Heritage building facade change detection for conservation compliance
Very-high-resolution optical satellites can detect facade alterations on listed buildings, from replaced sash windows to rear extensions, without a site visit. The method supports enforcement, insurance revaluation and heritage audit at portfolio scale.
Sensors
- Maxar WorldView-3: 31 cm panchromatic, 1.24 m multispectral (8 bands including coastal blue and SWIR). Single-pass revisit to any point within roughly 1 day at off-nadir. The panchromatic channel resolves individual window apertures on most residential facades.
- Airbus Pléiades Neo: 30 cm panchromatic, 1.2 m multispectral (6 bands). Stereo and tri-stereo tasking in a single pass enables 3-D surface models useful for detecting roofline changes and dormer additions. Revisit 1 to 2 days per location.
- Planet SkySat: 50 cm panchromatic, 1 m multispectral. Slightly coarser than WorldView-3 or Pléiades Neo, but the constellation of 21 satellites allows more flexible tasking windows, useful for monitoring multiple sites simultaneously. Resolves gross facade changes; marginal for window-level detail.
- Airbus Pléiades (original): 50 cm panchromatic. Deep archive from 2012 onward. Valuable for establishing the baseline state of a listed building before a dispute or insurance event, where no recent tasking exists.
What a 30-centimetre pixel actually sees on a building facade
At 30 to 31 cm ground sample distance, a single pixel on WorldView-3 or Pléiades Neo is smaller than a standard house brick. A typical sash window opening on a Georgian terrace spans roughly 60 to 90 cm in width, meaning it occupies two to three pixels across. That is enough to detect the presence or absence of a window, a change in glazing pattern, or the appearance of a new opening, but not enough to read the glazing bar count or confirm the exact material used. The distinction matters for setting client expectations: satellite imagery at this resolution can flag an anomaly and trigger an inspection; it cannot replace the inspector.
Roof modifications are more reliably detected than facade alterations at street level, because nadir or near-nadir satellite imagery looks predominantly downward. A new flat-roof extension, a dormer addition, or a change from slate to concrete tile produces a clear spectral and geometric signature in the panchromatic channel. Facade changes on the street-facing elevation are visible only when the satellite is tasked at a suitable off-nadir angle, typically 20 to 30 degrees, which reduces effective ground resolution by a corresponding factor. Operators should specify oblique collection geometry when the enforcement interest is the front elevation specifically.
The differencing workflow: from archive to alert
Change detection at facade scale rests on image differencing between a baseline acquisition and a monitoring acquisition over the same target. Both images must be orthorectified to a common coordinate frame; residual misregistration above roughly half a pixel will generate false positives at every edge. For buildings with complex rooflines, a digital surface model derived from stereo Pléiades Neo is used to improve orthorectification accuracy and to separate genuine structural change from parallax artefacts introduced by slightly different collection angles.
Once co-registered, per-pixel difference images are thresholded using a combination of spectral magnitude and spatial coherence filters. Isolated changed pixels are usually noise; changed regions that correspond to planar surfaces of building-scale extent are candidate alterations. The output is a polygon layer flagging changed zones, attributed with the date pair, the magnitude of change, and the affected building element (roof plane, wall section, or ground-floor addition). A human reviewer confirms or dismisses each flag before it enters an enforcement workflow. Automation reduces the review burden; it does not eliminate the reviewer.
Cloud, shadow and the limits of optical methods
The United Kingdom, where the density of listed buildings is among the highest in the world, is also among the cloudiest environments for optical satellite tasking. Cloud cover renders an acquisition unusable at the facade scale; unlike medium-resolution land-cover work, there is no cloud-gap-filling technique that recovers sub-metre detail. In practice, a monitoring programme over a portfolio of listed buildings in England or Scotland should budget for a cloud-clear acquisition rate of roughly 40 to 60 per cent of tasking attempts in winter months, rising to 60 to 80 per cent in summer. Programmes that require quarterly compliance checks should task monthly to ensure at least one usable acquisition per quarter.
Shadow is a secondary problem. A listed building in a dense terrace may have its rear elevation in shadow for much of the day at high latitudes in winter, making change detection on that elevation unreliable regardless of cloud. Scheduling collections in the two to three hours around solar noon, when shadow lengths are shortest, is standard practice. Even then, north-facing rear elevations in narrow plots may be partially obscured. These are honest constraints, not edge cases.
Portfolio-scale enforcement: where the method earns its cost
A single listed building inspection by a conservation officer costs time and travel. A local authority with several hundred listed buildings in its area cannot physically inspect every property on a meaningful cycle. Satellite-based monitoring changes the economics by allowing a desktop screening pass across an entire portfolio, concentrating physical visits on the subset of buildings that show a detected change.
The workflow is well suited to the period immediately following a change of ownership, when the risk of unpermitted alteration is statistically elevated. It is also used by heritage asset insurers to detect material changes that affect reinstatement value, particularly after a building has been subdivided into flats or converted to commercial use. For these applications, the archive depth of commercial sensors matters: WorldView-3 archive coverage of UK cities extends back to 2014, and Pléiades archive to 2012, providing a documented baseline against which any subsequent change can be measured. Satellize can structure a monitoring programme around existing local authority listed-building registers, cross-referencing detected changes against planning application records to identify alterations that lack consent.
The Satellize analytics team has applied similar change-detection pipelines in agricultural contexts, including the Kingdom of Tonga crop-estimation programme, and the same core differencing and polygon-attribution methods transfer directly to the built environment.
Minimum detectable change and what falls below the threshold
The practical minimum detectable alteration at 30 cm resolution is approximately 1 to 2 square metres of changed surface area, assuming good co-registration and adequate contrast between the new and original material. A single replaced window in a brick facade, where the new frame is a similar colour to the original, may fall below detection threshold if the spectral difference is small. A uPVC replacement in a stone-faced Georgian building will usually be detectable. A like-for-like timber replacement may not be.
Subsurface changes, internal alterations, and changes to features smaller than roughly 1 metre in any dimension are outside the method's reach entirely. The technique is a screening tool for gross external alterations. It is not a substitute for a condition survey, a measured building record, or a photogrammetric survey of fine architectural detail. Programmes that require that level of resolution should combine satellite screening with drone or terrestrial photogrammetry for confirmed change sites.
Typical figures
| Best available panchromatic resolution | 30 cm (Pléiades Neo) / 31 cm (WorldView-3) |
| Multispectral resolution | 1.2 m (Pléiades Neo) / 1.24 m (WorldView-3) |
| Tasking revisit (single site) | 1 to 2 days for Pléiades Neo and WorldView-3; same-day possible with SkySat |
| Minimum detectable altered area | Approximately 1 to 2 m² under good contrast and co-registration conditions |
| Spectral bands used | Panchromatic (primary); red, green, blue, NIR (change characterisation); SWIR (WorldView-3 material discrimination) |
| Archive depth | WorldView-3 from 2014; Pléiades from 2012; SkySat from 2016 |
| Cloud-clear acquisition rate (UK) | Approximately 40 to 80 per cent of attempts depending on season |
| Off-nadir capability | Up to 45° (reduces effective GSD proportionally; 30° recommended for facade work) |
| Typical delivery format | Orthorectified GeoTIFF; change polygon GeoPackage or Shapefile; PDF enforcement summary |
| Stereo surface model accuracy | Pléiades Neo stereo: vertical accuracy approximately 0.5 m CE90 in open terrain; degraded in dense urban canyons |
Analytics Satellize can run
| Facade change flag layer | Orthorectified image differencing with spatial coherence filtering; per-pixel magnitude thresholding | Polygon GIS layer attributed with change date pair, affected building element and magnitude score |
| Roofline modification detection | Digital surface model differencing from stereo Pléiades Neo acquisitions; height-change thresholding | Annotated raster and polygon report identifying new roof structures, dormer additions or removed chimneys |
| Listed building monitoring report | Automated change screening cross-referenced against local authority listed-building register; human-reviewed flags | Quarterly PDF enforcement summary per local authority area, with per-building change status and image evidence |
| Insurance revaluation trigger alert | Change detection against insured-property baseline image; area and element classification | Alert feed (email or API) flagging properties with detected alterations above a configurable area threshold |
| Baseline archive record | Best-available historical image selection and orthorectification from commercial archives (2012 to present) | Georeferenced baseline image set per property, stored as audit evidence with acquisition metadata |
| Material change characterisation | WorldView-3 SWIR band analysis for broad material class discrimination (e.g. original stone vs. modern render or metal roofing) | Spectral classification overlay on changed zones; narrative assessment of likely material substitution |
Who does the work
We can get this done for you. Satellize runs its own analyst desk and a strong science team. You do not buy a data feed and work out what it means; our people source the imagery, run the analysis described on this page, and hand you the answer with its confidence limits stated. Discuss this requirement.