Fast Measurement and Advanced Sensors of Soil Parameters with NIR Spectroscopy

Abstract:

Near infrared spectroscopy (NIRS) is a powerful tool in the analysis of the soil compositions as well as soil physical and chemical properties. Artificial neural networks (ANN), genetic algorithm (GA), wavelet transform, support vector machine (SVM) and other modern data processing algorithms have made estimation and analysis of the soil parameters more accurate and stable since they can eliminate spectral error caused by outside interference and efficiently extract spectral information. It is still necessary to eliminate the impact of soil moisture and particle size on soil spectra in the in-site measurement. Development of the portable and vehicle-mounted real-time soil spectral analyzer is an important measure to promote the practice of precision agriculture. The developed vehicle-mounted soil on-line spectrophotometers have realized the analysis of several soil parameters, and reached a very high accuracy. It is a trend to develop multifunctional vehicle-mounted real-time soil spectral analyzer in the future by which the correction of the spectra can be performed with the data fusing of the spectral data and other dielectric or mechanical properties. The modern spectroscopies, photoacoustic spectroscopy (PAS), laser induced breakdown spectroscopy (LIBA), and terahertz (THz), have shown the great potential in the analysis of soil compositions and parameters. It is needed to carry out basic researches to reveal the spectroscopy characteristics of soil samples in different types and conditions.