Comparison of two quantitative soil organic carbon models with a conceptual model using data from an agricultural long‐term experiment

Quantitative soil organic carbon (SOC) models are required for a better understanding of C sequestration in soils and for prognoses at different scenarios. However, it is unclear whether the major C‐stabilization mechanisms are included adequately. Objectives were (1) to test the performance of the CIPS (Carbon turnover In Pore Space) model and the Rothamsted Carbon (RothC) Model for a prediction of the SOC dynamics at the long‐term experiments at Halle (Germany) and (2) to compare the model structures of the quantitative models and a conceptual model in order to identify shortcomings of the quantitative models. Both quantitative models had a similar prediction performance: the C dynamics was predicted satisfactorily for the Halle sites under continuous rye with NPK fertilization or without any fertilization (1878–1953), but larger deviations between modeled and measured C contents were observed for the continuous rye and maize in the entire period from 1878 to 1996. The comparison of the conceptual model with the quantitative models revealed that the neglect of the black C dynamics and the interactions of SOC with mineral surfaces are not included explicitly. Site‐specific calibrations are required where these processes have a significant impact on soil organic matter dynamics. Furthermore, the number of pools and mechanisms in the conceptual model is higher than in the quantitative models independent of their level of abstraction. Despite the neglect of some important mechanisms in the quantitative models, it has to be noted that they reproduce the SOC data generally well in agricultural surface soils. Moreover, they need only few inputs which are generally easily obtainable.