Soil organic carbon estimation with topographic properties in artificial grassland using a state-space modeling approach

Abstract

She, D., Xuemei, G., Jingru, S., Timm, L. C. and Hu, W. 2014. Soil organic carbon estimation with topographic properties in artificial grassland using a state-space modeling approach. Can. J. Soil Sci. 94: 503–514. Knowledge of the distribution of soil organic carbon (SOC) in artificial grasslands in semiarid areas is helpful in optimizing management for soil fertility recovery and carbon sequestration. Accurate estimation of SOC with easy-to-obtain topographic properties can save considerable labor and cost as well as protect the grassland from being disturbed by intensive soil sampling. In our study, a total of 113 sampling points were setup within a patch of artificial grassland in a small catchment located in the north Loess Plateau of China. State-space modeling and traditional linear regression were used to estimate the localized variation of SOC in the 0- to 20-cm surface soil layer using five selected topographic properties (elevation, slope, aspect, plan curvature, and surface soil roughness). Soil surface roughness and plan curvature were identified as the most effective variables for SOC estimation in state-space models. Soil surface roughness and plan curvature explained 92.5% and 84.5% of the total variation of SOC, respectively. The best state-space model was the one using both plan curvature and surface soil roughness, explaining 94.5% of the total variation of SOC, whereas the best linear regression model could only explain 15.9% of the total variation of SOC. The results indicate that all the derived state-space models performed better than the equivalent linear regression models. Our study provides an insight into the possibility of accurate estimation of SOC only using one or two easy-to-obtain topographic properties with state-space modeling approach.