TL;DR: Aerial photogrammetry for mining works by flying a drone over a pit, dump or stockpile yard, capturing hundreds of overlapping geotagged photos, then using software to triangulate matching pixels into a dense 3D point cloud, surface model and orthophoto. With ground control points surveyed by total station or GNSS, the result is tied to GDA2020/MGA2020 and AHD with horizontal accuracy of 15-30 mm and vertical accuracy of 30-50 mm — accurate enough for end-of-month volumes, design conformance and rehabilitation reporting.
Key takeaways
- Photogrammetry reconstructs 3D geometry from 2D images by matching common features across photos taken with 70-80% forward and 60-70% side overlap; without that overlap, the model fails or warps.
- Ground control points (GCPs) surveyed to MGA2020 with RTK GNSS or a total station are what make a drone survey defensible — 5-9 well-distributed GCPs across a pit typically deliver 20-30 mm horizontal accuracy.
- RTK/PPK drones (DJI Matrice 350 RTK, Phantom 4 RTK) reduce GCP dependence, but a few check points are still essential to verify and report accuracy.
- A 50-hectare pit that takes a ground crew 2-3 days can be flown in 20-40 minutes and processed overnight — the saving is in field time and exposure, not processing time.
- All commercial mine flights in Australia fall under CASA Part 101; operators need a RePL and the business a ReOC, plus site-specific approvals from the mine's drone management plan.
How photogrammetry turns photos into a 3D mine model
Aerial photogrammetry is the science of measuring real-world geometry from photographs. When you fly a grid pattern over an open-cut pit and photograph the ground from many positions, every point on the surface appears in multiple overlapping images from slightly different angles. Processing software (Pix4Dmaper, Agisoft Metashape, Bentley iTwin Capture or DJI Terra) identifies tens of thousands of common features across those images, then uses the parallax between them — the same principle your two eyes use for depth perception — to calculate the 3D position of each point.
The workflow runs in four stages. First, structure from motion estimates where the camera was for each photo and stitches them into a sparse point cloud. Second, multi-view stereo densifies that into millions of coloured points — a dense point cloud describing the pit walls, benches, haul roads and stockpiles. Third, the software builds a triangulated mesh and a digital surface model (DSM). Fourth, it drapes the original imagery over the surface to produce an orthomosaic: a single, geometrically corrected, top-down map where you can measure distances and areas directly, free of the lens and perspective distortion of a raw photo.
The output is not a picture — it is survey data. Every pixel and point carries a coordinate, which is what lets a surveyor calculate stockpile tonnage, compare this month's pit floor against design, or measure how far a tailings embankment has crept since the last flight.
Why ground control is the difference between a map and a survey
A drone's onboard GPS alone places a model roughly — often a metre or more out, and on a local datum that does not match the mine's survey grid. For mining work that feeds reconciliation, grade control or statutory reporting, that is not good enough. The fix is ground control points.
Before or after the flight, the survey crew marks and measures targets across the site — typically high-contrast checkerboard panels or painted crosses on stable ground. Each GCP is positioned to MGA2020 (the Map Grid of Australia, projected from GDA2020) with vertical heights on AHD, using RTK GNSS connected to a CORS network or a local base, or a Leica/Trimble total station resection where satellite coverage is poor under pit walls. During processing, the surveyor identifies each target in the images and assigns its surveyed coordinate. The software then constrains the entire model to those known points.
Distribution matters more than raw quantity. Five to nine GCPs spread evenly across the site — including the pit floor, the crest and at least one at depth — anchor the model far better than a dozen clustered near the ramp. A separate set of check points, surveyed but withheld from processing, is used afterwards to independently verify accuracy. Reporting the residuals on those check points is what turns "the drone flew it" into a survey you can stand behind in an audit.
Equipment and accuracy you can expect
The capture platform depends on site size and required precision. For pit and dump monitoring across most Australian operations, ISS typically deploys:
| Sensor type | Typical platform | Best mining use |
|---|---|---|
| RGB photogrammetry | DJI Matrice 350 RTK, Phantom 4 RTK | Stockpile volumes, end-of-month pit progress, orthophotos |
| Fixed-wing photogrammetry | WingtraOne, senseFly eBee | Large tenements, haul road corridors, regional mapping |
| LiDAR (supplementary) | Riegl or Hesai payload on Matrice | Surveying through light vegetation on rehab and exploration ground |
With RTK/PPK positioning and a properly distributed GCP network, RGB photogrammetry delivers 15-30 mm horizontal and 30-50 mm vertical accuracy on bare, well-textured surfaces — comfortably inside the tolerance for stockpile reconciliation and earthworks conformance. That said, photogrammetry has hard limits a good surveyor will tell you about: it cannot see through vegetation (LiDAR is needed on rehab ground), it struggles on water, fresh black coal, wet ore and featureless surfaces, and it cannot survey vertical or undercut faces that the camera never sees from above.
It is worth being honest about where photogrammetry sits against ground methods. A Leica total station resolves a discrete point to 2-5 mm, and RTK GNSS to 10-20 mm. For dimensional control, blast pattern setout or deformation monitoring of a structure, those millimetre methods remain superior. Photogrammetry wins decisively on area, speed and safety — capturing millions of points across a whole pit in one flight — which is exactly the trade-off that suits volumetrics and broad-area monitoring.
What aerial photogrammetry is used for on a mine site
The same dataset answers several questions at once, which is why it has become routine across the Pilbara iron ore operations, the Bowen Basin coal mines and the WA Goldfields.
- Stockpile and ROM pad volumes. A monthly flight of a port or processing stockpile yard produces cut/fill volumes against the pad surface for reconciliation against weighbridge and survey records.
- End-of-month pit and dump progress. Comparing this month's surface to last month's gives mined volumes, remaining design, and waste movement for the mine plan.
- Design conformance. The as-built surface is differenced against the engineering design to flag where benches, batters or haul roads deviate from plan.
- Rehabilitation and compliance monitoring. Repeat surveys track landform settlement and revegetation, supporting reporting to state regulators.
- Tailings dam and embankment monitoring. Periodic capture detects movement and erosion across the wall, complementing instrumented monitoring.
- Safety and incident support. Highwall and slope geometry can be captured without putting anyone near an unstable face.
How the field-to-deliverable process actually runs
A typical end-of-month volumetric survey of a mid-sized pit follows a predictable sequence, and most of it is preparation rather than flying.
Step 1: Flight planning and approvals (Timing: before mobilisation)
The surveyor designs the flight in mission software — defining the area, flight height (commonly 80-120 m AGL for the right ground sample distance), overlap and camera settings — and confirms the flight sits within the mine's drone management plan, any CASA airspace conditions, and current site approvals.
Tip: Plan flight height to hit a ground sample distance around 2-3 cm/pixel for volumetrics. Flying too high speeds capture but coarsens the model; flying too low multiplies image count and processing time for little gain.
Step 2: Establish ground control (Timing: same mobilisation)
The crew sets and surveys GCPs and check points to MGA2020/AHD using RTK GNSS or a total station, choosing stable locations clear of dust and traffic. This is the step that determines whether the output is a survey or a sketch.
Tip: Place at least one GCP on the pit floor and one near the crest so the model is constrained through its full vertical range, not just at the top.
Step 3: Fly and capture (Timing: 20-40 minutes per ~50 ha)
The drone flies the planned grid autonomously, capturing geotagged images at the set overlap. The pilot maintains visual line of sight and monitors weather, traffic and battery.
Tip: Avoid flying into low sun and harsh shadow on deep pits — even illumination produces a cleaner point cloud and fewer holes on shaded benches.
Step 4: Process and tie to control (Timing: overnight)
Images are processed through structure-from-motion and multi-view stereo, the GCPs are marked, and the model is constrained to the survey grid. The point cloud is cleaned of vegetation, vehicles and noise.
Tip: Always review the processing report. Check-point residuals and GCP errors tell you whether the survey met spec before any volume is reported.
Step 5: Deliver (Timing: 1-3 business days)
The surveyor produces the agreed outputs — point cloud (LAS/LAZ), DSM, orthophoto (GeoTIFF), volume report and a surface (DTM) for the mine's software — with a short accuracy statement citing the check-point results.
Cost considerations
A single drone volumetric survey of a mine stockpile yard or a mid-sized pit typically runs $2,500-$10,000 in Australia, with large multi-pit tenements or LiDAR-equipped work running higher. The cost drivers below all reward good planning.
| Cost factor | Impact | How to manage |
|---|---|---|
| Site area and pit count | More ground means more flights, batteries and processing | Combine adjacent areas into one mobilisation |
| Ground control effort | GCP survey time is often the largest field cost | Use permanent marked control re-used each month |
| Accuracy and deliverables | Tighter tolerances and more outputs add processing | Specify only the accuracy the task actually needs |
| Travel and FIFO | Remote Pilbara/Goldfields sites carry mobilisation cost | Schedule with other site work; book ahead |
| Repeat frequency | Monthly programs amortise setup across visits | Establish a fixed survey program and reusable control |
Set against a ground crew spending 2-3 days walking a 50-hectare pit — at full day rates and exposed to traffic and unstable ground — a flight that captures the same area in under an hour is usually the lower-cost and lower-risk option for broad-area work.
CASA rules and site approvals in Australia
Commercial drone surveying on a mine is regulated under CASA Part 101 of the Civil Aviation Safety Regulations. In practical terms: the pilot must hold a Remote Pilot Licence (RePL), and the operating business must hold a Remotely Piloted Aircraft Operator's Certificate (ReOC). Standard operating conditions cap altitude at 120 m AGL, require visual line of sight, daylight operation, and separation from people and aerodromes — and operations near controlled airspace need additional CASA approval.
On top of CASA, every mine runs its own drone management plan with site-specific rules: exclusion zones around the pit, coordination with light-vehicle and haul-truck movements, spotter requirements and inductions. ISS holds the required CASA certifications and works to each site's procedures, so the flight is compliant before it leaves the ground.
Frequently asked questions
How accurate is aerial photogrammetry for mining surveys?
On bare, well-textured surfaces with RTK/PPK positioning and a good GCP network, expect 15-30 mm horizontal and 30-50 mm vertical accuracy tied to MGA2020/AHD. That comfortably meets stockpile reconciliation and earthworks conformance. For millimetre-level dimensional control, a total station is still the right tool.
Do I need ground control points if my drone has RTK?
Yes — at least check points. RTK/PPK positioning greatly reduces how many GCPs you need and can deliver good absolute accuracy, but independent check points are essential to verify and report the accuracy actually achieved. For statutory or audited volumes, a small GCP network plus check points remains best practice.
Can photogrammetry survey through vegetation on rehab ground?
No. RGB photogrammetry only measures the visible surface, so dense grass or scrub is captured as the top of the canopy, not bare earth. For vegetated rehabilitation or exploration ground, a LiDAR payload that penetrates light vegetation is the correct solution, often flown on the same platform.
How does photogrammetry compare to LiDAR for mine surveying?
Photogrammetry is cheaper, produces true-colour imagery and excels on open, bare pit and stockpile surfaces. LiDAR penetrates vegetation, performs better in low texture and low light, and gives cleaner ground returns — at higher cost. Many sites use photogrammetry for routine volumes and LiDAR where vegetation or face geometry demands it.
What weather stops a drone survey?
Sustained wind above roughly 35-40 km/h, rain, fog and heavy dust will ground or degrade a flight, and low sun creates shadow that leaves holes in deep pits. Photogrammetry also performs poorly on standing water and wet, dark surfaces, so timing around weather and light matters for data quality.
What deliverables do I get from an aerial mine survey?
Typically a coloured point cloud (LAS/LAZ), a digital surface model, a georeferenced orthophoto (GeoTIFF), volume reports, and a DTM ready for your mine software — all on MGA2020/AHD with an accuracy statement based on check-point residuals.
Request a quote
If you need defensible end-of-month volumes, design conformance or rehabilitation monitoring across your operation, ISS delivers CASA-compliant aerial photogrammetry tied to GDA2020/MGA2020 and AHD, with the ground control and accuracy reporting that statutory mining work demands. Talk to our team about a fixed survey program for your site and we will scope the flights, control and deliverables to suit. Call 0407 057 015 or request a quote through our contact page to get started.