From Earth Observation to Carbon Farming: Implementing the MRV Framework for Soil Carbon Modeling Across Croplands

Robust Monitoring, Reporting and Verification (MRV) systems are essential for quantifying soil organic carbon (SOC) changes and supporting carbon farming initiatives. However, the scalability of process-based SOC models is often constrained by the availability of detailed in-situ boundary conditions, here defined as the required model inputs including climate variables, soil properties, carbon inputs, and management practices. 

We present an operational MRV framework that integrates Earth Observation (EO) products with the RothC model to simulate SOC dynamics across croplands in the Walloon region of Belgium (≈50,655 ha). 

The framework is validated at a long-term experimental site (10 years). At the sub-landscape (field) scale, in-situ driven RothC simulations show a bias of −0.16 t C ha⁻¹, and EO-driven simulations achieve a bias of 1.76 t C ha⁻¹, demonstrating comparable performance. The model is then upscaled by replacing point-based inputs with EO-derived geospatial layers, enabling SOC simulations across all fields in the region.

Sensitivity analysis reveals that model uncertainty is dominated by carbon input estimation. Landscape-scale simulations indicate a mean SOC decline of −0.04 t C ha⁻¹ yr⁻¹ under business-as-usual management. In contrast, scenarios combining long-duration cover crops and reduced tillage reverse this trend, achieving sequestration rates of 3.8‰ yr⁻¹, consistent with the “4 per 1000” ambition.

This study demonstrates a transparent and scalable MRV workflow that links EO-based monitoring, process-based reporting, and multi-level verification, providing a practical foundation for soil carbon accounting and policy-relevant carbon farming assessments.