Bayesian calibration of the ICBM/3 soil organic carbon model constrained by data from long-term experiments and uncertainties of C inputs

Abstract

AbstractModels with various complexity can asses soil C sequestration in agriculture. In this study, we updated the Introductory Carbon Balance Model (ICBM) with 28 years of additional data and included multiple long-term bare fallow experiments for old pool kinetics. We validated the model with data from a new long-term sister experiment. The new calibration included uncertainty in the estimation of below-ground C inputs to soil. The model now considers above- and belowground and external C inputs separately (ICBM/3). The underlying mathematical approach is the same, with two state variables and a climatic parameter, and such simple structure remained robust enough to describe soil organic C dynamics over six decades. We tested including an inert soil C pool in the model structure, but it did not decrease the observed variance. Similarly, an intercept in the functions for estimating belowground C input from crop yield was not useful. Results suggest that root C contributes more to the old organic C pool than aboveground C inputs. We also evaluated parameters interactions, in particular between C inputs and their transformation into more stable soil C interacted and decomposition kinetics. We also describe new functions for estimating the ICBM climatic parameter in a more user-friendly way.

Keywords

• SOC decomposition
• SOC models
• Bayesian uncertainty estimation
• ICBM model