The Mine Disaster That Sent 25 Million Cubic Metres Into British Columbia's Watershed, and the Three-Year Window Nobody Caught

On August 4, 2014, the tailings dam at Imperial Metals' Mount Polley copper and gold mine in British Columbia failed. Approximately 17 million cubic metres of water and 8 million cubic metres of tailings material were released into Polley Lake, Hazeltine Creek, and Quesnel Lake, according to the BC government's official incident documentation. The discharge contained 134.1 tonnes of lead, 2.8 tonnes of cadmium, and 2.1 tonnes of arsenic, representing 92% of all lead released into Canada's environment in 2014 by any source. It remains the worst mine waste disaster in Canadian history.

The BC Chief Inspector of Mines and an independent panel of engineering experts found the collapse was caused by a design that failed to account for a weak layer of glacial silt beneath the embankment foundation. "It was a loaded gun," BC's Mines Minister said, "because the design of the dam was incorrect."

But the design error didn't cause the failure instantly. A peer-reviewed deformation analysis of the Mount Polley Tailings Storage Facility, published in the Canadian Geotechnical Journal, found that the embankment failed in a progressive manner, and that the foundation may have started failing locally as early as 2011. Three years before the dam breached.

Three years of developing deformation. Three years of a structural problem progressing from "beginning locally" to "catastrophic breach." The monitoring instrumentation in place during those years did not resolve the developing failure early enough to prevent it.

That gap, between "instrumented" and "fully characterized" is the problem this article is about.

How Tailings Dam Monitoring Actually Works Today

Modern tailings storage facility monitoring combines several methods, each with distinct roles and limitations.

Piezometers measure pore water pressure within the embankment — a critical indicator of seepage and liquefaction risk. They're installed at specific points within the dam structure and provide continuous readings at those locations.

Inclinometers measure lateral subsurface movement at specific installed locations. They're valuable for detecting movement at depth, but only at the points where they've been physically installed.

Survey prisms are optical targets placed on the dam face that are periodically measured by total stations or automated monitoring systems to detect surface movement. Each prism provides a precise reading of whether that specific point has moved, but only that point.

The Global Industry Standard on Tailings Management (GISTM), published in 2020 in the wake of Mount Polley and the Brumadinho disaster in Brazil, requires continuous monitoring and independent review for higher-risk facilities. The standard explicitly recognizes the need for monitoring systems that can catch both slow deformation and sudden change.

The limitation common to prism-based and inclinometer-based monitoring is instrumentation coverage: a prism measures the prism's location. An inclinometer measures what the inclinometer is encountering. A failure that initiates between instrumented points, in a zone the prisms weren't placed, in a foundation layer the inclinometers don't reach, can develop through its precursor stages without triggering any instrumented alarm.

This is precisely what the Mount Polley analysis describes: progressive failure initiating in a location and manner that the installed monitoring program did not resolve with sufficient spatial coverage to provide early warning.

What Brumadinho Added to the Record

If Mount Polley was the warning that Canadian and global mining regulation should have acted on more decisively, Brumadinho was the consequence of not acting fast enough.

On January 25, 2019, the Córrego do Feijão mine's tailings dam in Minas Gerais, Brazil, failed catastrophically. The collapse killed 270 people. The wave of mining waste destroyed a company cafeteria during lunchtime, submerged surrounding communities, and contaminated the Rio Paraopeba watershed. It was one of the deadliest industrial accidents in South American history.

Subsequent investigation found that the dam had shown deformation indicators in the months before failure. The deformation had been observed. The significance of what was observed was underestimated.

Two events, separated by five years, on two continents, with the same underlying pattern: deformation preceding failure, instrumentation present, monitoring program in place, and a gap between the data that was being collected and the complete spatial picture of what the structure was actually doing.

Why Drone-Based LiDAR Closes the Coverage Gap

The critical limitation of prism-based monitoring isn't that prisms are inaccurate, they're highly accurate at their installed locations. The limitation is that a tailings dam embankment covers a large, complex three-dimensional surface, and monitoring a finite set of points on that surface leaves the space between those points characterized only by inference.

Drone-based LiDAR removes that inferential gap entirely.

A LiDAR survey of a tailings dam captures millions of individual measurement points across the entire embankment face, crest, downstream slope, and surrounding terrain, not a network of installed instruments, but a continuous geometric record of the full structure. The resulting point cloud represents every measurable location on the dam's surface at the moment of the flight.

When that survey is repeated, monthly, quarterly, or at whatever frequency the facility's risk profile warrants, the change detection between surveys identifies every location where the geometry has changed. Settling. Bulging. Slope movement. Crest displacement. Erosion on the downstream face. Not at the prisms. Everywhere.

<cite index="83">Instead of individual points, drone photogrammetry, LiDAR, and satellite-based radar capture the entire dam face and surrounding terrain in a single dataset, so a small deformation anywhere on the structure is visible, not just at instrumented points.</cite> A documented Chilean mining operation flies their tailings dam monthly with drones specifically to track humidity levels, stability data, and structural change, catching problems before they become emergencies.

That is the monitoring program Mount Polley didn't have. A full-surface recurring survey that would have detected the developing deformation between 2011 and 2014 as a spatial pattern, not just as readings at instrumented points, a pattern that might have triggered an engineering review and a load-management decision before the addition of 2.5-4 metres of embankment material in 2014 triggered the final progression to failure.

What the GISTM Standard Now Requires

The Global Industry Standard on Tailings Management didn't emerge from a theoretical analysis. It emerged from Mount Polley, Brumadinho, and the pattern of documented failures that preceded them. The standard explicitly requires that higher-risk tailings facilities implement monitoring programs capable of detecting both slow, progressive deformation and rapid change, and that those programs include not just ground instrumentation but remote sensing capable of characterizing the full dam geometry.

Drone-based LiDAR surveys, combined with recurring photogrammetry and change detection, directly satisfy the remote sensing component of that requirement. Time-stamped, georeferenced datasets that capture the entire embankment in each survey cycle, not just the points where instruments were installed, are precisely what the GISTM standard is trying to mandate into routine practice.

The standard also requires independent review, which means the monitoring data needs to be defensible, transparent, and capable of being evaluated by engineers who weren't present at the site. A LiDAR point cloud and a photogrammetric model with documented change detection between surveys are exactly that kind of data.

How CropCopters Would Execute This

Mission planning: Tailings facility risk classification (using GISTM or equivalent framework), embankment geometry, downstream hazard assessment, and regulatory monitoring requirements determine survey frequency and coverage area. A higher-risk facility warrants monthly survey; a lower-risk facility may be adequately served by quarterly or semi-annual monitoring, supplemented by post-storm or post-significant-rainfall flights.

Data acquisition: The Zenmuse L3 LiDAR on the M400 RTK platform captures the full embankment geometry, dam face, crest, downstream slope, toe drainage, and surrounding terrain, with 2-4cm accuracy across the survey area. The Zenmuse H30T captures RGB photogrammetric imagery for visual condition documentation and thermal imaging to detect seepage indicators and moisture anomalies at the dam face and downstream toe. The L3 and H30T require separate flights on the M400 RTK platform.

Processing: DJI Terra processes LiDAR data into classified point clouds and elevation models. Change detection between sequential surveys identifies displacement at every location on the dam face, movement that would be invisible between individual prism measurement points. Photogrammetric processing produces orthomosaic documentation of surface condition, including crack indicators, erosion on downstream slopes, and toe seepage expression.

AI audit: AI-assisted change detection screening flags zones of anomalous displacement between survey cycles, identifying locations where the rate or direction of movement deviates from the expected consolidation and settlement pattern. Flagged zones are presented for geotechnical engineer review, not treated as autonomous alerts. The distinction matters: AI identifies the anomaly, a qualified geotechnical professional determines its significance.

Deliverable:

• Full-face LiDAR point cloud baseline and recurring comparison dataset

• Change detection report showing displacement magnitude and direction by zone across the full embankment, not just at instrumented points

• Thermal anomaly map highlighting potential seepage indicators at face and toe

• RGB photogrammetric documentation of current surface condition

• Displacement trend analysis across the monitoring history

• GISTM-compatible documentation package for independent engineering review

What the operator can do next: Have a full-surface geometric baseline of their tailings facility, not prism readings at instrumented points, but the complete embankment geometry at this date. At the next survey, every location on the dam that moved is identified and quantified. The three-year precursor deformation window that characterized Mount Polley becomes something that can be detected, documented, and acted on, not something discovered in the post-failure investigation.

Technical Reality

What this does well: Delivers complete-surface deformation monitoring across the entire embankment geometry, rather than coverage limited to instrumented points. Produces a defensible, time-stamped, georeferenced record suitable for GISTM-required independent review. Accesses the dam face and crest without putting personnel on potentially unstable ground. Supports monitoring frequency that traditional survey methods can't sustain at comparable cost.

What this doesn't replace: Ground instrumentation, piezometers, inclinometers, and survey prisms remain essential for subsurface measurement that surface LiDAR cannot provide. Geotechnical engineering judgment in interpreting deformation data, change detection identifies movement; a qualified geotechnical professional assesses its structural significance. Pore pressure monitoring, which requires in-place sensors rather than aerial capture.

The complementary relationship is the key: Drone LiDAR and photogrammetry cover the full dam surface; ground instrumentation covers subsurface conditions at specific installed points. Together, they produce a more complete picture than either provides alone. The GISTM standard recognizes this explicitly, the requirement is for both, not one replacing the other.

Regulatory note: Tailings facility monitoring requirements vary by jurisdiction. In Canada, provincial regulators (BC Ministry of Energy and Mines, Ontario Ministry of Mines) set specific requirements for tailings storage facility monitoring programs. GISTM adoption is voluntary but increasingly expected by international lenders and investors, and several major mining companies have committed to full GISTM implementation. Operators should confirm the specific applicable standard for their facility before designing a monitoring program.

The Human Story

When the Mount Polley dam failed, Doug Watt could hear the roar from kilometres away. The wave of tailings material snapped trees in a 50-metre wide flowpath as it moved toward Quesnel Lake. Ten years later, copper is still flowing down the Quesnel River at elevated levels, according to researchers monitoring the watershed.

The people who would have benefited from a monitoring program that could detect the three-year precursor deformation at Mount Polley aren't abstract. They're the fishing communities on the Quesnel watershed whose salmon runs were disrupted. The First Nations whose traditional territory was contaminated. The researchers still studying the long-term ecological consequences of a single morning a decade ago.

And they are the workers at the 1,700+ active tailings storage facilities operating globally right now, every one of them dependent on a monitoring program that accurately characterizes what their dam is doing, not just what it's doing at the points where instruments were installed.