Quarry blast and extraction volume estimation from optical stereo
Repeat tri-stereo photogrammetry from very-high-resolution optical satellites produces digital surface models accurate to 20–50 cm vertically, giving royalty auditors and regulators an independent record of how much material left the pit.
Sensors
- Airbus Pléiades Neo: 30 cm panchromatic resolution; native tri-stereo acquisition in a single pass collects fore, nadir and aft images simultaneously, minimising temporal parallax from moving equipment. Revisit of 1–2 days at mid-latitudes. The preferred sensor for sub-metre DSM work.
- Maxar WorldView-3: 31 cm panchromatic, 1.24 m multispectral. Agile enough to collect stereo or tri-stereo pairs in one pass. Published CE90 horizontal accuracy of around 3.5 m without ground control, improvable to under 1 m with GCPs. Strong contrast in the SWIR bands helps distinguish rock types.
- Airbus SPOT 6/7: 1.5 m panchromatic. Useful for larger quarry footprints where centimetre-level vertical precision is less critical than areal coverage. Stereo pairs available from archive; lower cost per km² than Pléiades Neo.
- Planet SkySat: 50 cm panchromatic. Can be tasked for stereo pairs; revisit is flexible given the constellation size. Vertical accuracy from SkySat stereo is typically 1–2 m without GCPs, adequate for coarse volume checks but not royalty-grade auditing at small quarries.
What a point-cloud difference actually measures
The method is straightforward in principle and unforgiving in practice. Two tri-stereo acquisitions, separated by weeks or months, each produce a dense point cloud via image matching. Subtract the earlier surface from the later one, cell by cell across a common grid, and every negative value is material removed. Integrate those depth differences over area and you have volume extracted.
The photogrammetric engine does the heavy lifting: semi-global matching or patch-based multi-view stereo algorithms correlate pixel intensities across the stereo images to assign elevation to each ground point. Pléiades Neo's 30 cm imagery, processed at native resolution, typically yields a DSM grid spacing of 50 cm to 1 m. At that density, a single blast that moves 50,000 cubic metres of rock is not a rounding error; it is a clear signal.
Volume figures are reported in cubic metres of bulk material. Converting to tonnes requires a bulk density assumption for the specific rock type, which the operator normally holds and which an auditor can cross-check against weigh-bridge records. The satellite does not know the density; it measures geometry.
Vertical accuracy: the honest numbers
Under good conditions, 20 to 50 cm vertical RMSE is achievable with Pléiades Neo or WorldView-3 tri-stereo, provided the scene has adequate texture contrast and at least three well-distributed ground control points (GCPs) or a high-quality reference DSM for co-registration. That figure comes from published validation studies and from Airbus and Maxar's own technical documentation.
Several things degrade it. Shadow zones on the shaded side of bench walls can be 10 to 30 metres wide in morning or late-afternoon acquisitions; image matching fails in near-zero contrast and the DSM interpolates across the gap, introducing errors that can exceed 1 m locally. Uniform pale limestone or chalk faces present a similar problem: the matching algorithm finds no texture to grip, and elevation estimates become unreliable. Wet pit floors with specular reflection from standing water cause the same failure mode.
The practical response is to acquire images near solar noon to minimise shadow, to flag shadow-affected cells in the delivered DSM, and to report volume uncertainty as a range rather than a point estimate. A 5% volume uncertainty on a 500,000-tonne extraction is 25,000 tonnes. That is worth stating clearly to a royalty authority before they sign off on the methodology.
Blast-event attribution and change cadence
A single DSM difference tells you how much rock moved between two dates. Repeat acquisitions, say monthly or after each scheduled blast, build a time series that can attribute volume to individual blast events. This matters for royalty schemes that charge per blast or per calendar quarter rather than annually.
Pléiades Neo's 1–2 day revisit means a tasking order placed within 24 hours of a blast notification can capture the pre- and post-blast surface before the next extraction cycle disturbs the geometry. In practice, cloud cover is the binding constraint in tropical and temperate climates. Optical stereo is genuinely limited here: a persistent cloud deck for two weeks can break the attribution chain entirely. SAR-based volume estimation is less precise but cloud-immune; for high-value royalty monitoring, a hybrid approach is worth considering.
Archive depth is also relevant. Pléiades (the predecessor to Neo) has collected imagery since 2012; WorldView-3 since 2014. An operator disputing a historical royalty assessment can, in principle, be checked against archived stereo collections if the site was tasked at the time.
Ground control and the GCP problem
Without ground control, even Pléiades Neo carries a vertical bias of up to 3–5 m from residual attitude and orbit errors. That bias is systematic across a scene and largely cancels in a difference operation between two acquisitions from the same sensor and similar geometry. But it does not cancel perfectly, and a systematic tilt in one acquisition that is absent in the other will appear as a false volume signal.
The cleanest solution is a small network of permanent GCPs at the quarry perimeter, surveyed once by GNSS to centimetre accuracy and visible in every acquisition. Four to six well-distributed points are typically sufficient. Where that is not possible, a lidar or RTK-drone survey of stable ground adjacent to the pit can serve as a reference surface for co-registration. The satellite DSM is then anchored to that reference before differencing.
Clients sometimes ask whether drone photogrammetry makes satellite stereo redundant. For a single quarry with good site access, a drone survey is faster and cheaper per epoch. Satellite stereo becomes the better choice when access is restricted, when the portfolio spans dozens of sites across multiple countries, or when an independent third-party measurement is legally required.
Regulatory and royalty applications
The primary buyers of this analysis are mining ministries, royalty streaming companies, and environmental regulators. Each has a slightly different need. A ministry wants to verify that declared extraction volumes match the royalty payments received. A streaming company wants to confirm that production at a royalty asset is tracking the operator's guidance. A regulator wants to check that the pit is not expanding beyond its permitted boundary.
All three can be served by the same DSM-difference workflow, with different reporting emphases. Boundary compliance adds a horizontal dimension: the permitted pit outline is overlaid on the DSM and any extraction outside the polygon is flagged. Volume verification compares the satellite-derived figure against the operator's declared tonnage, converted at the declared bulk density. Discrepancies above a threshold, say 10%, trigger a formal query.
Satellize structures this type of analysis as a periodic delivery tied to the client's royalty reporting calendar. The Tonga crop-estimation programme demonstrated the same principle in a different domain: satellite-derived measurements providing an independent check on ground-declared figures. The methodology transfers directly.
What this method cannot do
Stereo photogrammetry measures surface geometry. It cannot see underground workings, measure grade or ore quality, distinguish overburden from ore in a mixed stockpile, or detect subsurface voids. It also cannot resolve extraction volumes at underground mines where the surface disturbance is limited to portal areas and waste dumps.
At very small quarries, say under 2 hectares, the volume signal from a single blast may fall within the combined uncertainty of two DSM epochs. A 30 cm vertical RMSE across a 2-hectare pit implies a volume uncertainty on the order of 600 cubic metres per epoch, which may exceed the blast volume itself. The method is best suited to medium and large open pits where individual extraction events move tens of thousands of cubic metres or more.
Finally, the method produces a surface volume, not a mass. The conversion to tonnes is only as good as the bulk density figure applied, and bulk density varies with moisture content, fragmentation and rock type. An auditor who accepts satellite volumes as definitive without understanding this step is accepting a number with hidden uncertainty.
Typical figures
| Best achievable spatial resolution (DSM grid) | 50 cm to 1 m (Pléiades Neo tri-stereo at native resolution) |
| Vertical accuracy (RMSE, with GCPs, good contrast) | 20–50 cm typical; degrades to >1 m in shadow zones or on uniform rock faces |
| Vertical accuracy (RMSE, without GCPs) | Up to 3–5 m systematic bias; largely cancels in same-sensor differencing but not reliably |
| Revisit for tasking (Pléiades Neo) | 1–2 days at mid-latitudes under clear sky |
| Archive depth | Pléiades from 2012; WorldView-3 from 2014; SPOT 6/7 from 2012 |
| Minimum pit size for reliable volume estimation | Approximately 2–5 hectares active face; smaller pits risk signal falling within epoch uncertainty |
| Cloud sensitivity | Optical only; persistent cloud cover breaks acquisition cadence entirely |
| Delivery formats | GeoTIFF DSM and difference raster, GeoPackage change polygons, PDF audit report with volume table and uncertainty ranges |
Analytics Satellize can run
| Epoch DSM pair and difference raster | Semi-global matching or patch-based MVS photogrammetry on tri-stereo imagery; co-registered to GCPs or stable reference surface | GeoTIFF DSM (each epoch) and signed-difference raster with shadow-masked cells flagged; delivered within 5 working days of cloud-free acquisition |
| Extraction volume report | Cell-wise integration of negative elevation change over the permitted pit boundary polygon; uncertainty propagated from vertical RMSE and cell count | PDF report stating extracted volume in m³, uncertainty range, and bulk-density conversion table for operator-declared rock types |
| Blast-event attribution log | Repeat monthly or post-blast DSM differencing; volume assigned to date intervals between acquisitions | Tabular log (CSV or GIS layer) of volume-per-interval, suitable for reconciliation against weigh-bridge or blast-record data |
| Permitted boundary compliance check | Overlay of licensed pit outline on DSM difference raster; extraction pixels outside polygon flagged and area-summed | GeoPackage with flagged encroachment polygons and area/volume statistics; included in quarterly audit report |
| Multi-site royalty portfolio summary | Standardised DSM-difference pipeline applied across a portfolio of quarry sites; results normalised to common reporting period | Consolidated dashboard (PDF or data feed) comparing declared versus satellite-derived extraction across all sites, with per-site confidence ratings |
| Shadow and data-quality mask | Solar geometry modelling at acquisition time combined with image-matching confidence scores to identify unreliable DSM cells | Raster mask layer (GeoTIFF) delivered alongside each DSM; cells classified as reliable, interpolated, or failed |
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.