Accounting & Integrity

Model Validation

Independent testing of a carbon model against data it was not trained on, to confirm it performs as claimed.

Model validation checks whether a model produces reliable results on new data, separate from the data used to build it. This guards against overfitting and gives confidence in the estimates.

Validation reports metrics such as error and bias, and compares modelled values against independent measurements. A model that performs well in validation is more trustworthy for crediting.

Because models drive satellites-to-carbon conversions, transparent validation is a core part of defensible verification.

Frequently asked questions

Why validate a model?

To confirm it works on unseen data and is not merely fitting noise, ensuring credible estimates.

What evidence demonstrates validation?

Reported error and bias metrics against independent reference data not used in model training.

Related terms