Why drones replaced the tape and the walking survey for stockpiles
The traditional way to measure a pile is a GNSS rover or a total station. A surveyor walks the toe and the crest, picks up a few dozen shots on the faces, and builds a surface from those points. It works, but it has two weaknesses. The surface is only as good as the points someone was willing to climb for, and walking a live pile of sand, aggregate or wood chips next to loaders is a safety problem.
A drone flight replaces a few dozen shots with millions of surface points. Every face, every bench and every tire rut on top is captured in the same pass, and nobody climbs anything. A single battery covers a typical yard in 10 to 20 minutes, which means inventory can be measured on the same day every month without stopping the loaders for long.
The result is not magic. It is a digital surface model (DSM) built by photogrammetry, and the volume is still a mathematical difference between two surfaces. Understanding that difference is what separates a reliable number from a pretty map.
How a drone stockpile volume is actually computed
Every volume tool, desktop or cloud, does the same basic calculation. The DSM is a grid of cells, each with an elevation. You draw a polygon around the toe of the pile. The software defines a base surface under the polygon, then for every cell inside it computes the height of the DSM above or below that base and multiplies by the cell area. Summing those small prisms gives the volume.
- Cells where the DSM is above the base count as fill (material present above the reference).
- Cells where the DSM is below the base count as cut (a hollow or an excavation below the reference).
- Net volume is fill minus cut. For a normal pile on flat ground, cut should be close to zero.
Because the cell size is tied to your ground sampling distance, a 2 cm GSD flight gives a DSM with cells of a few centimeters. On a pile covering 1,000 square meters that is hundreds of thousands of prisms, which is why drone volumes capture irregular shapes far better than a handful of rover shots.
Choosing the base plane: the decision that moves the number most
Most disputes about stockpile volumes are not about the drone or the camera. They are about the base. The same pile, flown once, can give noticeably different volumes depending on how the bottom is defined. There are three common methods, and Dronemapping offers all three in the browser measurement tool.
- Lowest point: the base is a flat plane at the lowest elevation found on the polygon boundary. It is conservative in one direction: if the ground slopes, the low side drags the base down and adds material that is really just ground. Use it on hard, flat pads.
- Average of perimeter: the base is a flat plane at the average elevation of the polygon boundary. It balances high and low sides and is a good default for piles on gently sloping yards.
- Custom elevation: you type the base elevation yourself, for example the design grade of a concrete pad or a value taken from an as-built survey. Use it when you know the true floor.
A common rule of thumb is to pick one method per site, write it down in the site procedure, and never change it between months without a note. A consistent method produces consistent inventory changes, even if the absolute number carries a small bias. Switching methods mid-year creates phantom gains and losses that no one can explain later.
When piles sit on uneven ground or against a wall, a flat base is a compromise. In that case the better answer is a surface flown before the material arrived, or a custom elevation based on a known pad level. The polygon should also follow the actual toe of the pile closely: a polygon drawn too wide includes ground, and a polygon drawn too tight cuts off the skirt of the pile.
Flight settings for stockpile mapping
Good volumes start with good photos. Stockpiles are harder than flat ground because faces are steep and textures repeat. Aggregate, sand and salt look the same from one square meter to the next, which makes image matching harder.
- Overlap: fly at least 75 to 80 percent front overlap and 70 to 75 percent side overlap. Uniform textures need more overlap, not less. Our article on drone mapping overlap explains why.
- Altitude and GSD: a GSD of 1.5 to 3 cm per pixel is plenty for inventory work. Use the ground sampling distance calculator to find the altitude your camera needs for that value.
- Oblique images: for tall, steep piles, add a second pass with the camera tilted 60 to 70 degrees, or a circular orbit around the largest piles. Nadir-only flights tend to smooth the faces.
- Light: fly with even light. Low sun throws long shadows into the valleys between piles, and shadowed faces match poorly. Overcast days are often ideal.
- Wind and dust: loaders raise dust and wind moves loose material. Fly before the shift starts if you can.
Keep the drone, altitude and overlap identical from month to month. The easiest way to get comparable volumes is to repeat the same mission with the same settings every time.
Ground control points and checkpoints for inventory
Can you measure stockpiles without ground control? For relative work, yes. Volume is a difference between surfaces, so a small constant vertical offset has little effect when the base plane is derived from the same surface (lowest point or average of perimeter). A scale error or a tilt, however, does affect the volume, and those are exactly what ground control fixes.
For inventory that feeds financial statements, most teams use ground control points or an RTK or PPK drone with at least a few checkpoints. A practical setup for a yard is four to six control targets around the outside of the site plus two or three independent checkpoints in the middle. Our drone GCP guide covers placement and target design in detail.
Checkpoints are the part people skip, and they are the part auditors ask about. A checkpoint is a surveyed target that is not used to fit the model. The difference between its surveyed position and its position on the map is an honest measure of accuracy. When a custom elevation base is used, vertical accuracy matters directly, because an error in the model shifts the whole pile up or down against a fixed floor.
A worked example with hypothetical numbers
Here is how the method plays out on an invented site. The numbers are illustrative only, to show the arithmetic.
Suppose a conical sand pile sits on a pad that slopes gently from west to east. The polygon around the toe covers 850 square meters. The boundary elevations range from 101.20 m on the east side to 101.80 m on the west side, with an average of 101.50 m. The DSM gives a peak at 108.40 m.
- With a lowest point base at 101.20 m, every cell is measured from the lowest corner, so the software adds the wedge of ground between the slope and the flat base. The volume comes out highest.
- With an average of perimeter base at 101.50 m, the wedge on the high side is partly offset by a small cut on the low side. The net volume is lower and closer to the material actually present.
- With a custom elevation of 101.35 m taken from the pad as-built, the result lands between the two, and it reflects the pad level recorded in the as-built.
On a pile this size, a 0.30 m difference in the base elevation across 850 square meters is about 255 cubic meters of material. That single choice can outweigh every other source of error in the job, which is why the base method belongs in the report and in the site procedure.
From cubic meters or yards to tons
Most sites sell or account for material by weight, not volume. Converting requires a bulk density, which depends on the material, its moisture and how it was placed. Loose stockpiled aggregate is less dense than the same stone compacted in place.
- Use a density measured on your own material when you have one, for example from a weighed truckload divided by its surveyed volume.
- Record the density used, and the date it was measured, in the report next to the volume.
- Apply the same density across months unless the material or supplier changes.
Because density uncertainty is often larger than the survey uncertainty, the report should present volume first and tonnage second, with the conversion factor stated in plain sight.
Common sources of error and how to avoid them
- Vegetation and water: grass on old piles and standing water on the pad produce bad surface points. Clean the polygon or move the toe line inward.
- Equipment on the pile: loaders, conveyors and parked trucks become part of the surface. Fly when the pile is clear, or exclude those areas.
- Wrong polygon: draw the toe on the orthomosaic and check it against the elevation profile, where the break in slope is obvious.
- Changing method: a new base plane, a new altitude or a new drone between months creates a change that is not real.
- Missing checks: without checkpoints there is no evidence of accuracy to show anyone who questions the number.
Monthly inventory as a repeatable process
The highest-value use of drone volumes is not one measurement but a series. A yard flown on the last working day of each month, with the same mission, the same polygons and the same base method, gives an inventory curve that finance teams can trust. Reusing polygons from the previous month speeds the work and removes a large source of variation.
Keep a short site sheet: mission name, altitude, overlap, control used, base method per pile, density per material. Anyone on the team can then repeat the job, and a reviewer can see at a glance whether anything changed in the method.