The objective was to verify the performance of spectral techniques as well as validation models in the prediction of nitrogen, total organic carbon, and humic fractions under different cultivation conditions. Chemical analyses for the determination of nitrate, total nitrogen, total organic carbon, and the chemical fractionation of soil organic matter were performed, as well as spectral analyses by Vis-NIR-SWIR and X-ray fluorescence. The results of the spectroscopy were processed using RStudio v. 4.1.3, and PLSR and support vector machine learning algorithms were applied to validate the models. The Vis-NIR-SWIR and XRF spectroscopic techniques showed high performance and are indicated for the prediction of nitrogen, total organic carbon, and humic fractions in Ferralsols of medium sandy texture. However, it is important to highlight that each technique has its own characteristic mechanism of action: Vis-NIR-SWIR detects the element based on harmonic tones, while XRF is based on the atomic number of the element or elemental association. The PLSR and SVM models showed excellent validation results, allowing them to fit the experimental data, emphasizing that they are different statistical methods.

Rising global temperatures pose a significant threat to agricultural productivity, making the early detection of plant stress crucial for minimizing crop losses. While hyperspectral remote sensing is a powerful tool for monitoring plant health, the precise spectral regions and most effective machine learning models for detecting thermal stress remain an open research question. This study presents a robust framework that utilizes eight state-of-the-art machine learning algorithms to classify the reflectance response of thermal-induced stress in two cultivars of bean plants. Our controlled experiment measured hyperspectral data across two growth stages and three stress conditions (pre-stress, during stress, and post-stress) using a spectroradiometer. The results demonstrate the high performance of several algorithms, with the Artificial Neural Network (ANN) achieving an impressive 99.4% overall accuracy. A key contribution of this work is the identification of the most contributory spectral ranges for thermal stress discrimination: the green region (530–570 nm) and the red-edge region (700–710 nm). This framework is a feasible and effective tool for modelling the hyperspectral response of thermal-stressed bean plants and provides critical guidance for future research on stress-specific spectral indices.

This work aimed to develop suitable predictive models for ammonium, nitrate, total nitrogen, total organic carbon and soil humic fractions, for Ferralsols, using Vis-NIR-SWIR, MIR and X-ray fluorescence spectroscopic techniques in conjunction with machine learning algorithms, Cubist, PLSR, Random Forest and Support Vector Machine. Chemical analyzes were carried out to determine nitrate, total nitrogen, total organic carbon and chemical fractionation of soil organic matter, as well as spectral analyzes using Vis-NIR-SWIR spectroscopy, MIR and X-ray fluorescence. The spectroscopy results were processed using RStudio v. 4.1.3, applying Cusbist, PLSR, Random Forest and Support Vector Machine machine learning algorithms to create predictive models and describe spectral curves and Pearson correlation. Of the prediction models developed for nitrogen, total organic carbon and humic fractions, the PLSR and Support Vector Machine algorithms presented the best predictive performances. The descriptive analysis of the spectra identified the main absorption bands and the location of the bands sensitive to the attributes of interest. The correlation analysis proposed that the use of Vis-NIR-SWIR, MIR and XRF spectroscopic techniques were effective in predicting the contents of nitrogen, total organic carbon and humic fractions in soil with a medium sandy texture. However, it is important to highlight that each technique has its characteristic mechanism of action, Vis-NIR-SWIR and MIR detect the element based on overtones and fundamental tones, while XRF is based on the atomic number of the elements or elemental association.