Soil variability description using Fourier transform mid-infrared photoacoustic spectroscopy coupling with RGB method

- FTIR-PAS was firstly employed to capture the variability of paddy soils.
- Spectral information was extracted and expressed by the first three PCs.
- Mahalanobis distance and RGB method were used for descripting soil variability.

Fourier transform infrared photoacoustic spectroscopy (FTIR-PAS) has been successfully applied in recent years to predict soil properties due to its advantages as a rapid, non-destructive, and well-rounded characterization method. In this study, we investigated the potential of FTIR-PAS to characterize soil heterogeneity in an agricultural field in Lishui (China). On the basis that each FTIR-PAS spectrum represents a well-rounded soil characteristic from each sample, we aimed to explore a method to convert the spectral lines (multidimensional matrix) into data that can be used in soil mapping while retaining maximum soil information to display soil heterogeneity maps according to kriging interpolations. The Mahalanobis distance (MD) and red-green-blue tricolor (RGB) methods were employed to convert the soil information. A set of diverse soil samples (n = 597) was scanned with a FTIR-PAS spectrometer in the mid-infrared range (4000-400 cm− 1). The peaks showing soil organic matter content (C O from carboxylic acids at 1755 cm− 1; COO, NH, and CN from carboxylates and amides at 1540 cm− 1), clay minerals (quartz overtones at 1953 cm− 1: carbonates, quartz, and iron oxides lower than 800 cm− 1), and water (O-H: 3700-3000 cm− 1 and 1700-1600 cm− 1, respectively). The first three PCs (85.05%) were extracted using principal component analysis (PCA). The results showed that both methods could convert the multidimensional matrix based primarily on absorbance bands associated with characteristic soil attributes. However, comparison of the maps suggested that the RGB method could translate information on spectral heterogeneity with higher accuracy than MD method which lost the direction of soil attributions. Therefore, FTIR-PAS spectra data combined RGB method might provide a supplemental method for use in agricultural modelling and soil management.

145