Open-pit mine expansion monitoring
Multi-temporal very-high-resolution imagery reveals whether an open-pit mine is growing within its permitted boundary, by tracking bench cuts, haul-road extensions and waste-dump spread across successive satellite passes.
Sensors
- Sentinel-2 MSI: 10 m resolution in visible and near-infrared bands, 5-day revisit at the equator with both satellites. Provides the frequent baseline from which anomalous footprint growth is flagged before tasking a finer sensor.
- Planet SuperDove: 3 m resolution, 8 spectral bands, daily revisit over most land areas. Resolves haul-road extensions and new bench faces that are invisible at Sentinel-2 scale, and its archive from 2021 onward supports retrospective change analysis.
- Maxar WorldView-3: 0.31 m panchromatic, 1.24 m multispectral. The sharpest commercially available optical sensor for confirming individual bench widths, counting active faces and measuring waste-dump toe positions. Revisit is roughly 1–4.5 days depending on latitude and tasking priority.
- Airbus Pléiades Neo: 0.30 m panchromatic, 1.2 m multispectral, stereo and tri-stereo collection modes. Stereo pairs from a single pass allow a surface model to be derived, supporting vertical as well as lateral change detection without a separate SAR acquisition.
What the geometry of a pit actually tells you
An open-pit mine grows in two directions simultaneously: outward as the crest line advances across the permitted area, and downward as benches are cut into progressively deeper ore. Both movements leave unambiguous signatures in satellite imagery. A new bench cut appears as a fresh linear scarp with a characteristic shadow geometry that depends on sun angle and bench height. Typical bench heights in large hard-rock pits run from 10 m to 15 m; at WorldView-3's 0.31 m panchromatic resolution, the shadow cast by a 12 m bench face is roughly 4–8 pixels wide at a 45-degree sun elevation, which is comfortably detectable.
Haul-road extensions are equally legible. A road wide enough to carry a 300-tonne truck is typically 30–40 m across, which registers clearly even at Planet SuperDove's 3 m resolution. Waste dumps grow by progressive tipping, and their toe advances in a pattern that is both measurable and datable when imagery is collected at regular intervals. None of this requires any co-operation from the operator.
How the change-detection workflow is structured
The baseline layer is built from Sentinel-2 time series. At 10 m and a 5-day revisit, Sentinel-2 is not resolving individual benches, but it is reliably capturing the overall disturbed-ground footprint through spectral unmixing of bare soil and rock against surrounding vegetation. A change in that footprint triggers a tasking order for a finer sensor, rather than running expensive commercial imagery continuously over every site.
Once a higher-resolution image arrives, the comparison is made against the most recent cloud-free acquisition of equivalent resolution. Pixel-based change detection is supplemented by object-based analysis: the algorithm looks for new linear features consistent with road or bench geometry, and for convex-hull expansion of the waste-dump polygon. The permitted boundary, digitised from the regulatory document, is overlaid as a fixed reference. Any detected feature that falls outside it generates an exceedance flag.
Cloud cover is the principal operational constraint. In tropical mining regions such as parts of the Democratic Republic of Congo or Indonesia, consecutive cloud-free acquisitions at sub-metre resolution may be separated by weeks rather than days. Sentinel-2's cloud-mask product and Planet's daily revisit help bridge those gaps, but a period of persistent cloud cover will always introduce latency into the detection timeline. This is an honest limitation of passive optical methods that SAR can partially address, though SAR is handled on a separate page in this library.
Resolving the vertical dimension without a dedicated lidar pass
Stereo optical collection from Pléiades Neo or WorldView-3 allows a digital surface model to be derived photogrammetrically. Differencing two surface models collected months apart gives a volumetric change estimate for the pit and the waste dump. Published assessments of stereo-derived DSMs from WorldView sensors report vertical accuracies in the range of 0.3 m to 1 m root-mean-square error over flat terrain, degrading somewhat on steep pit walls where occlusion becomes a factor.
This is not a substitute for a precision survey, and regulators should not be told it is. What it does provide is an independent, dateable record of vertical growth that can corroborate or challenge an operator's own reported extraction volumes. For permit-compliance purposes, the question is usually not centimetric accuracy but whether the pit floor has advanced beyond the permitted depth limit, which is typically specified in tens of metres.
The permit boundary as the analytical anchor
Every other analytic in this workflow is only meaningful relative to a fixed legal reference. Permitted footprints are usually defined in national mining cadastre documents, environmental impact assessment maps or concession boundary shapefiles. The quality of those source documents varies considerably. Some are georeferenced to sub-metre accuracy; others are hand-drawn on 1:50,000 topographic sheets and carry positional uncertainty of 50 m or more.
Where the source boundary is imprecise, the analysis must be honest about the uncertainty buffer. A pit crest that appears to be 30 m outside a digitised boundary is a strong exceedance signal. One that appears 15 m outside a boundary with 50 m positional uncertainty is ambiguous and requires field verification or a higher-quality cadastre source. Satellize flags both, but categorises them differently in the output layer.
Archive depth and the retrospective audit
One underused capability is the retrospective audit. Sentinel-2 archive runs continuously from 2015; Landsat's archive extends to 1972 at 30 m resolution and to 1999 at 15 m pan-sharpened. Planet's SuperDove archive begins in 2021. Commercial very-high-resolution imagery from DigitalGlobe predecessors to Maxar is available through the USGS Earth Explorer interface with coverage of many major mine sites from the early 2000s onward.
This means that for a mine with a disputed expansion history, it is possible to reconstruct the footprint at roughly annual intervals over two decades, and at monthly or finer intervals from 2021 onward. That record is admissible evidence in regulatory proceedings in a number of jurisdictions. It is also useful to investors conducting environmental due diligence before an acquisition, where the question is not what the operator claims but what the ground actually shows.
Satellize's analytics pipeline, which has been applied in contexts ranging from Pacific island crop estimation (the Kingdom of Tonga programme) to mine-site monitoring, can ingest multi-source archive imagery and produce a dated change timeline as a GIS deliverable.
What this method cannot do
Sub-surface extraction is invisible to optical sensors. A mine that is deepening its pit while holding its surface footprint constant will not trigger a footprint-change alert, though stereo DSM differencing will detect the vertical change. Underground workings accessed through a surface portal are largely undetectable by this method; that is addressed by the ground-subsidence InSAR page elsewhere in this library.
Dust, haze and smoke from blasting or nearby fires can obscure imagery for days at a time, and atmospheric correction does not fully recover the radiometric quality needed for precise change detection in those conditions. Finally, very small expansions, say a 5 m advance of the pit crest over a three-month period, may fall below the reliable detection threshold of any sensor at the revisit frequencies that are practically affordable. The method is well suited to detecting significant, systematic expansion; it is not a substitute for a licensed surveyor conducting a boundary peg inspection.
Typical figures
| Baseline spatial resolution | 10 m (Sentinel-2 MSI) |
| High-resolution spatial resolution | 3 m (Planet SuperDove), 0.30–0.31 m panchromatic (Pléiades Neo / WorldView-3) |
| Baseline revisit | 5 days at equator (Sentinel-2 two-satellite constellation) |
| High-resolution revisit | Daily (Planet SuperDove); 1–4.5 days on tasking (WorldView-3, Pléiades Neo) |
| Stereo DSM vertical accuracy | 0.3–1 m RMSE on flat terrain (WorldView / Pléiades Neo photogrammetry); degrades on steep pit walls |
| Minimum detectable footprint change | Approximately 30–50 m lateral advance reliably at 3 m resolution; smaller changes require sub-metre imagery and are subject to boundary positional uncertainty |
| Spectral bands used | Visible, NIR and SWIR (Sentinel-2 13 bands); 8-band visible to NIR (SuperDove); panchromatic + 8-band VNIR-SWIR (WorldView-3) |
| Archive depth | Sentinel-2 from 2015; Landsat from 1972 at 30 m; commercial VHR from early 2000s at many sites; Planet SuperDove from 2021 |
| Cloud constraint | Passive optical only; persistent cloud cover in tropical regions can introduce weeks of latency in sub-metre detections |
| Delivery formats | GeoTIFF change rasters, vector exceedance polygons (GeoPackage / Shapefile), dated change timeline report (PDF), optional GIS web layer |
Analytics Satellize can run
| Footprint change map | Multi-temporal image differencing and spectral unmixing on Sentinel-2 time series to delineate disturbed-ground extent at each epoch | GeoTIFF raster and polygon layer showing footprint at each date, with net area change in hectares |
| Permit-boundary exceedance flag | Spatial intersection of detected pit-crest and waste-dump polygons against digitised permit boundary, with uncertainty buffer applied where source cadastre accuracy is low | Categorised alert layer (confirmed exceedance / ambiguous / within permit) delivered as GeoPackage with attribute table |
| Bench and haul-road extension detection | Object-based image analysis on Planet SuperDove or VHR imagery detecting new linear features consistent with road or bench geometry, compared against prior-epoch baseline | Vector layer of newly detected features with date of first appearance and confidence score |
| Waste-dump volume change estimate | Photogrammetric DSM generation from stereo or tri-stereo VHR pairs (Pléiades Neo or WorldView-3), differenced against prior-epoch DSM to compute volumetric change | DSM difference raster and tabular volume-change report in cubic metres, with stated vertical accuracy range |
| Retrospective expansion timeline | Multi-source archive compilation (Landsat, Sentinel-2, Planet, commercial VHR) processed through consistent change-detection pipeline to reconstruct footprint history | Dated GIS timeline layer and PDF report suitable for regulatory or due-diligence use |
| Monthly monitoring bulletin | Automated Sentinel-2 screening with triggered commercial tasking on anomaly detection; human-reviewed summary of findings | Monthly PDF bulletin with annotated imagery, change statistics and any exceedance flags raised in the period |
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.