Modelling long-term soil organic carbon sequestration under varying environmental drivers and internal protection mechanisms – towards a digital twin

Abstract

Soil organic carbon (SOC) plays a large role in sustainable soil management and climate change mitigation. To understand the potential of soils to sequester additional carbon requires detailed knowledge of the underlying processes and drivers. In this study, we use soil evolution model SoilGen3.8.2 to assess the effects of environmental drivers (bioclimate, erosion level and land use) and four protection mechanisms on long-term SOC dynamics. The protection mechanisms (aggregation, clay mineralogy, microporosity and metal oxyhydroxides, MOOHs) showed large differences with different temporal patterns, where aggregation and clay mineralogy dominated during 10 ka of pedogenesis and MOOHs had a negligible effect. Ranking internal and external controls on SOC stocks revealed a decreasing influence of bioclimate > protection mechanism > erosion rate > land use > time. Topsoil and subsoil SOC recovery after agricultural use revealed different dynamics, controlled by the history of environmental drivers and pedogenesis. Natural SOC recovery showed lowest rates for subsoils and highest rates for topsoils, with a strong control of erosion and pedogenetic history. The addition of ground rock of different mineralogies to enhance SOC sequestration had some effect, mainly for goethite, montmorillonite and a temporary effect of calcite. Our simulations demonstrate how SoilGen can improve understanding of soil processes, while also highlighting knowledge gaps, such as missing experimental insights in key SOC stabilization mechanisms. Our study shows that soil models such as SoilGen cannot act as digital twins of a soil that represent the entire soil, as not all processes and parameters of the complex soil system are represented. These models can, however, form the basis of topical digital twins that target specific processes or properties. We provide a roadmap for developing such topical digital twins and recommend to start from a complex model that accounts for pedogenetic history.