Comparing predictions of long-term soil carbon dynamics under various cropping management systems using K-model and CENTURY

There is a strong demand for accurate estimates of long-term changes in soil organic carbon(SOC)with different agricultural practices under different soil and climate conditions.A process and analytic model,K-model,including a non-compartmental algorithm of soil carbon decomposition,was developed to simulate the changes of SOC under different cropping and soil management practices.This study evaluates the performance of K-model by comparing its predictions on SOC with measurements and predictions of CENTURY model,which is widely used for the similar purposes.Both K-model and CENTURY can predict the dynamics of SOC when site-specific soil and climate data are used to initialize simulations.Very similar annual carbon decomposition rates were simulated by the single carbon pool K-model and the 3-carbon pool CENTURY model.However,compared with experimental measurements of SOC,K-model produces relative smaller errors than CENTURY(<0.1 kg C m-2 vs.0.08-0.48 kg C m-2,and within±5%vs.±5%-45%),mainly resulting from smaller biases of predicted crop production.When detailed site-specific soil and climate data are not available for initialization and feeding the running of model,K-model can still reasonably predict the dynamics of SOC with its auto-correction function,but CENTURY produces poor results.In comparison with measurements,K-model has improved capacities to predict the effects of chemical fertilizer,manure application,residue management and fallow on SOC dynamics.