28 September 2026 · precision: day · Kombolcha Airport (DSE/HADC) and adjacent industrial park, Amhara

Reported drone strike on a fuel depot at Kombolcha Airport

ACLED reports that a suicide drone strike believed to have been carried out by the Alliance for Survival targeted a fuel depot at Kombolcha Airport on 28 September. The targeting claim rests on that single source and is unresolved. The expected physical consequence is assessed separately: a Sentinel-2 before/after pair shows no burn scar at the airport or the adjacent industrial park, and NASA FIRMS records no thermal detection within 54 km on a day when a VIIRS pass demonstrably observed this latitude. A large fuel fire is therefore assessed as unlikely, which does not settle whether a strike was attempted.

Claim

15204555809599 ← ruled out established →

A suicide drone strike targeted a fuel depot at Kombolcha Airport on 28 September.ACLED (Dr Jalale Getachew Birru, East Africa Senior Analyst)

roughly even chance

A large fuel fire resulted at Kombolcha Airport on or about 28 September.Our own formulation, to test the reported strike against what the sensors could detect

unlikely

Satellite

Before and after

Satellite image after the reported event — 29 September, 07:56Z · after

5.1 km across · Sentinel-2 (S2B) MSI L2A · 10.0 m per pixel · identical extent and scale

What this can and cannot show. Sentinel-2 true colour at 10 m per pixel, acquired 34.9 h after the reported event time. One pixel is roughly the footprint of a small building, so damage to an individual structure, a crater, or munition debris would not be resolved. This imagery could show a burn scar, a large destroyed or removed structure, a plume caught at the moment of overpass, or substantial change to fuel or vehicle areas. Absence of visible change therefore carries little information about whether a small strike occurred. For THIS incident the floor matters less than usual: the reported target is a fuel depot, and a burning fuel installation produces a soot and burn scar tens of metres across that persists for weeks. That is well within what 10 m imagery resolves. A null result here therefore carries real weight, unlike a null against a single munition impact.

Observations

What the evidence shows

What we see

Comparing the 24 September and 29 September crops of the same 5.0 x 5.0 km extent, and inspecting the central 1.8 km around the airport, the runway, taxiway, apron, terminal buildings and the industrial park to the west all appear intact and unchanged. No burn scar, no blackened ground, no soot staining, no collapsed or removed structure and no disturbed earth is visible. No tank farm is identifiable in either image, and none is mapped in OpenStreetMap. Differences between the two dates are confined to field colour and overall brightness. The 24 September scene carries 19.7% cloud over the tile, but the cloud lies in the south-west corner and the airport and industrial park are clear in both images.

What it is consistent with

Consistent with no large fire or structural destruction at the airport or the adjacent industrial park between the two dates. It does not address whether a drone was launched at the site, whether a weapon impacted and failed to ignite, or whether a small fire was contained before it marked the ground. The location of the reported fuel depot is itself unestablished, so this observation covers the reported area rather than a confirmed target.

What we see

No FIRMS thermal detection occurred at or near Kombolcha Airport on 28 September or at any point in the 25 September to 2 October window; the nearest was 54.4 km away on 25 September. A VIIRS NOAA-20 daytime pass on 28 September at 11:25 UTC detected two fires 120 km away with fire radiative power of 4.12 and 2.56 MW. All nine detections in the 150 km neighbourhood range from 1.06 to 30.94 MW, the scale of ordinary agricultural burning.

What it is consistent with

Unlike the Addis case, this null is informative rather than empty. A burning fuel depot is a sustained, high-power thermal source — orders of magnitude above the 1-31 MW background seen here — and is precisely what a 375 m sensor is built to detect. The 28 September pass establishes the sensor was observing this latitude that day, so the absence is an absence of fire and not an absence of looking. The limits are real but narrower than usual: the time of the reported strike is unknown, the overpass is a snapshot rather than continuous watch, and a fire starting after 14:25 EAT and extinguished quickly could fall between passes. It would still be expected to leave a burn scar, and none is visible.

Assessments

Claim by claim

A suicide drone strike targeted a fuel depot at Kombolcha Airport on 28 September.

Claimed by ACLED (Dr Jalale Getachew Birru, East Africa Senior Analyst) · source rating B3

roughly even chance 45–55%

A mention in a list of reported incidents, reaching us second-hand. We have no target, no time of day, and no independent signal.

What we checked, and what a blank result means

No independent check run.

What would change our mind

Any geolocated imagery or video, a FIRMS thermal detection on the night in question, or an airport NOTAM. Establishing that the reports trace back to a single social-media post would lower it.

A large fuel fire resulted at Kombolcha Airport on or about 28 September.

Claimed by Our own formulation, to test the reported strike against what the sensors could detect · source rating F6

unlikely 20–45%

Two independent instruments that should have seen a large fuel fire did not. FIRMS records no thermal anomaly at the site on a day when a VIIRS pass demonstrably observed this latitude, and Sentinel-2 shows no burn scar or soot 35 hours later, when a fire of that kind would have left a mark persisting for weeks. Held at 'unlikely' rather than lower because the location of the fuel depot is not independently established, the time of the reported strike is unknown, and a small or quickly contained fire could escape both checks.

What we checked, and what a blank result means

This is the first null result in this ledger that carries real weight, and the reason is specific: the claimed effect is matched to what the instruments detect. Sentinel-2 at 10 m cannot resolve a single munition impact, but a burning fuel installation leaves a scar tens of metres across. VIIRS at 375 m cannot see a detonation, but a fuel fire is exactly its design case, and the 28 September pass confirms it was looking. Neither of those statements holds for the Addis incident, where the equivalent nulls mean almost nothing. The null still does not reach the targeting claim, which is assessed separately and unchanged.

What would change our mind

Ground photography or video of fire or damage at the site. Sub-metre imagery showing tank damage below the 10 m floor. Establishing that the fuel depot is somewhere other than the airport area covered here, which would make both checks irrelevant. A FIRMS archive query with a key, covering overpasses the rolling 7-day product may have missed. An Ethiopian Airports Enterprise or government statement either way.

Sources

Everything this rests on

research — ACLED expert comment, 'Blasts in Addis Ababa signal dangerous new phase in Ethiopia's conflict', attributed to Dr Jalale Getachew Birru, East Africa Senior Analyst. States that 'suicide drone strikes are believed to have been carried out by archived copy
thermal — NASA FIRMS active-fire detections within 150 km of Kombolcha Airport for 25 September to 2 October 2026, from the open 7-day global VIIRS (375 m, NOAA-20 and S-NPP) and MODIS (1 km) products. Nine detections in total. The nearest to the air regenerated from a recorded query