The oxidative degradation of anthocyanins in red wine was investigated under controlled conditions using hydroxyl radicals generated in the presence of Cu (II) as a catalyst. A full factorial experimental design with 23 replicates was used to evaluate the effects of hydrogen peroxide concentration, catalyst dosage, and reaction temperature on anthocyanin degradation over a fixed time. Statistical analysis (ANOVA and multiple regression) showed that all three variables and the main interactions significantly affected anthocyanin loss, with temperature identified as the most influential factor. The combined effects were described by a first-order polynomial model. The activation energies for degradation ranged from 56.62 kJ/mol (cyanidin-3-O-glucoside) to 40.58 kJ/mol (peonidin-3-O-glucoside acetate). Increasing the temperature from 30 °C to 40 °C accelerated the degradation kinetics, almost doubled the rate constants and shortened the half-life of the pigments. At 40 °C, the half-lives ranged from 62.3 min to 154.0 min, depending on the anthocyanin structure. These results contribute to a deeper understanding of the stability of anthocyanins in red wine under oxidative stress and provide insights into the chemical behavior of derived pigments. The results are of practical importance for both oenology and viticulture and support efforts to improve the color stability of wine and extend the shelf life of grape-based products.

This study focused on optimizing key extraction parameters (ethanol concentration, temperature, and time) to enhance the extraction of hydroxycinnamic acids from A. graveolens leaves, applying a replicated 2³ full factorial design. The experimental results demonstrated that extraction conditions significantly influenced the yield of hydroxycinnamic acids. The optimum conditions were 50% ethanol for 80 min at 50 °C, yielding 103.75 µg/g of chlorogenic acid (ChA), 6.05 µg/g of ferulic acid (FA), and 2.19 µg/g of sinapic acid (SA). Therefore, the extract obtained with 50% ethanol showed the highest levels of polyphenols, flavonoids, and antioxidant potential, highlighting its suitability for use in both food-related products and pharmaceutical formulations. Additionally, the applicability of different mathematical models (unsteady-state diffusion, parabolic diffusion, and power law models, as well as Ponomaryov’s and Elovich’s equations) to describe and better understand the kinetics of hydroxycinnamic acid extraction from dill leaves was evaluated. The fit of each kinetic model to the experimental data was assessed using the root mean square error and the coefficient of determination. Among the five kinetic models, the unsteady-state diffusion model and Ponomaryov’s equation provided the best fit, exhibiting the highest accuracy. The activation energy for the extraction process was determined to be 5.85, 6.46, and 7.59 kJ/mol for ChA, FA, and SA, respectively. The extraction of hydroxycinnamic acids from dill leaves was found to be endothermic, irreversible, and spontaneous.

Background/Objectives: Accurate land cover mapping from multispectral Sentinel-2 imagery is fundamental for environmental monitoring, efficient natural resource management, and spatial planning. While deep learning has become the dominant approach for semantic segmentation, the combined impact of spectral input selection and network architecture on cross-regional robustness remains insufficiently explored. This study systematically investigates multispectral land cover segmentation in Serbia and evaluates its transferability to Western Balkan regions using a structured experimental framework. Methods: A comprehensive band-combination ablation analysis (3–10 spectral bands and index-only inputs) was first conducted using Attention U-Net, followed by a comparative evaluation of representative convolutional and transformer-based architectures, including ResNet-UNet-50, ConvNeXt-UNet, DeepLabV3+ (ResNet-50), and DINOv2-S/14. Model performance is evaluated on an internal Serbian test split (Test SR), an external Serbian dataset (Ext SR), and a cross-regional Balkan dataset (Ext WB). Results: The results demonstrate that compact multispectral configurations (6–9 bands) provide the most stable performance, achieving mIoU values of approximately 0.72–0.74 under in-domain evaluation and remaining robust under external testing. The inclusion of near-infrared and shortwave infrared bands proved critical for effective land cover discrimination, whereas increasing spectral dimensionality beyond this range did not yield systematic improvements in external robustness. Notably, the magnitude of performance degradation under pronounced geographic domain shift exceeds the performance differences observed between architectures under in-domain conditions, indicating that distribution shift exerts a stronger influence on segmentation accuracy than model choice alone. Class-wise analysis revealed agricultural areas as the most domain-sensitive category, while Shapley-based explainability analysis provides additional insight into class-specific spectral dependencies and their role in generalization behavior. Conclusions: Although transformer-based models demonstrated competitive robustness, attention-enhanced convolutional architectures achieved comparable stability across evaluation scenarios. Overall, the findings emphasize the importance of balanced spectral design, class-aware robustness analysis, and explicit out-of-domain evaluation for developing transferable land cover segmentation models in remote sensing applications.