Quantifying the organic carbon content of soil over large areas is essential for characterising the soil and the effects of its management. However, analytical methods can be laborious and costly. Reflectance spectroscopy is a well-established and widespread method for estimating the chemical-element content of soils. The aim of this study was to estimate the soil organic carbon (SOC) content using hyperspectral remote sensing. The data were from soils from two localities in the semi-arid region of Brazil. The spectral reflectance factors of the collected soil samples were recorded at wavelengths ranging from 350–2500 nm. Pre-processing techniques were employed, including normalisation, Savitzky–Golay smoothing and first-order derivative analysis. The data (n = 65) were examined both jointly and by soil class, and subdivided into calibration and validation to independently assess the performance of the linear methods. Two multivariate models were calibrated using the SOC content estimated in the laboratory by principal component regression (PCR) and partial least squares regression (PLSR). The study showed significant success in predicting the SOC with transformed and untransformed data, yielding acceptable-to-excellent predictions (with the performance-to-deviation ratio ranging from 1.40–3.38). In general, the spectral reflectance factors of the soils decreased with the increasing levels of SOC. PLSR was considered more robust than PCR, whose wavelengths from 354 to 380 nm, 1685, 1718, 1757, 1840, 1876, 1880, 2018, 2037, 2042, and 2057 nm showed outstanding absorption characteristics between the predicted models. The results found here are of significant practical value for estimating SOC in Neosols and Cambisols in the semi-arid region of Brazil using VIS-NIR-SWIR spectroscopy.
The objective of this study was to evaluate the assimilated contents of mineral elements, yield characteristics and fruit quality of ‘Goldex’ yellow melon fertigated with different doses of nitrogen (N) and potassium (K) in two years of cultivation in the region of Apodi Plateau-CE. The design was randomized blocks in a 4x3 factorial scheme with four N doses (0%, 50%, 100%, 150% of the recommended dose) and three K doses (0%, 100%, 150% of the recommended dose). In the shoots of the melon plant, the highest levels of N (N Fert.) and K (K Fert.) fertilization promoted an increase in the assimilated N content up to 48 DAT and a reduction in the assimilated K content at the end of the cycle (62 DAT). The maximum shoot dry mass production was 3.23 t ha−1 with a fertilization of 150 and 100 kg ha−1 of N and K, respectively. The average marketable fruit yield obtained in the two years of cultivation was 34.68 t ha−1, reaching a maximum value of 40.28 t ha−1 under 150 kg ha−1 N and K fertilization. Total soluble solids increased as a function of the increase in N Fert. and K Fert., with a maximum value of 8.07 °Brix. The models indicated that the K Fert. factor was decisive in the increase of this variable. In future studies it is necessary to increase the concentrations of the two factors in order to optimize the maximum viability of yield and °Brix.
6 June 2023 in Revista Brasileira de Ciência do Solo
Understanding organic carbon and predominant humic fractions in the soil allows contributes to soil quality management. Conventional fractionation techniques require time, excessive sampling, and high maintenance costs. In this study, predictive models for organic carbon in humic substances (HS) were evaluated using hyperspectral data as an alternative to chemical fractionation and quantification by wet digestion. Twenty-nine samples of Neossolos Flúvicos (Fluvents) - A1, and 36 samples of Cambissolos (Inceptisols) - A2 were used. The samples were also analyzed jointly, creating a third sample group - A1&A2. Untransformed spectral reflectance factors were obtained using the FieldSpec Pro FR 3 hyperspectral sensor (350–2500 nm). Pre-processing techniques were employed, including Savitzky–Golay smoothing and first- and second-order derivative analysis. After selecting variables using the Backward method, which removes spectral variables that are not statistically significant for the regression. Estimation models were built by Principal Components Regression (PCR) and Partial Least Squares Regression (PLSR). The spectral data were evaluated individually for soil classes A1 and A2, and jointly for A1&A2. The PLSR was more efficient than PCR, especially for the estimation models that used the first derivative of reflectance employing the three sample groups. For samples of A1, the best estimate was seen for humic acid (RPD = 6.09) and humin (RPD = 2.38); for A2, the best models estimated the OC in fulvic acid (RPD = 2.35) and humin (RPD = 2.51); and for the joint spectral data (A1&A2), the prediction was robust for humin only (RPD = 2.01). The most representative wavelengths were observed using the first derivative with PLSR and PCR, centred on the region between 1600 and 1800 nm. The first-derivative of reflectance calculated more-robust predictive models using PLSR than PCR. The best predictions occurred for organic carbon associated with humic acid in Neossolos Flúvicos, with fulvic acid in Cambissolos, and with humin in both soil classes.
Plant pathologies significantly jeopardise global food security, necessitating the development of prompt and precise diagnostic methods. This study employs advanced deep learning techniques to evaluate the performance of nine convolutional neural networks (CNNs) in identifying a spectrum of phytosanitary issues affecting the foliage of Solanum lycopersicum (tomato). Ten thousand RGB images of leaf tissue were subsampled in training (64%), validation (16%), and test (20%) sets to rank the most suitable CNNs in expediting the diagnosis of plant disease. The study assessed the performance of eight well-known networks under identical hyperparameter conditions. Additionally, it introduced the GamaNNet architecture, a custom-designed model optimised for superior performance on this specific type of dataset. The investigational results were most promising for the innovative GamaNNet and ResNet-152, which both exhibited a 91% accuracy rate, as evidenced by their confusion matrices, ROC curves, and AUC metrics. In comparison, LeNet-5 and ResNet-50 demonstrated lower assertiveness, attaining accuracies of 74% and 69%, respectively. GoogLeNet and Inception-v3 emerged as the frontrunners, displaying diagnostic preeminence, achieving an average F1-score of 97%. Identifying such pathologies as Early Blight, Late Blight, Corynespora Leaf Spot, and Septoria Leaf Spot posed the most significant challenge for this class of problem.