Flood models are invaluable for strategic planning across large catchments, but they cannot account for every local factor that determines whether an individual building or critical infrastructure asset will flood. Combining predictive modelling with real-time monitoring, AI-assisted analytics and resilient communications can potentially provide a more effective, lower-risk, lower-carbon approach to flood resilience.
Flood modelling – A powerful planning tool
Flood modelling has transformed the way engineers, planners and emergency responders understand flood risk. Modern hydraulic models combine rainfall forecasts, river flow data, terrain mapping and hydraulic calculations to predict how water is likely to move across entire catchments. They support investment decisions, emergency planning and climate adaptation. Their greatest strength is providing a strategic picture of risk across wide geographical areas. However, they are not intended to predict with certainty whether one particular building, bridge or railway cutting will flood because local conditions change continuously and many of these locations can lie outside the scope of the model.
Why local conditions matter
Flooding can be influenced by antecedent rainfall, soil saturation, crop type, groundwater levels, overloaded sewers, blocked drains, blinded trash screens and partially or completely obstructed culverts. Even small changes in ground level, landscaping or new developments can alter the path of floodwater too. These dynamic factors are difficult to represent within regional models, meaning that predictions become less certain when applied to individual properties or assets.
Critical infrastructure
Railways, highways, and distributed utility assets illustrate these limitations well. A railway may lie outside a mapped floodplain yet still flood because bridges, culverts or siphons have finite hydraulic capacity or become blocked by vegetation and debris. Floodwater may overtop embankments, scour bridge foundations, destabilise ballast, carry debris onto the line and damage rolling stock travelling through deep standing water. Such events create operational disruption and potentially risk to life.
DIVE RIGHT IN
Sign up to our newsletter
From maintenance to intelligence
Traditional inspections and preventative maintenance are expensive and generate significant carbon emissions through repeated site visits. They also provide only a snapshot of an asset's condition, including flood prevention assets such as SuDS detention ponds and leaky dams. Sensors, cameras and AI now enable condition-based maintenance by continuously monitoring rivers, culverts, bridges, trash screens, drains, SuDS, pumping stations and groundwater. AI-assisted image analysis can identify debris accumulation and abnormal water levels, allowing engineers to prioritise the locations that genuinely require urgent intervention. This reduces unnecessary inspections while improving resilience.
A smarter future
Live river level measurements can also help to calibrate rainfall-runoff models during a storm, creating a dynamic representation of catchment behaviour rather than relying solely on theoretical assumptions. Level changes can also be used to trigger cameras to provide verification of levels and any debris deposited during and after the storm.
Forecast accuracy also improves as real-time observations are incorporated. However, real-time data also presents another opportunity: the creation of smart flood assets such as smart SuDS, wetlands, water gardens, and detention ponds, which are designed to control flows in real time to reduce flooding.
For example, our research working on an FCRIP program has shown that legacy SuDS/detention pond features may be designed to store a deluge from a 1:100 return storm, but at other times just fill to as little as 10% due to a fixed orifice outlet. By making this type of asset smart, closing off the outlet with a flow regulator in real-time based upon downstream levels, better use of the storage can be made, increasing the asset value and utilisation in the battle against flooding.
For critical infrastructure, resilient communications are also equally as important. Licensed radio telemetry, cellular services and low-Earth orbit satellite links provide multiple communication paths so alarms continue to reach operators even if one network fails during severe weather.
Conclusion
Flood modelling remains an indispensable engineering tool, but it is only one component of effective flood resilience. The greatest benefits will come from integrating predictive models with real-time monitoring, AI-assisted analytics, resilient communications and condition-based maintenance. This balanced approach reduces costs and carbon emissions while improving public safety, protecting critical infrastructure and enabling engineers to move from predicting floods towards preventing many of their consequences.