Topographic Controls, Spatial Heterogeneity, and Prediction Accuracies of SOC Stocks Across Geospatial and Earth System Models.

The ability to accurately represent the global distribution of existing soil organic carbon (SOC) stocks is a prerequisite for predicting carbon-climate feedbacks. Reliable estimates of regional SOC stocks and their spatial heterogeneity are essential to better understand environmental controls of SOC stocks and their vulnerability to changing climate. We investigated spatial scaling impacts on topographic controls of SOC stocks and compared SOC estimates from four geospatial (linear regression, geographically weighted regression, regression kriging, and local regression kriging) and four earth system model (ESM) estimates (Beijing Climate center, Canadian Center for Climate Modeling, Atmosphere and Ocean Research Institute, and Geophysical Fluid Dynamics Laboratory) to evaluate the spatial heterogeneity and prediction accuracy of SOC stocks for the State of Alaska, USA. Spatial scaling changed topographic indices, their controls on SOC, and predictions of SOC stocks. Topographic indices stabilized around 2 km spatial resolution and their controls on SOC became unrealistic as spatial resolution increased beyond 200 m. Spatial heterogeneity of SOC stocks was more realistic and in agreement with pedogenic knowledge in geospatial estimates in comparison to ESM estimates. Though variations in prediction errors observed within geospatial and ESM estimates, in general, higher prediction errors were obtained for the SOC stocks generated from ESMs in comparison to geospatial approaches. Results indicate the need of better process representations in ESMs about the environmental controls of arctic SOC stocks which may reduce the uncertainty in predicting permafrost carbon-climate feedbacks using ESMs.