Why the Drone Stockpile Volume and the Scale House Don't Match

It happens on almost every first drone inventory. The survey comes back, someone lays it next to the running total from the scale house, and the two numbers are not the same. Sometimes not close. The immediate reaction is that the drone must be wrong, because the scale is a scale and it has been there for years.

Usually neither number is wrong. They are measuring different things, through a conversion that has several soft assumptions in it, at two different moments in time. Understanding where the gap comes from is more useful than deciding which number to believe, because the gap itself is often the most valuable output of the first survey.

You are comparing a volume to a mass

This is the whole problem in one line. A photogrammetric or LiDAR survey produces a volume: the space between a measured top surface and a defined base surface. The scale house produces a mass: tonnes over a weighbridge.

Volume and mass are only related through density. So every reconciliation contains a step that looks like arithmetic and is actually an estimate, and that estimate carries more uncertainty than anything else in the chain. The survey can be excellent and the comparison can still be off by a wide margin, entirely because of the number in the middle.

Density is where most reconciliations break

The density figure used in a stockpile conversion is rarely measured on the day. It is usually a standing figure, inherited from a spec sheet, a lab result on a sample taken at some point, or simply what the site has always used. Several things move it.

• Loose versus in place. Material in a stockpile is not at the density it was in the bank, and not at the density it reaches compacted in a road base. Using an in situ or compacted figure against a loose pile volume produces a gap in a predictable direction.
• Moisture. Water adds mass without adding much volume. A pile after a wet week and the same pile after a dry one weigh differently across the scale while measuring near enough the same on a survey. In an Alberta spring this is not a rounding effect.
• Segregation and fines. Material placed by conveyor segregates, with coarse rolling to the toe and fines concentrating under the drop point. The average density of the pile is not the average density of a sample taken from where it is easy to reach.
• Self compaction. A pile that has sat through a season is denser at the bottom than one built last week, and traffic on and around a pad compacts the lower material further.
• Product variation. A washed product, a crushed product and a screened product from the same pit do not share a density, and piles that have been topped up from different runs are not uniform.

None of this means density figures are useless. It means that if the conversion factor has not been revisited in a few years, it deserves attention before the survey does.

The base surface question

This is the second big source of disagreement, and the one that produces the largest single errors.

A volume is never absolute. It is always the space between the measured top surface and a base surface someone chose. Change the base and you change the volume, sometimes dramatically, without anything about the flight changing at all.

The realistic options are a flat plane at a fixed elevation, a design surface from the site plan, a prior survey of the pad before material was placed, or a base interpolated from the visible toe of the pile at the time of the flight. They are all legitimate and they all give different answers.

The common failure modes:

• A stale base. A pad surveyed three years ago that has since settled, rutted, or accumulated a layer of compacted fines that nobody considers stock anymore. Everything measured against it inherits that error.
• Toe interpolation on a pile against a wall or another pile. Where the toe is not visible all the way around, the base under the pile is inferred. That inference is an assumption and it should be stated in the report.
• Material that is genuinely buried. On long lived pads there is often a layer at the bottom that has not moved in years and functionally is not inventory. Whether it is in or out of the number is a decision, and it needs to be the same decision every time the pile is surveyed.
• A base that changed between surveys. If year one used a design surface and year two used an as-flown toe, the year over year movement is partly an artefact of the method.

This is why the base surface belongs in writing in the deliverable rather than in the processor's head. Our stockpile measurement buyer's guide covers how to specify it, and the stockpile volume estimator is useful for a sanity check on order of magnitude before anyone flies.

The book number is not a measurement

The scale house total is often treated as the reference because it comes from an instrument. But the number being compared is usually not a weighing. It is a running balance: opening inventory, plus production, minus sales, carried forward since the last time anybody physically checked.

That balance accumulates every one of the following:

• Material that left the site without crossing the scale. Internal use, road maintenance, a load for the shop yard, a neighbouring job.
• Product moved between piles and recorded against the wrong one, or not recorded at all.
• Wash losses and screening rejects that leave the product without leaving the site.
• Contamination, freezing and material written down but never physically removed.
• Rounding on every ticket, in the same direction, over a long period.
• An opening inventory that was itself an estimate, inherited from the last reconciliation.

A book total carries all of that since the last true count. A drone survey carries only the day it was flown. When the two disagree, the interesting question is not which is right but how long the book number has been running unchecked, because the drift is cumulative and the survey is not.

The two numbers have to be for the same moment

Obvious, routinely missed. A pile flown Thursday morning and compared against a month end book total from the following Tuesday differs by everything that happened in between, on a working site where that can be a lot.

The fix is to fix a date. Fly on the cutoff, or reconcile the survey back to the cutoff using the tickets in between. Flying before shift start or during a break also removes the problem of a pile being drawn from while it is measured.

What genuinely is on the survey side

Not everything is a density or a bookkeeping problem. Aerial capture has its own real limits, and an honest operator will name them rather than presenting a single number with no context.

• Obstructions. A conveyor over the pile, a loader parked against the face, a bin or a truck on the pad. All of it is captured as part of the surface unless it is removed or edited, and editing is an interpolation.
• Snow and standing water. Both change the measured surface. On a winter inventory this is a planning decision, not something to fix in processing. See surveying stockpiles and sites under snow.
• Steep faces and undercuts. A pile cut back by a loader can have a face steeper than its natural angle of repose, and an overhang is not visible from above. Photogrammetry sees what the camera sees, and oblique passes exist to reduce this rather than eliminate it.
• Ground resolution. Flying higher covers more ground and resolves less. On a large pad that tradeoff is a real one, and the flight height should have been chosen from the accuracy the number needs.
• Method. LiDAR and photogrammetry behave differently on dark, wet, uniform or vegetated material. LiDAR versus photogrammetry for stockpile measurement goes through where each is stronger.

What a good deliverable does is state the base surface used, the obstructions removed and how, the conditions on the day, and the accuracy the method supports. That is what makes a number defensible when it disagrees with the book.

How to reconcile properly

A reconciliation that produces an answer rather than an argument looks like this.

• Fix the date. One cutoff for both numbers, with tickets in between applied to whichever number needs moving.
• Write down the base surface. Which surface, from when, and how the toe was handled where it was not visible. This is the single most useful line in the report.
• Check the density figure before anything else. When was it last measured, on which product, at what moisture, and was it a loose or in place figure. If nobody knows, that is the finding.
• Reconcile per pile, not in total. A total can net two large opposite errors into a small one and hide both. Per pile differences point at causes.
• Account for the untracked movements. Internal use, road maintenance, wash losses, transfers between piles. Sites that start recording these usually find the next reconciliation is much closer.
• Adjust the book, then re-survey on a cycle. The first survey resets the balance. The value is in the second and third, because a consistent method flown the same way each time makes the movement between them trustworthy even where the absolute number is debated.

Which number to use for what

For a physical inventory at a period end, a survey against a documented base is normally the stronger evidence, because it is a direct measurement of what is actually on the ground rather than a balance that has been running for months. For day to day operations, the scale remains the right tool, since every load is weighed and nobody is flying daily.

For royalty, audit or reporting purposes, what matters more than either is that the method is written down, repeatable, and the same next quarter. An auditor is rarely asking whether the number is perfect. They are asking how it was arrived at and whether it was arrived at the same way last time. That test is easier to pass with a documented survey than with a running balance.

If you want to see how the two methods compare on a real site, our stockpile drone versus ground survey case study lays out time, cost and accuracy, and where a drone survey fits in a project budget covers how the cost is usually justified. You can see the service itself on our aggregate and mining page.