Abstract Carbon farming is a nature‐based solution to capture atmospheric CO 2 and store it as soil organic carbon (SOC). Carbon farming trading schemes (CFTS) incentivize farmers to adopt these practices. Integral to CFTS is forecasting SOC changes, typically achieved using traditional multicompartmental soil carbon models (mSCM), and monitoring total SOC stocks. However, traditional mSCM simulate unmeasurable compartments, leading to overparameterization and indeterminable partitioning among carbon compartments, suggesting a need for structural improvements. The ProCarbon‐Soil (PROCS) model addresses this need by abstracting fundamental principles of mSCM, reducing SOC state variables to two (total carbon and decomposability), and employing only one stabilization parameter, compared to the four to eight state variables and 7–20 parameters typically required by mSCM. We mathematically derive methods that can use successive carbon measurements to estimate decomposability and initialize the model. PROCS can handle environmental modifiers and events such as crop rotations, tillage, and manuring events, and respond to soil characteristics and weather conditions. Tests show that PROCS can accurately reproduce synthetic SOC trajectories generated by an mSCM with perturbed parameters using short‐term data (12 years) with acceptable accuracy (median root mean square error <1.03 Mg ha −1 and absolute median of mean bias <0.55 Mg ha −1 ). In a cross‐validation test, the mean normalized root mean square error (NRMSE) closely aligns with the coefficient of variation of white noise introduced in the synthetic data (4.15% vs. 4.00%, respectively) for augmented carbon inflow scenarios, whereas the model exhibits higher errors for the no‐carbon‐inflow scenario (NRMSE = 5.48%, 7.25%, and 8.99% for 12, 24, and 50 years, respectively).
Barioni et al. (Fri,) studied this question.