Risk information for flood early warning and anticipatory action
A warning works when the people issuing it know what a river level means on the ground: how rare it is, how far the water will spread and who lives there. I produce that risk information. For ten incident weeks I also used it, as Flood Warning Duty Officer for south-east London.
- Under 1,000
- 1,000–5,000
- 5,000–20,000
- 20,000–60,000
- 60,000 or more
How to read this: each shape is a district of Ghana; the brighter it is, the more people live inside the modelled 1-in-100-year river flood, 1.18 million in all. That is the “who” a warning needs. Modelled, river flooding only; not official figures.
- incident weeks as Flood Warning Duty Officer for south-east London, Environment Agency
- 10
- people inside Ghana’s modelled 1-in-100-year river flood, about 421,700 of them children
- 1.18million
- years of annual maximum river flows analysed for one English village’s design flood
- 50
- sections of Ghana’s coast, each with its own extreme sea levels by return period
- 57
About the author. Hisham Nuhu is a UK-based flood and disaster risk specialist. He worked at England’s Environment Agency from 2019 to 2025 and holds an MSc in Water Science, Policy and Management from the University of Oxford. He works remotely or in-country with governments, development banks, UN agencies and consulting firms, and in England with councils, communities, landowners and developers.
How do you commission a flood early warning system?
Write all four pillars of Early Warnings for All into the terms of reference, each with an owner and a test: risk knowledge, monitoring and forecasting, warning dissemination, and preparedness to respond. My part is the first pillar and its hand-over to the second: the risk information that tells forecasters what a given level means on the ground.
Warning systems often fail at the joins between pillars rather than inside them. A forecast that nobody can translate into “these streets, these people” is ignored; a warning without a plan behind it frightens people without helping them.
So ask, for each pillar: who owns it, what it hands to the next one, and how you will test it before the flood season. The figure shows what I would put in the terms of reference, and where my own record sits.
The four pillars of early warning: what to specify, and my record on each
Disaster risk knowledge
Flood maps, return periods, and who and what each flood reaches, for every warning area. My record: Ghana’s 260-district screen; the Hamstall hydrology and model.
Monitoring and forecasting
Gauges, rainfall, forecasts and the thresholds they are checked against. My record: on duty I checked forecasts and live river levels against thresholds, including overnight alerts.
Warning dissemination
Who is warned, through which channels, in what words. My record: issued flood alerts and warnings through the Environment Agency’s warning service, which sends them by phone, text and email.
Preparedness and response
Plans that turn a warning into action by responders and residents. My record: calls with first responders; police and ambulance services kept involved through an escalation.
How to read this: the pillars light up in turn; tap, or tab to, any pillar to read what to specify and what I have done on it.
Pillar names shortened from the UN Early Warnings for All initiative: disaster risk knowledge and management; detection, observation, monitoring, analysis and forecasting; warning dissemination and communication; preparedness and response capabilities.
What risk information does a flood early warning system need?
For every warning area: how often each flood size happens, how far and how deep it spreads, who and what is in the way, and how much harm each depth does. Ghana’s model shows the scale. If every river in the country reached its 1-in-10-year level at once, the damage to homes and contents would be about GHS 38.5 billion; at the 1-in-500-year level, about GHS 80.8 billion.
This frequency–severity curve is what impact-based forecasting rests on: for a forecast flood of a given size, the likely harm. Ghana’s curve is national. A warning service needs the same curve for each warning area, and the method that built the national one can build those too.
People matter as much as money. About 1.18 million people live inside Ghana’s modelled 1-in-100-year flood. Ranked by people, Keta comes first; ranked by money, Ningo/Prampram. A warning service needs both lists.
The frequency–severity curve: rarer floods do more harm
How to read this: left to right, floods get rarer and bigger. The line is the median damage and the band its P5–P95 range; the dashed lines are two other flood edges. Tap a flood size, or use the arrow keys, to read it.
Modelled damage to homes and contents from one river flood across all 260 districts of Ghana, JRC Africa damage curve.
Show the data as a table
| Flood | Narrower flood edge, GHS bn (P5–P95) | Flood edge as mapped, GHS bn (P5–P95) | Wider flood edge, GHS bn (P5–P95) |
|---|---|---|---|
| 1 in 10 years (10% a year) | 19.5 (16.1–23.4) | 38.5 (31.8–46.1) | 58.5 (48.3–70.1) |
| 1 in 50 years (2% a year) | 35.5 (29.3–42.6) | 60.6 (50.1–72.6) | 83.5 (69.0–100.1) |
| 1 in 100 years (1% a year) | 40.5 (33.4–48.5) | 67.2 (55.5–80.5) | 90.7 (74.9–108.7) |
| 1 in 200 years (0.5% a year) | 44.4 (36.7–53.2) | 73.1 (60.4–87.6) | 97.5 (80.5–116.8) |
| 1 in 500 years (0.2% a year) | 49.5 (40.9–59.4) | 80.8 (66.7–96.8) | 106.2 (87.7–127.3) |
| Annual average (1 in 10 to 1 in 500) | 2.93 a year (2.42–3.52) | 5.19 a year (4.28–6.21) | 7.33 a year (6.05–8.78) |
How are warning thresholds set from return periods and exceedance probabilities?
A threshold is a river level or flow with a known chance of being exceeded and a known impact. The frequency curve gives the chance; the flood model gives the impact. At Hamstall Ridware, 50 years of annual maxima fix how rare each flow is, and the 2024 model shows 14 buildings flooded at 1% AEP, 29 at 0.1% AEP.
Return period and exceedance probability say the same thing two ways. A 1-in-100-year flood has a 1% chance of being exceeded in any year. Over 30 years, the chance of at least one is about 1 in 4.
Choosing the trigger is a trade-off between missed floods and false alarms. Set it low and people stop listening; set it high and the warning comes too late. The numbers inform that choice. The warning service and the communities it serves make it.
Check the gauge before you trust its curve. The Hamstall gauge is bypassed above about 13–14.5 m³/s, so its record flattens exactly where a warning threshold for a big flood would sit.
Every flow has a chance of being exceeded; the gauge flattens where water bypasses it
How to read this: each dot is one year’s highest flow, placed by how rare it is; read across from a return period to its flow. Switch the pooled estimate on and off; hover, tap or use the arrow keys to read each year.
River Blithe at Hamstall Ridware, gauge 28002, 50 water years: Gumbel fit to the gauge record and the pooled estimate for the whole catchment.
What is impact-based forecasting?
Forecasting what the flood will do, not only what the river will do: “these streets and these people” rather than “3 m at the gauge”. England’s Flood Guidance Statement works this way. Every day it rates flood risk by combining the likelihood and the impact of flooding, from green to red. I worked from it on every duty day.
Before a trigger goes live, run it against past floods. This is a hindcast. For Ghana’s national map I did it with the September 2020 White Volta flood: the model put 6,371 people inside the satellite-mapped flood, and NADMO, the national disaster agency, reported 7,561 affected in the Upper East Region. The satellite test areas lie further south, so the two counts do not cover the same ground: a sign the people layer is in a plausible range, not a match.
Impact data is usually the weak link: who was affected, where, and how badly. A hindcast tests the impact data as much as the forecast.
What does a flood warning duty officer need from the risk information?
Thresholds they trust, and a quick answer to “who will be affected?”, often at night. On duty for south-east London I watched rainfall and river telemetry, checked IMFS forecasts and the Flood Guidance Statement against thresholds, issued flood alerts and warnings, and kept police and ambulance services involved as the risk level moved between yellow and red.
An overnight telemetry alert needs a decision quickly. What helps then is a threshold with a clear impact attached and a record of what happened last time at that gauge. A number nobody can explain does not help.
After each event I wrote it up, so the decisions and timings were on record for review. Records like these are what a later review of thresholds draws on.
Where does this fit in Early Warnings for All and anticipatory action?
In pillar 1, disaster risk knowledge, and its hand-over to pillar 2, forecasting. Anticipatory action needs a trigger agreed in advance: a forecast chance of exceeding a level whose impacts are known. My work supplies the second half of that sentence: the levels, how rare they are and who they reach.
FrameworksUN Sendai Priority 4, preparedness and people-centred early warning; Early Warnings for All pillars 1 and 2ADB preparedness; flood forecasting, warning and gaugesWorld Bank hydromet and early warning: forecasting, warning and emergency response
Alignment of language and method, not certification or endorsement.
What can you do for a warning project, and what do you not do?
I produce return-period and exceedance-probability analysis, frequency–severity curves, impact estimates for each warning area, hindcasts against past floods, and a plain-language note on proposed thresholds. I do not configure forecasting systems: I was an operational user of the Environment Agency’s IMFS, which runs on Delft-FEWS.
- What I offer
- Flood frequency and extreme-value analysis; flood maps and exposure for each warning area; damage and people estimates by flood size; hindcasts against satellite or field records; training for the people who will use the thresholds.
- What sits with others
- Configuring forecasting software such as Delft-FEWS; telemetry networks; the warning service’s own decision on where to set triggers.
- My operational record
- Ten incident weeks on the Environment Agency rota for south-east London, 2023–25. Details on the duty officer page.
Questions people ask
Have you set national warning thresholds?
No. I have worked with thresholds operationally, as a duty officer, and I produce the risk analysis that informs where they should sit. The decision belongs to the warning service.
Does a 1-in-100-year flood happen once a century?
No. It has a 1% chance in any year, so it can happen two years running. Over 30 years the chance of at least one is about 1 in 4.
Can Ghana’s coastal sea levels be used for warning triggers?
Not yet. The coastal levels come from a screening with provisional forcing. Trigger levels need the next stage, 2D models of the priority sites. See coastal extreme water levels and inundation screening.
Can you train warning-service or disaster-agency staff?
Yes, as part of an assignment. My record so far is two internal Environment Agency courses, for 10 and 3 colleagues.
Sources
- United Nations. Early Warnings for All: the four pillars and the 2027 target.
- World Meteorological Organization. Early Warnings for All.
- UNDRR (2015). Sendai Framework for Disaster Risk Reduction 2015–2030.
- Flood Forecasting Centre. Flood Guidance Statement user guide.
- Environment Agency. Flood alerts and warnings: what they are and what to do.
- UK Centre for Ecology & Hydrology and Environment Agency. National River Flow Archive, station 28002.
- Copernicus Emergency Management Service. Global Flood Monitoring, Sentinel-1 flood extent, September 2020.
- FloodList (20 September 2020). Ghana: 5 dead after floods in Upper East Region, quoting NADMO: 7,561 people affected and 1,596 homes damaged in the Upper East Region.
- Nuhu, H. (2026). National fluvial flood-risk assessment and priority stability under uncertainty in Ghana. MSc dissertation, University of Oxford. Method and results.
Work with me
Planning a flood early warning system?
Send the terms of reference or the concept note. I will reply with where the risk information fits, what it needs and when I am available, remotely or in-country.