Get reliable rangeland science

Soils in the semi-arid rangelands of southeastern Australia are inherently low in organic carbon (SOC) content due to a combination of climatic and historic land degradation factors. Small increases in SOC attributable to improved management in such expansive landscapes offers an opportunity to restore rangeland function and play an important role in mitigating climate change. Soil Organic Carbon stocks and fluxes are influenced by complex interactions between plant growth, climate, soil type, topography and land management resulting in high spatial and temporal variability. Variability creates a challenge for designing soil sampling strategies to detect small, incremental changes in SOC. In the rangelands, this challenge is accentuated by low SOC stocks, low SOC sequestration rates, presence of soil inorganic carbon and cost of labour-intensive soil sampling programs across large pastoral properties. Optimal stratification by clustering homogenous areas within a paddock, combined with adequate sampling density can reduce variance and improve SOC stock estimations. Remotely sensed earth observation data can be used to determine stratification in soil carbon projects. This paper presents a basic stratification framework that integrates multiple sources of high-resolution landscape data. The study area is a dryland crop and grazing property located in the semi-arid rangelands of New South Wales (NSW). The method fuses a temporal ground cover raster classified by pixel-based analysis, with a segmented image processed by object-based image analysis. The success of the stratification can be judged by a moderately small variance in mean SOC within each carbon estimation area (CEA) and for the total project area. A validation baseline survey is planned for February 2025.