akar.earthConfidence interval: Definition
What is a confidence interval in carbon accounting? How 90% and 95% intervals express uncertainty in carbon stock estimates and trigger deductions for projects.
A confidence interval is a range around an estimate that is expected to contain the true value with a stated probability, such as 90% or 95%. In carbon accounting, it shows how precise an estimate of carbon stocks or removals is, given sampling and other uncertainty.
Why it matters
Methodologies use confidence intervals to decide how much to credit. A project might estimate removals of 10,000 tCO2e with a 90% confidence interval of plus or minus 8%. If the methodology’s precision target is 10%, no deduction applies. If the interval were plus or minus 20%, the project would likely have to credit a lower, conservative value. Wide intervals therefore cost money. Buyers and raters also read reported uncertainty as a signal of data quality.
How it is calculated
For a simple random sample:
CI = mean ± t × standard error
where t depends on the confidence level and the number of plots. Precision is usually expressed as the half-width divided by the mean, as a percentage. For stratified samples, stratum variances are combined with area weights. Uncertainty from several sources, such as sampling, allometry and remote sensing models, can be combined by error propagation or Monte Carlo simulation. Our guide to uncertainty deductions shows how programmes apply the result.
Indonesia angle
Indonesian forest and agroforestry projects tend to have variable carbon stocks and many small plots, which widens confidence intervals. Good stratification, enough plots per stratum, and consistent field protocols are the main remedies. Our carbon modelling service helps design inventories that hit precision targets without overspending on fieldwork.
Common mistakes
- Confusing the confidence level (90%) with precision (±10%).
- Reporting only sampling error and ignoring allometric model error where the methodology requires it.
- Computing intervals on per-plot values without area weighting across strata.
Related terms
See sampling error, stratification and uncertainty deduction, or explore the dMRV hub.
Frequently asked questions
What confidence level do carbon methodologies use?
Most use 90% or 95%, depending on the programme and methodology. Check the specific requirement before designing the inventory.
What happens if the confidence interval is too wide?
Projects typically apply an uncertainty deduction, crediting a conservative value such as the lower bound for removals, until precision meets the target.