Estimation of Organic Carbon in Anthropogenic Soil by VIS-NIR Spectroscopy: Effect of Variable Selection

Abstract

Visible and near-infrared reflectance (VIS-NIR) spectroscopy is widely applied to estimate soil organic carbon (SOC). Intense and diverse human activities increase the heterogeneity in the relationships between SOC and VIS-NIR spectra in anthropogenic soil. This fact results in poor performance of SOC estimation models. To improve model accuracy and parsimony, we investigated the performance of two variable selection algorithms, namely competitive adaptive reweighted sampling (CARS) and random frog (RF), coupled with five spectral pretreatments. A total of 108 samples were collected from Jianghan Plain, China, with the SOC content and VIS-NIR spectra measured in the laboratory. Results showed that both CARS and RF coupled with partial least squares regression (PLSR) outperformed PLSR alone in terms of higher model accuracy and less spectral variables. It revealed that spectral variable selection could identify important spectral variables that account for the relationships between SOC and VIS-NIR spectra, thereby improving the accuracy and parsimony of PLSR models in anthropogenic soil. Our findings are of significant practical value to the SOC estimation in anthropogenic soil by VIS-NIR spectroscopy.

Multivariate methods with feature wavebands selection and stratified calibration for soil organic carbon content prediction by Vis‐NIR spectroscopy

“…However, the feature wavebands of paddy soil (at 2,146–2,156 nm) are mainly influenced by the C‐O functional group, which agrees with the findings of L. Xu et al. (2020) and Hong et al. (2018) in the eastern part of Jianghan Plain (Hubei Province, China).…”

“…Shi et al, 2014). This technology is increasingly used to estimate SOC content because of its high spectral resolution, high efficiency, and environment-friendly procedure (Moura-Bueno et al, 2019;L. Xu et al, 2020).…”

“…In recent years, visible–near‐infrared (Vis‐NIR) spectroscopy has emerged as a powerful tool for quantitative research of SOC (Chen et al., 2021; Z. Shi et al., 2014). This technology is increasingly used to estimate SOC content because of its high spectral resolution, high efficiency, and environment‐friendly procedure (Moura‐Bueno et al., 2019; L. Xu et al., 2020).…”

Hyperspectral band selection and modeling of soil organic matter content in a forest using the Ranger algorithm

“…Pretreatment removes extraneous interference and improves the performance of estimation models. Common spectral pre-processing techniques include mathematical transformation [8][9][10][11][12], Savitzky-Golay (SG) [11,[13][14][15], continuum removal (CR) [15][16][17], multiplicative scatter correction (MSC) [11,17], and standard normal variate (SNV) [11,17]. Spectral band selection aims to select the optimal variables from the raw spectra, in order to enhance the spectral sensitivity of soil properties.…”

“…Spectral band selection aims to select the optimal variables from the raw spectra, in order to enhance the spectral sensitivity of soil properties. Several Machine learning (ML) methods have been proposed for spectral band selection, such as competitive adaptive reweighting sampling (CARS) [8,[16][17][18], principal component analysis (PCA) [14,15,19,20], locally linear embedding (LLE) [14], multidimensional scaling (MDS) [14], metaheuristic algorithms [21], and rough set algorithms [22]. In addition, partial least squares regression (PLSR) [13,20,23], artificial neural networks [4], random forest (RF) [2,17,24], and support vector machine (SVM) [24,25] are common methods of soil spectral modeling.…”