Surface water flood forecasting in England explained

If you picture flooding as a river bursting its banks, this story asks you to widen that picture. Surface water flooding happens when intense rain falls faster than the ground, drains and sewers can cope with. Water then rushes across roads, pavements and homes, often with very little warning. That matters because the numbers have shifted sharply. The Environment Agency's latest National Flood Risk Assessment says about 4.6 million properties are now at risk from surface water flooding, a 43% rise since the last assessment. It also says there are now three times as many properties at high risk from surface water flooding as from rivers and the sea. In the National Risk Register 2023, the UK government raised the impact score for this threat to significant. **What this means:** the kind of flooding some people once dismissed as a short-lived street problem is now being treated as one of England's most serious environmental risks.

Climate change is making short, intense summer downpours more likely, and that changes who is exposed. Places that have never thought of themselves as flood-prone can suddenly find water pouring into shops, schools and ground-floor flats, especially in built-up areas where more surfaces are sealed and more people live close together. Recent events show why this is not a remote risk. The flooding in Coverack in 2017 and London in 2021 showed how fast heavy rain can turn into danger. For readers, the key lesson is simple: surface water flooding is not only about where water starts. It is about where water cannot go. That is why better forecasting is being treated as a public-safety job, not just a weather task.

To respond to that, the Flood Forecasting Centre has been running the Surface Water Flood Forecasting Improvement Project, a three-year effort to strengthen national forecasting for rapid flood events. The clearest public result so far is the Rapid Flood Guidance service, which has moved into operational use and now has a commitment to continue until 2028. There is an important shift in thinking here. These services are not only asking whether heavy rain is likely. They are also asking what the rain is likely to do. Both the Flood Guidance Statement and the newer Rapid Flood Guidance service use impact-based forecasting. That means forecasters combine the chance of flooding with the likely effects on people, property, transport and infrastructure. For emergency planners, that can mean quicker and more confident decisions about warnings, road closures and where to send crews.

One of the main tools behind this work is the Surface Water Flooding Hazard Impact Model, or SWFHIM, developed by the National Hazards Partnership for the Flood Forecasting Centre in 2020. At the moment it helps forecasters look from roughly six hours ahead out to three days ahead, which is a useful window when councils, fire services and local resilience teams are getting ready. SWFHIM works by bringing several layers of information together. It uses runoff forecasts from the Grid-to-Grid hydrological model developed by UKCEH, and those forecasts depend on rainfall outlooks from the Met Office. It adds detailed flood maps from the Environment Agency's National Flood Risk Assessment, then matches that with a one-kilometre impact library produced with the Health and Safety Executive, covering properties, people, transport and infrastructure. The result is shown on a dashboard for hydrometeorologists, who can see county-level likelihoods for minimal, minor, significant and severe impacts. On 1 August 2024, for example, the model highlighted significant impacts in Hampshire and minor impacts elsewhere in south-east England during heavy thunderstorms.

But the hardest part of surface water flooding is often the last-minute part. User research found that responders want much more local detail in the two to six hours before a rapid flood event. That is the period when equipment may need to be moved, communities warned and emergency actions started. A forecast that is broadly right a day before an event may still be too blunt when a storm cell is forming over one town and not the next. The current Rapid Flood Guidance service already uses short-term storm forecasting, or convective nowcasting, from the Met Office's Expert Weather Hub, with Flood Forecasting Centre hydrometeorologists adding professional judgement. What it does not yet fully include is real-time hazard and impact modelling of the kind SWFHIM might provide in future. That gap matters, because knowing that a storm exists is not quite the same as knowing what damage it is likely to cause on the ground.

That is why the Flood Forecasting Centre tested two nowcasting tools for the zero-to-six-hour window. One was a nowcast version of SWFHIM. The other was FOREWARNS, a surface water flood model from the University of Leeds. Both used Met Office ensemble rainfall nowcasts, which combine radar with weather-model output to estimate what rain is likely over the next few hours. The trial was designed to answer three practical questions. Do these tools work well enough for everyday operational forecasting? Do they improve on the current way of issuing Rapid Flood Guidance? And what do forecasters actually need from a future tool if it is going to help when time is short? Phase 1 ran in the 2024 Met Office summer testbed and involved more than 50 meteorologists, modellers and academics from the UK and overseas. Phase 2 in 2025 focused more tightly on the Flood Forecasting Centre and Met Office teams who would use the tools in real operations.

The clearest result was that both models helped with more objective decision-making during Rapid Flood Guidance production. According to the trial summary, forecasts could have been improved on nearly half of the occasions examined. The models also helped forecasters judge when impact thresholds had been crossed, and in some cases they improved both location accuracy and the timing of warnings. There were differences between the two tools, and that is useful rather than awkward. SWFHIM performed best when forecasters needed detail on likelihood, impact severity and the type of impacts expected. FOREWARNS was valued for being simpler and for giving useful context around unfolding events. **What this means:** the question is not only which model wins on paper. It is which tool helps real people make the best decision quickly, clearly and with the least avoidable uncertainty.

The next steps are about turning trial results into something dependable. Recommendations include improving both SWFHIM and FOREWARNS with science partners, using the Met Office's next generation of rainfall nowcast data when it becomes available from 2027, and giving priority to a stronger SWFHIM nowcast tool that can offer impact-based, probability-led output in the zero-to-six-hour window. The trial also backed the idea of running FOREWARNS as an operational prototype and building automated checks so forecasters can measure whether the models are actually improving warnings. The plan, according to the government article, is for these recommendations to feed into a new three-year project with tools brought into operations by 2029. That will not stop intense rain from falling. What it can do is buy time, and in flood response time is often the difference between disruption and disaster. For readers, that is the main point to hold on to: better forecasting is not a technical extra. It is part of how a public service learns to protect people in a warmer, wetter and more built-up country.