Snapshot · observed July 31, 2026, 07:36 UTC

Jazan: two fires ignite 40 km apart, and burn out — July 2026

Jazan Economic City refinery complex, Saudi Red Sea coast — with a second site, a coastal fuel-distribution facility, 40 km south near Jizan city

Snapshots are curated, event-dated observations from open satellite data. They document specific events; they are not the systematic instrument records above. Every image carries its provenance in-frame.

Jazan: two fires ignite 40 km apart, and burn out — July 2026 — 23 July 2026, the morning pass — the refinery complex, its port and the coast south of it on a clear day. No plume anywhere. This frame and the next are the baseline every later claim is measured against — 23 July 2026, the morning pass — the refinery complex, its port and the coast south of it on a clear day. No plume anywhere. This frame and the next are the baseline every later claim is measured against · true color · 20 m/px · open full resolution ↗

Jazan: two fires ignite 40 km apart, and burn out — July 2026 — 23 July 2026, the same frame at 2.2 µm — the shortwave-infrared band where active combustion registers as bright points. Nothing in the complex is hot. This is what cold looks like: the cold half of a pair completed on the 26 July frames below — 23 July 2026, the same frame at 2.2 µm — the shortwave-infrared band where active combustion registers as bright points. Nothing in the complex is hot. This is what cold looks like: the cold half of a pair completed on the 26 July frames below · SWIR B12, fixed 1000–6000 DN stretch · 20 m/px · open full resolution ↗

Jazan: two fires ignite 40 km apart, and burn out — July 2026 — 23 July 2026 — the second site, 40 km south: the coast at Jizan city. A routine morning; nothing burning — 23 July 2026 — the second site, 40 km south: the coast at Jizan city. A routine morning; nothing burning · true color · 10 m/px · open full resolution ↗

Jazan: two fires ignite 40 km apart, and burn out — July 2026 — 23 July 2026 — the southern site at 2.2 µm, cold. On every earlier July pass this location is cold too: it is not a routine flare site — 23 July 2026 — the southern site at 2.2 µm, cold. On every earlier July pass this location is cold too: it is not a routine flare site · SWIR B12, fixed 1000–6000 DN stretch · 20 m/px · open full resolution ↗

Jazan: two fires ignite 40 km apart, and burn out — July 2026 — 26 July 2026 — the same water, the same coast, three days later. A dense black plume is rooted inside the refinery complex, with a separate white plume rising just east of the black root, and it trails south past the bottom of the frame: the full plume exceeds 40 km. Two emission points inside one complex, one event — 26 July 2026 — the same water, the same coast, three days later. A dense black plume is rooted inside the refinery complex, with a separate white plume rising just east of the black root, and it trails south past the bottom of the frame: the full plume exceeds 40 km. Two emission points inside one complex, one event · true color · 20 m/px · open full resolution ↗

Jazan: two fires ignite 40 km apart, and burn out — July 2026 — 26 July 2026 at 2.2 µm — the plume root saturates the detector. Bright at the root and dark down the plume is the signature of active combustion at a fixed source, not drifting smoke: compare the cold 23 July frame above — 26 July 2026 at 2.2 µm — the plume root saturates the detector. Bright at the root and dark down the plume is the signature of active combustion at a fixed source, not drifting smoke: compare the cold 23 July frame above · SWIR B12, fixed 1000–6000 DN stretch · 20 m/px · open full resolution ↗

Jazan: two fires ignite 40 km apart, and burn out — July 2026 — 26 July 2026 — the southern site, same morning: its own black plume, rooted at the coastal facility and trailing southwest over the sea. A second, separate fire, 40 km from the first. No wire report named this site — 26 July 2026 — the southern site, same morning: its own black plume, rooted at the coastal facility and trailing southwest over the sea. A second, separate fire, 40 km from the first. No wire report named this site · true color · 10 m/px · open full resolution ↗

Jazan: two fires ignite 40 km apart, and burn out — July 2026 — 26 July 2026 — the southern site at 2.2 µm: a hot point at the plume root. The same combustion test as the refinery frame, passed independently at both sites — 26 July 2026 — the southern site at 2.2 µm: a hot point at the plume root. The same combustion test as the refinery frame, passed independently at both sites · SWIR B12, fixed 1000–6000 DN stretch · 20 m/px · open full resolution ↗

Jazan: two fires ignite 40 km apart, and burn out — July 2026 — 28 July 2026 — the southern site, two days later: no plume, a clear morning — 28 July 2026 — the southern site, two days later: no plume, a clear morning · true color · 10 m/px · open full resolution ↗

Jazan: two fires ignite 40 km apart, and burn out — July 2026 — 28 July 2026 — and no hot point at 2.2 µm. The southern fire is out, not merely smokeless. The refinery cannot be checked on this date: the 28 and 30 July passes fly an orbit whose swath edge excludes the refinery entirely, so its state between 26 and 31 July is unobserved — 28 July 2026 — and no hot point at 2.2 µm. The southern fire is out, not merely smokeless. The refinery cannot be checked on this date: the 28 and 30 July passes fly an orbit whose swath edge excludes the refinery entirely, so its state between 26 and 31 July is unobserved · SWIR B12, fixed 1000–6000 DN stretch · 20 m/px · open full resolution ↗

Jazan: two fires ignite 40 km apart, and burn out — July 2026 — 31 July 2026 — the refinery complex at 10 m, the first pass able to see it since the 26th. The plume is gone. The complex, port and jetty read intact at this resolution; a dark scar marks the southern tank-farm block — 31 July 2026 — the refinery complex at 10 m, the first pass able to see it since the 26th. The plume is gone. The complex, port and jetty read intact at this resolution; a dark scar marks the southern tank-farm block · true color · 10 m/px · open full resolution ↗

Jazan: two fires ignite 40 km apart, and burn out — July 2026 — 31 July 2026 at 2.2 µm — no extended hot zone anywhere in the complex. What remains is a handful of compact bright points along the eastern edge of the plant — 31 July 2026 at 2.2 µm — no extended hot zone anywhere in the complex. What remains is a handful of compact bright points along the eastern edge of the plant · SWIR B12, fixed 1000–6000 DN stretch · 20 m/px · open full resolution ↗

Jazan: two fires ignite 40 km apart, and burn out — July 2026 — 31 July 2026, close view — the structures under those bright points are flare stacks: a derrick, a long thin shadow, and no smoke — 31 July 2026, close view — the structures under those bright points are flare stacks: a derrick, a long thin shadow, and no smoke · true color · 10 m/px · open full resolution ↗

Jazan: two fires ignite 40 km apart, and burn out — July 2026 — 31 July 2026, the same close view at 2.2 µm — each hot point is one to four pixels wide, sitting exactly at a stack tip. That is the signature of a flare system operating, not of a fire front. The refinery fire burned five days or fewer — 31 July 2026, the same close view at 2.2 µm — each hot point is one to four pixels wide, sitting exactly at a stack tip. That is the signature of a flare system operating, not of a fire front. The refinery fire burned five days or fewer · SWIR B12, fixed 1000–6000 DN stretch · 20 m/px · open full resolution ↗

Caption

Sentinel-2 images, 23–31 July 2026: two energy facilities 40 km apart on Saudi Arabia's Red Sea coast — the Jazan Economic City refinery complex, and a coastal fuel-distribution site south of Jizan city. On the morning of 23 July both sites are cold: no plume in true color, no hot point at 2.2 µm. On the morning of 26 July both are burning — the refinery feeds a dense black plume rooted inside the complex and extending beyond 40 km, with a second white plume just east of the black root, and the southern site has its own rooted plume trailing over the sea. Houthi forces claimed a strike package against the region early on 25 July (an attributed claim by an interested party; the imagery brackets ignition between the 23 and 26 July passes and cannot date it finer). By 28 July the southern fire is out. The refinery is unobservable from 27 to 30 July — the intervening passes fly an orbit that structurally excludes it — and on the morning of 31 July the plume is gone: the complex shows compact bright points at flare stacks along its eastern flare line and a dark burn scar in the southern tank farm. The refinery fire burned five days or fewer. Industry reporting dates a refinery shutdown to 27 July with a restart targeted around 15 August (consultancy-sourced; no official damage assessment has been published); flaring on 31 July is consistent with a complex operating its flare system and does not establish production status. Imagery: Copernicus Sentinel-2 (open data) via Microsoft Planetary Computer, rendered by NewsPlanetAI.

What this imagery does not establish

This is a record of a bounded event, not a standing measurement. It shows combustion state at two named sites on named pass dates — nothing more. The imagery does not establish cause: a claimed strike and an industrial accident look alike at 10 m, and the attribution of these fires is external reporting, not a pixel finding. It does not establish what burned or which units were damaged — unit-level damage assessment sits below this resolution, and 'intact at 10 m' does not rule out precision damage. It does not establish production status: flare-stack points show the flare system operating, which is what both a restarting and a purging refinery do. Between 26 and 31 July there is no refinery observation at all — the 28 and 30 July passes fly an orbit whose swath edge excludes the refinery box (verified on every scene of that orbit, not assumed) — so the refinery's state on 27–30 July is unobserved, and 'out by 31 July' brackets the end of the fire without dating it. The ignition is likewise a bracket: after the 23 July pass, burning by the 26 July pass. SWIR brightness is sensor response at 2.2 µm, not a temperature measurement; 'saturated' describes the detector, not the fire. The identification of the southern site as a fuel-distribution facility rests on its location and external reporting, not on anything readable in these pixels. Every frame of a given box uses one fixed stretch, so frames are directly comparable across dates; no frame is rescaled to its own contents.