Every scientific source must have a stable identity, access basis, licensing or reuse note and provenance trail. Sources that do not permit the intended production use are rejected or kept reference-only instead of being scraped silently.
Record identity and provenance
Every scientific record should be traceable to source_id, station_id or another stable source identity, observation time and processing information. Derived values and authored indices additionally preserve algorithm version, parameter hash and exact input-record identities so the transformation can be audited.
Reuse rights before automation
A source does not become an automated dependency merely because it is visible in a browser. Production ingestion requires an access basis and reuse terms compatible with the intended GeoLira function. When rights are unclear the source can remain a bibliographic or validation reference rather than being silently scraped.
Canonical history and user-facing attribution
Canonical storage may retain successive revisions of one event for audit. User-facing views select the newest accepted revision of the logical event so, for example, repeated USGS corrections do not appear as separate earthquakes. Source links help users verify provenance, but an institutional name or DOI does not replace freshness and quality metadata for the specific record. GeoLira does not present external material as its own measurement.