Tristan Grupp
Tristan Grupp is an Agricultural Data Scientist in the Food, Land, and Water Program and Data Lab at the World Resources Institute. He collaborates closely with Land and Carbon Lab. His current research focuses on applying remote sensing and machine learning to monitor deforestation and natural land conversion driven by agricultural supply chains, supporting commodity traceability and corporate sustainability compliance, including under the EU Deforestation Regulation (EUDR). His work spans forest change monitoring, climate adaptation, and the intersections of food systems and natural landscapes. Beyond WRI, Grupp has contributed to research on climate change adaptation tracking in support of national adaptation planning under the UNFCCC, protected area policy evaluation in the EU, and tropical forest dynamics in the Peruvian Amazon. He has presented his work at international venues including AGU, COP, and the UN National Adaptation Planning Conference.
Sessions
Accurate crop field boundary delineation is foundational for agricultural mapping, yield estimation, and decision support systems. Yet existing AI models, trained predominantly on data from the Global North, perform poorly in underrepresented farming systems such as those in Sub-Saharan Africa (typically under 2 hectares, irregularly shaped) and South America (characterized by shifting cultivation and complex morphologies). This data gap misleads agricultural statistics, weak policies, and inefficient resource allocation.
We leveraged AI and open-source remote sensing data to automatically delineate field boundaries in both regions using transfer learning, adapting pretrained global models to local contexts. In South America, we annotated over 46,000 field boundaries for model training and generated more than 10 million boundaries continent-wide. In East Africa's Great Rift Valley, we automatically detected over 400,000 farms from just 6,000 samples, incorporating multi-stakeholder annotation workflows and quality assurance pipelines refined from lessons learned in South America.
Our results show that models trained on limited but high-quality local annotations scale effectively to out-of-sample regions. In Africa, delineated fields have enabled field level crop type and yield data collection, in preparation for field level crop type mapping, yield estimation and monitoring of agroecological and regenerative agriculture practices. In South America, they have supported supply chain auditing for deforestation-free commitments, EUDR compliance, country-level crop forecasting, and scope 3 emissions estimation. Across both regions, the approach has strengthened national and subnational agricultural data systems and climate resilience frameworks.
By demonstrating AI model transferability across contrasting geographies, this work charts a pathway toward open, inclusive, and scalable Earth observation systems that close critical data gaps in the Global South, positioning AI as a core enabler of sustainable agricultural monitoring at national and subnational scales.