Key takeaways
- Start by assuming both surveys are competent. Most gaps between two volumes of the same pile are differences in definition, not differences in skill.
- Check the base surface first. It is the single largest cause and the easiest to confirm, because a different base surface produces a difference that scales with footprint area.
- Check the toe boundary second. Where the pile is judged to end is a human decision, and on a flat-sided pile a small boundary shift moves a large volume.
- A vertical datum or geoid mismatch produces a smooth, uniform offset. That signature is distinctive and it is a frame problem rather than a measurement problem.
- Compare the surfaces, not just the numbers. A difference map tells you immediately whether the gap is spread evenly across the footprint or concentrated in one area, and those have different causes.
- Capture timing matters on a working pile. Two surveys days apart on an active site are measuring two different piles, and that is not a disagreement.
- Neither volume is a measurement until its reference is stated. A volume without a base surface, boundary and datum written down cannot be audited or reproduced.
- If both surveys hold up and the gap is within their combined uncertainty, the honest answer is that the pile is known to that precision and the decision needs a tolerance rather than a tiebreaker.
This comes up more often than anybody expects. A site gets measured by two different vendors, or by the same vendor before and after a method change, or by a vendor and the client's own crew, and the two volumes do not match. Sometimes the gap is small enough to shrug at. Sometimes it is large enough that somebody wants to know who is wrong.
The useful starting assumption is that nobody is wrong. Two competent surveys of the same pile disagree because they answered slightly different questions, and the work is to find out which part of the question differed. There is an order to check things in, and it runs from the causes that explain large gaps cheaply to the ones that explain small gaps expensively.
First, confirm it is the same pile
Before touching the surveys, check the capture dates and what happened between them. On an active site a pile can change materially in days. A loader working one face, a delivery, a shipment out, or a push to consolidate the stack all move real material, and two surveys taken a week apart on a working pile are not a disagreement about measurement. They are a correct record of two different states.
This sounds obvious and it is still the first thing to rule out, because it is free to check and it closes a meaningful share of cases outright. If the dates are close and the site was idle between them, move on.
Second, the base surface
A stockpile volume is the space between the pile surface and a base surface underneath it. The pile surface is measured. The base surface is chosen, and the choice is made by a person. The common options are:
- A surveyed pad surface captured when the pad was empty, which is the strongest option.
- A plane fitted through the toe of the pile, which is reasonable on a flat pad and wrong on a sloped or dished one.
- A triangulated surface across the toe points, which follows the pad slope but still guesses at the ground under the middle of the pile.
- A design or assumed elevation, which is a convenience rather than a measurement.
Two vendors who used different options here will produce different volumes, and the difference scales with the footprint area. On a large footprint, a base surface sitting a short distance higher or lower than the other one produces a volume difference that looks alarming and is fully explained.
This is the first thing to ask each vendor for in writing: what base surface was used, how it was derived, and will they deliver it as a file. If one of them cannot answer, that is the answer.
Third, the toe boundary
Deciding where the pile stops and the pad starts is a judgement call. On a cleanly swept pad with a sharp toe, two people will draw nearly the same line. On a pile with spilled material around the base, a feathered edge, a pile that runs into a wall or berm, or two piles that have grown into each other, they will not.
The boundary matters most where the pile is thin, which is around the whole perimeter. Material at the edge is shallow, so a boundary shift adds or removes a thin layer over a long perimeter, and on a flat-sided pile that adds up faster than people expect.
The fix is the same as for the base surface. Both vendors deliver the boundary they used as a file, and the two boundaries get overlaid. If they differ, the question stops being whose volume is right and becomes which boundary matches what the client means by the pile, which is a question the client can answer.
Fourth, the vertical frame
If the base surface and the boundary match and the gap persists, look at the datum. A vertical datum difference, a different geoid model, or a base station coordinate entered in one reference and processed in another all produce the same signature: a smooth, nearly constant vertical offset across the whole site.
That signature is diagnostic. A uniform offset is a frame problem. It is not an accuracy problem, and it is not a sign that either survey was sloppy, because a survey can be internally excellent and sit in the wrong vertical frame. It is also the easiest kind of gap to fix once identified, because the correction is a single shift rather than a reflight.
What makes it findable is a difference map rather than a pair of numbers. Two volumes cannot tell you whether the gap is uniform. Two surfaces can.
Fifth, what the surface is actually the top of
Photogrammetry returns the top of whatever reflected light. On a stockpile that usually means the material, but not always:
- Snow and ice read as material, and on a worked pile the cover is uneven so a flat correction does not hold.
- Standing water at the toe reads as a surface at water level, hiding whatever is under it.
- Vegetation on a long-dormant pile reads as volume, and the error is always in the direction of too much.
- Equipment, tarps and material stored against the pile read as part of it unless they were removed in processing.
LiDAR behaves differently from photogrammetry on some of these, particularly vegetation, which is one reason two surveys using different sensors can disagree on a pile that has been sitting. If one survey was photogrammetric and the other was LiDAR, this moves up the list.
Sixth, accuracy, and only now
By the time the first five causes are ruled out, the remaining gap is usually small. This is where the comparison becomes a real accuracy question, and the honest handling of it is to combine the two uncertainties rather than to pick a winner.
Each survey has an uncertainty, which should have been established against independent checkpoints rather than quoted from a sensor specification. Two volumes whose difference sits inside the combined uncertainty are not in conflict. They are two consistent estimates, and the pile is known to that precision and no better.
That is an unsatisfying answer when money depends on the number, and it is the correct one. The response is not to keep measuring until two numbers match. It is to decide what precision the decision requires and to specify a method that delivers it, which may mean better ground control, a lower flight, a different sensor, or accepting a tolerance in the contract.
What to ask each vendor for
None of the above is possible with only two volumes in hand. The comparison needs the inputs, and these are reasonable things to ask any vendor for:
- The capture date and time, and the sensor and method used.
- The base surface as a file, with how it was derived.
- The toe boundary as a file.
- The coordinate system, the vertical datum and the geoid model, written out rather than implied.
- The ground control or base station coordinates, with their source and datum.
- The achieved accuracy and how it was established, meaning checkpoints held out of the solution rather than a nominal figure.
- The processed surface in an open format, so the two can be differenced.
A vendor who supplies all of that is making their work auditable, which is a reasonable thing to expect on a number that settles an inventory or a payment. A vendor who supplies a volume and nothing else has delivered an opinion.
Preventing the next one
Once a disagreement is resolved, the cheap move is to write the resolution down as the site's standard. A site that has a captured pad surface on file, a defined toe boundary for each stockpile, a named coordinate system and vertical datum, and a required accuracy demonstrated on every capture will not have this argument again, including across a vendor change.
That package is also what makes a measurement series comparable over time. The same decisions that resolve a two-vendor disagreement are the ones that stop a vendor change from producing an apparent volume step that is really a change in method.