Most growth and yield models in tropical forestry were built to answer a timber question: how much merchantable volume will this stand produce, and when should it be cut? Carbon projects borrow them because nothing better exists. The borrowing works, up to a point, and the point is worth knowing before a projection is built on one.
What the Models Are Good At
| Strength | Why |
|---|---|
| Even-aged, single-species stands | What they were calibrated on |
| Merchantable volume | The quantity they were built to predict |
| Thinning and rotation scenarios | Management levers they were designed around |
| Relative comparison | Which regime yields more, even if absolute values drift |
Within that envelope they are genuinely useful, and for a plantation-style ARR project on a commercial species with local yield tables, they answer the growth question well.
Where the Transfer to Carbon Breaks
Merchantable volume is not biomass. Yield models predict the part of the tree a mill would buy, which is the stem to a top diameter. Carbon accounting needs the whole tree, including branches, foliage, roots, and the non merchantable stem.
Bridging the gap requires expansion factors, and each one adds uncertainty that the yield model's own precision statistics do not include. A model quoted as accurate to a few percent on volume is not accurate to a few percent on biomass.
Mixed-species systems. Models calibrated on monocultures struggle with the species mixtures typical of restoration plantings and agroforestry, where competition dynamics differ and no single species trajectory describes the stand.
The management assumption. Yield models usually assume a management regime, such as spacing, thinning, and tending, that the calibration stands received. A restoration planting with minimal tending after establishment will not follow a curve derived from a managed plantation. This is the assumption most often carried across unnoticed.
The Simplest Honest Check
Compare the model against your own plots at every monitoring event.
If measured stock tracks the model, the transfer is working. If it comes in consistently below, which is the common direction for under tended restoration, the model was calibrated on better managed stands and the projection needs adjusting, not just refitting.
Projects that do this catch the divergence within two monitoring events. Projects that refit each time without comparing never learn that the model was wrong, only that it keeps needing adjustment.
What to Use When Nothing Fits
For mixed restoration plantings in Indonesian conditions, often no calibrated model exists. Realistic options, in order of preference:
Build a local growth curve from your own permanent plots. It is slow, given that it requires several monitoring events, but it is the only approach that is actually about your stand.
Use a model from the most similar available system and validate against local plots, documenting the transfer as a known uncertainty.
Use conservative published increment rates. Crude, defensible, and it will understate a well-performing project.
The third option is unpopular and it is the honest choice for a young project with two monitoring events and no local curve, with the alternative being a confident projection with nothing underneath it.
The Framing That Helps
A growth model is a hypothesis about the future of a stand, tested at every monitoring event. Treating it that way, with a projection, followed by a check, and then a revision, produces a model that improves. Treating it as a settled input produces a number that drifts away from the forest without anyone noticing.
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.
