What a linked 1D-2D flood model shows, and what it does not

A 1D 2D flood model in planning is a hydraulic model that treats the watercourse as a one-dimensional network of cross sections and the floodplain as a two-dimensional grid, with the two linked so that water spills out of the channel and spreads across the ground in a way that reflects real topography. You need one when the published Flood Map for Planning is too coarse to answer the question, when the site sits in a mapped zone but the mapping is clearly out of date, or when you are proposing development that will itself change flood behaviour. You do not need one for most sites, and commissioning a model that was never necessary is one of the more expensive mistakes in flood risk work.

Agree the scope with the Environment Agency before anybody builds anything. Everything below assumes that step has happened.

When a site specific model is justified

Four situations make a model worth the money.

The published mapping shows the site partly in Flood Zone 3 and partly in Zone 2 with an arbitrary-looking boundary, usually because the national model used coarse topography and did not represent a local structure.

There is a detailed model held by the Environment Agency that predates significant change on the ground, such as a new culvert, a defence or a highway embankment.

The proposal itself alters conveyance or storage. Raising ground, building in the floodplain, replacing a culvert or changing a bridge all require you to show the effect, and a before-and-after comparison can only come from a model.

The site requires an exception test and you need to demonstrate that the development will be safe for its lifetime, including the effect of climate change allowances, with finished floor levels and access routes justified rather than asserted.

Outside those, a flood risk assessment using published mapping, the Environment Agency's product 4 data and sensible freeboard will usually do the job.

What the linked representation actually does

The one-dimensional part solves flow along the channel using surveyed cross sections, with structures given their own units: bridges by afflux equations, culverts by inlet and barrel relationships, weirs by discharge coefficients.

The two-dimensional part solves flow across a grid derived from a terrain model, usually built from airborne survey data, with each cell assigned a roughness value.

The link between them is a set of lateral connections, typically along the channel bank line, which spill water out when the channel level exceeds the bank and draw it back when levels fall. That linkage is the part most often got wrong. If the bank line is in the wrong place or the spill coefficients are left at defaults, the model will put water in the wrong field.

Boundary conditions complete it. Inflows come from a hydrological estimate at the upstream limit, and the downstream boundary is usually a normal depth or a rating relationship. Both are assumptions, and both are places a reviewer will look.

Topography and roughness

Terrain is the single biggest control on 2D results. National airborne survey data at one metre resolution is the usual starting point, and it is good, but it was not flown for your site.

Three corrections matter. Buildings need to be built in, either as raised cells or as removed from the flow area, and the choice changes depths on adjacent land. Channel bathymetry is not captured by airborne survey through water, so surveyed cross sections have to be burned in. And small features that control flow, such as a kerb line, a wall or a farm track embankment, are often below the resolution of the data and have to be added from survey.

Roughness values are assigned by land cover: channel bed, grassland, woodland, urban surfaces. The values come from published tables and are a matter of judgement within a range. That is why sensitivity testing matters more than the headline choice.

Calibration, verification and their limits

Calibration adjusts model parameters so that simulated results match observed data. Verification then tests the calibrated model against a different event.

The honest position is that most small watercourses in England have neither a gauge nor a well-recorded historic flood outline, so there is nothing to calibrate against. A model of an ungauged ordinary watercourse is a structured set of assumptions, not a validated prediction.

When that is the case, say so. State what data was available, what was used, and what sensitivity tests were run instead. A model presented as calibrated when it is not is the fastest route to losing credibility with a reviewer.

Where data does exist, usual sources are gauged flow records, recorded flood levels and wrack marks, historic flood outlines, local authority flood investigation reports and resident accounts. The last of those is anecdotal and should be labelled as such, but a photograph with a date on it is evidence.

Reading the outputs

A model produces three primary outputs and one derived one.

Hazard is the one that decides whether a layout works. The published hazard formula combines depth with velocity plus a constant, multiplied by a debris factor, and the result falls into bands from low hazard through to danger for all. A route out of a site that passes through a danger for most cell is not a safe access route, whatever its depth alone suggests.

Compare events, not just the design event. The difference between the 1 in 100 annual probability event, the same event with the relevant climate change allowance, and the 1 in 1000 event tells you how sensitive the site is. A site where all three give similar depths is behaving predictably. A site where a small increase in flow triples the depth is on a threshold, and that needs saying in the report.

Agreeing the scope first

Scope agreement is not a courtesy. It determines whether the model will be accepted.

The items to settle in writing are the model extent, the structures to be included and how, the source of the inflow hydrographs and the hydrological method, the climate change allowances and epoch, the events to be run, how defences and their failure or overtopping are handled, the downstream boundary, and the format of the deliverables. Add the sensitivity tests expected.

Climate change allowances are updated, and the applicable peak river flow allowance depends on the river basin district, the allowance category and the development's lifetime. Check the current published allowances rather than copying a figure from a previous report, and state the allowance and date used.

Where a blockage scenario is required, agree the blockage assumption too, as our note on modelling a partially blocked flow control discusses.

Review and the usual criticisms

Five criticisms account for most review comments.

The model extent is too short, so the upstream or downstream boundary sits close enough to the site to influence results.

Structures are modelled optimistically. Culvert entry losses, screen effects and bridge afflux all reduce conveyance in reality, and defaults tend to flatter the design.

Climate change allowances are wrong, out of date or applied to the wrong epoch.

No sensitivity testing on roughness, inflow or the downstream boundary, which leaves the reviewer unable to judge how much the result depends on a judgement call.

The site's own development is modelled with compensatory storage that is not level for level or does not drain, which is a drainage design failure dressed up as a modelling result.

A model is not a substitute for a clear drainage strategy and a sequential approach to layout. It tells you where the water goes. Deciding what to build and where remains a planning judgement, and the sequential test comes before any of this.

This is general guidance, not legal advice. Check the position with your local planning authority or a qualified professional 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.