Reducing uncertainty in long-term (1923-2024) SOC change estimates through simultaneous calibration of C inputs and soil parameters

Robust estimation of soil organic carbon (SOC) changes in mineral agricultural soils is essential for consistency when implemented in national greenhouse gas inventories. In Denmark, SOC changes in mineral soils are quantified using the C-TOOL soil carbon model, aligned with IPCC Tier 3 methodologies. However, substantial uncertainties remain related to SOC parameterization and the estimation of carbon (C) inputs to soil. In this study, we compared alternative strategies for calibrating C-TOOL by combining different approaches for estimating C inputs with adjustments of SOC pool distribution and degradation, using an ensemble-based uncertainty analysis. SOC observations (1923–2024) from a continuous crop rotation receiving mineral fertilization (1 NPK) at the Askov long-term experiment (field B5) were used as the reference dataset. Three C-input approaches: (i) yield-based allometric functions, (ii) modified allometric functions, and (iii) fixed crop-specific C-input values were combined with recalibration of soil parameters controlling carbon pool size distribution and organic matter decomposition rates. Results showed that model performance was strongly influenced by both C-input approach and SOC parameterization. The best performance of CTOOL simulation compared with observed data was obtained using fixed C-input values combined with recalibrated SOC parameters set (f_HUM = 0.55, f_ROM = 0.48, k_HUM = 0.024 yr⁻¹); increasing the D-index from 0.75 to 0.82 and reduced RMSE from 4.52 to 2.98 Mg C ha⁻¹ when compared with standard parameterization and C-inputs from allometric functions. Accurate SOC-change estimates require jointly refining C-input methods and soil parameterization to reduce uncertainty for reliable inventories, supporting credible mitigation decisions.