Filling gaps in models simulating carbon storage in agricultural soils: the role of cereal stubbles

Carbon (C) input is a prerequisite for the formation of soil organic matter and thus for soil organic C (SOC) sequestration. Here we used the C-TOOL model to simulate SOC changes in a long-term field experiment (1932-2020) at Askov, Denmark, which involved four different levels of nutrients added in mineral fertilizer (0, 0.5, 1, 1.5 NPK) and a four-crop rotation. The C input into soils consists of belowground and aboveground plant biomass and was estimated using allometric functions. The simulation showed that modelled SOC based on standard allometric functions of C input from crop residues did not adequately matched measured SOC contents. However, applying modified allometric functions based on current and the previously measured results for aboveground and belowground C inputs in winter wheat and grass clover in rotations provided much better match between simulated and measured SOC contents for fertilized treatments at normal and high level of fertilization. This improved indicators of C-TOOL model performance (e.g. yielding RMSE of 2.24 t C ha^-1 and model efficiency of 0.73 in 1.5 NPK treatment). The results highlight that standard allometric functions greatly overestimates the amount of C in winter wheat stubble left after harvest in treatments dressed with NPK compared with modified functions. The results also highlight further needs for improvement of allometric functions used in simulation models for C-accounting in agroecosystems.