Skip to content

Digest the retired DSCI Confluence space into the KB#233

Open
t-downing wants to merge 1 commit into
mainfrom
confluence-digest
Open

Digest the retired DSCI Confluence space into the KB#233
t-downing wants to merge 1 commit into
mainfrom
confluence-digest

Conversation

@t-downing

Copy link
Copy Markdown
Collaborator

Companion to ds-knowledge-base-internal#1 (the raw archive) and #232 (the pointer page).

All 132 archived Confluence pages were triaged against the live KB: 64 historical (meeting notes, work plans, landing pages), 35 superseded (mostly by the generated db-schema.md/pipeline-registry.md and existing pipeline/app pages), 30 with still-relevant content, 3 sensitive (stay internal). Verdicts + superseded_by pointers are stamped on every archived page in the private repo (confluence/TRIAGE.md is the table), so the archive can't be mistaken for current guidance.

This PR ingests the public-safe keepers (~325 lines across 24 files):

New pagesdatasets/ecmwf.md (incl. the valid_date-is-END-of-window and ERA5-Land accumulation gotchas, C3S anomalies product), datasets/glofas.md (forecast≠reforecast initialization gotcha), datasets/ibtracs.md (provisional/best-track semantics, 1-min vs 10-min winds, La Réunion gap), datasets/jrc-asap.md (full season/warning methodology, distinct from FAO ASI), datasets/chirps-gefs.md, methods/disputed-territories-mapping.md (India IND+xAP / J&K convention).

Extended — raster-stats (methodology rationale + admin-boundary quirks), raster-pipelines (ERA5T/expver, SEAS5 MARS→AWS, v/i/lt semantics; one flagged 10-vs-11-day baseline discrepancy TODO), hdx-signals (box testing gotchas, console triage, new-indicator checklist), comms-listmonk (deployments/roles/upgrade dance/design history), databricks (cluster how-tos), storage (four-account model, raw/processed/tmp, COGs-over-Zarr ADR), conventions (git/notebook/R/results-review/repo-hygiene), datasets README (derived-product shareability matrix), emdat (manual refresh recipe), license lines on floodscan/imerg/nhc/storms, trigger-design (CERF reporting table, spec-form timing), mmr-cyclones (IMD table, rainfall-join method), database (consolidation note).

Sanitized: no personal names, internal Drive/Sheets URLs, subscription IDs, or access matrices; nothing propagated from the archived public.floodscan page (its column descriptions are wrong); stale DSCI_AZ_SAS_* env names not carried forward.

🤖 Generated with Claude Code

All 132 archived Confluence pages were triaged against the live KB
(verdicts stamped on the archive in ds-knowledge-base-internal):
64 historical, 35 superseded, 30 with still-relevant content, 3 sensitive.
This ingests the public-safe keepers:

New pages:
- infrastructure/datasets/: ecmwf.md (SEAS5/ERA5 + valid_date & ERA5-Land
  gotchas + C3S anomalies product), glofas.md (forecast-vs-reforecast
  initialization gotcha), ibtracs.md (provisional/best-track semantics,
  wind-averaging, quirks), jrc-asap.md (season/warning methodology),
  chirps-gefs.md
- methods/disputed-territories-mapping.md (OCHA brand rule + the India
  IND+xAP / J&K implementation decision)

Extended:
- pipelines/raster-stats.md (methodology rationale: whole-pixel vs
  exactextract, rasterize-once caching; admin-boundary quirks)
- pipelines/raster-pipelines.md (ERA5T/expver, SEAS5 MARS->AWS rationale,
  v/i/lt filename semantics; FloodScan baseline 10-vs-11-day TODO)
- pipelines/hdx-signals.md (box testing/lint gotchas, console triage,
  adding-a-new-indicator checklist)
- infrastructure/comms-listmonk.md (two deployments, roles, image-pin +
  upgrade procedure, design history)
- infrastructure/databricks.md (cluster how-tos, operating conventions)
- infrastructure/storage.md (four-account model, raw/processed/tmp,
  access tiers, COGs-over-Zarr format decision)
- infrastructure/conventions.md (git/PR workflow, notebooks, R tooling,
  results review, repo-hygiene checklist)
- infrastructure/datasets/README.md (derived-product shareability matrix)
- emdat.md (manual refresh recipe), floodscan/imerg/nhc/storms pipelines
  (license lines), methods/trigger-design.md (spec-form timing, CERF
  reporting table, portfolio bookkeeping), frameworks/mmr-cyclones
  (IMD classification table, rainfall-join method), database.md (DB
  consolidation note)

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant