In many tropical systems the soil holds more carbon than everything growing on it. It is also the pool where measured change is most easily swamped by measurement error, which is why most forestry methodologies either exclude it or require an unusually rigorous protocol before it can be claimed.
Why the Signal-to-Noise Ratio Is Bad
Soil carbon stocks are large, and their annual change is small relative to the stock.
A soil holding 60 tC/ha that gains 0.5 tC/ha/year is changing by under 1% annually. Meanwhile, spatial variability in soil carbon across a single field routinely exceeds 20%. You are trying to detect a small change against a large, noisy background.
This is the reverse of the above-ground situation, where stock is smaller but growth is proportionally fast and visible.
Bulk Density Is Where Errors Enter
Carbon stock is concentration times bulk density times depth. Two of those three are easy to get wrong.
| Input | Typical issue | Consequence |
|---|---|---|
| Carbon concentration (%) | Lab method differences (dry combustion vs Walkley-Black) | Systematic offset between labs |
| Bulk density (g/cm³) | Compaction during coring; stones not excluded | Direct proportional error in stock |
| Depth | Inconsistent between visits | Compares different volumes of soil |
Bulk density deserves particular attention because it is the input most often measured casually and most directly proportional to the answer. A 10% bulk density error is a 10% stock error, full stop.
The Equivalent Soil Mass Problem
Here is the subtlety that catches otherwise careful projects.
If a management change alters bulk density, and tillage, compaction, and organic matter addition all do, then sampling to a fixed depth compares different masses of soil between visits. A soil that becomes less dense will appear to lose carbon at fixed depth even if it gained carbon, simply because the same 30 cm now contains less soil.
The correction is equivalent soil mass accounting: compare equal masses rather than equal depths. Methodologies working seriously with SOC require it. Projects that sample to fixed depth and report a change without this correction are reporting a density artefact alongside the carbon change, and cannot separate the two.
Depth: How Deep Is Deep Enough
Most protocols specify 0 to 30 cm because that is where change concentrates and where IPCC defaults are calibrated. Deeper carbon exists, often a great deal of it, but changes more slowly and is far more expensive to sample.
Sampling to 30 cm is defensible. Sampling to 30 cm and then implying the result represents total soil carbon is not.
When to Claim SOC and When Not To
The honest assessment: SOC is claimable when the project has a genuine reason to expect substantial change, a paired or permanent-plot design, equivalent-soil-mass handling, and a budget for enough samples to beat the variance.
Absent those, excluding SOC is the better decision and in ARR on degraded land it is usually conservative anyway, since the soil is more likely to gain than lose.
The failure mode to avoid is claiming SOC with an underpowered design, producing a confidence interval so wide that the conservative deduction eliminates the claim. That is the worst of both worlds: the cost of measurement without the benefit.
Where science meets climate action
TREEO Indonesia pairs expert consulting with dMRV technology across afforestation and reforestation, agroforestry, regenerative agriculture, and biochar, moving from project design through to verified issuance.



