Inside the camps the aid reports don't mention

Nairobi — The clearest sign that something is missing from the official record is a grid. On a high-resolution image captured in March, a flat stretch of semi-arid scrub near the eastern border resolves into rows: hundreds of rectangular shelters laid out with the unmistakable regularity of a planned settlement, footpaths worn pale between them, a darker smear of standing water at the low end. The coordinates fall inside a district that three separate humanitarian situation reports, published the same month, describe as hosting "no formal displacement sites." The grid was not there in an image from the previous September. By the time the reports went out, roughly four thousand people appear to have been living on it.

Meridian spent two months comparing commercial satellite imagery, publicly filed coordinates and the situation reports issued by the agencies coordinating the response. The exercise is deliberately narrow. We are not alleging that anyone is being hidden on purpose. We are documenting a gap — places visible from orbit that do not appear in the documents meant to direct food, water and medicine toward them.

Reading the ground from orbit

Satellite analysis of displacement is now routine, but it is not magic. A shelter is a roof, and a roof is a few pixels of contrast against bare earth. To count people, analysts count structures and apply an occupancy estimate; to confirm a structure is occupied, they look for the surrounding evidence — cooking-fire scarring, latrine pits, the braided tracks that thousands of feet leave on dry ground, the way a market stall throws a shadow at the right hour. None of this is conclusive on its own. Combined, and checked across several dates, it is hard to fake and hard to miss.

We catalogued every cluster of more than fifty apparent shelters within the response area, then cross-referenced each against the coordinates listed in the official site directories. The directories are not secret; they are the backbone of the system, the master list against which deliveries are planned. Most clusters matched a registered site cleanly. Eleven did not.

A roof is a few pixels of contrast against bare earth — but thousands of feet leave tracks that are hard to fake and hard to miss.

The eleven clusters that do not appear

Of the eleven, three sit close enough to a registered camp that they are plausibly informal overspill — people settling at the edges of an official site faster than the registers can record them. That is common, and broadly understood. The other eight are different. They are kilometres from any listed site, in several cases astride a seasonal road, and they grew over the same months that the formal camps were reported as full.

Applying the same occupancy estimate the response itself uses — between four and six people per shelter — the eight detached clusters together suggest a population in the low tens of thousands. That figure carries a wide margin, and we present it as a range, not a headline. The point is not the precise number. The point is that the documents record it as zero.

What the statements say

The situation reports are careful, bureaucratic documents, and their language repays close reading. They do not claim no one lives in these areas. They claim no formal sites exist there — which is technically accurate and practically misleading, because the formal/informal distinction governs who counts as a beneficiary.

Access constraints are real. The roads are bad, the rains close them for weeks, and security clearance for field teams can take longer than the dry season lasts. But a constraint that prevents a ground visit does not prevent a satellite pass. The imagery we used is commercially available and was captured on schedule throughout the period the assessment was deferred.

The honest answer is that if a site is not assessed, it does not enter the figures, and if it is not in the figures, no one is formally responsible for it. That is not a conspiracy. It is how the machinery is built, and the machinery has blind spots we know about and rarely fix.Former humanitarian information officer, who worked on three regional responses

Why the gap persists

Several people who have worked inside these systems described the same loop, in nearly the same words. Verification requires physical access; access is blocked; without verification a population cannot be entered into the planning figures; and resources are allocated against the planning figures. A settlement can be entirely visible and entirely unfunded at once.

• None of the eight detached clusters appears in any site directory we obtained for the response period.
• All eight are visible across at least three separate imaging dates, ruling out cloud artefacts or transient activity.
• Each shows ground-level signatures of habitation — track networks, latrine pits, fire scarring — not merely roofs.
• The largest predates the March report by roughly five months.

We are not the first to flag this. Independent verification groups have published similar findings in other responses, and the coordinating bodies have, in places, begun to integrate satellite layers into their site tracking. The shift is slow and uneven, and it tends to follow public attention rather than precede it.

What remains open

What our imagery cannot tell us is who these people are, where they came from, or how they are surviving without the formal channels. It cannot distinguish a family that fled fighting from one moving with livestock, and it cannot read a death rate off a roofline. Those answers require the ground visit that has not happened. We have shared our coordinates with two agencies operating in the area; one acknowledged receipt, and said an assessment was "under consideration."

The narrow finding stands regardless. Settlements large enough to register from space, populated for months, do not appear in the documents that decide where aid goes. The reports are not false. They are incomplete in a way that is structurally invisible to the people reading them — and the gap falls precisely on those least able to make themselves seen. Whether the next round of reports closes it is, for now, the open question.