Item

Abstract

Soil organic carbon changes (ΔSOC) are regulated by climate and land use change. Here, we analyze regional and global ΔSOC from 1861 to 2099 based on five terrestrial biosphere model (TBM) simulations of the Inter‐Sectoral Impact Model Intercomparison Project Phase 2b. The TBMs were driven by harmonized gridded land use change and bias‐adjusted climate forcing data from different general circulation models (GCMs) for climate scenarios RCP 2.6 and RCP 6.0. Between 2005 and the end of this century, we estimated an increase of SOC for two scenarios with large uncertainty, which is dominated by differences between TBMs. We present a new emergent constraint approach to constrain future modeled ΔSOC over natural vegetation from RCP 6.0 simulations using recent observed trends of net primary productivity as a proxy of litter inputs to soil pools. Our results showed that the uncertainties in constrained ΔSOC can be reduced in comparison with the original model ensemble, but constrained values of ΔSOC depend on the choice of a GCM and climate regions. For the reduction of the SOC density in areas where cropland expanded (Δsoccropland expansion) over natural vegetation as a result of land use change, the constrained Δsoccropland expansion still features large uncertainties due to uncertain observed data. Our proposed emergent constraint approach appears to be valuable to reduce uncertainty on SOC projections, but it is limited here by the small number of models (five) and by the uncertainty in the observational data. Applications to larger ensembles from Earth System Models should be tested for the future.