Predicting anthropogenic soil organic carbon redistribution in the Midwestern United States

Abstract:

Soil erosion diminishes agricultural productivity by driving the loss of soil organic carbon (SOC). The ability to predict SOC loss, transport, and deposition is important for guiding sustainable agricultural practices and determining the influence of soil erosion on the carbon cycle. However, models that predict the influence of farming practices on the distribution of SOC across large agricultural regions are lacking. Recent remote sensing observations indicate carbon rich soil has been lost across nearly one-third of the American Midwest. Comparison of the soil loss patterns and LiDAR-derived digital elevation models reveals a consistent correlation between topsoil loss and topographic curvature, indicating that diffusion-like erosional processes are important in driving the loss of SOC. Here, we develop a landscape evolution model that couples soil mixing and transport to predict soil loss and SOC patterns within agricultural fields. Our reduced complexity numerical model requires the specification of only two physical parameters: a tillage mixing depth, L_t [L] and a hillslope diffusion coefficient, D [L²/T]. Using topography as an input and independently determined values of D, the model predicts spatial patterns of surficial SOC concentrations and complex 3D SOC pedostratigraphy as the landscape evolves. We use soil cores from native prairies to determine initial SOC-depth relations and the spatial pattern of remote sensing-derived SOC in agricultural fields adjacent to the prairies to evaluate the model predictions. The model reproduces spatial patterns of topsoil loss similar to those observed in the satellite images. Our results indicate that the distribution of soil erosion and SOC in agricultural fields can be predicted using a simple geomorphic model where hillslope diffusion plays a dominant role. Such predictions can aid estimates of carbon burial and evaluate potential for future soil loss in agricultural landscapes.