Chandrakant Singh, U. Martin Persson

Achieving global climate and biodiversity targets and ensuring future food security will require halting agriculture-driven deforestation. Accurate data on the commodities driving deforestation across time and space is crucial for informing policy development, implementation and evaluation. However, such information is currently hampered by limited and heterogeneous data availability (in both comprehensiveness and scope), computational challenges, and lack of updates to the existing databases, that diminish their accuracy and relevance over time. To tackle these challenges, we introduce the Deforestation Driver and Carbon Emission (DeDuCE) model, a framework that merges remotely sensed datasets with comprehensive agricultural statistics to enhance the quantification of agriculture and forestry-driven deforestation globally. Developed using Google Earth Engine and Python, DeDuCE is designed to integrate new and emerging datasets, ensuring the model remains efficient and relevant despite increasing data volumes. This approach also ensures adherence to FAIR data principles, emphasising replicability, adaptability and utility. DeDuCE reports over 9,100 unique country-commodity deforestation footprints across 176 countries and 184 commodities from 2001-2022, surpassing existing databases in scope and detail. The insights from DeDuCE are crucial for governments, companies, and financial institutions aiming to undertake deforestation and emissions accounting, risk assessments, and sustainability evaluations of investments.