Topsoil mapping using hyperspectral airborne data and multivariate regression modeling

The spatial variability of topsoil texture and organic matter across fields was studied using airborne hyperspectral imagery to lead towards improved fine-scale soil mapping procedures. Two important topsoil features for precision farming applications, soil organic matter and texture, were correlated with spectral properties of the airborne HyMap scanner. Sand, clay, organic carbon and total nitrogen contents can be predicted quantitatively and simultaneously by a multivariate calibration approach using Partial Least Square Regression or Multiple Linear Regression. The suite of topsoil parameters can be determined simultaneously from a single spectral signature since the various features are represented by varying combinations of wavebands across the spectra.