It is known that carbon distribution in the earth crust and soil mantle is uneven; in contrast to the atmosphere and hydrosphere, the redistribution of carbon in the soil mantle does not lead to leveling of carbon concentrations. The current rise in the atmospheric carbon concentration poses the problem of quantitative determination of carbon sinks and sources. The reserves of organic carbon concentrated in soils by far exceed carbon pools in other components of the biosphere (except for World Ocean); soils may serve as the most significant terrestrial sinks and sources of carbon. The agricultural development of soils may lead to considerable changes in the organic carbon concentration. The losses of soil organic matter upon the long-term soil cultivation may be caused by different processes, including direct mineralization-induced carbon emission into the atmosphere. To calculate and predict these losses, models of the carbon status of soils are being developed. One of the crucial factors for successful modeling is the adequate input information.

The analysis of different models suggests that soil and climatic parameters and land use data are needed for most of them. Spatially distributed three-dimensional data (including the vertical or depth coordinate) are required. Different models have different demands for the number of input parameters. It is feasible to develop the general methodology of geographic information systems for the purposes of modeling the carbon dynamics.

The GIS developed in the Dokuchaev Soil Science Institute for the territory of Russia is based on the digitized Soil Map of the Russian Federation, 1:2.5 M scale. As it is necessary to compare different layers of spatially distributed data with the initially different scales, all of them were georeferenced to the topographic map of 1:1 M scale. To display the character of land use, the Map of Land Uses on a scale of 1:4 M was used. The soil map contains 205 mapping units, which is still insufficient to characterize the diversity of soil conditions. An additional layer displaying information on the soil texture was added. In order to reduce variability in soil properties within a given mapping unit, the map of Natural-Agricultural regionalization (1:8 M scale) was added. The units for further modeling of carbon dynamics are obtained via superposition of these four maps. Thus, they are characterized by similarity in the genetic type (subtype) of soil, natural zone, soil texture, and the type of land use. Each of such combinations is characterized by a typical soil profile; soil physical, physicochemical, and chemical properties are given in the attribute tables for separate soil horizons.

The database on soil horizons includes the following characteristics: (1) Horizon depth (thickness), (2) bulk density, (3) content of fraction <0.01 mm, (4) content of fraction <0.001 mm, (5) content of fraction <0.002 mm, (6) content of fraction <0.0063 mm, (7) humus content, (8) type of humus, and (9) soil acidity.

The climatic database is developed on the basis of the world grid 0.5°; for every cell, the following data are included with the time step of 10 days: (1) air temperature, (2) precipitation, (3) water vapor pressure, (4) solar radiation, (5) evaporability, (6) mean daily amplitude of temperatures, and (7) Cloudiness.

The database on land use was developed with due account for typical crop rotation systems. For every crop in the rotation, the following parameters were included: (1) rates of organic fertilizers, (2) crop yield, and (3) amount of postharvest residues left in the field.

After the creation of the geographic information system, special work on data aggregation was performed in order to reduce the computation time of modeling the carbon dynamics for vast areas. For the European part of Russia, the resulting map consists of 200 more or less homogenous areas suitable for further model calculations.

The GIS created was used to calculate the reserves of organic carbon in natural and agricultural soils for 1990. The carbon dynamics were simulated using the CANDY and ROTHC models. It is supposed that additional information from interpretation of remote sensing data and from the soil map of 1:1 M scale will make it possible to calculate soil carbon dynamics from 1990 to 2005 taking into account the transfer of a part of cultivated lands into long-term fallow lands.