Spatiotemporal Scaling of Soil Organic Carbon using Soil Survey, Ecological Site Concepts, and Land Use Land Cover Data in the Southern Coastal Plain (MLRA 133A)

Soil organic carbon (SOC) represents the largest terrestrial carbon stock globally, but it is susceptible to management changes that make regular monitoring of SOC stock crucial. This study presents the DSP-Scale approach to predict SOC in space and time using soil survey, ecological site concepts, and land cover data. We used 1441-point measurements of SOC taken between 2000 and 2018 to predict SOC stock in the upper 20 cm (SOC20) for four-time periods (2001, 2006, 2011, and 2016) in the Southern Coastal Plain. Our random forest model explained 68.46% and 55.6% of the variation for the training and validation datasets. Total SOC20 stock in 2016 was estimated at 1305 Tg. Mean annual precipitation, elevation, erosional classes, and land cover classes were the most important predictors of SOC20 variability. The ability to monitor SOC in space and time will assist policy makers and land managers with decisions related to soil health and SOC stocks over time at regional scales.