Scale effect on soil attribute prediction in a complex landscape region

Abstract

Total 283 soil samples were collected in 1220 km2 area of Dengfeng county, Henan province, according to a nested sample strategy. Several scenarios were designed to research the scale effect on mapping soil organic material (O.M) with regression kriging interpolation. It was found that the trend of soil O.M on the elevation factor was macroscopical and that could be fitted optimally using only large scale data. If small scale data were added in simulation the precision of trend equation would decrease. However small scale data was contributive to the prediction of the residue in regression kriging, which could reveal not only spatial variability of residue in small scale but also enhanced the spatial structure in large scale and improved effectively the prediction. Therefore, the optimal way of soil O.M in regression kriging was that extracting the trend using only large scale data and simulating the residual using data of the both scales. © 2012 Springer-Verlag.