TerraKeepKeeping Earth within its limits
Cloud-free satellite view around Phoenix Islands Protected Area — tile 2/2, with the finding area outlined
3.73°S 171.24°WNASA Blue Marble via GIBS · outline = finding area
Published

1872.7 hours of apparent fishing logged inside Phoenix Islands Protected Area over 30 days

Phoenix Islands Protected Area — tile 2/2 · seen · published

Global Fishing Watch's AIS-based model recorded 1872.7 hours of apparent fishing inside the legal boundary of the Phoenix Islands Protected Area within this watch box, between 30 August and 25 September 2026, on 27 of 30 days, by vessels of 4 flag states. Across the whole reserve the model logged 3589.3 hours. A second pass using a tighter box, trimmed 15% in from each side, found 835 hours, or 45% of the total, well inside the area rather than hugging the edge, so this is not a boundary-mapping artefact.

What this is not: it is not proof of illegal fishing. The reserve was closed to commercial fishing in 2015, but Kiribati announced a reopening in 2021, so effort here may be licensed, and zoning inside the reserve is not modelled. "Apparent fishing" is inferred from vessel movement, not observed catch, and transiting, drifting or waiting vessels can be misread as fishing. The period also falls in an El Niño phase, which can shift where fleets work.

What would change our minds: licence or zoning records covering this area, a revised effort model, or the activity being gone at the seven-day revisit.

Written by AI, then checked by a separate reviewer before publication. Every claim points at the evidence below.

What we saw

  • globalfishingwatch-fishing-effort — 1872.7 h of apparent fishing inside the legal boundary of MPA 309888 (GFW polygon clip) within this watch box, 2026-08-30..2026-09-25; whole reserve by flag state: CHN 3282 h, NRU 221 h, KIR 59 h. Global Fishing Watch, 2026. www.globalfishingwatch.org. (Sep 25, 2026) See the source ↗
  • globalfishingwatch-fishing-effort — 835 h (45% of the 1872.7 h) lie in 0.01° cells well inside the area (box inset 15% per side) and inside the legal polygon — not a boundary artefact (bar 25 h). (Sep 25, 2026) See the source ↗

What it might not be

  • Fishing here may be lawful: a reopening of the reserve was announced in 2021, and internal zoning is not modelled.
  • "Apparent fishing" is model-inferred from vessel movement, not observed catch; it can be wrong for any single track.
  • Only vessels broadcasting AIS are counted; those with AIS off, absent or spoofed are invisible, so the total is a floor, not a census.
  • Flag state is where a vessel is registered, not who owns or operates it; this finding assigns no blame.
  • Data lag is about 4 days, and the effort model is revised between dataset versions.
  • The boundary used is a copy of the WDPA polygon on 0.01° cells with a centre-in rule.
  • The check is method-level (a tighter inset box on the same dataset), not a separate data source; the seven-day revisit is still pending.
  • Only vessels broadcasting AIS are seen: 'dark' vessels (AIS off, absent or spoofed) are invisible, and most small-scale boats carry no AIS at all.
  • Apparent fishing is inferred by a GFW model from vessel movement, not observed catch — it can be wrong for a given track.
  • Transiting, drifting or waiting vessels can be misread as fishing (and slow fishing as transit).
  • The legal boundary is GFW's copy of the WDPA polygon (0.01° cells, centre-in rule); without a WDPA id the watch box stands in for the boundary. Zoning inside a reserve is not modelled.
  • Some MPAs allow fishing legally in zones or for local/artisanal fleets (e.g. Galápagos, Phoenix Islands since 2021) — effort there may be lawful.
  • Data lag ~4 days, and GFW's effort model is revised between dataset versions.
  • Says nothing about who is responsible; flag state is where a vessel is registered, not who owns or controls it.

What would change our mind

Newer satellite images that show no change, a correction from the data source, or a reply with evidence from the ground. If that happens, the case is marked “Turned out wrong” — in public — and it stays on this site.

Why this was published

Two separate sources agreed — globalfishingwatch-fishing-effort, then globalfishingwatch-fishing-effort — and a reviewer checked the evidence before it went up.

Replies

Know this place? Say what you see. Replies are checked before they appear. No account, no email.

Replies are open on the interactive site (it needs JavaScript).

Ledger id 01a0ead8-3141-7832-a91c-0b51b88b7294

Technical details rule, evidence ids, history, proof
Rule
mpa_fishing v1.0
Tier
T2
Tags
marine, marine-protected-area, fishing, kir, iucn-not-reported, fishing, marine-protected-area
Confirmed
Yes — independent method signal, 2026-09-29
Rule FP rate
40 % · 2 of 5 decided
Location
3.73°S 171.24°W
Observed
2026-09-25
Opened
2026-09-29 by automatic rule mpa_fishing v1.0
  1. candidate
  2. confirmed
  3. published
  4. resolved

Evidence2

recordglobalfishingwatch-fishing-effort

1872.7 h of apparent fishing inside the legal boundary of MPA 309888 (GFW polygon clip) within this watch box, 2026-08-30..2026-09-25; whole reserve by flag state: CHN 3282 h, NRU 221 h, KIR 59 h. Global Fishing Watch, 2026. www.globalfishingwatch.org.

activeDays 27days 30flagStates 4hours 1872.7reserveHours 3589.3
· fishing_activity v1.0 · gfw-fishing-mpa-phoenix-islands-2-2026-08-30-2026-09-29
Source ↗
recordglobalfishingwatch-fishing-effortConfirming signal

835 h (45% of the 1872.7 h) lie in 0.01° cells well inside the area (box inset 15% per side) and inside the legal polygon — not a boundary artefact (bar 25 h).

insetHours 835insetSharePct 45
· fishing_activity v1.0 · gfw-fishing-inset-mpa-phoenix-islands-2-2026-08-30-2026-09-29
Source ↗

Context

ENSO El Niño · ONI +1.8
  • Apparent fishing is model-inferred from AIS; vessels without AIS are not counted. Flag state ≠ owner.
  • Closed to commercial fishing 2015; Kiribati announced a reopening in 2021 — fishing here may be licensed.

Publication gates

  • Evidence — 1 piece from globalfishingwatch-fishing-effort — met
  • Independent confirmation — a different method (globalfishingwatch-fishing-effort), 2026-09-29 — met
  • Human approval — 1 of 2 needed at tier 2 — not met
  • Narration — AI-drafted by claude-opus-5 — met
Published
2026-09-30 by AI model
Narrated By
AI model
Reviewed By
AI model
Policy
2026-09-26-autonomous
Note
autonomous publish: narrated by model:claude-opus-5, reviewed by model:claude-sonnet-5 (policy 2026-09-26-autonomous)

History5

  1. status changed → publishedAI model
  2. reviewed → confirmedAI model
  3. narrated → confirmedAI model
  4. confirmed → confirmedautomatic rule mpa_fishing v1.0
  5. created → candidateautomatic rule mpa_fishing v1.0

Verify

Checkpoint
earthdeck.dev/findings/v1 · 971 entries
Signed root
65becba35e7b…cfb1ac
This case
5 signed events
Leaf
#955 of 971 — this case’s latest event
Leaf hash
7ca4b9814a20…417de5
Proof
10-hash audit path
  1. 337cbc3d005c2cb92aa02c8a2488635db0155a97f2be3457e9d81300ef2875ca
  2. f3bc0f79cb25d5622409e71eee8b5a10f9fe01da094dac577f3bee1a428d4adc
  3. 4f99f736066a1d65d44475ffba8d8c0c5aeac86377deee569e0d5298830cab5a
  4. dd2c9fe6ae688c5475aff3e46cbdfeaf464df870ad3a941c7f428943a0ab02f3
  5. 0045955efc1ce8cbac08a8733861af830ac685c100ede195c387e074963da34e
  6. 881a017e2581f97da1c46804177dfd5f924565433e1166705255b96348abe25d
  7. a72c3339b408a45e95e937a87116707abc006bf01b9d03fe3f1a13fbcc11992a
  8. d4a69eb9bd4aba2b04a0c06f8365d5c370a5591fa80d0e8a9962a94eea222b01
  9. b151de5eaefddf3d1cfc86f94cdbc3995ed5d406a6c0acfc99649a5d48d50794
  10. 5f7580b7f5a7f09ff7675565fb36c19876079a6c101f721419a5a07a7a23b1c1
Last event
76ed690f136d…4b99e2
Note
Checkpoint text
earthdeck.dev/findings/v1
971
Zb7Lo157ettQqPmu0tm88BnWSkMC3ZJMEFFRbmvPsaw=

— earthdeck TkCXvy4ab1W8aoaOb+pzCGCDnG5hglritooG+HSh5E2eK5bX/GOa8ngRawRBFFq9xkoAkQHY2WV8CTtkhuUX2a2EhA8=

Re-derive every entry, signature and trust rule yourself:

npx -y earthdeck ledger verify --remote https://vitalearth.io
Case events + proof (JSON)Signed checkpointPublic keyFull log (JSONL)