Where the East-West Pipeline Actually Burned

On the morning of Thursday, September 10, drones struck Saudi Arabia’s East-West Pipeline, the 1,200-kilometer line from Abqaiq to Yanbu that has carried most of the kingdom’s exports since Iran shut the Strait of Hormuz. The Ministry of Energy shut the line “as a precautionary measure” and reported injuries. The Ministry of Foreign Affairs then said the drones came from Iraq and that Riyadh, at the request of Iraqi Prime Minister Ali Falih al-Zaidi, had “chosen not to respond at this stage.” Early on Saturday, Baghdad said its investigation had traced the launches to Maysan province on the Iranian border and dismissed the commander of Maysan Operations Command. That is the whole public record.

Aramco has not said which facilities were hit, how badly, or when flows will resume. So the Middle East Forum went to the satellites. The VIIRS instruments on Suomi NPP and NOAA-20 record any 375-meter pixel hotter than its surroundings on every pass. We pulled every detection within the pipeline corridor from August 12 to September 12, and compared the two attack days against the 29 days before them. The question was simple: which fires were new? Then, on Saturday morning, a 10-meter optical frame of the worst-hit site came in, and we could see the answer.

Seven. All sit on the pipeline right-of-way along 110 kilometers of line southeast of Medina, and none registered a single detection in the previous four weeks. Undated commercial basemap imagery shows what sits at each one, and a pre-attack optical frame from September 7 confirms the layout.

Site 3, at 23.959 N 40.054 E, is a full pump station complex: tank farm, turbine halls, and pipe manifolds inside a fenced compound about 700 meters across. It was burning at 13:09 local time on September 10, the first satellite pass after the attack, at 61 megawatts, and was still burning at 31 to 45 megawatts on the two night passes at 01:33 and 01:54 the next morning. By the next daytime pass, at 12:47 on September 11, the station showed no detection. A fire of that size, burning for at least 13 hours, points to stored crude or process equipment, not the buried line.

.

The optical frame acquired at 11:03 local on September 12 removes the doubt. A soot and burn scar covers the whole compound and spreads about 800 meters to the northeast along the access track, some 12 hectares in all by our pixel count. Five days earlier, the same frame showed a tank farm and turbine halls on clean ground. At 10 meters a pixel it is not possible to say whether tankage is gone, but a scar of that size does not come from a line break; it comes from the contents of the station burning where they stood. The Ministry of Energy’s own statement, which placed attacks in the Medina and Riyadh regions, fits that reading.

Sites 1 and 2, 18 and 66 kilometers east of the pump station, share an identical layout: a small valve installation on the corridor beside a large bermed square with a central pad. They read as relief pits, where crude is burned off when a section of line has to be emptied in an emergency. Both lit up on September 10. Site 1 was still producing 93 megawatts at 12:47 local on September 11 and 31 megawatts at 14:28, on the same pass at which the pump station had gone dark. Either that is a very long controlled burn, or the line was hit twice in 48 hours, which is exactly what Iraqi sources told Alhurra that Riyadh had complained of in private. Neither reading is confirmed. By the night passes of September 11, none of the seven sites registered a detection. The September 12 optical frame shows a small glowing spot inside each of the two pits at 11:03 local, consistent with the last of a burn-off; thermal passes at 12:30 and 14:10 that afternoon registered nothing at any of the seven sites.

Sites 4 through 7 are four small installations spread along 20 kilometers of right-of-way 25 to 45 kilometers west of the pump station, each within 300 to 800 meters of the mapped line. They appear only on night passes, at 6 to 18 megawatts. A 375-meter pixel cannot tell an impact point from a vent.

The seven new fires sit on one 110-kilometer segment of a 1,200-kilometer line. Everything else that burned along the corridor that night was burning the week before.

That last point matters because the first 24 hours of open-source commentary counted every hot pixel near the pipeline as damage. Our baseline shows nine sites along the corridor that flare routinely, lit on five or more of the 29 pre-attack days. The largest, near 25.26 N 48.1 E on the eastern half of the line, produced 189 megawatts on September 10, which looks alarming until one sees that it produced 221 megawatts on an ordinary day in the preceding month. We found no new fire anywhere on the eastern half of the line, including the Riyadh region the ministry named; if drones struck there, they left nothing a 375-meter thermal pixel could see. The Payne Institute at the Colorado School of Mines reached a compatible conclusion from its Nightfire data: most detected activity was at known flare sites, with two entering an emergency flaring regime, one of them at a rate approaching 0.4 billion cubic meters a year, while combustion temperatures stayed inside each site’s prior range.

The two readings are not in conflict, and the difference is worth spelling out because it decides what the fires mean. A flare catalog built over years lists every installation that has ever burned; our four-week baseline lists only those burning in the month before the attack. A relief pit that sits dormant for years and is lit once in an emergency is “known” in the first sense and “new” in the second. That is what sites 1 and 2 are: existing infrastructure doing what it was built for, which is to take the contents of an isolated line segment and burn them. A steady combustion temperature with a surge in volume is the signature of exactly that. It is also the strongest argument against the early claims of eight rupture points along the line. What the catalog cannot settle is site 3. A flare burns at a stack or in a pit and leaves the plant around it clean; the September 12 frame shows a burn that covered the compound itself. The flaring at sites 1 and 2 was the response. The fire at site 3 was the cause.

One more signal appears some 900 kilometers to the east. Three processing sites in the Ghawar producing area, which feed crude toward Abqaiq and the pipeline intake, show daytime detections at 13:09 local on September 10 of 35 to 70 megawatts, against a prior-month maximum of 1 to 10 megawatts at the same spots. The likeliest reading is upstream flaring: the line was shut and production had nowhere to go. That is inference, not measurement. Saudi Arabia reported crude output of 6.24 million barrels a day in August, its lowest since 1990; a shut pipeline pushes that lower still.

Baghdad’s statement that the drones came from Maysan is the most consequential fact of the week, and the map shows why. From Amarah, the provincial capital, to the burning pump station is about 1,100 kilometers in a straight line, roughly 800 of them inside Saudi airspace after crossing the border near the Kuwait tri-point. The drones flew over or past the Riyadh region, the airspace around the capital, to reach a target in the Hejaz. In May 2019, when Houthi drones hit pump stations 8 and 9 near Afif and Dawadmi, they flew about 950 kilometers from Sanaa; that attack was treated as a watershed. This one came from the opposite direction, flew farther, and reached deeper.

The range and the fixed-installation targeting narrow the field. Open-source analysts matched the July attacks from Iraq to Iranian-designed one-way attack drones of the Shahed 131 and 136 class. Of the types in the factions’ hands, only that class covers 1,100 kilometers with margin, and within it the smaller Shahed 131, rated at about 900 kilometers, does not; the Shahed 136, rated at 2,000 kilometers or more with a warhead of 20 to 50 kilograms, does. That is an inference from range, not an identification. Iraqi intelligence sources told Alhurra that non-Iraqi operators who entered Iraq in March and April run the launches from the eastern border, and that the aim is to implicate Iraqi factions rather than to act through them. Credit the second claim or not; the first fits Maysan, where the Iranian border and the Popular Mobilization Forces’ logistics overlap.

Saudi oil infrastructure has now been hit from the north, from Iraq, in July and again in September, on top of April’s strike and the Houthi strikes from the south. In July Riyadh struck back with the United States. In September it chose to wait.

That restraint is a policy choice with a price. The April 8 strike, hours after the ceasefire, was attributed by industry sources to Iran itself. On July 28, Saudi and American aircraft struck Iran-aligned groups in Iraq after drones from Iraqi territory hit oil facilities in the Eastern and Riyadh regions; the PMF said at least 20 were killed. Riyadh then worked to repair the relationship: on August 7 its intelligence chief flew to Baghdad. Thursday’s attack tested that repair, and the Saudi response, a threat delivered by phone according to Alhurra and a public decision not to retaliate, hands al-Zaidi the problem. He has responded with the tools available to a weak state: a sacked general, an investigative council, a police raid in Anbar that seized 47 drones and 46 munitions from a former tire shop, and, on Saturday, the removal of Maysan’s police chief, the referral of the province’s security directors for investigation, and the closure of the Shalamcheh crossing with Iran. None of it touches the factions. The U.S.-led coalition’s mission in Iraq ends on September 30. The same date is al-Zaidi’s own deadline for bringing weapons outside state control under it, and the factions are negotiating to meet it by transferring arsenals to the Popular Mobilization Forces, which would leave them in the same hands under a state label. A launch from Maysan three weeks before that deadline is a statement about who controls the border, and it is not the prime minister.

The pipeline is the kingdom’s answer to Iran’s closure of the Strait of Hormuz. It was built for that purpose in 1981, during the Iran-Iraq war, as a 56-inch crude line and a 48-inch natural gas liquids line; the second was converted to crude after the 2019 Abqaiq attack, which is how nominal capacity rose from 5 to 7 million barrels a day. Aramco says it can move up to 7 million barrels a day, of which about 5 million are for export from Yanbu and 2 million feed west-coast refineries. Since March, Yanbu has gone from a secondary outlet loading under 800,000 barrels a day to the main one; tanker trackers put September loadings at 2.9 to 3.7 million barrels a day before the attack, depending on the vendor. Riyadh has discussed adding 1 to 2 million barrels a day of capacity. The constraint is not the pipe but the port: Yanbu’s two crude terminals have a nominal loading capacity of 4.5 million barrels a day, 1.5 million at Yanbu North and 3 million at Yanbu South, and market sources put the effective ceiling nearer 4 million. As our colleague Umud Shokri argued in July, the line relieves pressure without removing risk: it is a fixed target 1,200 kilometers long that ends at a port sized for peacetime.

The April 8 attack showed how fast Aramco can recover when one pump station is damaged: throughput fell by 700,000 barrels a day and full capacity was declared restored on April 12. This time the ministry shut the entire line, and the thermal record shows a station still burning into a second day plus relief burns at two sites. A longer outage is possible. It comes in the week the Houthis took Mokha, Dhubab, and Perim Island and put themselves astride the Bab al-Mandab, the exit Yanbu’s tankers must use to reach Asia. The Houthis’ spokesman said on Friday that navigation remained safe for all shipping except Saudi vessels, which are under a ban announced in July. Saudi crude transits through Bab al-Mandab had already fallen close to zero in August. The barrels went north instead: tankers now discharge at Ain Sokhna into the SUMED pipeline and reload at Sidi Kerir on the Mediterranean, where loadings averaged about 2.2 million barrels a day over five weeks, nearly triple normal, with 43 percent bound for Asia the long way round through Gibraltar and the Cape. Hormuz is shut and Bab al-Mandab is now closed to Saudi hulls: both of the kingdom’s export routes are under pressure from Iranian clients at once, and the one line that fed the workaround has been hit from a third direction. Brent crossed $100 on Thursday.

Four things will show whether Riyadh’s patience with Baghdad holds. First, the pump station. The 12-hectare scar is visible from orbit; imagery at higher resolution will show whether tankage is missing, and that will say more about repair time than any statement. Second, Maysan: arrests, or only a dismissal. Third, Saudi aircraft, and whether they return to Iraqi airspace. Fourth, Washington. In July, Saudi and American aircraft struck Iran-aligned groups in Iraq together; the U.S.-led coalition’s mission in Iraq ends on September 30, and after that date the questions of who polices Maysan and whether the July model returns fall to Washington as much as to Riyadh. Washington has its own reason to care: U.S. diesel passed $6 a gallon for the first time on Friday, at $6.05, up more than $2 on a year earlier, with gasoline at $4.29, and the midterm elections are in November. The kingdom’s foreign ministry “reserves the right to take all measures” to protect its infrastructure. The satellite record says what that infrastructure is: one pump station, two burn pits, and four smaller sites on 110 kilometers of line southeast of Medina, hit from 1,100 kilometers away. Riyadh has the same picture. The question is what it does with it.

Method note: Detections are VIIRS 375-meter active-fire records for Suomi NPP and NOAA-20, obtained from NASA’s GIBS service and processed by the Middle East Forum. A “new” site is one with no detection within 2 kilometers between August 12 and September 9, 2026. Radiative power is measured at the moment of overpass and is not a measure of damage or of gas volume. Site functions are read from undated Esri/Maxar basemap imagery and have not been confirmed by Saudi Aramco. Records were pulled at 01:00 UTC, 09:30 UTC and 14:30 UTC on September 12; the later pulls completed the September 11 and 12 records and changed none of the seven sites. Optical frames are 10-meter true-colour scenes acquired at 11:03 local on September 7 and September 12, 2026; the burn-scar area is a pixel-difference estimate by the Middle East Forum. The pipeline route is from OpenStreetMap, sourced to EIA and Bing imagery. All figures are unaudited MEF measurements offered alongside, not instead of, official statements.