Machine learning-based geospatial applications offer unique opportunities for environmental monitoring due to domains and scales adaptability and computational efficiency. However, the specificity of environmental data introduces biases in straightforward implementations. We identify a streamlined pipeline to enhance model accuracy, addressing issues like imbalanced data, spatial autocorrelation, prediction errors, and the nuances of model generalization and uncertainty estimation. We examine tools and techniques for overcoming these obstacles and provide insights into future geospatial AI developments. A big picture of the field is completed from advances in data processing in general, including the demands of industry-related solutions relevant to outcomes of applied sciences.

Redox moieties, which are present in the molecular backbone of humic substances (HS), govern their antioxidant properties. We hypothesized that a directed modification of the humic backbone via incorporation of redox moieties with known redox properties might provide an efficient tool for tuning up antioxidant properties of HS. In this work, hydroquinonoid and hydronaphthoquinonoid centres were used, which possess very different redox characteristics. They were incorporated into the structure of coal (leonardite) humic acids CHA) and peat fulvic acids (PFA). For this goal, an oxidative copolymerization of phenols was used. The latter was induced via oxidation of hydroquinones and hydroxynapjtaquinones with a use of Fenton’s reagent. The structure of the obtained products was characterized using NMR and FTIR spectroscopy. H/D labelling coupled to FT ICR mass spectrometry analysis was applied for identification of the reaction products as a tool for surmising on reaction mechanism. It was shown that covalent -C-C- bond were formed between the incorporated redox centers and aromatic core of HS. The parent humic acids and their naphthoquinonoid derivatives have demonstrated high accepting capacity. At the same time, fulvic acids and their hydroquinonoid derivatives have possessed both high donor and high antioxidant capacities. The kinetic studies have demonstrated that both humic acids and their derivatives showed much slower kinetics of antioxidant reactions as compared to fulvic acids. The obtained results show, firstly, substantial difference in redox and antioxidant properties of the humic and fulvic acids, and, secondly, they can serve as an experimental evidence that directed chemical modification of humic substances can be used to tune and control antioxidant properties of natural HS.

Accurate information about growing crops allows for regulating the internal stocks of agricultural products and drawing strategies for negotiating agricultural commodities on financial markets. Machine learning methods are widely implemented for crop type recognition and classification based on satellite images. However, field classification is complicated by class imbalance and aggregation of pixel-wise into field-wise forecasting. We propose here a Bayesian methodology for the aggregation of classification results. We report the comparison of class balancing techniques. We also report the comparison of classical machine learning methods and the U-Net convolutional neural network for classifying crops using a single satellite image. The best result for single-satellite-image crop classification was achieved with an overall accuracy of 77.4% and a Macro F1-score of 0.66. Bayesian aggregation for field-wise classification improved the result obtained using majority voting aggregation by 1.5%. We demonstrate here that the Bayesian aggregation approach outperforms the majority voting and averaging strategy in overall accuracy for the single-image crop classification task.