// Research
Remote Sensing and Soil Organic Carbon
The soil is by far the largest terrestrial carbon pool, largely exceeding the amounts stored in the atmosphere and the vegetation. At the same time, the carbon stored in soils permits long-term storage. Substantial amounts of additional atmospheric carbon can be absorbed by soils, and in particular by the most degraded soils, thereby enhancing its productivity, bio-diversity, and resistance against soil erosion.
The fact that soils can sequester atmospheric CO2 in the form of soil organic carbon (SOC) offers a cost-efficient means to mitigate climate change. This has attracted the attention of environmental research and policy agendas. SOC is crucial to preserving soil quality and health and assuring food security. For example, maintaining or increasing SOC concentration improves other chemical and physical soil properties, such as nutrient storage, water holding capacity, aggregation and aggregate stability, and sorption of organic and/or inorganic pollutants. Increased carbon stocks also often go hand-in-hand with increased microbiological diversity and activity. Together, this underpins the growing need to spatially estimate and monitor SOC and also other soil properties. Ultimately, such a continuous monitoring at different spatial and temporal scales provides invaluable support and guidance for soil and land management as well as decision and policy-making.
Unfortunately, high-resolution maps that indicate areas where SOC levels are critically low, and where improved management options could be most effective, are scarce. Current SOC maps at regional to global scale consist of rather coarse grid cells and are based on legacy data that is generally not up to date, as collecting and analysing new soil cores is expensive. Too little effort was spent in the past to extract the soil inherent optical properties, unaffected by remnants of crops and stubbles, changes in soil water content and roughness, which permit a robust SOC mapping across large areas. With the launch of the European Copernicus program and other complementary space activities, the availability and quality of remote sensing data has dramatically changed the paradigm. Together with the upcoming imaging spectrometers, remote sensing of (top) soils becomes feasible in a coherent manner from local to global scales – the use of high spatial and spectral resolution optical images from orbital platforms enables the development of soil monitoring and mapping techniques from the local to the regional, continental and global scales in an effective, fast, frequent, and economical way. Only satellite remote sensing offers the necessary high revisit cycles, to permit a comprehensive monitoring of large areas at fine spatial resolution.
SOC shows a close relationship with the reflected electromagnetic radiation in visible (VIS), near‐infrared (NIR), and short wave-infrared (SWIR) regions. Therefore, several recent studies and scientific papers demonstrated the capability of satellite images, including the Copernicus Sentinel‐2 multi‐spectral instrument (MSI) and NASA’s Landsat‐8 operational land imager (OLI), for modelling SOC content and prediction and mapping of it, obtaining encouraging results. The spectral models can be further improved by including auxiliary predictor variables such as related to vegetation, topography, climate, and geology, which are often also highly correlated with SOC. The acquired data and retrieved information can be combined using physically-based, statistical, geostatistical, machine learning, and hybrid techniques to predict and monitor SOC at a fine spatial resolution with a consistent specification and uncertainty, for sustainable soil management and decision-making.
Based on the Geotree team’s deep understanding of the physical interaction of electromagnetic radiation with matter, we have developed a robust SOC modelling approach based on time-series of high-resolution Sentinel-2 data. The developed approach is robust against changes in soil wetness and roughness whilst also detecting and removing spectral effects of residual vegetation cover and stubble. In this way, Geotree obtains from the Sentinel-2 time-series soil inherent optical properties (SIOP) that are closely related to SOC as well as other soil properties. This revolutionary approach will allow for scalable and regular global soil carbon monitoring.
