Current Research on Geography, Earth Science and Environment Vol. 10

Soil organic carbon (SOC) sequestration in managed land occupies an unusual position in climate policy. The biogeochemical rationale is well established, yet the quantitative claims made on its behalf are settled not in the soil but in the measurement, reporting and verification (MRV) systems that convert soil processes into tradable or reportable units. Over the past decade these systems have been reconstructed around digital components, including soil spectroscopy, satellite reflectance compositing, machine learning prediction, process-based simulation, data assimilation and distributed ledger infrastructures. This review critically appraises what that digital reconstruction has and has not achieved. Peer-reviewed literature published between January 2015 and 4 July 2026 was identified through open scholarly indexes, supplemented by citation searching and authoritative institutional documents, and appraised for methodological adequacy, replication, and alignment between evidence and claim. Five findings emerge. First, digital sensing has substantially reduced the marginal cost of estimating SOC concentration, but the accuracy gains reported in cross-sectional calibration studies do not transfer reliably to the detection of change at field scale. Second, the dominant sources of uncertainty in stock change have shifted away from laboratory analysis towards bulk density, sampling depth, spatial variability and soil mass correction, which remain comparatively under-digitised. Third, model-based crediting achieves scalability by importing structural assumptions that empirical resampling would test, and the evidence that ensembles or Bayesian calibration resolve this dependence remains preliminary. Fourth, permanence is not primarily a data problem: saturation behaviour, mineral association and reversibility constrain what any verification architecture can guarantee, and current accounting instruments handle this unevenly. Fifth, protocol heterogeneity produces materially non-equivalent credits from identical management. Confidence is strongest for the descriptive mapping of SOC concentration and weakest for verified, permanent, additional removals at project scale. Priorities include instrumented reference networks, transparent uncertainty propagation, and evaluation of digital tools under smallholder conditions.