Flood Frequency in a Douala Wetland District

The cause is the catchment, not the rainfall. The frequency has risen from two to seven damaging floods a year while totals are flat, which points to changes in how water reaches the channel and how much channel capacity is available.

Runoff volume and timing. Take a 40 mm storm over 100 ha, giving 40,000 cubic metres of rain. At 35% impervious cover a mixed runoff coefficient of about 0.35 produces roughly 14,000 cubic metres; at 68% impervious, near 0.6, it produces about 24,000 cubic metres. That is a 70% increase in volume, and the peak arrives sooner because paved surfaces and roofs deliver water in minutes rather than hours. The 1990s drains were sized for the older, slower hydrograph.

Loss of storage. Filling 12 ha of wetland removes a natural reservoir and the slow release it provided. Infilling also raises ground levels on some plots, so water that once ponded harmlessly now runs onto neighbouring land.

Reduced outlet capacity. The canal is tide-locked daily, so during high tide the hydraulic gradient is small and water backs up. Blocked culverts at the market road reduce capacity further, effectively narrowing the bottleneck that the whole network depends on.

Exposure. More households occupy the lowest ground than in the 1990s, so the same water depth now counts as a damaging flood. Part of the rising count is a measurement effect.

Conclusion. The district has been caught by a land-use change of its own making combined with a fixed drainage standard and a tidal outlet. Blaming rainfall would misdiagnose the problem and lead to the wrong fix.

Option 1: widen the outfall. Directly increases the bottleneck capacity and helps the largest storms, which upstream measures cannot. But it is expensive, requires resettling 240 households, and resettlement costs and political resistance are real. It also treats the symptom and may encourage more building in the wetland.

Option 2: retention plus permeable paving plus waste service. About one-third the cost, no displacement, and it attacks the cause by cutting the peak and keeping culverts open. Its limit is important: if the outfall is the binding constraint, upstream storage will not prevent flooding in the biggest events, and the benefits depend on the waste service actually running.

Option 3: relocate 300 households. Removes the most exposed people from danger and can be the cheapest way to eliminate risk for those households, but it breaks social networks and livelihoods, requires land that is not itself flood-prone, and is often resisted unless compensation is credible.

Recommendation. Combine options 2 and 3 with strict maintenance of existing culverts. Upstream storage and paving reduce the frequency of moderate floods, the waste service restores lost capacity at negligible cost, and relocating the lowest 300 households handles the residual risk that storage cannot remove. Defer widening the canal unless monitoring shows the outlet is still the binding constraint after those works.

Who bears the burden. The beneficiaries of wetland filling are developers and wealthier plot-holders; the burden falls on the poorest, who occupy the lowest ground. Funding should therefore come from a district levy on new development and from the waste service charge, not solely from general taxation. That reasoning is as important as the engineering choice.

What the standard means. A 10-year design means a 10% chance of exceedance in any year, and roughly a 65% chance over a 10-year life. That was tolerable where assets were low-value and land was available to flood. In a district of 4,200 households, with a water table 1.5 m down and a tide-locked outlet, exceedance means homes flooded, not fields.

Why it has aged badly. Three things changed after the drains were built: impervious cover roughly doubled, so the same storm now produces a much larger peak; the wetland storage that used to absorb exceedance was filled; and exposure rose sharply. Sea-level rise and continued infilling add further pressure. A standard set for a 1990s catchment is therefore no longer a statement about acceptable risk.

What should replace it. Not simply a 100-year pipe, which is unaffordable in this setting and creates a false sense of security that attracts more building in the wetland. A risk-based approach is better:

• Zoning. Map depth and frequency and prohibit occupation of the lowest, most frequently flooded land, so the drainage standard governs an area where damage is limited.
• Layered defences. Design the piped network for a modest event, and accept controlled ponding in designated areas for larger ones.
• Performance monitoring. Record water levels and flood depths, and review the standard when land use changes, rather than assuming a fixed return period forever.
• Maintenance budget. Capacity is only real if culverts are cleared; a design standard without a maintenance line is aspirational.

Judgement. The defensible answer is that the standard should follow the consequences of failure, not a fixed number. In the lowest zone, no affordable drain can guarantee protection, so land use must do the work.