Stratification is the one sampling decision standards examine most closely, because it determines whether a plot count means anything. The requirements are consistent across standards even where the wording differs.
The Four Expectations
| Requirement | What it means in practice |
|---|---|
| Documented basis | Strata defined by stated, carbon-relevant variables — not convenience |
| Internal homogeneity | Variance within a stratum materially lower than across the whole area |
| Complete, non-overlapping coverage | Every hectare in exactly one stratum |
| Re-stratification on change | Strata revisited when disturbance or growth alters the landscape |
The fourth is the one projects most often miss. Strata drawn at validation describe the landscape as it was; a decade of growth, harvest or fire can make them meaningless, and continuing to sample against stale strata produces confident numbers about a landscape that no longer exists.
Acceptable and Unacceptable Bases
Standards do not prescribe which variables to stratify on, because the right answer depends on what drives carbon density at a given site. They do consistently reject stratification by variables with no carbon relevance — ownership parcels, administrative boundaries, or blocks drawn for field-crew convenience.
The defensible test is simple and worth applying before validation: can you state, with evidence, why carbon density differs between these strata? If not, the strata are decoration.
Post-Stratification Has Limits
Some methodologies permit post-stratification — assigning plots to strata after measurement, using remote-sensing data to define boundaries retrospectively.
This is legitimate when the plot locations were themselves selected probabilistically. It is not a remedy for a sample that was located by convenience: reorganising a biased sample into strata does not remove the bias, it distributes it.
Turn uncertainty into a plan
Sampling design and growth assumptions decide how much measured carbon survives conservative deduction. TREEO Carbon Simulator models sequestration scenarios and biomass growth before the field season.
