EO Capability Benefits
Estimating Soil Organic Carbon (SOC) through Earth Observation (EO) enables cost-effective, repeatable assessments of soil health, land degradation, and carbon sequestration potential. SOC is a key indicator of soil fertility, productivity, and climate mitigation capacity. EO-derived SOC maps support sustainable land management, carbon accounting, and nature-based solutions for climate finance. In many regions, EO offers the only viable alternative to expensive and spatially limited ground sampling campaigns, empowering stakeholders with data for monitoring land restoration, regenerative agriculture, and carbon offset programs.
EO Capability Description
Soil is made of a mix of organic and inorganic material, with different physical and chemical properties, the proportions of which change from place to place. Soil Organic Carbon (SOC) refers specifically to the carbon component present in the organic portion of soils, originating from plant and animal material decomposition. SOC is considered to be the largest terrestrial carbon sink, and thus, has a major role in the regulation of CO2 levels in the atmosphere. Remote sensing is used for predicting SOC levels in topsoils globally. Organic matter present in soils will influence their spectral reflectance, especially in the visible, near-infrared and shortwave-infrared regions of the electromagnetic spectrum.
EO-based SOC estimation typically combines optical satellite imagery (e.g., Sentinel-2, Landsat-8) with machine learning models calibrated using ground-truth soil samples. Reflectance values in visible, red-edge, and near-infrared bands are sensitive to surface soil properties, allowing spatial prediction of SOC levels using regression models such as Random Forest or LightGBM.
Products are usually delivered as continuous SOC percentage maps (e.g., %C in topsoil) at 10–30 m spatial resolution, covering areas from field to regional scale. Timeseries analysis enables tracking of SOC trends and potential impacts of land use changes or soil management interventions. Accuracy depends on the density and distribution of calibration samples, surface conditions (e.g., bare soil visibility), and model robustness.
Advances in Earth Observation (EO) technology, such as the integrated use of optical, hyperspectral, thermal, and radar data, have significantly improved the capacity to assess and monitor soil properties at different scales and in a consistent manner. Machine learning algorithms and different techniques like spectral unmixing are used to optimise SOC levels prediction by improving the identification and modelling relationships among variables.
To ensure transparency and usability, SOC products should be accompanied by metadata on ground sampling, model validation (e.g., R², RMSE), and uncertainty estimates. Integration with land use, vegetation, and topographic data further enhances model performance.
EO-derived SOC estimation is a valuable tool for climate-smart agriculture, land degradation neutrality reporting, and carbon market readiness, offering a scalable pathway for soil carbon monitoring in support of global sustainability goals.
