This study aimed to predict soil properties using visible–near infrared (VIS-NIR) spectroscopy combined with partial least square regression (PLSR) modeling. Special emphasis was given to evaluating effect of pre-processing methods on prediction accuracy and important wavelengths. A total of 114 samples were collected and involved in chemical and spectral analyzes. PLSR model of each soil property was calibrated for all pre-processing methods using all samples, and leave-one-out cross-validation was used to make comparisons between them. Then, PLSR model of each best pre-processing method was calibrated using a 75% of all samples and correspondingly validated with the remaining a 25%. Model accuracy was evaluated based on coefficient of determination (R²), root mean-squared errors (RMSE), and residual prediction deviations (RPD). The high correlation coefficients were found between the tested soil properties and reflectance spectra. The pre-processing methods considerably improved prediction accuracy and filtering methods outperformed linearization methods, and the latter outperformed normalization methods. The performance of cross-validation, calibration and independent validation was similar. An excellent prediction (RPD>2.5) model was obtained for soil organic carbon (SOC) and calcium-carbonate (CaCO₃), good quantitative (2.0< RPD<2.5) prediction for sand, silt, and clay, fair prediction (1.4< RPD<1.8) for pH, and poor prediction (1.0< RPD<1.4) for hygroscopic water content (WC). Important wavelengths varied depending on soil property, but some wavelengths were common. This study can be a precursor to building a pioneering soil spectral database, calibrating satellite data, and hyperspectral image mapping of soils as well as digital soil mapping, environmental, and erosion modeling in the Caucasus Mountains.

Visible-near infrared (Vis-NIR) and mid-infrared (MIR) spectroscopy are increasingly being used for the fast determination of soil properties. The aim of this study was (i) to test the use of MIR spectra (Agilent 4300 FTIR Handheld spectrometer) for the prediction of soil properties and (ii) to compare the prediction performances of MIR spectra and Vis-NIR (ASD FieldSpecPro) spectra; the Vis-NIR data were adopted from a previous study. Both the MIR and Vis-NIR spectra were coupled with partial least squares regression, different pre-processing techniques, and the same 114 soil samples, collected from the agricultural land located between boreal forests and semi-arid steppe belts (Kastanozems). The prediction accuracy (R² = 0.70–0.99) of both techniques was similar for most of the soil properties assessed. However, (i) the MIR spectra were superior for estimating CaCO₃, pH, SOC, sand, Ca, Mg, Cd, Fe, Mn, and Pb. (ii) The Vis-NIR spectra provided better results for silt, clay, and K, and (iii) the hygroscopic water content, Cu, P, and Zn were poorly predicted by both methods. The importance of the applied pre-processing techniques was evident, and among others, the first derivative spectra produced more reliable predictions for 11 of the 17 soil properties analyzed. The spectrally active CaCO₃ had a dominant contribution in the MIR predictions of spectrally inactive soil properties, followed by SOC and Fe, whereas particle sizes and hygroscopic water content appeared as confounding factors. The estimation of spectrally inactive soil properties was carried out by considering their secondary correlation with carbonates, clay minerals, and organic matter. The soil information covered by the MIR spectra was more meaningful than that covered by the Vis-NIR spectra, while both displayed similar capturing mechanisms. Both the MIR and Vis-NIR spectra seized the same soil information, which may appear as a limiting factor for combining both spectral ranges. The interpretation of MIR spectra allowed us to differentiate non-carbonated and carbonated samples corresponding to carbonate leaching and accumulation zones associated with topography and land use. The prediction capability of the MIR spectra and the content of nutrient elements was highly related to soil-forming factors in the study area, which highlights the importance of local (site-specific) prediction models.

Soil spectroscopy is a promising alternative to evaluate and monitor soil and water quality, particularly in mountainous agricultural lands characterized by intense degradation and limited soil tests reports; a few studies have evaluated the feasibility of VIS-NIR spectroscopy to predict Mehlich 3 (M3) extractable nutrients. This study aimed to (i) examine the potential of VIS-NIR spectroscopy in combination with partial least squares regression to predict M3-extractable elements (Ca, K, Mg, P, Fe, Cd, Cu, Mn, Pb, and Zn) and basic soil properties (clay, silt, sand, CaCO₃, pH, and soil organic carbon-SOC), (ii) find optimal pre-processing techniques, and (iii) determine primary prediction mechanisms for spectrally featureless soil properties. Topsoil samples were collected from a representative area (114 samples from 525 ha) located in the mountainous region of NW Azerbaijan. A series of pre-processing steps and transformations were applied to the spectral data, and the models were calibrated and evaluated based on the coefficient of determination (R²), root mean square error (RMSE), and the residual prediction deviation (RPD). The leave-one-out cross-validated predictions showed that the first derivative spectra produce higher prediction accuracies (R² = 0.51–0.91; RPD = 1.20–2.29) for most soil properties. The evaluation of the model performance with optimal pre-processing techniques revealed that both calibration and validation models produce considerable differences in RPD values associated with sample size and the random partition of the calibration or validation subsets. The prediction models were excellent or very good (RPD > 2.0) for CaCO_3, SOC, sand, silt, Ca, and Pb, good or fair (1.4 < RPD < 2.0) for clay, K, Cd, pH, Fe, Mn, and Cu, and poor (1.0 < RPD < 1.4) for Mg, P, and Zn. Principal component and correlation, stepwise regression analysis, and variable importance in projection procedures allowed to elucidate the underlying prediction mechanisms. Unlike the previous studies, the spectral estimations of pH, Ca, Mg, P, Fe, Pb, and Cd concentrations were linked to their correlation with CaCO₃ rather than soil organic matter, whereas Mg and P concentrations were also connected to Fe-oxides. Soil particle sizes contributed to predicting K concentration but confounded the prediction of P and Zn concentration. The weaker correlations of Mn, Cu or Zn with CaCO₃, particle sizes, SOC, Fe, and spectral data yielded to their lower prediction accuracy. The major prediction mechanisms for M3-extractable elements relied on their relations with CaCO₃, pH, clay content and mineralogy, and exchangeable cations in the context of their association with land use. The results can be used in mountain lands to evaluate and control the effect of management on soil quality indices and land degradation neutrality. Further studies are needed to develop most advantageous sampling schemes and modeling.

The increasing demand for precise and dependable models has led to the development of both sensors and statistical algorithms. However, numerous studies have demonstrated that model performance is highly dependent on a range of environmental factors, such as spatio-temporal fluctuations of moisture, sensor type, sample variability, preprocessing methods, and model selection. These factors can impact prediction results, leading to erroneous comparisons across lab, field, or imaging models. Samples for this study were collected from a tailing settling basin of a porphyry copper deposit near Erdenet, Mongolia. The database contains lab and field spectra and hyperspectral imagery from a HySpex imaging sensor. In this study we propose a workflow that includes a simulation that yields an appropriate regression threshold while addressing data-driven uncertainty. The workflow consists of two regression models and five classification models at different scales for quantitative geochemical, mineralogical, and textural prediction of tailing samples. Each model is compared to the acquisition space's performance potential. Acceptable R² values for regression models are 0.58 for laboratory, 0.40 for field, and 0.31 for hyperspectral airborne data. Results of this study are not limited to tailing samples but can be applied on other fields of research such as geology, pedology or agriculture.