Quantitative Study of the Effect of Water Content on Soil Texture Parameters and Organic Matter Using Proximal Visible—Near Infrared Spectroscopy

Continuous monitoring of soil quality is a challenging task in agricultural activity. To meet this need, scientists have succeeded in developing a quick and inexpensive method to characterize soil properties. Thus, spectroscopy has become a promising method for quantifying soil parameters. However, this method remains sensitive to several factors such as water content (WC). The present study aims to quantify the effect of WC on the estimation of soil texture parameters (sand, silt, and clay) and organic matter (OM) using spectroscopy. Reflectance measurements in the laboratory on 68 soil samples were performed by varying the WC in each sample. The analysis revealed a significant influence of WC on spectra acquired from visible to near infrared (V/NIR) spectroscopy data and that spectra can be divided into two classes. To quantify the effect of WC, calibration/validation steps were performed on soil texture parameters and OM with and without taking WC into account. Calibration was performed using the partial least square regression algorithm, and the validation was assessed using four statistical evaluation indices (R2, Nash criterion (Nash), root-mean-square error (RMSE), and BIAS). Results showed a systematic increase in the accuracy of all studied soil particles when the WC is considered. Clay and OM were less influenced, while silt and sand were much more influenced by the WC. The study also highlighted that estimates of soil texture parameters using V/NIR data achieved relatively higher levels of accuracy (R2 > 0.80 and Nash > 0.80) than OM estimation (R2 = 0.83 and Nash = 0.78).