Monument height change detection using shadow length photogrammetry
Shadow length photogrammetry extracts height change in standing monuments and earthworks from very-high-resolution optical imagery at known solar angles, offering non-contact structural monitoring where ground access is impossible or politically fraught.
Sensors
- WorldView-2 / WorldView-3: Panchromatic ground sampling distance of 0.31 m (WV-3) and 0.46 m (WV-2) enables sub-pixel shadow boundary extraction; off-nadir tasking to angles below 30° preserves geometric fidelity needed for shadow-length measurement. Revisit roughly 1–4.5 days depending on latitude and tasking priority.
- Pleiades Neo: 30 cm native panchromatic resolution with a daily revisit capability per satellite (four-satellite constellation). Precise attitude knowledge and rigorous sensor model support the sub-pixel orthorectification accuracy that shadow photogrammetry requires.
- SPOT-6 / SPOT-7: 1.5 m panchromatic resolution. Adequate for large earthworks and mounds where height changes exceed roughly 0.5 m; less suitable for fine masonry monuments where centimetre-level change is the target. Useful for archive baseline acquisition back to 2012.
- Pléiades 1A / 1B (archive): 0.5 m panchromatic resolution, archive from 2011. Multi-date archive pairs at matched or precisely known solar angles allow retrospective change detection over more than a decade, which is important for slow erosion processes.
What a shadow length actually encodes
The geometry is straightforward trigonometry. If the solar elevation angle at the moment of image acquisition is known, and the shadow cast by a monument can be measured in the image plane with sufficient precision, the object height follows directly: H = L × tan(α), where L is shadow length and α is solar elevation. The technique requires no ground control on the monument itself, which is its principal advantage in restricted-access contexts.
The catch is that every term in that equation carries an error budget. Solar elevation at the exact acquisition time and location must be computed to better than 0.1° to keep height error below 10 cm at typical monument scales. Image orthorectification must be accurate to a fraction of a pixel, because a one-pixel error in shadow tip location at 0.31 m GSD translates directly into a height error. And the shadow boundary must be a clean, hard edge: diffuse penumbra from a low solar elevation, or shadow falling on uneven terrain, degrades precision substantially.
Where the published record stands
The method is not speculative. Published work in remote sensing literature has applied shadow-length photogrammetry to Egyptian pyramid complexes, demonstrating height retrieval within roughly 1–2 m of ground-truth values using commercial VHR imagery. Studies on Mesoamerican temple mounds have used the same principle to estimate mound volumes and detect differential subsidence between construction phases. The technique appears regularly in the MDPI Remote Sensing journal and related outlets.
Change detection, as opposed to single-epoch height estimation, is harder. It requires two or more images with independently precise solar angle records, matched orthorectification, and ideally similar off-nadir viewing geometry so that any residual parallax error is consistent across epochs. When those conditions are met, height changes of 20–50 cm have been reported as detectable in published studies using sub-metre imagery. Smaller changes remain ambiguous unless the shadow boundary can be localised to better than half a pixel, which is achievable with careful sub-pixel edge detection algorithms but is not guaranteed in operational conditions.
Honest limits: what this method cannot do
Cloud cover is an absolute blocker. Unlike SAR-based deformation methods, shadow photogrammetry needs clear optical conditions at acquisition. In humid tropical environments with persistent cloud, reliable multi-year time series are difficult to assemble. Scheduling acquisitions around cloud windows at sites in Mesoamerica or Southeast Asia demands patience and sometimes years of tasking attempts.
Low solar elevation angles, which produce long shadows and therefore higher theoretical sensitivity, also produce wider penumbral zones and increase the risk of shadows falling across complex terrain. High solar elevation angles shorten shadows and compress the height signal into fewer pixels. The practical sweet spot for most mid-latitude monuments is solar elevation between 25° and 45°. Below 20°, penumbra dominates. Above 55°, the shadow is too short to measure reliably at monument scales under 20 m.
The method measures the height of the shadow-casting surface, not necessarily the monument's original design height. Rubble accumulation at the base, vegetation growth on earthworks, or debris aprons all shift the effective base datum. Interpreting height change as structural loss requires careful baseline documentation and, ideally, corroboration from field survey or a sibling InSAR analysis.
Acquisition planning is half the analysis
Getting the image geometry right before tasking is not a preliminary step, it is most of the analytical work. The solar elevation at a given site varies by season and time of day; tasking windows must be specified to within a narrow time band to hit the target elevation range. For a site at 25°N, a solar elevation of 35° occurs at predictable local times in spring and autumn but may not be achievable at all in midsummer without accepting a very short shadow.
Off-nadir angle matters independently of solar angle. Steep off-nadir views introduce parallax that displaces shadow tips in the image plane by an amount proportional to monument height. This displacement is predictable and correctable if the sensor model is rigorous, but it must be accounted for explicitly in the measurement pipeline. Nadir or near-nadir acquisitions (off-nadir below 15°) are preferable for shadow photogrammetry unless the site geometry makes them impractical.
From measurement to monitoring programme
A single shadow-length measurement gives a height estimate with an uncertainty range. A time series of measurements, acquired at matched or precisely characterised solar angles over months or years, gives a change trajectory. The value of that trajectory is proportional to the consistency of the acquisition conditions: same sensor, similar viewing geometry, documented solar ephemeris for each epoch.
For heritage authorities managing monuments under active threat from erosion, subsidence, or unauthorised modification, a twice-yearly tasking cadence is usually sufficient to detect structurally significant change before it becomes irreversible. Annual tasking can miss rapid events. The archive depth of WorldView-2 (operational since 2009) and Pleiades 1A/1B (2011 onwards) means that retrospective baselines are often available without new tasking, which is worth checking before commissioning fresh acquisitions.
Satellize structures these workflows as part of its satellite-data analytics service, combining commercial tasking coordination with the geometric processing pipeline. The approach is directly analogous to the acquisition-planning logic used in its Tonga crop-estimation programme, where solar angle and sensor geometry are pre-computed for each tasking window rather than treated as post-hoc metadata.
Typical figures
| Best available panchromatic GSD | 0.31 m (WorldView-3); 0.30 m (Pleiades Neo) |
| Typical shadow-length measurement precision | 0.5–1.5 pixels after sub-pixel edge detection; translates to ~15–50 cm height uncertainty at 35° solar elevation |
| Minimum detectable height change (reported in literature) | ~20–50 cm under optimal geometry; larger under adverse shadow or terrain conditions |
| Solar elevation working range | 25°–45° preferred; below 20° penumbra degrades shadow tip; above 55° shadow too short for monuments under 20 m |
| Off-nadir angle limit for reliable measurement | Prefer below 15°; up to 30° acceptable with rigorous sensor model correction |
| Revisit (WorldView-3 tasked) | 1–4.5 days depending on latitude and priority; cloud-free acquisition may require multiple scheduling attempts |
| Archive baseline depth | WorldView-2 from 2009; Pleiades 1A/1B from 2011; SPOT-6/7 from 2012 |
| Delivery formats | Orthorectified GeoTIFF (16-bit); height-change GIS layer (GeoPackage or Shapefile); PDF measurement report with solar ephemeris metadata |
| Cloud cover constraint | Optical method; acquisition fails under cloud. No cloud-penetrating fallback within this technique alone |
Analytics Satellize can run
| Single-epoch monument height estimate | Shadow-length photogrammetry using solar ephemeris and sub-pixel edge detection on orthorectified VHR panchromatic imagery | GIS point or polygon layer with height value, uncertainty range and solar angle metadata; PDF summary report |
| Multi-epoch height change time series | Consistent shadow-length measurement across archive and newly tasked imagery; change computed as differenced height estimates with propagated uncertainty | Time-series chart with confidence intervals per epoch; GIS layer flagging statistically significant change events |
| Acquisition geometry optimisation plan | Pre-computation of solar elevation, azimuth and off-nadir angle windows for a given site and target measurement precision; output defines tasking parameters | Tasking specification document with permitted acquisition windows by calendar date and time band |
| Shadow boundary quality assessment | Automated penumbra width measurement and terrain-shadow contamination check per acquired image; images failing quality thresholds flagged before measurement | Per-image quality score; pass/fail classification for inclusion in change time series |
| Volume change proxy for earthworks | Shadow-derived height change combined with planimetric footprint from image segmentation to estimate volume delta; method follows published mound-volume studies | Volume change estimate (m³) with stated uncertainty; suitable for condition reporting to heritage authorities |
| Archive retrospective baseline extraction | Mining existing WorldView and Pleiades archive for historical acquisitions at acceptable solar and viewing geometry; height measurements extracted without new tasking cost | Retrospective height time series extending up to 15 years; gap analysis identifying periods with no usable archive imagery |
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.