Evidence & sources

Every dataset used anywhere on Pacific Pulse is listed here, including its licence, update frequency, resolution, and known limitations. If a field is blank, that source hasn't been validated yet — see docs/data-sources.md.

Sample / Placeholder Data (local development only)

Update frequency: n/a
Resolution: n/a / n/a
Licence: Not applicable — synthetic values for local UI development.
Limitations: This is placeholder data used to build and test the UI locally.
It has not been validated against docs/data-sources.md and must not be treated as a real climate observation.
As of 2026-07-04 only the Australia regional impact card still reads this placeholder — BOM provides no sanctioned automated channel for rainfall/drought data (see the bom-rainfall-investigated entry). Every other panel uses a real, validated source.

NOAA CPC Weekly Niño Indices

https://www.cpc.ncep.noaa.gov/data/indices/wksst8110.for

Update frequency: weekly
Resolution: regional index / weekly
Licence: U.S. government work — public domain (NOAA data policy).
Limitations: Superseded by the noaa-oni source for v0.1: ONI already provides matching-resolution values on the same 3-month running mean the site's thresholds use, so this weekly file isn't ingested separately.
See docs/learnings/2026-07-03-noaa-oni-ingestion.md.

NOAA ONI / RONI

https://www.cpc.ncep.noaa.gov/data/indices/oni.ascii.txt

Update frequency: monthly
Resolution: regional index / monthly (3-month running mean)
Licence: U.S. government work — public domain (NOAA data policy).
Limitations: ONI is a 3-month running mean; it smooths out short-term weekly fluctuations.
The most recent 1-2 periods may still be revised as more observations come in.

IRI/CPC ENSO Forecast Probabilities

https://www.cpc.ncep.noaa.gov/products/analysis_monitoring/enso_advisory/ensodisc.shtml

Update frequency: monthly (published mid-month, per the discussion's own schedule)
Resolution: regional index / monthly, rolling forecast periods
Licence: U.S. government work — public domain (NOAA data policy).
Limitations: Not a structured El Nino/Neutral/La Nina probability table — IRI/CPC do not publish one for free scraping (checked 2026-07-03).
The single extracted probability statement is whichever one the discussion happens to lead with; it is not necessarily the full picture across all lead times.
Some months may not contain a matching 'X% chance...' sentence at all; the card falls back to alert status and synopsis only when that happens.

Natural Earth 1:110m Country Boundaries

https://www.naturalearthdata.com/

Update frequency: rarely (country borders change infrequently; refetched manually via scripts/fetch_basemap.py)
Resolution: 1:110m (coarse/simplified — suitable for small web maps, not precise boundary disputes) / n/a (static reference data)
Licence: Public domain — no attribution required (Natural Earth terms of use).
Limitations: 1:110m resolution simplifies coastlines significantly — not suitable for precise geographic analysis.

NOAA OISST v2.1 SST Anomaly Grid

https://coastwatch.pfeg.noaa.gov/erddap/griddap/ncdcOisst21Agg.html

Update frequency: weekly regeneration (.github/workflows/refresh-oisst.yml, Mondays 02:17 UTC); underlying OISST fields are daily with ~2-week latency
Resolution: 0.25° native, sampled at stride 8 → 2.0° cells for the web map / daily fields, ~6 per month averaged into a monthly estimate
Licence: U.S. government work — public domain (NOAA data policy). Served by NOAA CoastWatch ERDDAP (official government endpoint, checked not to be a community mirror).
Limitations: Monthly values are a mean of ~6 sampled daily fields, not NOAA's official monthly mean product; differences are small because SST anomalies evolve slowly, but the two are not identical.
The most recent month is usually incomplete (OISST publishes with roughly two weeks of latency) and is flagged complete:false in the dataset.
The Niño 3.4 box mean here is an unweighted mean of 2° cells for display context — the site's authoritative index remains NOAA ONI (see noaa-oni source).
2° cells are for visual/educational use, not precise local analysis.

Australian Bureau of Meteorology SOI

ftp://ftp.bom.gov.au/anon/home/ncc/www/sco/soi/soiplaintext.html

Update frequency: monthly
Resolution: index (basin-wide) / monthly
Licence: BOM default copyright terms apply (see bom.gov.au/other/copyright.shtml) — free to access via anonymous FTP. Full copyright text not independently verifiable by automated fetch, since BOM's own copyright page blocks automated requests too.
Limitations: Coupling classification follows the rule validated against real data in docs/methodology.md.
SOI is one indicator of atmospheric response; it does not capture the full atmospheric picture.

BOM Rainfall / Drought Data (investigated, not ingested)

https://www.bom.gov.au/climate/current/

Update frequency: not yet recorded
Resolution: /
Licence: n/a — not ingested.
Limitations: CHIRPS/GPM IMERG gridded rainfall was ruled out (would require raster subsetting/processing, out of scope for v0.1).
BOM's anonymous FTP channel (the sanctioned alternative used for SOI) does not carry rainfall decile/anomaly figures in usable text form — only static maps and a raw per-station archive that would require computing deciles ourselves.
BOM's own website (www.bom.gov.au) 403s every scripted request regardless of User-Agent, including an honest self-identifying custom UA — only a browser-spoofed UA gets through, which this project will not do. See docs/learnings/2026-07-04-bom-rainfall-fully-blocked.md.
Australia's regional impact card therefore still shows placeholder rainfall/land-stress figures pending a sanctioned BOM data channel.

NIWA Seasonal Climate Outlook

https://niwa.co.nz/climate-and-weather/seasonal-climate-outlook/seasonal-climate-outlook-july-september-2026

Update frequency: quarterly
Resolution: regional (per-region probability statement) / 3-month outlook period
Licence: NIWA/Earth Sciences New Zealand permits reproduction of Seasonal Climate Outlook content with attribution, issue date, and validity period retained.
Limitations: Probability statements, not measured anomalies — a forecast, not an observation.
Only the two eastern regions matching this card's existing framing are extracted; NIWA publishes six regions in total.
Quarterly update cadence — the figure can be up to ~3 months stale between NIWA's own outlook releases.