Daylight model accuracy: survey, massing and reveals

A daylight result is only as good as the geometry behind it, and most of the geometry in a typical assessment was never measured. Neighbouring building heights come from mapping data, storey counts come from photographs, internal room layouts are assumed, and the window positions on the proposed building come from a drawing that has been revised twice since. When a vertical sky component result sits at 26.4 per cent against a 27 per cent target, the question a reviewer should ask is not whether the scheme complies but whether the model is accurate to better than half a percentage point. It usually is not.

That does not make the assessment worthless. It makes disclosure of the geometry sources part of the assessment rather than an appendix nobody reads.

Sources of geometry and their error

Five sources appear in practice, and they carry different kinds of error.

Ordnance Survey mapping gives building footprints with good relative accuracy and known positional tolerance, but the polygon may follow the roof outline rather than the wall line, and recent extensions are often missing.

Airborne lidar and derived surface models give heights quickly across a wide area. The height returned is the surface, so a pitched roof, a parapet or a plant enclosure all read as building height unless the model separates them.

Photographs and storey counting are how most neighbouring buildings get their heights. Multiplying storeys by an assumed floor-to-floor height is defensible for a consistent terrace and unreliable for a Victorian villa next to a 1960s block.

Planning portal drawings for a consented neighbouring scheme give exact intended geometry, subject to the scheme having been built as approved.

A measured topographic survey, with levels to Ordnance Datum and building corners picked up on the ground, is the only source with a stated accuracy. On a constrained site where results will be close to the thresholds, it is the one that matters.

Ground level error is the sleeper. A daylight model built on a flat plane at an assumed level, on a site that falls a metre and a half across its width, will misplace every obstruction angle.

Measured survey against mapping-derived massing

The honest position is that mapping-derived massing is adequate for early testing and inadequate for a submitted assessment where results are marginal.

Early massing work exists to tell the design team where the problem is, and our note on obstruction angles and the early massing test sets out what that stage can and cannot answer.

For a submitted assessment, specify the survey to cover:

• The site boundary, levels and existing structures.
• The elevations of every neighbouring property whose windows are being assessed, with window positions, cill and head heights, and reveal depths.
• Ground levels at the base of each assessed facade, since the obstruction angle is measured from the window, not from a nominal datum.
• Roof forms of obstructing buildings, including parapets, dormers and plant.

Where access is refused, say so and state what was assumed instead. An assessment that silently substitutes a photograph for a survey of a neighbour's rear elevation is the one that falls apart under cross-examination.

Window positions, reveals and frame allowances

Reveal depth is where reviewers most often find inconsistency, partly because it affects different metrics differently.

Vertical sky component is calculated at the centre of the window on the outside plane of the window wall. A deeper reveal therefore does not change the vertical sky component at all. It does change internal illuminance and the no-sky line, because the reveal shades the room. A report that claims a reveal change improved the vertical sky component has an error in it somewhere.

The frame allowance matters for any illuminance-based metric. Whether the calculation uses the structural opening, the clear glazed area or a stated frame factor changes the result, and the assumption has to be declared. So does the glazing transmittance: a specification change from clear double glazing to a solar control unit with a lower light transmittance will move internal daylight results without anything moving in the geometry.

Three checks on any model:

1. Does the window schedule in the model match the current drawing revision, by window reference.
2. Are cill heights measured from finished floor level or from structural slab, and is that consistent across the model.
3. Are balconies above windows modelled as geometry, including the soffit and the balustrade, or omitted.

Balconies are the single largest source of understated obstruction in proposed-scheme internal daylight results.

Reflectance and maintenance assumptions

Internal reflectances follow the defaults in the BRE guidance unless the finishes are known, which for a neighbouring property they never are. The conventional set is a light ceiling, mid walls and a dark floor, and stating the values used is part of a reviewable report.

External surfaces matter too. Obstruction reflectance affects reflected light components in illuminance calculations, and using a bright value for a dark brick flank wall flatters the result.

Maintenance factors, glazing dirt allowances and the treatment of net curtains are the assumptions most often left out entirely. None of them changes an answer dramatically. Together they can move a borderline room across a threshold, which is why they belong in the assumptions register rather than in the modeller's head.

Where a neighbouring room layout is unknown, the convention is to assume a plausible layout and to test the sensitivity, and the mirrored-plan approach is discussed in our note on mirror image massing tests.

Sensitivity of results to small changes

The BRE targets are numbers with decimal places, and treating them as bright lines invites exactly the wrong argument.

Run a sensitivity test wherever a result sits within a few per cent of a threshold. Vary the obstruction height by the stated survey tolerance, vary the assumed room depth, and report the range rather than a single figure. A reviewer given a range trusts the report. A reviewer given 26.9 per cent with no tolerance assumes the modeller chose inputs until it passed.

Verification before results go out

A short verification routine catches most errors and takes an hour.

• Reproduce the existing condition and check that a known window returns a plausible baseline value against a manual calculation.
• Check the north point against the survey and against the site plan, not against the architectural drawing's convention.
• Check the model's ground plane against surveyed levels at three separate points.
• Confirm the proposed geometry matches the current drawing revision by comparing three dimensions: overall height, distance to the nearest boundary, and the position of the closest facade.
• Confirm window references in the results table match those on the survey drawing.
• Have a second modeller open the file and check the units, the location and the time zone or orientation settings.

Version control belongs in the same routine. Every issued results table should name the model file, its date, and the drawing revisions it was built from. Where the design changes afterwards, that record is what tells the team whether the results still stand.

Disclosing what the model cannot know

The limitations section is the part of the report that a specialist reviewer reads first, and writing it properly protects everyone.

State the source of each element of geometry and its accuracy. Name the properties surveyed internally, those surveyed externally only, and those modelled from mapping alone. Say which room uses were assumed and on what basis. List the windows whose reveal depth was not measured. Identify the results that sit within the sensitivity range of the inputs and say so in the same table as the results themselves, not in an appendix.

Committees and inspectors respond better to an assessment that admits a tolerance than to one that presents false precision, and the reasoning is easier to summarise, as our note on presenting daylight results a committee can read sets out. The scope of the exercise as a whole belongs in the Daylight and Sunlight Assessment rather than in a covering email.

This is general guidance, not legal advice. Assessment methodology should follow the current BRE guidance and the relevant British Standard, so check both editions and take advice from a qualified practitioner.

Draft the daylight and amenity documentation in minutes, then send it for professional review through the professional directory. PlanScope provides AI-assisted draft reports for planning applications. Reports should be reviewed before submission. We do not guarantee planning approval.