# Lux Data License Policy

Code and static-site glue are MIT licensed. Data records keep their own license fields and provenance.

## Included in lux-data 1.3.0

- `ossl-visnir-soils-10k`: measured OSSL VisNIR soil spectra, CC-BY-4.0.
- `plant-chlorophyll-crops`: measured crop chlorophyll/index records, CC-BY-4.0.
- `lux-computed-minerals`: Lux computed mineral priors, CC0-1.0.
- `lux-tddft-small-molecules`: Lux PySCF TDDFT calculations, CC0-1.0.
- `lux-uvvis-compound-id-v0-training`: frozen 1,060-record computed UV/VIS training-input release, CC0-1.0.

The compound-ID training release is computed UV/VIS data, not measured Raman data. It is research-only and carries the exported-model leakage warning in its manifest.

## Active source campaigns

- `lux-tddft-santander-field`: active Santander PySCF TDDFT source lane; its compound-ID v0 snapshot is frozen in the training-input release.
- `lux-tddft-santander-gpu-refined`: active DGX GPU4PySCF source lane; its compound-ID v0 snapshot is frozen in the training-input release.

## Listed but not mirrored by default

- Open Specy is CC-BY-NC-4.0, so it remains a non-commercial research connector.
- Mendeley Coffea arabica chlorophyll data is CC-BY-4.0 but API/file access returned 403 from this host, so it is listed for manual/user-auth import.
- GLCC Figshare yearly LCC archives are CC-BY-4.0 but many gigabytes per year, so they are listed as source metadata rather than bundled by default.
- MuSyQ Global LCC is a Google Earth Engine workflow target rather than a static local mirror in this release.
- USGS SPLIB07 is public domain and remains the next measured mineral import once the current ScienceBase file endpoint is reachable.

Every normalized spectrum must include `provenance.source_id`, `provenance.source_url`, `provenance.license_id`, `provenance.citation`, `provenance.redistribution`, and `provenance.not_measured`.