Estimating Soil Organic Matter Content by Regression Kriging

Abstract

In Mediterranean countries soil organic matter (SOM) depletion is a key factor in land degradation. Here, climate (temperate winter/dry summer) and water scarcity give rise to faster mineralization rates and lower accumulation intensities, particularly in association with intensive and non-conservative agronomic practices. The study area is located in central Italy, in the Soil Region 61.3 as defined by the European Soil Bureau, where soil erosion is the main cause of the low SOM content. In this area, about 250 georeferenced samples were collected from the surface horizon (plough layer) of agricultural soils. These samples have been analyzed for particle size distribution and soil organic carbon (SOC) content. The use of regression kriging (RK) is proposed to predict SOC content and soil texture using the following attributes as predictors: (a) soil subsystems map (1:250,000) derived from pedological survey, (b) terrain parameters derived from DEM (elevation, slope, plan and profile curvature, TWI, incoming solar radiation), and (c) other indexes derived from Landsat TM imagery (e.g. NDVI, Grain Size Index, Clay Index). Since the same level of SOM differently influences soil functions depending on soil texture, the values of SOM obtained from SOC were classified in four classes (very low, low, medium, high) based on the estimated USDA texture, and RK was applied. This map was compared with the soil subsystems map to evaluate the influence of the prevalent land use on SOM levels.