TerraKeepKeeping Earth within its limits
Published

Seven steady night-time heat sources burn offshore Campeche; top site lit 17 of 30 nights

Flare field MEX 19.45, -92.09 (18 VNF sites) · seen · published

Night-time satellite passes over the flare field at roughly 19.45N, 92.09W picked up 7 clusters of heat that reached at least 5 MW and reappeared on 5 or more of the last 30 nights (28 August to 26 September 2026). The most persistent was lit on 17 nights, with a peak fire radiative power of 28.89 MW, and 252 hot-night detections were logged in total. All 7 clusters sit on sites listed in an independent 2024 annual flare summary from the Earth Observation Group, the largest of them 0.14 km away and credited with 0.217 billion cubic metres for that year.

What this is not: heat alone cannot separate a gas flare from a refinery furnace, a kiln or another industrial source — only the repetition and the known-flare match point to flaring. The brightness figure is not a volume; the sensor saturates on the biggest flares. Unlit venting and cold methane leaks would not show up at all. The finding names a place, not an operator.

What would change our mind: a month of heavy cloud can hide nights and mimic a shutdown, so a gap is not proof of stoppage. Daytime imagery, methane sounding, or an updated flare inventory could recast these clusters.

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

Cloud-free satellite view around Flare field MEX 19.45, -92.09 (18 VNF sites), with the finding area outlined
19.48°N 92.15°WNASA Blue Marble via GIBS · outline = finding area

What we saw

  • NASA fire detections — 7 night-time cluster(s) with FRP ≥ 5 MW lit on ≥ 5 of 30 nights (2026-08-28…2026-09-26); top cluster 17 nights, max 28.89 MW at 19.3948, -92.0437. (Sep 25, 2026) See the source ↗
  • Gas-flare detections (VIIRS Nightfire) — 7 persistent cluster(s) coincide with sites in the VNF 2024 annual flare summary (top site 0.217 BCM, 0.14 km away). Credit: Earth Observation Group, Payne Institute for Public Policy, Colorado School of Mines — VIIRS Nightfire global flare summary. (Dec 31, 2024) See the source ↗

What it might not be

  • Thermal sensors cannot distinguish a flare from a furnace, kiln or other industrial heat source.
  • Cloud or smoke can hide nights and make a steady flare look intermittent.
  • FRP is not a flared volume; the sensor saturates on the largest flares.
  • Unlit venting and cold methane releases are invisible here; methane sounding would be needed.
  • The confirming flare inventory is an annual aggregate from 2024 — it shows the site is a known flare, not that it flared this week.
  • Two passes per night around 01:30 local time; short or intermittent flaring between passes is missed.
  • No operator or ownership attribution — the clusters are locations only.
  • Cannot tell a gas flare from an industrial furnace, refinery, steel or cement kiln, or a volcano — only persistence and the VNF site match argue 'flare'.
  • Cloud and heavy smoke hide nights; a cloudy month under-counts persistence and can look like a flare was turned off.
  • FRP is not flared volume: the VIIRS I-band saturates on the largest flares and the FIRMS FRP is a single-pixel estimate — volumes come only from the VNF annual summary.
  • Venting and unlit (cold) methane releases are invisible to thermal sensors — pair with S5P / EMIT methane.
  • No ownership or operator attribution: a cluster is a place, not a company; attribution needs asset registries (GEM, GGFR) and human review.
  • Two VIIRS passes per night at ~01:30 local; short or intermittent flaring between passes is missed. S-NPP ends Nov 2026, so NOAA-20/21 only.
  • The VNF confirmation is an annual aggregate from a past year: it proves the site is a known flare, not that it flared this week.

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 — NASA fire detections, then Gas-flare detections (VIIRS Nightfire) — and a reviewer checked the evidence before it went up.

Right of reply

If a case names or affects you, you can reply. A reply channel that keeps both sides on record is being set up; until then, every case page carries its ledger id so a reply can be attached to it.

Ledger id 01a0df8b-e045-7fa4-9428-e7f599938fa1

Technical details rule, evidence ids, history, proof
Rule
flaring v1.0
Tier
T2
Tags
flaring, oil-gas, discovered, mex, mex_ups_92.0866w_19.4475n
Confirmed
Yes — independent provider signal, 2026-09-26
Rule FP rate
9.1 % · 1 of 11 decided
Location
19.48°N 92.15°W
Observed
2026-09-25
Opened
2026-09-26 by automatic rule flaring v1.0
  1. candidate
  2. confirmed
  3. published
  4. resolved

Evidence2

alertfirms-viirs_noaa20_nrt+viirs_noaa21_nrt

7 night-time cluster(s) with FRP ≥ 5 MW lit on ≥ 5 of 30 nights (2026-08-28…2026-09-26); top cluster 17 nights, max 28.89 MW at 19.3948, -92.0437.

hotNightDetections 252maxFrp 28.89maxNights 17persistentClusters 7vnfMatched 7vnfTopBcm 0.217vnfTopKm 0.14vnfYear 2024
· flaring v1.0 · firms-flaring-flare-mex-19p4n-92p1w-2026-08-28-2026-09-26
Source ↗
recordeog-vnf-annualConfirming signal

7 persistent cluster(s) coincide with sites in the VNF 2024 annual flare summary (top site 0.217 BCM, 0.14 km away). Credit: Earth Observation Group, Payne Institute for Public Policy, Colorado School of Mines — VIIRS Nightfire global flare summary.

bcm 0.217vnfMatched 7
· flaring v1.0 · eog-vnf-annual-2024-flare-mex-19p4n-92p1w
Source ↗

Context

ENSO El Niño · ONI +1.8

Publication gates

  • Evidence — 1 piece from firms-viirs_noaa20_nrt+viirs_noaa21_nrt — met
  • Independent confirmation — a different provider (eog-vnf-annual), 2026-09-26 — met
  • Human approval — 1 of 2 needed at tier 2 — not met
  • Narration — AI-drafted by claude-opus-5 — met
Published
2026-09-26 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 flaring v1.0
  5. created → candidateautomatic rule flaring v1.0

Verify

Checkpoint
earthdeck.dev/findings/v1 · 164 entries
Signed root
dee55129d595…ff9240
This case
5 signed events
Leaf
#139 of 164 — this case’s latest event
Leaf hash
1f120fb8c5aa…bc62f8
Proof
7-hash audit path
  1. cf5b82f56e908b531e0d428a7496f5e68669d7b8b86c256a2164258a7f7de629
  2. 7c1d03843e3c7f66c4f0d8b0c89d320ae95e70a79c60af6d6a815fed88b8f655
  3. b13dc0db5f1134c95f7da180b64ee68e411bf0d20ddf2d22a212de896380aa9c
  4. 78d4a217127a8d785683c1be031cba2e901cbe093b1ae77f14424b62c88b2c67
  5. c5791be43a93742281091b42839887266fcf328111a065fdb199bd6029e8a58f
  6. ea90af85b57fc2296d71cae3061ddb7573bdaad07fb8fd70a326eb6bfa8a0b0d
  7. 1e6b35b77ca24da32136c1f20f9724e4c9e881569d9550e8b5024cf584b97d34
Last event
46b49ee25c73…bb3c2d
Note
Checkpoint text
earthdeck.dev/findings/v1
164
3uVRKdWVACCMvWGLLrloPsdgo2FcJrOlejRp8Ir/kkA=

— earthdeck TkCXv1RvXauCpUiWuk11ZCqwdoBJwbds1eETDZCkAWnWOFvcgHuHpllz0Q9kEbLx6iBG7M0UPs+Ejc/MfFN6WPWxNQA=

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

npx -y earthdeck ledger verify --remote https://vitalearth.io