Soutenance

The growing interest in the study of carbon stored in soils is due to two main reasons. First of all, soils are the major components of the terrestrial ecosystems and the largest organic carbon reservoir on Earth, being very reactive to human disturbance and climate change. Second, despite its importance within the carbon reservoirs, soil carbon dynamics is an important source of uncertainties for future climate predictions. The aim of the thesis was to explore different aspects of soil carbon studies (Experimental measurements, modeling, and database evaluation) at different spatial scales (from the scale of a profile to the global scale).

In fact, we first estimated global and regional soil carbon stocks on 1m depth given by three existing databases (SoilGrids, the Harmonized World Soil Database (HWSD), and the Northern Circumpolar Soil Carbon Database (NCSCD)). We observed that total stocks predicted by each product greatly differ. The estimation of the global soil carbon stocks is therefore still quite uncertain. Consequently, the role of soil carbon in the climate dynamics becomes one of the major uncertainties in the Earth system models (ESMs) used to predict future climate change. Regarding the relatively simple representation of the soil organic matter dynamic in such model, the comparison with data is very complex. One of the promising approaches emerging from literature is the use of carbon isotopes data (¹³C and ¹⁴C) that can be considered as integrators of the soil carbon dynamics for different time scales (year to century). Therefore, isotopic measurements were performed in the laboratory on a soil latter used to evaluate a new version of the Land Surface Model ORCHIDEE. Indeed, the soil studied is an Argentinean soil under two different land covers. Total organic carbon content, ¹³C and ¹⁴C activity were measured. ¹³C data of both top layers clearly exhibits input of new carbon that derived from C4 Plant into the upper layers. Then, the lower ¹⁴C activity of top layer of the C4 plant profile highlights a loss of new carbon that was under Native Forest. A loss of the old carbon that dates from the Holocene was also hilighted. The last part of thesis deals with the presentation of a new version of the IPSL-Land Surface Model called ORCHIDEE-SOM, incorporating the ¹⁴C dynamics in the soil. After implementing the ¹⁴C in the model, we evaluated model outputs against observations of soil organic carbon and modern fraction from four sites with different vegetation covers. The model managed to reproduce the soil organic carbon stocks and the modern fraction along the vertical profiles at the four sites. Several tests done assume that model improvements should focus more on a depth dependent parameterization, mainly for the diffusion, in order to improve the representation of the global carbon cycle in Land Surface Models, thus helping to constrain the predictions of the future soil organic carbon response to global warming.