3D Mapping for Mining Operations

A modern mine is one of the most spatially documented environments on the planet. Drone flights over the pit happen daily. The high wall gets a LiDAR scan most weeks. Underground development is surveyed every shift. The processing plant has been scanned and re-scanned for two decades. Stockpiles are measured monthly for reconciliation. Bench faces are captured for geotechnical mapping. Rehabilitation areas are flown for environmental reporting.

The volumes are enormous. The use cases are operationally critical. And the long-term consequences of poor data management are severe — mines run for decades, and the spatial record of an operation is, more than the corporate documents or the engineering drawings, the actual record of what happened on that ground.

Here’s how 3D mapping fits into mining operations in 2026, what’s actually being captured, and the data problem that compounds every year an operation runs.

Open-cut vs underground vs plant

Mining is not a single environment. It is at least three, each with distinct capture needs.

Open-cut

Open-cut operations are dominated by aerial work. A weekly or fortnightly drone flight over the active pit produces orthomosaics and point clouds that feed into volumetrics, slope monitoring, and progress reporting. Larger operations may run multiple captures per week — daily during high-activity periods, less frequently when the pit is in a stable state.

The pit is not the only target. Stockpile areas, ROM pads, infrastructure corridors, haul roads, and waste rock dumps all need routine capture, typically on a different cadence to the pit itself. Stockpile flights for reconciliation are usually monthly. Infrastructure scans are episodic — triggered by works, inspections, or asset condition assessments.

Underground

Underground operations capture by SLAM. A handheld or backpack LiDAR system (GeoSLAM, Emesent Hovermap, NavVis VLX) walks the development heading and produces a point cloud of the freshly mined drive. The capture is done by the surveyor, frequently per shift in active development areas.

In stoping operations, the additional category is post-blast and post-firing scans of the stope itself — frequently done with cavity monitoring systems (CMS) that scan inaccessible voids from the access drive. These produce the spatial data behind stope reconciliation, which is the foundation of grade control and reserve accuracy.

Processing plant and infrastructure

The plant is captured episodically. Terrestrial LiDAR scans of conveyors, crushing circuits, flotation cells, tailings infrastructure, and structural steel are commissioned around major shutdowns, capital projects, or asset integrity reviews. The captures are large, dense, and used for engineering design — typically by external engineering firms working on plant expansions or upgrades.

The pattern across all three environments is the same: high-volume, high-frequency capture, by multiple parties, over a very long operational life. For more background see survey data management for mining.

What gets captured, and how often

A mid-sized open-cut operation typically generates 300–800 GB per month of new spatial data, between internal teams and external contractors. A large underground or multi-pit operation can comfortably exceed a terabyte per month.

For an operation running for 30 years, that compounds. Even at a conservative 5 TB per year, the spatial archive at the end of operations is 150 TB. In practice, most long-running operations have a fragmented, partial, undocumented spatial archive distributed across dozens of file shares, contractor handovers, and decommissioned servers — but the underlying volume of what was captured at some point is far larger than what is currently retrievable.

The core use cases

Stockpile volumetrics

The most-cited use case, and for good reason. A drone flight over a stockpile area produces a point cloud and an orthomosaic. Comparing the surface against a known base level (or a previous capture) gives a volume. The volume gives a tonnage. The tonnage feeds into reconciliation and reporting.

The accuracy is good enough for monthly inventory work — typically ±1–2% on a well-flown stockpile with proper ground control. For statutory reporting, this has been operationally accepted across most major jurisdictions for several years.

Slope stability monitoring

Repeated LiDAR scans of high wall and bench faces feed into slope movement detection. The technique is straightforward: capture the same wall at fortnightly intervals, register the point clouds, compute the displacement field. Any movement above noise threshold is flagged for the geotechnical engineer.

For active monitoring of known unstable areas, this is supplemented by radar (IBIS, GroundProbe) and prism monitoring — but the LiDAR record provides the volumetric context that the point-monitoring techniques lack. A 3 mm displacement at a prism point is data; a 3 mm displacement vector field across the entire wall is understanding.

Geotechnical mapping

Bench faces are scanned for structural mapping — identifying joint sets, joint spacing, persistence, and orientation. The point cloud is processed in specialist software (Maptek PointStudio, Cloudcompare with geomechanical plugins) to extract structural data that feeds into slope design and risk assessment.

This work was historically done by a geotechnical engineer with a compass and a notebook standing in a moderately dangerous place. The shift to remote structural mapping from drone or terrestrial LiDAR is one of the clearest safety wins of 3D capture in mining.

Stope reconciliation

In underground operations, the difference between the planned stope and the actual mined void is reconciliation. The planned stope comes from the design. The actual void comes from a cavity scan after firing. The reconciliation drives everything from grade estimation to dilution analysis to next-stope design.

A modern underground operation generates hundreds of cavity scans per year. The historical archive of these scans is one of the most operationally valuable datasets the mine owns — and one of the most commonly lost in contractor handovers and server migrations.

Rehabilitation and closure

Rehabilitation areas are flown quarterly or annually to document revegetation, landform stability, and progressive closure work. The orthomosaic time series is the primary evidence presented to regulators for bond release.

Closure planning extends this further. Modern mine closure plans increasingly require 3D documentation of the final landform, including walk-throughs of permanent infrastructure and 3D capture of any remaining open voids. This data is expected to persist long after the operation has closed.

Environmental compliance

Tailings dam inspections, water management infrastructure, dust monitoring, biodiversity offset areas — all increasingly include a 3D documentation requirement. The regulator’s expectation is shifting from “show us the report” to “show us the data the report came from.”

For context on time-series environmental capture more generally, see environmental monitoring site records.

Data residency

Mining operations cross borders in three directions. The operation is in one country. The corporate office is in another. The engineering consultants are in a third. And the data centre storing the spatial record might be in a fourth, depending on which cloud platform someone signed up to in 2019.

This matters more than most operations acknowledge. Several jurisdictions have specific rules around mineral exploration data, geological survey records, and resource estimation inputs. Indigenous land councils and traditional owner groups frequently negotiate data sovereignty terms that require host-country storage of spatial records taken on country. Foreign investment review boards have been known to inspect data residency arrangements as part of joint venture approvals.

Generic cloud storage gives little or no control over which data centre region a file sits in. SharePoint will store your point cloud somewhere. Dropbox will store it somewhere else. Whether either of those locations meets your obligations is something nobody on the operations team is in a position to verify.

For operations subject to data residency rules, the storage region needs to be explicit, configurable per dataset, and auditable. Australian operations need a Sydney option. Canadian operations need a Toronto option. UK operations need a London option. South African, Chilean, and Indonesian operations need locally compliant equivalents.

This is the part of the spatial data infrastructure that is rarely visible during procurement and consistently expensive to retrofit after the fact.

The long-term archive problem

Mines run for decades. A new copper operation coming online today will likely be operating in 2070. Anything captured in the first ten years of operations needs to remain accessible — searchable, viewable, usable — for the next forty.

Generic file storage does not survive this. Folder structures get reorganised. Servers get decommissioned. Cloud providers change their pricing and the cold archive becomes uneconomic. Contractor handovers drop data. Personnel changes drop context. Five years after a capture, the surveyor who flew it has moved firms, the project manager who commissioned it has retired, and the data either sits in an undocumented folder or has been quietly deleted in a cleanup.

The pattern repeats every five to ten years. A new operations geologist arrives, looks at the spatial archive, declares it unusable, and commissions a fresh baseline survey. The historical record is effectively re-started. The cost of this — in lost context, in re-survey work, in regulatory delay — is rarely calculated.

The discipline that prevents this is not technical. It is organisational. Every capture needs to be anchored to a specific spatial unit (a pit, a bench, a heading, an asset), time-stamped to a specific date, attributed to the capturing party, and stored in a location that survives staff turnover and contractor changes. The capture metadata is as important as the capture itself.

Where Swyvl fits

Swyvl is built around exactly this organisational structure. Each spatial unit of the operation becomes a site. Every capture lands on the correct site, time-stamped to the survey date, viewable in the browser, and accessible to anyone with current permission. The Australian operations store in Sydney, the Canadian operations store in Toronto, the UK operations store in London — multi-region storage is explicit, not implicit. The audit trail records every access event with IP and approximate location, which matters when a regulator asks who looked at the slope scan before the wall failure.

The pattern works particularly well for the contractor problem. A typical operation has three to six different spatial data contractors active at any time — internal surveyors, external drone providers, specialist scanning firms, plant inspection consultants. Each delivers to the same site record. The mine ends up with a coherent record of what was captured at each location, regardless of who captured it.

Getting structured before the volume becomes unmanageable

The decision to bring structure to spatial data in mining is usually made too late. It is made at the point where someone — a new GM, a new chief surveyor, a regulator with an awkward question — looks at the existing state and concludes that it cannot continue.

The transition from that point is harder than it would have been if structure had been in place from the start. Three years of legacy captures sitting in unstructured folders is a problem; ten years is a much bigger problem.

For operations not yet at that point, the cheapest version of this work is to put the structure in place now and let it accumulate captures going forward. The legacy backfill can happen selectively. The active data starts being organised properly from today.

For more on the contractor multi-source delivery problem in mining, see survey data management for mining. For the operational case more broadly, see how to deliver drone survey data.

Mining is one of the most spatial-data-intensive industries on the planet, and one of the worst at retaining the spatial data it captures. The capture technology is excellent. The operational use cases are clear. The processing software is mature. The gap — and it is the gap that determines whether the data delivers long-term value — is in how it is organised, stored, and made accessible across a thirty-year operational life.

That is the part worth building properly the first time.