Spatial detection of topsoil properties using hyperspectral sensing

The spatial variability of topsoil texture and organic matter across fields was studied using fieldspectroscopy and airborne hyperspectral imagery with the aim of improving fine-scale soil mapping procedures. Two important topsoil parameters for precision farming applications, organic matter and clay content, were correlated with spectral properties. Both parameters can be determined simultaneously from a single spectral signature since organic carbon largely responds to wavebands in the visible range and clay responds to wavebands in the Near Infrared. Because of cross-correlations, one has to consider iron oxides and high amounts of coarse sand in order to infer clay content from the spectral signature. The composition of the organic matter should be considered in order to infer the organic matter content from the spectral signature. It is shown that the clay and organic matter content can be predicted quantitatively and simultaneously using Partial Least Squares Regression by a multivariate calibration approach.