The exchange of Carbon between the atmosphere and terrestrial ecosystems represents one of the major uncertainties of current models of global atmospheric circulation aimed at predicting the effects of greenhouse gas emissions on the climate of the 21. century. A key issue is the correct assessment of the amount and stability of organic matter in soils. Unlike vegetation, soil carbon is difficult to assess by remote sensing. It is also subject to slow changes of the ecology within the critical zone in response to both climate and land use change. The heterogeneity, both in time and space, of natural forests therefore represents a major challenge to inventorying soil carbon in the natural environment. This project focuses on the development of geostatistically sound sampling strategies for carbon inventories in some of the most dynamic forest environments. The effects of forest management, land use and climate are integrated in the rationale for selecting field sites in Canada, the Swiss Alps and Israel. At each site, sampling techniques and spatial patterns adapted to the local geo-ecologic conditions are developed, aimed at identifying appropriate sampling strategies as well as establishing a reliable soil Carbon inventory.