Spatial variability of soil depth and organic carbon stocks of Alaska

Despite a large number of global soil organic carbon (SOC) stock estimates for a variety of depth intervals, very few estimates of the magnitude and spatial distribution of permafrost region carbon pools exist. Further, because of polar amplification under expected climate change, degradation of permafrost carbon may lead to strong feedbacks with the atmosphere. The objective of this study was to predict the fine-resolution spatial variability of soil depth and whole profile SOC stocks across Alaska. We used spatially referenced environmental variables (topographic attributes, land cover, climate, and bedrock geology) and legacy pedon description data (n=472) in a local regression kriging framework to predict the spatial variability of soil depth and SOC stocks at 60-m spatial resolution. Data trends were modeled using environmental variables by using a geographically weighted regression approach and the residuals were kriged using a moving window approach. Estimates of soil depth and SOC stocks were validated using independent datasets (n= 62). SOC stock of Alaska ranged from 1 to 270 kg m-2, and the total was estimated to be 56.5 Pg. The predicted SOC stocks had a mean estimation error (MEE) of -2.56 kg m-2 (average =48.5 kg m-2), and a root mean square error (RMSE) of 33.8 kg m-2 (standard deviation = 38.7 kg m-2). Similarly, the predicted soil depth of the study area ranged from 16 to 234 cm, and had a MEE of -17.5 cm (average =115 cm) and RMSE of 62.4 cm (standard deviation =74 cm). Fine-scale spatial heterogeneity in soil depth and SOC stocks was large, but followed a broadly increasing trend from South to North. Further results of active layer depths and its associated SOC stocks will be presented at the meeting.