What does forest inventory data analysis involve?

Turning measurements into estimates through a chain of choices. Tree diameters, heights and counts from plots go through volume or taper equations, defect deductions, merchandising rules and volume rules, then get expanded from plots to stands, summarised with sampling statistics, and often grown forward to a later date. Each step uses equations and defaults someone chose. Change them, and the reported volume, biomass or value changes with nothing different on the ground. Good analysis records those choices with every figure.

What the usual answer says

Descriptions of inventory analysis list the steps: collect plot and tree data, calculate volume, basal area and stocking, analyse species and size classes, project growth and report. That is the sequence. It presents the calculations as settled arithmetic. They are models, and the choice of model matters.

A national change of equations

The clearest recent example comes from a national forest inventory programme. In September 2023 it replaced its regional volume equations and its method for estimating biomass with a new national system, the National Scale Volume and Biomass Estimators. The programme reports that on the same inventory data, estimated aboveground biomass rose from 30.28 to 34.71 billion tons, "an increase of 14.6% in aboveground tree biomass". Estimated carbon storage "increased 11.6%".

The change was not uniform. The programme notes "only very minor differences in total merchantable bole volume and biomass" nationally, while "some species having 15-20% increases in aboveground biomass while others decreased." Nationally, then, merchantable volume barely moved, though that says nothing about a particular species mix or region. Biomass rose by about a seventh and carbon by about a ninth, with nothing changed on the ground.

Formulas carry their own error

The same applies at the scale of a single tree. Maria Triantafyllidou and colleagues, publishing in Forests in October 2025, felled 60 trees of one species in one forest and measured them at one-metre intervals to test a quick single-section volume formula against reference volumes; we read the abstract. The formula "underestimated true volume" with a mean absolute percentage error of 21%, and "Under-estimation increased with diameter". Correction multipliers by diameter class cut the residual error to 20% or less overall and under 10% in the best-sampled class. It is a small study, but it shows the direction: the uncorrected formula understated the largest trees most.

The choices in the chain

Every step from compiled tree records to a projected stand table involves a choice of equation, rule or default, and a report that gives only the results hides those choices. For a timber business the national change above is the rare case. The common one is closer to home: a compilation program updated between cruises, a new defect rule, a different volume rule for a new buyer, a consultant who uses different defaults. Each can move a stand's figures, and none shows in the printed report unless someone wrote it down.

Why the choices must travel with the figures

Two inventories analysed with different equations cannot be compared directly. A rise in volume between cruises may be growth, or it may be a new equation set or a changed default. The same holds for carbon claims, valuations and harvest plans built on past inventories.

So record, with every estimate, the equation set and its version, the volume rule, the defect and merchandising rules and the growth model. Before comparing years, recompute the older inventory with the current choices. AI can help here by reading old cruise reports and compilation outputs into data, so they can be recomputed rather than compared as printed.

Where AI adds to the chain

AI adds steps as well as helping with them. Inventories built from LiDAR or satellite imagery use machine learning models to predict plot-level attributes across every hectare. Those models are another choice in the chain, with their own version, training plots and error, and they need recording like any volume equation.

AI also helps with the dull, error-prone parts. It can scan thousands of tree records for species codes that don't exist, heights impossible for a diameter, or plots whose totals jump between cruises, and send each to a person, which makes it practical to check every record instead of a sample.

When it doesn't apply

A single cruise for a single decision, such as pricing one sale, needs sound equations but no history of them. Businesses that have used the same compilation software and settings for years can compare within that history, though not with outside figures. And the national model change mainly moved biomass and carbon, so volume-based decisions were less affected; for them, the recording discipline matters more than that one change.

Quarri for forest management is built around how a forest operation runs, from the cruise to the settled account.

Quarri is an AI-native data platform for the timber supply chain. It connects buying, production, sales and inventory for forest management, sawmill, wood products and pulp, paper and packaging operators.