CE-DYNAM (v1), a spatially explicit, process-based carbon erosion scheme for the use in Earth system models

Soil erosion by rainfall and runoff is an important process behind the redistribution of soil organic carbon (SOC) over land, hereby impacting the exchange of carbon (C) between land, atmosphere and rivers. However, the net role of soil erosion in the global C cycle is still unclear as it involves small-scale SOC removal, transport and re-deposition processes that can only be addressed over selected small regions with measurements and models. This leads to uncertainties in future projections of SOC stocks and complicates the evaluation of strategies to mitigate climate change through increased SOC sequestration. In this study we present the parsimonious process-based Carbon Erosion DYNAMics model (CE-DYNAM) that links sediment dynamics resulting from water erosion with the C cycle along a cascade of hillslopes, floodplains and rivers. The model simulates horizontal soil and C transfers triggered by erosion across landscapes and the resulting changes in land-atmosphere CO2 fluxes at a resolution of about 8 km at the catchment scale. CE-DYNAM is the result of the coupling of a previously developed coarse-resolution sediment budget model and the ecosystem C cycle and erosion removal model derived from the ORCHIDEE land surface model. CE-DYNAM is driven by spatially explicit historical land use change, climate forcing, and global atmospheric CO2 concentrations affecting ecosystem productivity, erosion rates and residence times of sediment and C in deposition sites. The main features of CE-DYNAM are (1) the spatially explicit simulation of sediment and C fluxes linking hillslopes and floodplains, (2) the low number of parameters that allow running the model at large spatial scales and over long-time scales, and (3) its compatibility with any global land surface model, hereby, providing opportunities to study the effect of soil erosion under global changes. We present the model structure, concepts, and evaluation at the scale of the Rhine catchment for the period 1850–2005 AD. Model results are validated against independent estimates of gross and net soil and C erosion rates, and the spatial variability of SOC stocks from high-resolution modeling studies and observational datasets. We show that despite local differences, the resulting soil and C erosion rates, and SOC stocks from our rather coarse-resolution modelling approach are comparable to high-resolution estimates and observations at sub-basin level. The model also shows that SOC storage increases exponentially with basin area for floodplains in contrast to hillslopes as is seen in observations. We find that soil erosion mobilized 159 Tg (1012 g) of C under changing climate and land use, assuming that the erosion loop of the C cycle was in near steady-state by 1850. This caused a net C sink equal to 1 % of the Net Primary Productivity of the Rhine catchment over 1850–2005 AD. This sink is a result of the dynamic replacement of C on eroding sites that increases in this period due to rising atmospheric CO2 concentrations enhancing the litter C input to the soil from primary production.