Proving no increase in flood risk to third parties
The burden sits with the applicant. Comparable baseline and proposed runs, difference plots at named receptors and proper sensitivity testing are what satisfy the test.
Proving no increase in flood risk to third parties is the applicant's job, and the standard is a demonstration rather than an assertion. You model the baseline, model the developed case, compare them at every location where somebody else's land or property could be affected, and present the difference in a form an objector can follow. National flood risk policy puts the burden on the applicant, and the Environment Agency and the lead local flood authority both read this section before anything else.
The test is not that the site is safe. It is that the development does not push water onto someone else, in any event up to and including the design case with climate change.
The policy requirement and who has to satisfy it
National policy requires development to be safe for its lifetime without increasing flood risk elsewhere. That last phrase is the one that generates the work, and it applies whether or not the site itself is at risk.
Three consultees have an interest and they look at different things. The Environment Agency deals with main river and tidal flood risk, and will object where a scheme reduces floodplain storage or alters conveyance. The lead local flood authority deals with surface water and ordinary watercourses, and looks at discharge rates, exceedance routes and drainage. The sewerage undertaker looks at discharge to its network.
Where the site is in Flood Zone 2 or 3, or is a hectare or more, the Flood Risk Assessment has to address this directly. On smaller sites outside the flood zones the requirement still applies through drainage policy, and a surface water assessment carries it.
The person who has to satisfy the test is the applicant, using their own modelling. An objection that the scheme will flood the field next door does not have to be disproved by the council.
Baseline and proposed model runs compared
The comparison needs to be like for like, and most disputes come from the two runs not being comparable.
Build the baseline from existing ground levels, existing buildings and existing drainage, using survey data with a stated datum and date. Then build the proposed case with the same model, the same boundary conditions, the same events and the same roughness assumptions, changing only what the development changes.
Four things to hold constant and state explicitly: the hydrological inputs, the downstream boundary condition, the model extent, and the topographic base outside the site. A proposed run that quietly uses a finer grid, a different downstream boundary or updated rainfall data is not a comparison, and a reviewer will find it.
Report the events the policy requires: the design annual probability for the source in question, the same event with the climate change allowance for the development lifetime, and a more extreme sensitivity case. On a fluvial site that usually means the 1 per cent event and the 1 per cent plus allowance. On a tidal site the 0.5 per cent event.
Difference plots and the thresholds used
The output that persuades is a difference plot: proposed water level or depth minus baseline, mapped across the model extent, with a colour scale that distinguishes increases from decreases.
Two conventions to fix before producing them. The threshold below which a difference is treated as model noise, which for a well-built two-dimensional model is commonly in the range of a few millimetres, and which has to be stated and justified rather than chosen to make the map look clean. And the receptors at which any increase is reported numerically, which should be named properties and named parcels rather than a general area.
Then report three things for each affected receptor: the change in peak level, the change in depth, and the change in the extent of flooding. A scheme can hold levels constant and still extend the flooded area sideways, and an objector standing in the extended part will notice.
Where an increase is predicted, say the number. A difference plot showing a 40mm rise across a neighbouring paddock with the figure stated is a far better position than one where the colour scale conveniently starts at 50mm.
Sensitivity to model assumptions
Sensitivity testing is what separates a defensible model from a persuasive-looking one, and it is the part most often omitted.
Run and report at least four tests:
- Roughness, varying channel and floodplain values across a credible range.
- The downstream boundary condition, since a modelled water level or a normal depth assumption can dominate results in the lower part of the reach.
- Blockage, particularly at culverts, trash screens and bridges, because a partial blockage scenario often produces the worst third-party effect.
- Climate change, using the allowance range for the region rather than the central estimate alone.
Where the site is influenced by groundwater or by a culverted watercourse, the assumptions need separate treatment, and the evidence expected for groundwater is covered in groundwater flooding evidence and how to assess it. Where a culvert is involved, policy starts from a presumption against burying watercourses, as set out in culverts and the presumption against burying a stream.
State which sensitivity run produces the worst third-party effect and design to that rather than to the central case. A design that works only on the central estimate will be objected to on exactly this point.
Where a small increase is argued to be acceptable
The policy language is about not increasing flood risk elsewhere, and in strict terms any increase is a problem. In practice three arguments are made, and only two of them work.
The increase is within model uncertainty. This works where the threshold is properly justified and applied consistently, and where the increase really is at that scale. It fails where the threshold has been set to accommodate the result.
The increase affects land where it causes no harm. A few millimetres of additional depth in an already deeply flooded agricultural field, with no property, no access route and no habitat interest affected, is a different matter from the same increase against a residential threshold. This argument requires you to identify what is actually there, and it requires the landowner's position to be considered rather than assumed.
The increase is offset by a betterment elsewhere. This works only where the betterment is secured, delivered before or at the same time as the development, and located where it benefits the same receptors. Offsetting an increase at property A with an improvement at property B is not an answer to the owner of property A.
What does not work is arguing that the increase is small in percentage terms, or that the receiving land already floods so it does not matter.
Neighbouring land and the objection it attracts
Third-party objections on flood risk are among the most effective objections in the system, because they engage a statutory consultee and a policy test rather than a matter of judgement.
Two situations recur. The downstream neighbour who has flooded before and produces photographs, which is powerful evidence of what the baseline actually is and sometimes more reliable than the model. And the upstream or adjacent landowner whose land is proposed to hold compensatory storage or a betterment, without their agreement.
That second situation defeats schemes regularly. Compensatory floodplain storage on land you do not control is not deliverable, and a model that relies on it is a model of a scheme that cannot be built. Secure the land or the easement before the application goes in, or design the mitigation within the site.
Where a neighbour has flood history, engage with it directly in the assessment. Include the reported events, compare them with the model, and explain any discrepancy. A model that predicts no flooding at a property that flooded in 2020 is a model with a problem, and pretending otherwise damages everything else in the report.
Presenting the results so an objector can follow them
Write this section for a resident with a copy of the report and a grievance. The Environment Agency is the easier audience.
Lead with a plain statement of the conclusion and the evidence for it. Then the difference plots at a legible scale, with north, a scale bar, property names or numbers and the colour scale explained in words. Then a short table for each named receptor giving baseline level, proposed level and the change, for each event modelled.
Avoid two things. Presenting results only as extent maps, which hide level changes. And burying the numbers in an appendix that the summary does not reference.
Then set out the mitigation in the Drainage Strategy or the flood risk assessment: compensatory storage with its volume and location on a level-for-level basis where required, discharge rates with their derivation, exceedance routes shown on a drawing, and the maintenance arrangements that keep all of it working.
Finally, state the residual risk plainly, including what happens in an event beyond the design case. A report that names the route water takes when the system is overwhelmed is more credible than one that stops at the design event, and it is what the Exception Test needs where that applies, as covered in exception test evidence.
Enter your postcode on PlanScope to see which flood and drainage reports your site is likely to need.
This is general guidance rather than legal advice. Check the position with your local planning authority, the lead local flood authority, the Environment Agency or a qualified flood risk consultant before relying on it. PlanScope provides AI-assisted draft reports for planning applications. Reports should be reviewed before submission. We do not guarantee planning approval.
Frequently asked questions
Who has to prove that flood risk elsewhere does not increase?
The applicant, through their own modelling. National flood risk policy places the burden on the applicant to demonstrate the development is safe for its lifetime without increasing flood risk elsewhere. A third-party objection that the scheme will flood neighbouring land does not have to be disproved by the council.
What makes baseline and proposed model runs comparable?
Holding everything constant except what the development changes: the hydrological inputs, the downstream boundary condition, the model extent, the grid, the roughness assumptions and the topographic base outside the site. A proposed run using updated rainfall data or a finer grid is not a comparison and a reviewer will identify it.
What difference threshold counts as model noise?
For a well-built two-dimensional model it is commonly a few millimetres, but the figure has to be stated and justified rather than chosen so the difference plot looks clean. Apply it consistently across both increases and decreases, and report numerical changes at named receptors rather than only as mapped colour bands.
Can a small increase in flood risk ever be acceptable?
Only on limited grounds. Where the increase falls within a properly justified model uncertainty threshold, or where it affects land with no property, access or habitat interest affected, the argument can work. Offsetting an increase at one property with a betterment at another is not an answer to the first owner.
What if a neighbour has flooded and the model says they should not have?
Engage with it directly. Include the reported events, compare them with the model output and explain the discrepancy. A model predicting no flooding at a property that demonstrably flooded has a problem, and ignoring the evidence undermines the credibility of the rest of the assessment with both the consultee and the committee.
Sources and further reading
This article is general guidance about the planning system in England and, where stated, the rest of the UK. It is not legal or professional advice. Check your local planning authority’s validation requirements and take professional advice for your specific site.