Agriculture

Due to the intensive cultivation of various crops, the surface soil layer is depleted. This leads to a decrease in fertility, losses of organic matter and nutrients, and an overall decrease in soil health. We aimed to investigate the role of green manure application and organic fertilization on winter fodder barley (Hordeum vulgare L., Zemela cult.) in terms of agronomic and soil parameters. The cultivation was carried out in two fields, the predecessors of which were oats–vetch green manure (field 1) or fallow (field 2). In each field, five treatments were prepared: a control without fertilization, mineral fertilization, vermicompost, mineral fertilizer + vermicompost, and biochar. The green manure incorporation led to a decrease in grain yield of barley by 10.8–20.0% depending on the treatment. A similar tendency was observed for the rest of the studied agronomic parameters (thousand-grain mass, hectolitre weight, ear number, plants per hectare). Additionally, the vermicompost application had the most substantial effect, accounting for a 20.1% increase compared to the control, while the smallest was expressed by biochar—1.6%. Nevertheless, the photosynthesis intensity was higher in treatments after green manure. The microbiome’s activity was boosted in the vermicompost treatments, while amino acids, carboxylic acids, and polymers were the most fully metabolised compounds by the soil communities. In conclusion, the type of predecessor influenced mainly grain protein, carotenoids, and chlorophyll contents, as well as microbial activities, respiration, and dehydrogenase, while the fertilization impacted primarily on soil water and organic content, total soil N, and photosynthetic pigments of barley plants.

Accurate profiling of deep (20–300 cm) soil moisture is crucial for precision irrigation but remains technically challenging and costly at operational scales. We systematically benchmark eight regression algorithms—including linear regression, Lasso, Ridge, elastic net, support vector regression, multi-layer perceptron (MLP), random forest (RF), and gradient boosting trees (GBDT)—that use easily accessible inputs of 0–20 cm surface soil moisture (SSM) and ten meteorological variables to non-invasively infer soil moisture at fourteen 20 cm layers. Data from a typical agricultural site in Wenxi, Shanxi (2020–2022), were divided into training and testing datasets based on temporal order (2020–2021 for training, 2022 for testing) and standardized prior to modeling. Across depths, non-linear ensemble models significantly outperform linear baselines. Ridge Regression achieves the highest accuracy at 0–20 cm, SVR performs best at 20–40 cm, and MLP yields consistently optimal performance across deep layers from 60 cm to 300 cm (R² = 0.895–0.978, KGE = 0.826–0.985). Although ensemble models like RF and GBDT exhibit strong fitting ability, their generalization performance under temporal validation is relatively limited. Model interpretability combining SHAP, PDP, and ALE shows that surface soil moisture is the dominant predictor across all depths, with a clear attenuation trend and a critical transition zone between 160 and 200 cm. Precipitation and humidity primarily drive shallow to mid-layers (20–140 cm), whereas temperature variables gain relative importance in deeper profiles (200–300 cm). ALE analysis eliminates feature correlation biases while maintaining high predictive accuracy, confirming surface-to-deep information transmission mechanisms. We propose a depth-adaptive modeling strategy by assigning the best-performing model at each soil layer, enabling practical non-invasive deep soil moisture prediction for precision irrigation and water resource management.

Accurate numerical simulations of soil-tire interactions are essential for optimizing agricultural machinery to minimize soil compaction and enhance crop yield. This study developed and compared two approaches for identifying and validating parameters of a LS-Dyna soil model. The laboratory-based approach derives parameters from triaxial, consolidation, and cone penetrometer tests (CPT), while the optimization-based method refines them using in-situ CPT data via LS-OPT to better capture field variability. Simulations employing Multi-Material Arbitrary Lagrangian–Eulerian (MM-ALE), Smoothed Particle Hydrodynamics (SPH), and Hybrid-SPH methods demonstrate that Hybrid-SPH achieves the optimal balance of accuracy (2% error post-optimization) and efficiency (14-h runtime vs. 22 h for SPH). Optimized parameters improve soil–tire interaction predictions, including net traction and tire sinkage across slip ratios from −10% to 30% (e.g., sinkage of 12.5 mm vs. 11.1 mm experimental at 30% slip, with overall mean-absolute percentage error (MAPE) reduced to 3.5% for sinkage and 4.2% for traction) and rut profiles, outperforming lab-derived values. This framework highlights the value of field-calibrated optimization for sustainable agriculture, offering a cost-effective alternative to field trials for designing low-compaction equipment and reducing yield losses from soil degradation. While sandy loam soil at 0.4% moisture content was used in this study, future extensions to different soil types with varied moisture are recommended.