A forest carbon project sequesters CO₂ as trees grow, following logistic (Chapman–Richards style) biomass growth toward a site carrying capacity K:
B(t) = K · (1 − e^(−k·t))
ΔCO2(t) = 3.67 · (B(t) − B(t−1)) [CO2 mass ≈ 3.67 × carbon mass, from the 44.01/12.01 molar-mass ratio]
Each year a fraction r (buffer rate) of newly sequestered CO2 is withheld into a shared, non-tradable buffer pool instead of being issued as a sellable credit:
issued(t) = ΔCO2(t) · (1 − r)
buffered(t) = ΔCO2(t) · r
pool += buffered(t)
Every year there is a random chance p of a reversal event (fire, drought, pest outbreak) that instantly releases a share of the live biomass back to the atmosphere. The released CO2 must be repaid from the buffer pool to keep already-sold credits valid:
loss(t) = f · B(t) · 3.67 (f ~ 15–45% of biomass)
if pool ≥ loss(t): pool -= loss(t) // absorbed, credits stay valid
else: shortfall = loss(t) − pool; pool = 0
retire(shortfall) credits from the market // permanence broken
The market price responds to scarcity and to confidence in permanence — each retirement (a credit invalidated after already being sold) spikes price and drags the running "permanence" score down, exactly as it does in real voluntary carbon markets when a project under-reserves its buffer.
- Buffer contribution rate — how much of each year's credits are withheld as insurance vs. sold immediately. Higher = safer but less revenue now.
- Reversal probability — annual chance of a fire/drought/pest event this cycle.
- Plot size — number of plotted trees in the 2D stand, sets the sequestration ceiling K.
Real-world relevance: this buffer-pool mechanism is exactly how registries like Verra's VCS and the California Air Resources Board insure forest offset permanence — an under-funded buffer is one of the most common reasons voluntary carbon credits get invalidated after purchase.
Note on the underlying model: the 3.67 CO2/carbon mass ratio and the logistic growth form were checked numerically against the 44.01/12.01 molecular-weight ratio (≈3.6645) and standard Chapman–Richards growth curves — both are correct and were carried over unchanged from the source model.