C-DRONE GUIDE · 1 SEPTEMBER 2026

Measuring a Bulk Stockpile by Drone for the Annual Inventory: Year-End Accounts, Auditors, Price

In a quarry, a port, a composting platform or a scrapyard, the year-end stock is often the heaviest line in current assets — and the worst justified. Nobody counts a pile of aggregate: they estimate it. The site manager says "around 12,000 tonnes", the accountant books it, the external accountant accepts it for want of anything better, and the statutory auditor asks what that figure rests on. That is precisely where drone measurement changes nature: it stops being an operations tool and becomes dated, traceable, reproducible audit evidence. This guide is not about the technique of volume computation — we cover that elsewhere — but about what the figure becomes once it enters the balance sheet: the French Commercial Code's inventory obligation, what an auditor expects, the fragile chain running from volume to tonnage to value, the tolerance to write into a protocol, and repeatability from one financial year to the next.

Published on 1 September 2026, reviewed on 11 September 2026 — regulations in force as of September 2026.

A pile is not a pallet: why bulk resists inventory counting

The obligation is simple and long-standing. Article L. 123-12 of the French Commercial Code requires every trader, individual or corporate, to "check by inventory, at least once every twelve months, the existence and value of the assets and liabilities of the undertaking", then to draw up annual accounts at year-end "on the basis of the accounting records and of the inventory". Article R. 123-177, as worded since decree no. 2015-903 of 23 July 2015, states that the inventory is "the annual check on the existence and value of all assets and liabilities at the closing date", and that inventory data must be retained and "organised so as to justify the content and the valuation method of each balance-sheet item". The same decree removed the formal obligation to keep an inventory book: this is not a relaxation of substance, it shifts the burden of proof onto the quality of the supporting file.

For warehouse stock, that justification is mechanical: you count pallets, scan barcodes, reconcile unit quantities. That is the subject of our guide to warehouse stock inventory by drone, where the drone replaces the forklift and the platform to read labels at height. Bulk does not work like that. A pile of aggregate, compost, road salt, grain, wood chips or scrap metal cannot be counted: it has no unit, no identifier, no container. It can only be measured — and, historically, it was not really measured at all.

Common practice remains expert estimation: the quarry manager looks at the pile, compares it with what he knows, cross-checks against tonnages produced and dispatched since the last count, and states a figure. That estimate is not absurd — it embeds site knowledge nobody else has — but it is unverifiable and non-reproducible. Two site managers will give two figures. The same manager, faced with two piles of similar shape but different height, will err one way on the first and the other way on the second, with nothing in the file able to detect it. And errors do not cancel out: on a conical pile, underestimating the height by one metre in ten produces a volume error well above 10%, since volume varies with the square of the radius and with height.

The financial stake follows directly. A change in closing stock feeds straight through the income statement via the change in inventories, and therefore into taxable profit and equity. On a site holding 50,000 tonnes of aggregate valued at a few euros per tonne, a 10% misjudgement shifts tens of thousands of euros of profit from one year to the next. That is exactly the kind of gap an auditor seeks to bound — and that a figure produced by eye cannot bound.

What a statutory auditor expects: NEP 501 and attendance at the physical count

The reference text on the French statutory audit side is NEP 501, "Evidential value of items collected (specific applications)", approved by the order of 22 December 2006. It provides that, where inventories are material, the statutory auditor attends the physical stock count in order to collect sufficient and appropriate evidence on their existence and their physical condition. Their attendance has a second purpose, often overlooked by companies: to verify that the procedures defined by management for recording and checking count results are actually applied, and to assess their reliability. In other words, the auditor does not merely come to look at a pile: they come to audit your method.

This is decisive for anyone considering drone measurement. If the company has no written procedure, the contractor's flight remains an isolated technical service, hard to tie into the audit file. If the company has formalised a bulk-stock inventory protocol — who measures, when, with what equipment, over what extent, with what checks, who signs off — the survey becomes the documented execution of a procedure the auditor can observe, test and, where appropriate, rely on in later years. NEP 501 also provides that, where inventories are spread across several sites, the auditor determines the locations where attendance is necessary according to the risk of material misstatement per site: a uniform protocol applied across all a group's sites multiplies the reach of a single visit.

The standard also provides two fallbacks directly relevant here. Where the auditor cannot attend on the planned date but the entity operates a perpetual inventory, they may attend on another date, performing or observing counts, with tests on the intervening movements. And where attendance is impossible, notably because of the nature or location of the inventory, they determine whether alternative audit procedures can provide evidence of equivalent value. A dated, time-stamped drone survey accompanied by a processing report and ground-surveyed check points fits naturally into both configurations — which is also what makes it useful on sites where walking on a stockpile is simply forbidden for safety reasons.

One clarification avoids a common confusion: this evidence belongs to the audit world, not to judicial findings. If the issue is not the certification of accounts but litigation, an insurance assessment or a commercial dispute over a delivered quantity, the evidentiary vehicle changes and falls under a drone survey with a judicial officer, which we cover separately. Both approaches share the same traceability requirement, but not the same recipient or formalism.

Volume, tonnage, value: density decides how reliable the result is

This has to be said plainly, because the sector's marketing systematically leaves it out: a drone measures a volume, and nothing else. Yet stock does not go onto the balance sheet in cubic metres. It goes on in tonnes, valued at a cost price. The full chain therefore has three links — volume, density, unit cost — and the link the drone handles is the least problematic one.

Conversion runs through the material's bulk density in the pile, in tonnes per cubic metre. It is anything but a constant. It depends on grain size and shape, on compaction — a pile dropped from a conveyor, a pile rehandled by loader and a pile compacted by machine traffic do not have the same density —, on bulking, and above all on moisture. The American test standard ASTM C29/C29M, "Standard Test Method for Bulk Density (Unit Weight) and Voids in Aggregate", is explicit about these limits and is worth quoting to anyone offering you a density figure taken from a table: it warns that the relationship between the degree of compaction of aggregate in a hauling unit or stockpile and that achieved in the standard test is unknown, and that aggregates in stockpiles usually contain absorbed and surface moisture — the latter being precisely what causes bulking — whereas the test determines bulk density on a dry basis.

The practical consequences are large and rarely quantified in inventory files. Dry sand and saturated sand, at equal volume, do not weigh the same; a wood-chip stockpile measured after three weeks of rain has gained mass without gaining dry matter; a fermenting compost pile settles on its own. On a 20,000 m³ pile, a 0.1 t/m³ hesitation over the density used — hardly extravagant between a catalogue value and a measured one — shifts 2,000 tonnes. No improvement in survey precision, however spectacular, offsets that: you can measure the volume to within 1% and be 10% out on tonnage.

The operational conclusion is clear: density must be measured, not assumed, and its measurement must be dated and documented just like the survey. The usual routes are a laboratory test on a sample taken at the time of the flight, or — often more convincing to an auditor because it bears on the actual material — weighing a reference batch: a known volume taken from the pile on the day of measurement is loaded and weighed on a weighbridge. That value becomes the year's conversion assumption, defensible and repeatable. On heterogeneous, moisture-sensitive materials — wood fuel, compost, scrap — it is legitimate to use a density per product type rather than a site-wide density; our guide to log pile volume measurement by drone describes the same logic applied to a stacking coefficient calibrated per species.

Finally, measuring volume says nothing about the quality of the stored product, and therefore nothing about its unit value. Aggregate contaminated with fines, downgraded grain, scrap whose composition has changed all keep their volume and lose their value. Any write-down remains an operator decision, supported by analyses and market knowledge: the survey never substitutes for it.

Traceability of the computation: extent, pile base, reference DTM, check points

A volume never exists in the absolute: it is always the difference between two surfaces. That is why the most important question in a volume report is not "how much?" but "relative to what?". The digital terrain model that serves as the floor — the reference surface — can be defined in three ways, which do not give the same answer: a horizontal plane at a fixed elevation, a surface interpolated from the toe-of-pile outline (the "pile base method"), or a DTM surveyed earlier, when the pad was empty. On sloping ground, on an area cut or filled between two financial years, or on a pile backed against a bund, the gap between those three definitions easily runs into hundreds of cubic metres. A report that does not state which was used is not auditable.

The second parameter is the extent used: the exact outline within which the volume is computed. On a continuous-batter stockpile, on adjoining piles touching at the base, or on stock partly under a canopy or against a dividing wall, drawing the polygon is an operator decision, and that decision must be reproduced identically from year to year. Good practice is to freeze the extent polygons in a versioned reference file annexed to the protocol, rather than redrawing them each year: that is what makes the year-on-year comparison honest.

The third parameter is the computation software itself, and research documents this. The study by G. Tucci, A. Gebbia, A. Conti, L. Fiorini and C. Lubello published in 2019 in Remote Sensing measured stockpiles of separately collected waste material by drone photogrammetry, comparing two imaging configurations — vertical and oblique — with a terrestrial laser scanner used both to survey the control points and as ground truth, then recomputed the volumes with two different software packages and compared the results across configurations and across packages (see the study on Google Scholar). The lesson for an inventory file is direct: the processing chain is part of the method, and changing software or flight configuration between two years introduces variation that has nothing to do with the real stock.

That leaves accuracy proper, which is checked exactly as on any topographic survey: through independent check points, excluded from the adjustment and surveyed on the ground by a method more accurate than the product being tested. Our guide to quality control of a photogrammetric deliverable covers how to read this — and why the "RMS" printed on page one of a processing report proves nothing on its own. Georeferencing matters too: on a site with no permanent network of markers, an RTK or PPK drone removes part of the ground-target logistics but never removes the need for a few independent check points. Finally, it is the deliverables described in our guide to photogrammetry mission deliverables — point cloud, DTM, DSM, orthophoto — that must be handed over and archived with the report: without them no later recomputation is possible, and article R. 123-177 requires inventory data to be retained and organised so as to justify the valuation method.

Writing the protocol: tolerance, materiality threshold, handling discrepancies

A bulk-stock inventory protocol runs to three or four pages and is worth far more in an audit file than an isolated volume report. It sets out: the list of stocks concerned and their reference extent polygons; the measurement method (photogrammetry or LiDAR, flight height and target ground resolution, overlap, oblique imagery, georeferencing mode); the reference surface used, pile by pile; the control arrangements (number and distribution of independent check points, ground survey method); the conversion rule from volume to tonnage, with the origin and date of the density used; the accepted tolerance; and finally the procedure for handling discrepancies and the signatories.

On tolerance, two symmetrical temptations must be resisted. The first is to write nothing, which sends any later discussion back to subjective judgement. The second is to write an unrealistic figure for reassurance. Documented orders of magnitude give an honest frame: a study by H. He, T. Chen, H. Zeng and S. Huang published in 2019 in Sensors, on estimating the volume of bulk loads carried on barges by drone photogrammetry without ground control points, reports a deviation of about ±2% against traditional manual measurement, with vertical positioning errors below 8 cm (see the study on Google Scholar). For their part, P. L. Raeva, S. L. Filipova and D. G. Filipov, in a comparison of GNSS measurements and drone photogrammetry on an open-pit quarry stockpile published in 2016 in the International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, note that mining regulations commonly require an accuracy of 3% on the total volume, and stress that ground GNSS, although very accurate, quickly becomes unworkable in field time over large areas (see the study on Google Scholar).

These figures concern the geometric volume and nothing else. The protocol must therefore carry two distinct tolerances: one on volume, verifiable through independent check points or by cross-checking with a second method; the other on tonnage, wider, absorbing the density uncertainty. Year-end reconciliation then compares measured tonnage with the theoretical tonnage rebuilt from materials accounting (opening stock + production or receipts − dispatches). The resulting gap is never nil, and that is normal: it aggregates handling losses, fines blown away, moisture uptake, weighbridge errors and measurement uncertainty itself. What matters is to position it against the materiality threshold set for the year, and to document it: below it, adjust the book stock to the measurement and explain why; above it, investigate before adjusting, because a large gap often signals something other than survey imprecision — a badly chosen density, a modified extent, or a genuine materials-accounting problem.

This framework connects with neighbouring reporting obligations where the site is a classified installation. Quarry operators file an annual production declaration whose measurement and traceability logic is very close; our guide to 3D quarry modelling for the annual production declaration covers that regulatory strand, which it is economically rational to handle in the same flight as the accounting inventory.

Repeatability: comparing years, not methods

The value of a drone inventory does not appear in the first year: it appears in the second. The first campaign produces a figure; the second produces a measured change, and it is that change which feeds the accounts, management analysis and the auditor's confidence. But the observed difference must reflect the stock, not a change of protocol.

The rule is therefore easy to state and demanding to keep: same extent, same reference surface, same processing chain, same conversion, same contractor where possible. Every altered parameter must be flagged and, ideally, quantified: if you move from a horizontal floor to an interpolated base, or from one software package to another, good practice is to recompute the previous year's stock once with the new method, publish both values and document the methodological gap. It costs little, it is done from archived data, and it prevents a purely technical variation from being read as a loss or gain of material.

This continuity has a contractual corollary: the raw data belong to the client. A contract providing only for a volume PDF locks the company into dependence on the contractor and makes any recomputation impossible. Insist on delivery and archiving of the georeferenced imagery, the point cloud, the DTM, the coordinates of control and check points, and the processing report — which is also what the obligation to retain inventory data implies. The general principles for commissioning a survey, from choosing accuracy to choosing deliverables, are set out in our guide to drone surveying: the price of a topographic survey.

Finally, frequency is better framed economically than by regulation. The Commercial Code requires only one inventory per financial year. But on a high-turnover site a quarterly measurement costs little compared with the first campaign — the flight plan and extent polygons already exist — and turns the inventory from an imposed duty into a management tool: early detection of materials-accounting drift, informed decisions on stock levels, calm preparation for year-end. It is the same logic described, from the operations side, in our guide to stockpile and volume computation by drone in quarries and civil works.

Honest limits: what the drone does not measure

Three limits deserve to be stated in black and white in a protocol, because they condition the credibility of the whole exercise.

The first is geometric. The drone only sees what is visible from above. Stock under a canopy, in a silo, in a covered bay or inside a closed building escapes classic aerial photogrammetry; it calls for scanner surveying, indoor flight or gauging methods specific to the container. Likewise, a pile whose toe is hidden by machinery, bulk bags or vegetation will have a poorly reconstructed outline, and manually redrawing the base polygon becomes a source of variance that must be owned and documented. Finally, a very dark, highly reflective or very uniform surface — fresh coal, salt, dry fine sand under low sun — degrades photogrammetric matching: on such materials, airborne LiDAR or acquisition in diffuse light gives a better result.

The second is physical: the drone measures an apparent volume, not a mass. We developed this above; the consequence simply has to be drawn in the allocation of responsibilities. The drone contractor is responsible for the volume, its method and its traceability. Density, quality, product grading and valuation are the operator's responsibility, supported where necessary by a laboratory. A report that states a tonnage directly without setting out the density used and its origin mixes two responsibilities and weakens both.

The third is thermal and safety-related: some bulk stocks self-heat — coal, wood chips, compost, certain wastes — and that self-heating is monitored separately from the inventory, with a thermal camera. Our guides on coal and ore stockpiles at port terminals and on wood fuel storage cover that strand. It does not replace volume computation, but it combines very well with it in a single visit: same flight plan, two sensors, two reports.

Method and 2026 prices

The sequence of an inventory campaign is stable: scoping with the accounting department and, where possible, with the statutory auditor (list of stocks, target date, tolerances, expected deliverables); drafting or updating the protocol and freezing the extent polygons; placing and ground-surveying the control and check points; flying, avoiding days of rain, strong wind or heavy handling on site; sampling or weighing for density, on the same day; processing, computing volumes pile by pile and issuing the control report; conversion to tonnage, reconciliation with materials accounting and a variance note. Allow half a day to a full day on site for a standard yard, and three to ten working days for the complete deliverables.

Orders of magnitude observed in France in 2026 (excl. VAT):

• Annual inventory campaign, site up to 5 ha (around ten piles; flight, check points, processing, volume and control report): €900 to €1,800.
• Large site (quarry, port terminal, recycling platform, 10 to 30 ha): €1,800 to €3,500 depending on area and number of stockpiles.
• Control and check points surveyed by GNSS + acceptance note (reinforced geodetic tie, independent check): €400 to €900.
• Drafting the inventory protocol (first year only, reusable thereafter): €400 to €900.
• Quarterly monitoring visit (same flight plan, same extents): from €500 per visit, on a degressive rate against the initial campaign.
• Combined volume + thermography campaign (stocks prone to self-heating): €1,500 to €3,000.
• Density determination (laboratory test or weighbridge weighing of a reference batch): outside the drone service, to be quoted by a laboratory or organised in house; it is nonetheless the best-value spend in the whole file.

Our guide to how much a drone service costs details the parameters that make these amounts vary. This service falls under our drone surveying and photogrammetry offering; request a quote stating the nature of the materials, the number and approximate footprint of the piles, your financial year-end date and the name of your statutory auditor if they are to be involved.

Frequently asked questions

Request a free quote

Also worth reading

• Olive orchards and drones: water stress and yield forecasting, price
• Prune d'Agen and drones: water stress and ripeness, price
• Asparagus and drones: spotting weeds on the ridge before emergence, price