Dynamic Baselines for Avoided Unplanned Deforestation
Carbon project baselines are business-as-usual scenarios of emissions that would have occurred in the absence of carbon finance. These counterfactuals cannot be measured directly, meaning that the truth of carbon credit quality is fundamentally unobservable. Risk exists on a spectrum, is probabilistic, and requires sophisticated models that combine qualitative and quantitative data, human and artificial intelligence, to identify and translate into actionable insights.
For each project claiming to avoid unplanned deforestation, we construct an independent dynamic baseline using satellite monitoring, land stratification, machine learning, statistical matching, and expert review. Comparison with project emissions informs our view on additionality. Comparison with project-reported baselines and issuance calculations informs our view on over-crediting risk.
Our approach confronts local complexities and deep uncertainties inherent in baseline assessment. We continue to innovate, and will periodically update our methods to reflect the latest advances in data, frameworks and application. We welcome feedback from our clients, partners, market participants, and the wider scientific community.
Contents
Introduction
Overview
Land stratification
Covariate construction
Covariate importance
Statistical matching
Risk and uncertainty