RESUMO
Hyperspectral technology offers significant potential for non-invasive monitoring and prediction of morphological parameters in plants. In this study, UV-VIS-NIR-SWIR reflectance hyperspectral data were collected from Nicotiana tabacum L. plants using a spectroradiometer. These plants were grown under different light and gibberellic acid (GA3) concentrations. Through spectroscopy and multivariate analyses, key growth parameters, such as height, leaf area, energy yield, and biomass, were effectively evaluated based on the interaction of light with leaf structures. The shortwave infrared (SWIR) bands, specifically SWIR1 and SWIR2, showed the strongest correlations with these growth parameters. When classifying tobacco plants grown under different GA3 concentrations in greenhouses, artificial intelligence (AI) and machine learning (ML) algorithms were employed, achieving an average accuracy of over 99.1% using neural network (NN) and gradient boosting (GB) algorithms. Among the 34 tested vegetation indices, the photochemical reflectance index (PRI) demonstrated the strongest correlations with all evaluated plant phenotypes. Partial least squares regression (PLSR) models effectively predicted morphological attributes, with R2CV values ranging from 0.81 to 0.87 and RPDP values exceeding 2.09 for all parameters. Based on Pearson's coefficient XYZ interpolations and HVI algorithms, the NIR-SWIR band combination proved the most effective for predicting height and leaf area, while VIS-NIR was optimal for optimal energy yield, and VIS-VIS was best for predicting biomass. To further corroborate these findings, the SWIR bands for certain morphological characteristic wavelengths selected with s-PLS were most significant for SWIR1 and SWIR2, while i-PLS showed a more uniform distribution in VIS-NIR-SWIR bands. Therefore, SWIR hyperspectral bands provide valuable insights into developing alternative bands for remote sensing measurements to estimate plant morphological parameters. These findings underscore the potential of remote sensing technology for rapid, accurate, and non-invasive monitoring within stationary high-throughput phenotyping systems in greenhouses. These insights align with advancements in digital and precision technology, indicating a promising future for research and innovation in this field.
RESUMO
Os levantamentos pedológicos são amplamente utilizados nos mapeamentos de solo por serem métodos confiáveis, no entanto, apesar de tal vantagem são demorados e trabalhosos. Dentro deste contexto, surge o sensoriamento remoto como uma técnica rápida e promissora capaz de auxiliar nos levantamentos, de forma a tornar o processo mais dinâmico. O objetivo deste trabalho foi avaliar a possibilidade de discriminação de cinco classes de solos localizadas no planalto de Apucarana por meio de suas respostas espectrais. Foi estabelecido um grid de 500 m x 500 m em uma área com dimensão de 2500 ha, a partir do qual foram coletadas amostras a 0-0,2 m e 0,8-1,0 m de profundidade. As reflectâncias foram obtidas com o FiedSpec 3 JR, na faixa de 350 a 2500 nm. Equações discriminantes e simulações foram geradas a partir das respostas espectrais das amostras de solo. Das 88 variáveis avaliadas, apenas 8 foram selecionadas pelo procedimento STEPDISC para fazerem parte dos modelos. As equações discriminantes geradas foram testadas, obtendo-se matrizes de confusão, as quais apresentaram acerto acima de 70% para cada classe de solo. Da mesma forma, equações discriminantes simuladas foram geradas, obtendo-se resultados mais significativos para reclassificação quando utilizados dados que fizeram parte da geração do modelo (60%) em comparação com os dados independentes do modelo (40%). As respostas espectrais das amostras de solo empregadas na análise discriminante foram capazes de dar subsídio para separação das cinco classes de solo da área de estudo, comprovando ser uma ferramenta valiosa mesmo em condições de elevada variabilidade pedológica e de atributos como em regiões transicionais.
The pedological surveys are widely used in soil mapping because they are reliable methods, however, although this advantage are time consuming and laborious. Within this context, remote sensing appears as a quickly and promising technique able to assist in the surveys in order to make the process more dynamic. The objective this work was to evaluate the possibility of discrimination of five classes of soils located in the plateau of Apucarana through their spectral responses. Was established a grid of 500 m x 500 m in an area with dimensions of 2500 ha, from which samples were collected at 0 to 0.2 and of 0.8 to 1.0 m deep. The reflectances were obtained with the FiedSpec 3 JR, in the range 350 to 2500 nm. Discriminant equations and simulations were generated from the spectral responses of soil samples. Of the 88 variables evaluated, only 8 were selected by the procedure STEPDISC to be part of the models. The discriminant equations generated were tested, resulting in confusion matrices, which showed accuracy above 70% for each class of soil. Likewise, simulated discriminant equations were generated, obtaining most significant results for reclassification when used data that were part of the model generation (60%) in comparison with the model independent data (40%). The spectral responses of soil samples used in discriminant analysis were able to give support for separation of five classes of soil in the study area, proving to be a valuable tool even in conditions of high pedological and attribute variability as in transitional regions.
