
sonora_2024.zarr exists and sampled flux values are finite.
wavelength unit is micron.
flux unit is W/m2/m.
cgs logg spans approximately 3.49..5.50.
spectra.zip is still present.
validation JSON records valid_count=1440, skipped_count=4, readback_ok=true, and raw_deleted=false.

Runtime status¶

Sonora is processed and visible to AtmosphereLibrary.from_local("data") as sonora_2024. At the time of this document, it is artifact-ready but does not yet have an implemented SonoraBackend. A library query can report Sonora coverage, but AtmosphereLibrary.spectrum(...) should not delegate to Sonora until the backend interpolation policy is added and tested.