Stockpile Volumetrics by Drone: What the Numbers Are Worth on a Cochise County Site

Johnson Camp is producing copper cathode again, out near Dragoon, with a projected 25 million pounds a year over fifteen years and a hundred-plus jobs attached to it. Aggregate operations up and down the I-10 corridor move with construction, and construction in this county is not slowing down.

All of which means more people out here now have material sitting in piles that somebody wants a number for — for inventory, for a monthly reconciliation, for a reclamation bond, or because a buyer asked.

Drone volumetrics is a good answer to that. It is also routinely oversold. This is what the method actually delivers, and what it needs to deliver it.

What the flight produces

A volumetric survey is not a clever camera trick. It is photogrammetry: several hundred overlapping images flown on a grid at a fixed altitude, processed into a dense point cloud and a digital surface model. Once you have a surface, volume is arithmetic — the software integrates the difference between the pile surface and a base plane.

The deliverable that matters is not the point cloud. It is:

• A volume figure per pile, in cubic yards, with the base plane definition stated.
• An orthomosaic of the yard, so every pile is located and identifiable rather than "the one by the crusher."
• Toe outlines — the polygon each volume was computed inside, so the next survey uses the same footprint.
• A stated accuracy basis, which most reports skip.

The base plane is where the error lives

This is the single biggest source of disagreement between a drone number and a truck count, and it has nothing to do with the drone.

A pile sitting on flat, clean pad is easy — the base is the pad, and the pad is visible all the way round. A pile that has been pushed up against a wall, built on ground that slopes, or grown on top of last quarter's fines has a base surface nobody can see. The software will happily fit a plane through the visible toe and give you a confident number, and that number will be wrong by however much the ground underneath departs from the plane.

What accuracy is realistic

We publish this rather than quoting a marketing number:

Our aircraft is a non-RTK DJI Mavic 4 Pro. Absolute accuracy comes from ground control we place and survey, not from the aircraft. With 5 to 9 surveyed points on a site we deliver roughly 3–5 cm horizontal and 5–8 cm vertical checkpoint RMSE — and those are RMSE figures measured at independent checkpoints withheld from processing, not the optimistic residuals the processing software reports on its own control.

That distinction is worth knowing generally. Software-reported control residuals typically run two to three times better than what independent checkpoints show, because the control points were used to fit the solution. A report quoting residuals as accuracy is not lying, exactly, but it is not telling you what you think it is.

Material properties nobody warns you about

Photogrammetry needs texture. Three things on a Southern Arizona site work against it:

Uniform fines. Very fine, very uniform material — screened sand, some tailings — can be nearly featureless at altitude in flat light. The matcher has nothing to lock onto and the surface goes soft. Flying lower fixes it; so does flying when the sun is low enough to give the surface relief.

Bright, high-albedo material. Caliche and light aggregate in full midday sun will clip highlights and lose the same texture. Mid-morning is better than noon.

Wet or steaming material. A pile that is actively being watered, or leach material with visible vapour, will produce noise and holes in the cloud. Fly before the water truck, not after.

And the obvious one: a pile being worked is a moving target. If a loader takes six passes off it during the flight, the model is a blend of before and after. Volumetric flights happen at shift change, at first light, or with the yard held for twenty minutes.

Repeatability beats accuracy

For most operational inventory this is the point people miss. If you are reconciling month over month, what you need is not the perfect absolute number — it is the same number-making process every time.

Same flight altitude. Same overlap. Same control points, physically marked and left in place. Same toe polygons. Same base plane method. Do that and your month-to-month delta is far more accurate than either individual absolute figure, because the systematic error mostly cancels.

Which is why we push monthly or quarterly programmes over one-off surveys for anyone doing this for inventory. A one-off tells you what is there today. A programme tells you what moved, and what moved is usually the number the accounting actually turns on.

The site-specific things out here

Airspace is easy. Johnson Camp, the I-10 corridor pits and most of the valley floor sit in Class G. No LAANC, no waiting, a survey can happen the week you decide you want one. That is not true in Tucson and it is not true near Sierra Vista.

Line of sight is the real constraint. Part 107 § 107.31 requires unaided visual line of sight. A large yard with a highwall or a plant structure in the middle is not one launch point, and the quote reflects that.

Site access and MSHA. On an MSHA-regulated property we come in under your site's requirements — hazard training, escort, PPE, and whatever your safety department needs before anyone launches anything. Tell us up front; it changes scheduling more than it changes price.

Wind and heat. Late spring afternoons in this valley regularly exceed what makes a good survey flight. Early is better for wind, for light, and for the aircraft.

What it costs

Stockpile volumetrics as an add-on to a construction visit is $200 per pile. A standalone mapping call-out starts at $750 covering up to 10 acres with an orthomosaic and surface model, then $15/acre to 100 acres, $9/acre to 500, $5/acre beyond. Surveyed ground control — required for any stated accuracy figure — is +$450. Bare-pad baseline surveys are quoted the same way and are cheap relative to their value.

There is no travel charge anywhere in the Sulphur Springs Valley. Full rate card.

What this is not

It is not a survey in the legal sense. Boundary determination, ALTA/NSPS work and anything requiring a sealed document needs an Arizona Registered Land Surveyor. That is a licensing question rather than an accuracy question, and no drone changes it.

It also is not a substitute for a scale house on material you sell by weight. Volume times an assumed density is an estimate, and the density assumption will move your tonnage more than the volumetric error will. If you need tons, we can give you a defensible volume and you should apply your own measured density, not ours.

If you have piles that need a number and you want to know what is realistic on your particular site, send us the yard and we will tell you before you commit — including where the base plane problem is going to bite.