SPATIAL-ANALYSIS AND MODELING OF TOPSOIL CARBON STORAGE IN TEMPERATE FOREST HUMIC LOAMY SOILS OF FRANCE

Organic matter (OM) is an important component of soils because of its influence on cation exchange capacity, water retention, soil structure, and ecology and as a source of plants nutrients. Recent attention to rising levels of atmospheric CO2 has directed attention to the stores of organic C in soils and to changes resulting from conversion of forest to cropping. However,the spatial distribution of carbon pools in forest soils is difficult to estimate because of the unavailability of reliable data. In southwest France, thick humic acid soils have developed from Quaternary silty alluvial deposits. The area of study is characterized by textural and climatic gradients. The objective of this work was to determine if relationships between these gradients and organic matter contents could be established, in order to make a spatial prediction of organic pools in forest soils, and to simulate future evolution under corn cropping. Soil samples were collected hom an oceanic zone of the French Pyrenean piedmont, ancient terraces of Pyrenean streams (southwest France), and from 11 sites. On each site, from 13 to 27 topsoil (0-30 cm) samples were collected from mature forests. A total of 194 samples were collected. Correlations between all climatic, geomorphological, and pedological data were calculated. The area of the terraces was delimited using a traditional geomorphologically based survey and stored into a geographical information system (ARC/INFO). This map was overlayed with a 1-km X 1-km grid, and probability level maps of organic C amounts in forest soils down to 30 cm were produced using a multiple linear model. This study shows that relating OM contents to spatial available parameters that might influence OM distribution can provide a useful tool to improve geographical prediction of was found to be the most important soil parameter influencing OM distribution. Another important point is the concept of probability level associated with spatial prediction. This study gives an example of a spatial model taking this variability into account.