What Hurricane Harvey can teach Ghana before the next dam spillage. Ghana’s recurring floods and dam spillages do more than damage property. They pull children out of school, out of routine, and often out of any institution’s clear view. Research on Hurricane Harvey offers a warning worth heeding before the next spillage: when several agencies each recognise a displaced child differently, who is responsible for following that child once the water recedes?

By Kwadwo Ampofo-Agyemang

Somewhere in a flood-affected household in northern Ghana right now, there is a child who has been pulled out of school, moved into a relative’s house or a church hall, and told simply to wait for the water to go down. As I write this, the 2026 Bagre Dam spillage is underway in Burkina Faso, and NADMO has already issued flood alerts along the White Volta, urging residents to move to higher ground. This cycle repeats most years. The scene repeats itself with the water.

A far larger, better-documented version of this happened during the 2023 Akosombo Dam spillage, when the Volta River Authority opened the spillways at Akosombo and Kpong after weeks of heavy rain — the highest-magnitude displacement since the dam’s construction, according to the Ghana News Agency. The controlled release displaced an estimated 31,000 to almost 36,000 people across North, South and Central Tongu and surrounding districts. At an October 31 press conference, Child Rights International (Ghana) reported that 19,743 school children — up from an estimate of roughly 10,900 two weeks earlier, as assessments continued — were unable to attend class across 71 affected schools, and its director estimated it would take the state more than three months to get them back into classrooms.

A UNICEF Ghana account from Mepe captured what that gap looked like on the ground: a school headmaster let a flooded-out family store their belongings on the compound, and lessons resumed with children simply “learning under a tree on the school compound.” That image is worth sitting with, because Ghana already had roughly 5,400 schools nationally operating under trees or in dilapidated buildings before the flood ever hit, according to a 2023 civil-society report. The problem wasn’t only that schools closed. It’s that the system meant to absorb displaced children had very little spare capacity to begin with.

What Hurricane Harvey teaches Ghana about displaced children

I recently completed research examining a comparable flood on the other side of the world: Hurricane Harvey’s displacement of families across Houston, Texas, in 2017. Working through more than a dozen official U.S. documents, my research found that each institution recognised a different displaced child. The school system recognised the enrolled student. Child welfare recognised the child already inside its caseload — displacement alone, for a family with no open case, rarely triggered anything. Emergency management recognised the damaged household, with children counted as members of it rather than as people with their own distinct needs. One detail from the Texas record: more than sixteen thousand children were displaced when licensed childcare facilities shut down during the storm, and the record does not show a clear institutional mechanism for tracking what happened to them next.

Fragmented recognition is what I call this: institutions recognising the same child through different administrative categories, with none of them assuming responsibility for that child’s longer-term recovery. They weren’t hidden. They were simply nobody’s job. A child can be known to four different agencies during a flood and still be owned — in the sense of someone tracking their schooling, health and safety through the months of recovery — by none of them.

My Houston research does not establish that Ghana’s institutions operate the same way. It does leave a useful question for Ghana: when several agencies encounter the same displaced child, who is responsible for following that child after the emergency ends? It’s worth asking whether the same gap opens up between NADMO’s household-level displacement counts, the Department of Social Welfare’s case-based mandate, and the Ghana Education Service’s enrolment records — three different views of the same child, with nothing connecting them once the emergency phase ends.

Why recurring flooding matters for children in Ghana

This isn’t only a big-disaster problem. Akosombo was a single dramatic event, but the same pattern shows up in Ghana’s ordinary rainy-season flooding, which returns to the same places most years. On a single weekend at the end of June 2026, flooding displaced 7,761 households and affected more than 38,800 people across Greater Accra, according to a parliamentary report, while the Central Region recorded roughly 9,000 people affected and 377 internally displaced, with the Department of Social Welfare listed among the responding agencies. Reporting on that flood traced Accra’s worst-hit neighbourhoods — around the Odaw River, Alajo, Kaneshie and Weija — back through flood years in 2010, 2015, 2016, 2022, 2023 and 2025. These aren’t outlier catastrophes. They’re the same neighbourhoods, the same families, the same institutional gaps, on a loop.

Here is the detail from my research that matters most for Ghana. Harvey was, in the ordinary sense, unpredictable — nobody could say in June 2017 that a storm would hit in August. Yet the recognition gaps I found were not improvised in the moment; they were fixed years earlier, in a 2015 state emergency-planning document written well before Harvey existed. Ghana has something Houston did not have before Harvey: advance warning that a major release was coming. NADMO announces the Bagre spillage weeks ahead — in 2026, alerts began in early August for a spillage that didn’t reach full discharge until the end of the month — with the Water Resources Commission publishing reservoir levels as it monitors them; the Volta River Authority gave public notice before Akosombo’s gates opened in 2023. If a storm nobody could predict still justified building anticipatory, pre-agreed child-protection arrangements, a dam release Ghana can see coming on a spreadsheet is a harder case to leave to improvisation, year after year.

What NADMO, Social Welfare and GES should change

• Create a single child-displacement tracking protocol. When a household registers as flood-affected — as part of the routine registration NADMO already runs — that registration should trigger a basic, non-investigative check on the children in it: are they enrolled and attending, are they safe, do they need medical follow-up. One protocol, a clearly designated responsible agency, one shared record, agreed between NADMO, the Department of Social Welfare and the Ghana Education Service before spillage season starts, not rebuilt from scratch at every flood.
• Follow children beyond the emergency phase. The clearest harm in the Akosombo record — three-plus months out of school — happened well after the acute emergency ended. A response that stops counting once the flood recedes will miss exactly the damage that matters most.
• Remove documentation barriers. Rural, riverine communities carry a disproportionate share of Ghana’s flood displacement, and they are also where child registration gaps run deepest: UNICEF Ghana estimates that roughly 40 per cent of Ghanaian children under five have no registered birth, concentrated heavily in rural and deprived communities. If assistance quietly requires paperwork many affected families don’t have, the eligibility rule itself becomes the reason some children disappear from the record — a version of what my research found happening in Texas.

The water is not the unpredictable part. What remains undecided is whether, by then, any institution can say it followed one child all the way through — or whether we will simply count them again, in four different places, and call that recognition.

Kwadwo Ampofo-Agyemang recently completed an MA by Research at the University of Huddersfield examining institutional recognition of climate-displaced children after Hurricane Harvey.