Modelling soil organic carbon using vegetation indices across large catchments in eastern Australia

Area of Science:

• Environmental Science
• Soil Science
• Remote Sensing

Background:

• Soil organic carbon (SOC) is crucial but difficult to measure at large scales.
• Traditional methods require extensive labor, limiting catchment-scale analysis.
• Grazing lands present unique challenges for SOC assessment.

Purpose of the Study:

• To examine soil organic carbon distribution at large catchment scales (>500 km²).
• To assess the utility of remote sensing-derived vegetation indices (VI) for predicting SOC.
• To compare VI data from MODIS and Landsat satellites for SOC mapping.

Main Methods:

• Field-sampled SOC data from eastern Australian catchments (Krui and Merriwa).
• Comparison of SOC data with DEM-derived elevation, insolation, and VI (NDVI, EVI) from MODIS and Landsat.
• Analysis of aggregated, prior VI data sets for correlation with SOC.

Main Results:

• Single, immediate VI data showed poor correlation with SOC.
• Multiple, aggregated prior VI data sets provided a good match with SOC.
• Landsat 8 EVI yielded the strongest correlation (R² up to 0.41), indicating higher resolution VIs improve accuracy.

Conclusions:

• Vegetation indices are reliable surrogates for historical vegetation growth and soil carbon inputs in pasture landscapes.
• SOC can be reliably predicted at large catchment scales using remote sensing data.
• The developed method is robust, applicable globally, and offers a new approach for environmental assessment.