Field-derived spectral characteristics to classify conventional and conservation agricultural practices

Abstract

Field-derived hyperspectral reflectance of soil and wheat were collected during a wheat-growing season (February, March and April) using a portable spectrophotometer in an experimental field receiving conventional and conservation agricultural practices. These practices included two types of tillage, conventional and conservation, and two types of nutrient sources, inorganic and a combination of inorganic–organic fertilizers. This research aims at discriminating conventional and conservation practices and identifying the most applicable bands for discrimination. Spectral characteristics of soil and wheat indicated that soil spectra were useful to discriminate tillage practices, while wheat spectra were useful to discriminate nutrient sources. Cluster and principal component analysis (PCA) revealed that the best time to collect soil and wheat spectra for tillage and nutrient sources differentiation was at the beginning of the season. According to the best time for spectra collection, wavelengths more suitable for discriminating tillage practices were in the VIS and NIR ranges, and in the NIR range for nutrient sources. Furthermore, spectra and band-wise normalization were applied indistinctly to the data sets. It was understood that soil or wheat spectra could not be independently used to discriminate treatments (interactions between tillage and nutrient sources), unless the spectra are normalized and a method such as PCA used.