BAU deforestation scenario generated by a machine learning algorithm

To prevent deforestation, decision-makers need reliable information about where it’s likely to occur and why. Using historical data to create a single baseline deforestation rate across a whole country or even within a jurisdiction fails to accurately capture the spatial heterogeneity.

Instead, we calculate a deforestation risk score for every pixel in the landscape. The result is a BAU deforestation scenario that uses data on real local drivers, seamlessly ‘nesting’ estimates from the pixel to national scale using a methodology that can be applied consistently in any country.

This approach will help policy-makers to create incentives for project developers to focus on areas at most risk.

Please find more info here and the progress to date in this presentation.

Many thanks to all collaborators!