7 October 2026 · Adigrat and the Zalambessa road, eastern Tigray

True?likely~70%

Eritrean troops reported inside Tigray at Adigrat

Updated 12 October 2026

Reuters, AFP and the BBC each spoke to residents of Adigrat who describe soldiers in Eritrean uniform moving south through the town. ACLED's East Africa expert says they crossed at Zalambessa and moved down the Adigrat-Idaga Hamus road. Addis Ababa has publicly demanded their withdrawal, which is not something a government does about troops it says are not there. Asmara's information minister did not deny the crossing; he called Ethiopia's account a pretext for a war of aggression.

Why they came is the contested part. The Ethiopian government says Eritrea is backing the TPLF-led alliance. ACLED's regional expert says close to the opposite — that Eritrea wants to weaken Addis Ababa and to eliminate the TPLF over older grievances, and moved to stop Ethiopian forces reaching its border. Both cannot be right and we cannot settle it.

We checked what we could check. The satellite imagery cannot resolve a lorry, so it neither supports nor contradicts any of this. No thermal sensor recorded a fire anywhere near the reported drone strike, which establishes less than it sounds: a fire is missed if it burns between passes or under cloud, and this area was 41 per cent cloud that morning. The strike location itself we could not find on any map.

Claim

15204555809599 ← ruled out established →

Eritrean soldiers were present in and around Adigrat, eastern Tigray, in the first week of October 2026.Adigrat residents and a local official, separately to Reuters, AFP and the BBC; ACLED's East Africa regional expertlikely

likely

Eritrea sent those forces to support the TPLF-led alliance against the Ethiopian federal government.The Ethiopian government, including minister Abraham Belayroughly even chance

roughly even chance

Ethiopian drones struck vehicles carrying Eritrean soldiers at Guada, on the road towards Idaga Hamus.ACLED's East Africa regional expert, and a resident of Edaga Hamus who described strikes nearbylikely

likely

Satellite

A closer look

Sentinel-2 (S2A) MSI L2A over Adigrat and the Zalambessa road, eastern Tigray, 1 October 2026 — 1 October, 07:56Z · Sentinel-2 (S2A) MSI L2A

Sentinel-2 (S2B) MSI L2A over Adigrat and the Zalambessa road, eastern Tigray, 4 October 2026 — 4 October, 07:55Z · Sentinel-2 (S2B) MSI L2A

Sentinel-2 (S2B) MSI L2A over Adigrat and the Zalambessa road, eastern Tigray, 7 October 2026 — 7 October, 08:06Z · Sentinel-2 (S2B) MSI L2A

What this can and cannot show. Sentinel-2 true colour at ten metres per pixel, acquired about five hours after the time residents describe. A military lorry covers about a quarter of one pixel, so no column of vehicles on a road can be resolved at this scale whatever the weather. On top of that limit, the scene classification band flags 41.4 per cent of this area as cloud or cloud shadow, including the road corridor north-east of the town — the scene is listed as 11.8 per cent cloud, but that figure is for the whole hundred-kilometre tile, not for here. Nothing should be read into this image either way.

Observations

What the evidence shows

What we see

The accounts differ by an order of magnitude on size. Reuters witnesses describe two lorries carrying dozens of soldiers through Adigrat on Wednesday morning. BBC witnesses describe convoys of between twenty and sixty lorries, one estimating up to sixty, identified by Eritrean registration plates. A local official told AFP that Eritrean soldiers had been in the city for about four days.

What it is consistent with

The accounts are consistent with a movement spread over several days being seen by different people at different moments, and they are equally consistent with the number growing as it is retold. Nothing we hold separates those. The range matters for what could have been checked: two lorries leave no trace any public sensor can detect, while sixty parked together would occupy enough ground to register.

What we see

Zalambessa, Adigrat, Edaga Hamus and Wukro resolve in OpenStreetMap at 14.5256/39.3871, 14.2774/39.4615, 14.1860/39.5634 and 13.7863/39.6015. They lie in that order along the main road running south from the Eritrean border, about 28 km from Zalambessa to Adigrat and 15 km on to Edaga Hamus. Guada, named as the place vehicles were struck, returns no result under Guada, Gu'ada, Gwada, Goada, Gaada, Gueda or Guda within the corridor.

What it is consistent with

The route described is the route that exists, in the order described, which is a weak check an invented account could still pass but a careless one would fail. The unresolvable strike location is the reason no instrument was pointed at it; small Tigrayan settlements are often absent from open map data or carried under a different transliteration, so this is not evidence the place does not exist.

What we see

Images are circulating on social media which appear to show loaded military lorries carrying uniformed personnel towards Adigrat, Edaga Hamus and possibly Wukro. The BBC states it has not independently verified them. We have not obtained a copy with recoverable provenance and have not examined any of them.

What it is consistent with

We are recording that these images exist and that nobody has verified them. We are not treating them as evidence in either direction. Pictures of foreign soldiers are among the most frequently recycled material in this conflict, and an unverified image is as capable of being old, mislabelled or from another country as it is of being what it claims.

What we see

Sentinel-2 covered Zalambessa on 21, 24 and 27 September and on 1, 2, 4 and 7 October. Over a 5 km box on the town, the scene classification layer flags nothing at all as cloud on 27 September and 1 October, 6.6 and 8.8 per cent on the two 4 October passes, and one hundred per cent on 7 October, the day the crossing is reported.

What it is consistent with

The only day that would have mattered is the one day with no view at all. That is not bad luck so much as the season: this is the tail of the kiremt rains over the Tigray highlands, and it has now blocked us at Mekelle airport, at Adigrat and here. It is worth stating plainly because the alternative is a reader assuming that if satellites exist, somebody must have seen it. On the clear days either side there is nothing to report: a lorry is a quarter of one sample, so a column crossing on any of those dates would not have been visible either.

Assessments

Claim by claim

Eritrean soldiers were present in and around Adigrat, eastern Tigray, in the first week of October 2026.

Claimed by Adigrat residents and a local official, separately to Reuters, AFP and the BBC; ACLED's East Africa regional expert · source rating B2

likely 55–80%

Three news organisations found residents describing the same thing, and a named conflict researcher places the crossing at Zalambessa. Ethiopia has demanded a withdrawal and Eritrea did not deny the crossing.

Supporting evidence

Al Jazeera Yahoo ACLED Africanews

What complicates it

Every civilian account is anonymous, and Adigrat is small enough that three outlets may have spoken to overlapping people. Identification rests on uniform colour and lorry plates seen from a distance. Nothing we can measure from orbit confirms or contradicts it.

What would change our mind

Photography or video with recoverable location and date, a named on-the-record witness, a clear Eritrean denial, or reporting that the accounts trace back to a single source.

Eritrea sent those forces to support the TPLF-led alliance against the Ethiopian federal government.

Claimed by The Ethiopian government, including minister Abraham Belay · source rating C4

roughly even chance 45–55%

Nobody has established this. Ethiopia says Eritrea is backing the TPLF-led alliance; ACLED's regional expert says Eritrea wants the TPLF eliminated and moved to keep Ethiopian forces away from its border.

Supporting evidence

ACLED

What complicates it

Troops being present says nothing about why. The two accounts are close to opposites and both come from interested or hedged sources. Treat any confident statement of Eritrea's intentions, including ours, as unsupported.

What would change our mind

Documented co-ordination between Eritrean and alliance units, captured orders, or Eritrean forces engaging the TPLF rather than alongside it.

Ethiopian drones struck vehicles carrying Eritrean soldiers at Guada, on the road towards Idaga Hamus.

Claimed by ACLED's East Africa regional expert, and a resident of Edaga Hamus who described strikes nearby · source rating B3

likely 55–80%

Satellite imagery on 8 October shows a destroyed lorry on the named road, and analysts geolocated video from 7 October to the same stretch and dated it no earlier than 2024, so it is not recycled from the last war.

Supporting evidence

ACLED Yahoo Criticalthreats

What complicates it

The imagery is theirs, not ours — a single lorry is far below what the free imagery we hold can resolve, so we cannot reproduce this. It establishes that vehicles were struck on that road. It does not establish that they were Eritrean, and CTP says the same.

What would change our mind

A resolvable location for Guada, imagery of destroyed vehicles, or an Ethiopian or Eritrean statement about the strike.

Sources

Everything this rests on

report — Al Jazeera carrying Reuters, 7 October. At least three Adigrat residents describe dozens of soldiers in two lorries driving through the town on Wednesday, identified by off-white camouflage uniforms. Reuters states it could not independentl archived copy
report — Africanews, 7 October, on the reported deployment. Carried here as a second publication of the Reuters account, not as corroboration of it. archived copy
report — BBC reporting of 7 October, retrieved via Yahoo syndication. Residents describe convoys of between twenty and sixty lorries carrying Eritrean registration plates; a resident of Edaga Hamus reports troops passing and drone strikes nearby. Qu archived copy
research — ACLED expert comment of 7 October by Dr Jalale Getachew Birru, Regional Expert for East Africa. States that Eritrean forces crossed at Zalambessa and moved along the Adigrat-Idaga Hamus road, and that Ethiopian drones targeted them at Guada archived copy
thermal — NASA FIRMS active-fire detections within 75 km of Adigrat for 30 September to 7 October 2026, from the open seven-day global VIIRS products (375 m, NOAA-20 and S-NPP) and MODIS (1 km). Total detections in that radius across all three sensor regenerated from a recorded query
research — Critical Threats Project Africa File, 8 October 2026. States that satellite imagery from 8 October showed a destroyed truck on the road between Adigrat and Idaga Hamus, "confirming ACLED and AFP reporting and at least one of the videos clai archived copy

Captured at: 2026-10-01 07:56:00+00:00
Timestamp basis: platform
Licence: Copernicus Sentinel data 2026, processed by ESA. Free and open under the Copernicus licence; redistribution permitted with attribution.
Reproduction procedure: earth-search sentinel-2-l2a, scene S2A_37PER_20261001_0_L2A, windowed to a 12 km box centred on 14.2774, 39.4615.
Upstream sources: ["Element 84 Earth Search v1 / AWS Open Data sentinel-cogs"]
Artifact type: satellite_rgb
Sensor: Sentinel-2 (S2A) MSI L2A
Acquisition start: 2026-10-01T07:56:00Z
Source scene id: S2A_37PER_20261001_0_L2A
Source url: https://earth-search.aws.element84.com/v1/collections/sentinel-2-l2a/items/S2A_37PER_20261001_0_L2A
Source catalog: Element 84 Earth Search v1 / AWS Open Data sentinel-cogs
Artifact path: media/artifacts/61586af185aca5fb.png
Sha256: 61586af185aca5fbf7cc640f0078389db345377fa2195aff56829390d52dcd1f
Processing: ["earth-search STAC query (sentinel-2-l2a)", "windowed read of asset 'visual' at native resolution (EPSG:32637)", "bands 1-3 as RGB", "PNG encode"]
Bbox: [39.405, 14.223, 39.518, 14.331]
Crs: EPSG:4326
Resolution m: 10.0
Cloud cover pct: 0.6
License: Copernicus Sentinel data 2026, processed by ESA. Free and open under the Copernicus licence; redistribution permitted with attribution.
Redistributable: True
Detectability note: Sentinel-2 true colour at ten metres per pixel. A military lorry is roughly two and a half metres by nine, about a quarter of the area of one pixel, so neither a single vehicle nor a column of them can be resolved. This imagery could show a large vehicle park, new earthworks, a burn scar more than about thirty metres across, destroyed buildings, or a smoke plume caught at the moment of overpass. It cannot show a convoy on a road, and absence of change in it is not evidence that no convoy passed.

Satellite image · 7 Oct 2026

Sentinel-2 true colour over Adigrat, 7 October 08:06Z, the morning residents describe Eritrean lorries passing through the town. Published because we looked, not because it settles anything.

Technical provenance

Captured at: 2026-10-07 08:06:00+00:00
Timestamp basis: platform
Licence: Copernicus Sentinel data 2026, processed by ESA. Free and open under the Copernicus licence; redistribution permitted with attribution.
Reproduction procedure: earth-search sentinel-2-l2a, scene S2B_37PER_20261007_0_L2A, windowed to a 12 km box centred on 14.2774, 39.4615.
Upstream sources: ["Element 84 Earth Search v1 / AWS Open Data sentinel-cogs"]
Artifact type: satellite_rgb
Sensor: Sentinel-2 (S2B) MSI L2A
Acquisition start: 2026-10-07T08:06:00Z
Source scene id: S2B_37PER_20261007_0_L2A
Source url: https://earth-search.aws.element84.com/v1/collections/sentinel-2-l2a/items/S2B_37PER_20261007_0_L2A
Source catalog: Element 84 Earth Search v1 / AWS Open Data sentinel-cogs
Artifact path: media/artifacts/cca159a46658dae7.png
Sha256: cca159a46658dae7d77fcda459654c24584f272b880ee8407b577c5f71c7cd09
Processing: ["earth-search STAC query (sentinel-2-l2a)", "windowed read of asset 'visual' at native resolution (EPSG:32637)", "bands 1-3 as RGB", "PNG encode"]
Bbox: [39.405, 14.223, 39.518, 14.331]
Crs: EPSG:4326
Resolution m: 10.0
Cloud cover pct: 11.8
License: Copernicus Sentinel data 2026, processed by ESA. Free and open under the Copernicus licence; redistribution permitted with attribution.
Redistributable: True
Detectability note: Sentinel-2 true colour at ten metres per pixel, acquired about five hours after the time residents describe. A military lorry covers about a quarter of one pixel, so no column of vehicles on a road can be resolved at this scale whatever the weather. On top of that limit, the scene classification band flags 41.4 per cent of this area as cloud or cloud shadow, including the road corridor north-east of the town — the scene is listed as 11.8 per cent cloud, but that figure is for the whole hundred-kilometre tile, not for here. Nothing should be read into this image either way.

Fire-detection data

NASA FIRMS active-fire detections within 75 km of Adigrat for 30 September to 7 October 2026, from the open seven-day global VIIRS products (375 m, NOAA-20 and S-NPP) and MODIS (1 km). Total detections in that radius across all three sensors and the whole window: zero. The same files carry 8,573 detections inside Ethiopia over the window, 1,336 of them on 7 October, and 89 inside a wider Tigray box.

Technical provenance

Timestamp basis: unknown
Reproduction procedure: Stream the open seven-day global CSVs for noaa-20-viirs-c2, suomi-npp-viirs-c2 and modis-c6.1 from firms.modaps.eosdis.nasa.gov/data/active_fire/, filter to Ethiopia (lat 3.0-15.5, lon 32.5-48.5), then to within 75 km of 14.2774, 39.4615. No API key is needed. The window is ROLLING: after 2026-10-14 these files no longer cover 7 October.
Upstream sources: ["NASA FIRMS"]

Satellite image · 1 Oct 2026

Sentinel-2 true colour over Zalambessa, the border town where ACLED places the crossing, 1 October 07:56Z. Nothing in this box is flagged as cloud or shadow.

Technical provenance

Captured at: 2026-10-01 07:56:00+00:00
Timestamp basis: platform
Licence: Copernicus Sentinel data 2026, processed by ESA. Free and open under the Copernicus licence; redistribution permitted with attribution.
Reproduction procedure: earth-search sentinel-2-l2a S2A_37PES_20261001_0_L2A, 2.5 km radius on 14.5256, 39.3871
Upstream sources: ["Element 84 Earth Search v1 / AWS Open Data sentinel-cogs"]
Artifact type: satellite_rgb
Sensor: Sentinel-2 (S2A) MSI L2A
Acquisition start: 2026-10-01T07:56:00Z
Source scene id: S2A_37PES_20261001_0_L2A
Source url: https://earth-search.aws.element84.com/v1/collections/sentinel-2-l2a/items/S2A_37PES_20261001_0_L2A
Source catalog: Element 84 Earth Search v1 / AWS Open Data sentinel-cogs
Artifact path: media/artifacts/02d9fdd6216083b9.png
Sha256: 02d9fdd6216083b97781b3a26bd97f4146d390e10f5adc49c40b7dc7b9d6d573
Processing: ["earth-search STAC query (sentinel-2-l2a)", "window centred on Zalambessa, 2.5 km radius", "windowed read of asset 'visual' at native resolution (EPSG:32637)", "bands 1-3 as RGB", "PNG encode"]
Bbox: [39.363901, 14.503142, 39.410299, 14.548058]
Crs: EPSG:4326
Resolution m: 10.0
Cloud cover pct: 5.4
License: Copernicus Sentinel data 2026, processed by ESA. Free and open under the Copernicus licence; redistribution permitted with attribution.
Redistributable: True
Detectability note: Sentinel-2 true colour at ten metres per pixel over a 5 km box on Zalambessa, the border town where ACLED places the crossing. These are the nearest clear views to the event, not views of it: on 7 October, the day Eritrean forces are reported to have come through, the scene classification layer flags one hundred per cent of this box as cloud. There is no imagery of the crossing and there will not be. What these frames show is the ground: the town strung along the road south, and the road itself. A lorry is about a quarter of one sample here, so no column of vehicles could be counted even on a clear day. The dark wedges at the edges are outside the satellite's swath for these granules and carry no data; they differ between the two dates, so nothing at the frame edges should be compared. The town itself is fully covered on both.

Satellite image · 4 Oct 2026

The same box on 4 October 07:55Z, three days before the reported crossing. 6.6 per cent of the box is flagged as cloud or shadow; the town is clear.

Technical provenance

Captured at: 2026-10-04 07:55:00+00:00
Timestamp basis: platform
Licence: Copernicus Sentinel data 2026, processed by ESA. Free and open under the Copernicus licence; redistribution permitted with attribution.
Reproduction procedure: earth-search sentinel-2-l2a S2B_37PES_20261004_0_L2A, 2.5 km radius on 14.5256, 39.3871
Upstream sources: ["Element 84 Earth Search v1 / AWS Open Data sentinel-cogs"]
Artifact type: satellite_rgb
Sensor: Sentinel-2 (S2B) MSI L2A
Acquisition start: 2026-10-04T07:55:00Z
Source scene id: S2B_37PES_20261004_0_L2A
Source url: https://earth-search.aws.element84.com/v1/collections/sentinel-2-l2a/items/S2B_37PES_20261004_0_L2A
Source catalog: Element 84 Earth Search v1 / AWS Open Data sentinel-cogs
Artifact path: media/artifacts/16bab5b7efa76dff.png
Sha256: 16bab5b7efa76dff4c5577bdcf8589088c6d5795e1a7463b3496239ed34dec18
Processing: ["earth-search STAC query (sentinel-2-l2a)", "window centred on Zalambessa, 2.5 km radius", "windowed read of asset 'visual' at native resolution (EPSG:32637)", "bands 1-3 as RGB", "PNG encode"]
Bbox: [39.363901, 14.503142, 39.410299, 14.548058]
Crs: EPSG:4326
Resolution m: 10.0
Cloud cover pct: 27.9
License: Copernicus Sentinel data 2026, processed by ESA. Free and open under the Copernicus licence; redistribution permitted with attribution.
Redistributable: True
Detectability note: Sentinel-2 true colour at ten metres per pixel over a 5 km box on Zalambessa, the border town where ACLED places the crossing. These are the nearest clear views to the event, not views of it: on 7 October, the day Eritrean forces are reported to have come through, the scene classification layer flags one hundred per cent of this box as cloud. There is no imagery of the crossing and there will not be. What these frames show is the ground: the town strung along the road south, and the road itself. A lorry is about a quarter of one sample here, so no column of vehicles could be counted even on a clear day. The dark wedges at the edges are outside the satellite's swath for these granules and carry no data; they differ between the two dates, so nothing at the frame edges should be compared. The town itself is fully covered on both.

Criticalthreats

Critical Threats Project Africa File, 8 October 2026. States that satellite imagery from 8 October showed a destroyed truck on the road between Adigrat and Idaga Hamus, "confirming ACLED and AFP reporting and at least one of the videos claiming that Ethiopian drone strikes had targeted suspected Eritrean forces in the area". CTP and ISW geolocated media posted by pro-Ethiopian government accounts on 7 October to Idaga Hamus and that road, and determined the media was from no earlier than 2024, so not reused footage from the last war. CTP states it could not verify the exact time most of the media was captured, or whether the personnel and vehicles shown are Eritrean military.

Original source Archived copy

Technical provenance

Timestamp basis: unknown
Archived at: 2026-10-09 17:39:27+00:00
Archive method: wayback-save-page-now
SHA-256: 4b78b03b2e703693a6a462e1a772819c8bf4a4549b22c9e11b37d96eb3555253
Upstream sources: ["Critical Threats Project", "ISW", "ACLED", "AFP"]