Reviewing a hydraulic modelling report line by line

Reviewing a hydraulic modelling report is a question of whether the model answers the question that was asked. Most reports that fail review do not fail on hydraulics. They fail because a 1D model has been used to predict depths across a floodplain, because the downstream boundary was placed too close to the site, or because nobody tested what happens when the culvert screen blocks.

Work through the report in the order the model was built, not in the order the chapters appear. The results section is the last thing to read.

Model choice and whether it suits the question

Start with the question. A planning application needs to know flood extent and level at the site, depth and velocity across the development area, the effect of the proposal on third parties, and the residual risk after mitigation. Different questions need different models.

The common error is a 1D model used to produce floodplain depths by projecting water levels horizontally. That works on a narrow valley floor and gives nonsense on a wide floodplain with flow paths, embankments and road corridors.

Ask two questions of the choice. Does the model represent the mechanism the site is actually exposed to? And where more than one mechanism applies, such as fluvial and surface water, has each been modelled or has one been assumed to dominate without a check?

Topography, schematisation and the model build

Check the source and vintage of the ground model. National terrain data is suitable for the floodplain at large and is rarely good enough at the site itself, where a ground survey of levels, defences, walls and thresholds is needed. A model that relies on terrain data alone for finished floor level comparisons is not fit for that purpose.

The items to verify:

1. Grid or mesh size against the features it has to represent. A 5 m grid cannot resolve a 2 m wide flow route between buildings.
2. Breaklines or elevation modifiers along walls, embankments, kerbs and road crowns, because these control where water goes.
3. Cross section spacing and location in the 1D domain, and whether sections extend far enough onto the floodplain.
4. Structure geometry from survey rather than from assumption, with soffit levels, spans and invert levels recorded.
5. Treatment of buildings, whether as blockages, as raised cells or through roughness, with the choice stated and justified.
6. Roughness values by land use, and whether they fall within the ranges used in published hydraulic references.
7. Channel bed levels, including any bathymetry where the water was too deep to survey on foot.

Where the site drains to an ordinary watercourse and works are proposed to it, consents sit alongside the modelling, as we set out in land drainage consent for works to an ordinary watercourse.

Boundary conditions and inflows

This is where most of the uncertainty lives and where reports are thinnest.

Inflows should state the method, the catchment descriptors used, the return periods modelled and the hydrograph shape. For a small catchment, check whether a revitalised rainfall runoff approach or a statistical approach was used and why, because the two can give materially different peak flows. For an urban catchment, check whether urbanisation has been accounted for in the descriptors.

Climate change allowances should name the allowance applied, the river basin district, the epoch and the vulnerability classification. The allowances are revised periodically, so check the date of the figures against the current Environment Agency guidance rather than accepting a value from an earlier report.

The downstream boundary deserves specific attention. A normal depth boundary with an assumed bed slope, placed a short distance below the site, will influence levels at the site. The report should show a sensitivity test with the boundary moved or the slope varied. If it does not, ask for one.

Where fluvial and tidal influences combine, check whether joint probability has been addressed rather than the two peaks simply being added or the worst single case taken.

Calibration, verification and sensitivity

Calibration against a recorded event is the strongest evidence a model can carry. Where gauged levels, recorded flood outlines or surveyed wrack marks exist, the report should compare modelled and observed, state the differences and explain them.

Many small catchments have no usable record. That is not a defect in itself, provided the report says so and compensates with sensitivity testing. What is a defect is silence on the point.

Sensitivity tests to look for:

• Channel and floodplain roughness varied upwards and downwards
• Inflow varied by a stated percentage
• Blockage of culverts, screens and bridge openings, which the Environment Agency expects to be tested where a structure could obstruct
• Downstream boundary condition
• Climate change allowance at the higher central and upper end figures where relevant to the vulnerability class
• Defence breach and overtopping where the site sits behind a defence

A report that presents a single design run with no sensitivity work is telling you that nobody tested the assumptions.

Results presentation and mapping

What a reviewer needs to see, and frequently does not:

Levels at named nodes, tabulated, for each return period and each scenario, so the numbers can be traced. Depth and velocity mapping, and a hazard classification where people will be present during a flood. A baseline and a proposed scenario for the same events, mapped at the same scale and the same colour bands, so the comparison is visible rather than asserted.

Model health indicators matter too. Mass balance error should be small, commonly within one per cent, and reported. Instabilities, oscillations and dry or wet cell warnings should be acknowledged rather than smoothed out of the output.

Where floodplain storage is lost, the compensation should be demonstrated level for level and volume for volume, with a table rather than a statement. And where finished floor levels are set with freeboard, the freeboard should be stated and justified against the modelled level including the climate change allowance.

The results then feed the Flood Risk Assessment and the Drainage Strategy, and a reviewer should check that the numbers quoted in those documents match the model output rather than an earlier version of it. That mismatch is more common than it ought to be.

Assumptions recorded against those omitted

Ask for the assumptions register. A good report has one, listing each assumption, its basis, its direction of conservatism and what would change if it were wrong.

The omissions that matter most, from experience:

• Surface water or groundwater mechanisms set aside without a check. Groundwater in particular is often dismissed, and evidencing it properly is its own exercise, covered in evidencing groundwater flooding in a flood risk assessment
• Future development in the catchment, where a local plan allocation upstream changes the runoff
• Maintenance assumptions, such as a clear channel in perpetuity
• Operation of pumps, sluices or flap valves assumed to work during the event
• The lifetime of the development, which for residential is normally taken as 100 years and sets the climate change epoch

A model is a set of assumptions with arithmetic attached. The reviewer's job is to find the assumptions that the result depends on.

Questions to put to the modeller

Six questions that usually produce the information a review needs:

1. What question was the model built to answer, and what is it not suitable for?
2. Where did the topography come from, at what date, and what was surveyed on the ground?
3. How was the downstream boundary set, and what does a sensitivity test on it show?
4. Which blockage scenarios were run, and at what percentage?
5. What is the mass balance error, and were there instabilities?
6. Which three assumptions would change the conclusion if they were wrong?

Put them in writing, and give the modeller a deadline that allows a real answer rather than a defensive one. Most modellers will give a straight answer to question six, and it tells you where to look.

Councils expect flood modelling to be carried out by suitably qualified and experienced practitioners, and professional review is strongly advised before a report is relied on in a decision.

Enter your postcode on PlanScope to see which flood and drainage documents a site of this kind is likely to need.

This is general guidance rather than legal advice. Check the position with your local planning authority or a qualified professional before relying on it. For specialist technical reports, qualified professional input may be required. PlanScope provides AI-assisted draft reports for planning applications. Reports should be reviewed before submission. We do not guarantee planning approval.