Surveying a Containment Pond, Cell or Berm by Drone

A containment facility is one of the few assets where the question the owner has is genuinely a geometry question. How much room is left before it overtops. Whether the berm is still the shape it was built as. How much material or water is actually in there against what the cell was designed to hold.

Those are measurable from the air, with one important exception that most of this article is about. Here is what a drone survey of a water or tailings containment facility establishes, what it cannot, and how to set it up so the result stands as a record rather than an estimate.

Freeboard is a distance to the lowest point, not to the nearest berm

Freeboard is the vertical distance between the stored liquid surface and the lowest point at which the facility would begin to discharge. That controlling elevation is sometimes the crest of the containing berm, sometimes a spillway or overflow invert, sometimes a low point in an embankment several hundred metres from where anyone stands to read the gauge.

This is where a staff gauge quietly misleads. A gauge reads water level at one location against a mark installed at one time. It says nothing about whether the structure has settled somewhere else, and it is referenced to the elevation someone recorded when it was installed. A drone survey produces a continuous surface across the whole crest, so the lowest controlling elevation is found rather than assumed, and the freeboard figure is a real minimum instead of a local one.

Two things have to be right for that number to mean anything. The survey has to be tied to the same vertical datum the facility's approval and as-built drawings use, which for Alberta work is a question worth settling before the flight and is covered in which coordinate system your drone survey should be in. And the water surface has to be captured in a way photogrammetry can actually handle, which is the next problem.

What the water surface does to photogrammetry

Water is the hardest surface in aerial survey. Photogrammetry works by matching the same point across overlapping images, and a water surface has no fixed points to match: it reflects the sky, it moves between frames, and glare changes with sun angle and camera position. The result in a raw photogrammetric surface over open water is usually noise, holes, or a plausible-looking surface that is some distance from reality.

The practical answers are not exotic:

• Measure the shoreline instead of the surface. The waterline where the liquid meets the bank is a sharp, matchable feature. Its elevation is the water surface elevation, and it can be extracted around the whole perimeter rather than at one gauge.
• Fly conditions that suit it. Calm water, overcast or high sun rather than low-angle glare, and a flight plan with enough overlap to keep the bank well reconstructed.
• Use LiDAR where it is available and the surface matters. A LiDAR sensor returns from the water surface more reliably than image matching does, and gives a cleaner result over a large pond. Which method suits which job is the subject of LiDAR versus photogrammetry and our LiDAR survey service.
• Tie to a gauge reading taken at the time of flight. Simple, cheap, and it gives an independent check on the aerial water surface elevation.

What none of these do is see through the water. A standard drone sensor, LiDAR included, does not penetrate turbid liquid, and tailings and process water are not clear.

Stored volume against design capacity

This is the request owners most often make and the one that needs the most care. A volume is the space between two surfaces. A drone gives you the top one. The bottom one has to come from somewhere else:

• The as-built basin survey from before the cell was filled, if it exists and if you trust it. This is the usual answer for an engineered cell.
• A bathymetric survey, by boat-mounted sonar or equivalent, where the bottom has changed or was never surveyed. Sediment accumulation, settled solids and deposition all move the floor away from the design.
• The design drawings, which tell you what the basin was supposed to be, not what it is. Acceptable for a rough capacity check, not for a stored-volume number anyone will rely on.

Say which of those three you used, because the stored volume inherits all of its uncertainty from the floor rather than from the aerial work. The same base-surface problem drives every stockpile measurement too, and is explained in base surface. Where a drone survey is unambiguously strong is remaining capacity to the crest: the space between today's liquid surface and the structure above it is bounded by two surfaces the drone actually measured.

Deposited solids above the waterline are straightforward volumetric work, and behave like any other stockpile measurement.

Berm crest, slope and settlement

Containment geometry is where aerial capture needs no caveats, and it is often more useful than the water number.

• Crest elevation along the entire length. Not spot elevations at survey monuments but a continuous profile, which is how a settled low point gets found before it becomes the controlling elevation nobody knew about.
• Slope angles, inside and out. Extracted from the surface and compared against the design section at any chainage, which catches oversteepening from erosion or from material pushed over the edge.
• Crest width and alignment against the as-built, including encroachment from traffic, stockpiling on the crest or vehicle rutting.
• Change between captures. Two surveys to the same control, differenced, show settlement, bulging or loss of section as a surface rather than as a set of points. This is the single most valuable product of a repeating program.
• Visual condition in the imagery. Seepage scarring, vegetation changes that follow a wet line, animal burrowing, erosion gullies and rutting are all visible in a high-resolution ortho and are all things somebody has to walk the berm to find otherwise.

Vegetation is the one limit worth naming here. A photogrammetric surface sits on top of grass and brush rather than on the ground, which on a vegetated berm biases crest elevations upward, exactly the wrong direction for a freeboard calculation. Cut the grass, fly before growth or in the autumn, or use LiDAR, which gets returns through light cover. The general version of this problem is in what a drone survey cannot tell you.

Control, accuracy and why this one needs ground control

A freeboard figure is a vertical measurement, and vertical accuracy is the weakest axis in drone photogrammetry. For a containment survey that is going into a file somebody may be held to, that means ground control rather than an unchecked RTK capture.

In practice: permanent or semi-permanent control points around the facility, surveyed once and reused on every flight so that successive captures sit on the same reference; independent check points that are not used in processing, so the achieved accuracy is measured rather than claimed; and the vertical datum matched to the approval and the as-built drawings. How many points and where they go is covered in how many ground control points you actually need, and the positioning methods behind the capture in RTK versus PPK.

Reusing the same control across every visit is what turns a series of surveys into a monitoring record. Without it, each capture is internally consistent and the differences between them are not trustworthy.

Cadence, and what makes it a record

How often to fly follows from what is changing and what the facility's approval requires. A commonly workable pattern on an Alberta site is a capture after spring runoff when levels are at their seasonal high and the berms have been through freeze and thaw, one in late summer when vegetation and access are at their most cooperative, and one before freeze-up. Active deposition, a construction stage or a facility under heightened scrutiny justifies more.

What makes the series a record rather than a folder of surveys is consistency: the same control, the same method, the same deliverable set, the same reporting format, with the conditions at capture noted each time. Add a water level reading at the time of flight and a note of recent precipitation, and the comparison between any two visits means something. The general argument for a fixed cadence is in how often should you re-fly a site, and what makes any inspection deliverable defensible is in how to tell a good drone inspection report from a bad one.

The regulatory side, stated carefully

Containment facilities in Alberta sit under approvals and directives that depend on what the facility holds and who operates it. Water and wastewater containment, oilfield facilities, mine tailings and agricultural manure storage are governed by different instruments, administered by different bodies, and carry different required freeboards, inspection frequencies and reporting triggers.

No article should tell you what your number is. The required freeboard, the inspection frequency and the circumstances that trigger a report are in your own approval, your operating licence and the directive your facility falls under, and if there is a professional of record for the structure, their requirements govern the survey too. What a drone survey provides is the measurement against that standard, in a form that is dated, repeatable and traceable to control.

One consequence worth stating: if the survey is supporting a submission or a regulated inspection, confirm before the flight whether the deliverable needs sign-off by a professional, because that changes who has to be involved and when. The surveyor sign-off question is covered in does a drone survey need a professional land surveyor's sign-off.

What this does not do

Plainly, so it is not discovered later. Aerial capture does not see the bottom of a liquid-filled cell, does not assess the internal condition or integrity of an embankment, does not detect seepage except where it has produced a visible surface effect, and does not replace a geotechnical assessment, a piezometer network or an engineer's inspection. It measures external geometry extremely well and it does so without putting anyone on a slope.

Used alongside those other things rather than instead of them, it is usually the cheapest way to turn an annual walkover into a dated, measurable record of the whole structure. If you want to scope one against your own facility and its approval, get in touch with the approval conditions to hand and we will work out what the survey has to deliver.