A pixel-twinning approach for assessing additionality: a case study of human-induced regeneration projects

The impact of past management changes on environmental variables is sometimes assessed by comparing actual and counterfactual scenarios, with the differences between these scenarios referred to as additionality. For investigating the impact of management changes expected to have driven increased growth of woody vegetation, a counterfactual scenario must represent changes that would still have occurred naturally due to climatic variation. Remote sensing datasets from long time-series earth-observation data (e.g. Landsat) offer the possibility of building counterfactual scenarios and investigating additionality. We present and apply a pixel-twinning approach to build counterfactual scenarios from remote-sensing time-series datasets. The approach builds a counterfactual scenario for a project (where past management changes have been implemented). It does this by twinning each project pixel with a pixel peripheral to the project area (within a distance of 3 km, excluding other known project areas), with the twin selected by matching the pre-project time-series data. Since both the target pixel and its twinned reference pixel responded similarly to climate before management changes were implemented, it would seem reasonable to assume that in the absence of any management change, those pixels should continue to respond similarly to climate, providing a rationale for using the twinned pixel to build a counterfactual. The approach is applied in a case study to investigate the additionality of 84 Queensland-based human-induced regeneration projects in the Australian Carbon Credit Unit Scheme, using Landsat-derived persistent green time-series data as an indicator of changes in woody vegetation. Results demonstrate significant statistical evidence ( p < 0.001) of additional persistent green, but with a small magnitude (0.07 % per year). Further analysis demonstrates that this additionality occurs within land that is likely to have been cleared in the past. The method offers potential to be applied in other studies and with other datasets for investigating additionality associated with past management changes. • Remote-sensing data can be used to build counterfactual scenarios. • We present a pixel-twinning approach to do this and assess additionality. • We demonstrate on a case study of regeneration projects using persistent green data. • Results show small additionality, located mainly in previously cleared project land. • The approach offers potential to be applied in other studies and with other datasets.