Terceira event-source discovery

Companion to the scan runbook. Last researched: 2026-09-04, Atlantic/Azores. This was a public-source discovery and quality pass, not an exhaustive signed-in Instagram/Facebook scan. A source is useful when it supplies new, verifiable local events or corrects existing details, not simply when its homepage loads.

Search procedure

1. Load existing events, recurring schedules, open PRs, and the last scan’s access gaps. Search the complete event file; its final date is not a coverage watermark.
2. Check official calendars first, then organizer programmes, then discovery services. Paginate until the requested date range is covered or record where access stops. Retain ongoing exhibitions and verified later announcements.
3. Use the query matrix below to look beyond the existing account inventory. Pair place names with the current year and each month in the next 90 days. Portuguese month names usually find more local announcements; repeat in English for visitor-facing and science sources. Date-search filters find recently indexed pages, not necessarily upcoming events.
4. Follow the organizer, venue, co-author, and collaborators from each useful result. Read the original detail page or flyer. Record the source’s role before promoting it to the runbook/resources pages.
5. Measure useful yield: new events, corrections, duplicates, held leads, and access gaps. A duplicate with a corrected distance or registration deadline still improves the data.

Query matrix

Replace {year} and {month} at scan time; do not keep a hard-coded previous year. Use quotes around place names, not an unqualified Terceira which also matches the Portuguese word for “third”. Start broad, then narrow by domain/account discovered in results.

On engines without Boolean support, split alternatives into separate searches. If results are sparse, try unaccented names and individual parish/venue names. A low-yield query is not evidence that no events exist.

Newly useful sources and recovered routes

These recommendations are based on observed pages, not just plausible source names.

Add AIR Centre, Bravos Trail, and Festing to the public EN/PT resource lists. Keep byAçores as secondary and use the recovered Portuguese route. None of these observations proves complete coverage of the platform.

Tested leads not promoted to a regular calendar

Additional source graph to investigate next: BioBlitz’s named partners include Gê-Questa, Marine Waste on Terceira Island, and BioMUST4All/Universidade dos Açores. These are leads to locate through the organizer’s links, not accounts claimed as scanned. Rotate parish committees and sports associations on subsequent passes rather than repeatedly searching only the existing bar list.

Evidence and acceptance rules

• Separate publication date, event date, registration deadline, and scan date. Resolve relative words from the original post timestamp in Atlantic/Azores. Require a year or unambiguous original context.
• Use the event’s actual location. Do not default a missing venue to Angra, and do not copy an organization’s footer address into an event.
• Prefer current organizer programmes over older tourism descriptions. Keep the original URL even when an aggregator supplied the lead. Two mirrors of one feed are one source of evidence.
• A closed registration period does not cancel a public race already listed, but its description must say registration is closed. Do not imply a selected-attendee workshop is a public drop-in event.
• Preserve confirmed weekly programming when fresh posts are inaccessible. Removing a recurrence requires affirmative evidence, not absence of search results.
• Qualify a source by a clear first-party role or at least two dated local event examples. Record access and usefulness separately: an authoritative source can still be inaccessible or have no verified upcoming programme.
• Keep unresolved candidates in the scan report, not in published event data. Do not add unverified dates, venue addresses, admission prices, or recurring schedules.

Coverage report template

For each actual scan, record:

Checked at: YYYY-MM-DD HH:MM Atlantic/Azores
Publication window: ... (overlap the previous successful scan by at least 14 days)
Event window: today through +90 days; ongoing and later verified announcements included
Source URL / role / access method / observed date range:
Status: reachable | partial | blocked | empty-with-window | untested | stale
Evidence: example event/detail URL; exact access failure if any
Yield: new N / corrections N / duplicates N / held N
Next action: original organizer to verify, unread page range, or retry route

empty-with-window means the relevant calendar/filter actually loaded and showed no events in the stated range. A 403, login wall, parser returning zero, or an unread page is not an empty calendar. For automated sources also record fetched/parsed/local-candidate counts when available; a green workflow alone does not establish source health.

Suggested ingestion follow-up, not implemented here: replace guessed CMPV feeds and homepage-only discovery with verified live-agenda adapters, report partial/source failures separately from zero events, and test island filtering, multi-session dates, and exclusive iCal end dates. Keep automated touradas with their existing workflow.