Soil Carbon Modeling Across the Continental U.S.

Soils may act as a source or sink for atmospheric CO₂ and climate and human activities, such as land use change impact soil carbon stocks.  Hence, insights on soil C modeling at continental scale are critical to provide significant improvements in understanding the carbon cycle. The objectives of this study were to 1) determine the historic soil C stock across the U.S.; 2) investigate the interrelationships of climate change and land use/land cover and SOC stocks within and across different ecoregions. We used soil carbon data in the topsoil (0-20 cm) and aggregated 0-100 cm depth from the U.S. National Soil Survey Database (Natural Resource and Conservation Service). Environmental covariate sets were assembled according to the STEP-AWBH model (S: soils, T: topography, E: ecology, P: parent, A: atmosphere/climate, W: water, B: biota, and H: human) as predictor variables to develop digital soil models. We used spline function to determine soil carbon concentration down to 1m and used a set of pedotransfer functions to estimate bulk density. Random forest was applied to predict soil C and analysis of variance to evaluate land use/land cover and climate change effects on soil C. Our results suggest that the prediction models have capability to estimate soil C storage accurately. The variation of soil C across the continental U.S. was governed in some regions by climate gradient while in others land use effects predominated. The soil C stock estimates can be used to infer possible changes to soil C in terms of C sequestration potential and potential C loss. Furthermore, our results provide further guidance to measure ecological variables influencing C dynamics at continental scale.