A hundred small defenses; Why Accra’s floods demand distributed resilience, not another great pond

When the rains came to Accra

On April 18, 1960 the Daily Graphic‘s front page read: When the rains came to Accra. Sixty-six years later they came again. On June 29, 2026 the Odaw overflowed into Odawna, Alajo, Nima and Circle; Kaneshie market drowned; at least 12 died, seven vanished, nearly 40,000 were displaced. The same system killed 20 more in Côte d’Ivoire.

What struck was the familiarity, not the scale. The names flooded in 2026 fill the reports of 2023, 2016, 2015, 1997, 1968. “When it rains like this,” one resident said, “we already know the trouble coming.” Accra holds the diagnosis, signed the contracts, and buried its dead, yet unready each June.

We made a modest intuitive inquiry into this issue which resulted in a proposed model. Our non-engineering model (doi.org/10.13140/RG.2.2.19813.77280) asked one question: would large upstream basins stop the flooding? Tested against the calibrated Odaw study (Acheampong, Gyamfi and Arthur, 2023) and first-principles hydrology, both pointed the same way: upstream detention trims the flood marginally, least in the biggest storms, because required storage rises with the square of the cut, and the bottleneck is the choked exit to the sea.

We widened the frame, from the Odaw corridor to all five Greater Accra systems (458 km²), sizing about 120 small basins on mapped depressions and restoring coastal wetlands as a self-cleansing buffer.

Three claims: Accra’s flooding is a solvable systems problem, not weather; the two great ponds are the wrong remedy, the answer nearer its opposite; and, least comfortably, it should not have taken an outsider to say so. All planning-grade, checked against government models before ground breaks.

An honest accounting of the deluge

June 2026 was extreme: Ghana’s 593.2mm was its wettest month since 1995. True, no system erases that. But that’s the wrong test. Resilience means failing gracefully, not preventing. Accra also drowns in ordinary storms, May 2025’s 132mm killed four, including a child in Nima. Extremes are becoming ordinary. “Unprecedented” argues for building now, not less.

The seduction of the great pond

This inquiry did not begin with its conclusion. It began with the intuition almost everyone reaches for: too much water arrives at the bottom of the catchment too fast, so intercept it at the top, one or two large basins in the hills, releasing slowly, flattening the peak that hits the streets. It is elegant, monumental, and it feels like decisive action.

In Accra it has a concrete form: two proposed detention basins in the upper Odaw at Kwabenya, in the Atomic area, named due to its proximity to the Ghana Atomic Energy Commission that sits there. Their history is the history of the modern response to these floods. They were conceived after the 3 June 2015 flood-and-fire that killed more than 150 people at Circle; from that catastrophe came the World Bank-backed Greater Accra Resilient and Integrated Development project (GARID), prepared from 2016, approved in 2019, meant to run to 2025 and now extended to 2027. Within it, the two Atomic ponds were the flagship storage element, valued at roughly US$44 million. And it is only fair to say that Accra has not flooded for want of thought: its engineers have understood the mechanics since drainage master plans of 1963, 1991 and 1995; GARID has done real work on dredging, waste and early warning; and the science this argument leans on is Ghanaian and recent. The trouble is not a failure of intelligence. It is a failure of framing; and the frame was the great pond.

Here is the first hint of that failure, after a decade, the ponds are not built. They reached design and procurement, then largely stalled. But set the delay aside and ask the deeper question this inquiry eventually learned to ask, not whether the ponds can be delivered, but whether they would work.

How the question changed

Two findings answer it; one from Ghana’s calibrated science, one from physics.

First, empirically: Acheampong, Gyamfi and Arthur’s 2023 Odaw model didn’t assume detention would work; it simulated the two ponds across worsening storms. They cut the flood peak by only 8–13% and its volume by 2–4%, and the gain shrank as storms grew, sliding from 13% (mild) to 8% (severe), because a fixed basin fills and then spills over. The defense fades when needed most.

Second, the crux: storage rises with the square of the fraction cut 30% costs 12% of the storm, half a third, 70% two-thirds. At one point, storage faces the whole basin’s runoff (22 million cubic metres in June), so even a 30% cut needs 2.5 million; the ponds hold barely 0.2 million. Yet the law rewards spreading: small basins where the city floods, Circle, Kaneshie, Alajo ,clip local peaks at the gentle end, not the savage one.

Third, the reframing: even a perfectly delayed flood still arrives. If the channel and lagoon can’t carry it out, delay just buys hours. The Odaw ends in a silted, waste-choked lagoon behind a tidal sandbar, and Ghana’s studies show a modelled trash barrier there reproduces the upstream flooding alone. The binding limit isn’t storage but conveyance—getting water out.

The scrutiny GARID deserves

Arguing for a different approach means scrutinising the dominant one fairly, GARID has done real work, but without flinching. Scope first: GARID centres on the Odaw, carrying about 60% of Accra’s storm water, leaving the other 40% Densu and Weija west, Kpeshie, Sakumo, Ashaiman, Chemu east, outside the flagship. Weija, Tse Addo, Ashaiman, Kpone flood too. Then the storage: a US$44 million marquee built around an instrument the science shows marginal, worst when stakes are highest, the central weakness.

On execution we stay even-handed. After the flood, a bitter dispute erupted: some analysts read the data as nine of every ten GARID cedis going elsewhere; others, citing the World Bank, said funding was secured and 40 per cent of works done while the current government left funds idle. We take no side, both describe one pathology: concentrated capital is vulnerable to delay and friction. Witness the US$7 million early-warning system, confirmed non-operational when the 2026 rains came. Don’t abandon GARID, but its centre of gravity is misplaced.

The model, and where the basins go

Our modest proposal is one system of four layers, applied across all five basins and listed in the order in which they bind.

Conveyance comes first, nothing works behind a blocked exit: finish dredging the Odaw and primary drains, remove channel encroachment (Old Fadama and Agbogbloshie the starkest), enlarge culverts at the 163 mapped chokepoints, and keep the lagoon mouths open against the sand. Unglamorous, and the highest-value work there is.

Distributed storage swaps the two ponds for about 120 small basins on natural low points, each a hectare and a few metres deep, cheap, quick, parallel, redundant when one silts up. The square law turned to the city’s advantage.

Terminal storage restores the self-cleansing wetlands, Korle, Densu Delta, Sakumo, Kpeshie, Chemu, returning some 22 million cubic metres, a whole central storm, without perpetual dredging.

Source control shrinks runoff at origin. Around all four: a warning system switched on, enforcement holding boundaries.

June yields 36.6 million cubic metres across 458km²; a 30% cut needs 4.3 million, about 120 basins. The 2024 screening mapped 5,018 depressions, land isn’t the constraint. Candidates: doi.org/10.13140/RG.2.2.19813.77280.

Why it took an outsider to say so

A competent client hiring a lawyer doesn’t nurse vague unease; she says, I want this land, give me the strongest legal case. No insult, the precondition for it. A clear aim lets him bring everything; fog in, fog out.

Two jobs blur: framing; what to optimise for, the binding constraint, what success means—and execution, hydraulics, structures, survey. Accra handed engineers both, called flooding technical, blamed them when floods returned. At the Odaw, he builds a bigger channel and two ponds. The mandate forbids challenging the frame: the river is only 60 per cent of the problem, storage the wrong instrument, the real constraints the blocked exit, encroached wetland, and an institution that won’t maintain what it builds. Without reframing, brilliant engineering hits the wrong target.

An honest outsider frames it right: the constraint is conveyance and distributed storage across five systems; the target, clipping the peak by a third with a hundred basins and restored wetlands. Now engineers make it real, elevating them. It’s democratic: cost against safety, concrete against wetland, monument against network—public, not technical. We don’t design the basins; sizing to calibrated models, placement to survey, structures to engineers. The essay stops at the frame, handing execution back.

Designed for the deluge

A great storm defeats any system, but that was never the test. Success is a city that bends, not breaks: water shallower, exits open, lives kept. Accra’s failure isn’t diagnosis or money; its reserves dwarf the fix. It’s the wrong picture, one grand pond, not a hundred small basins. The rains return.

BY AZOSKA SIMEONE, & ENGR. WORMENOR EMMANUEL

Follow our WhatsApp Channel now! https://whatsapp.com/channel/0029VbAjG7g3gvWajUAEX12Q