The rapid increase in fertilizer use has led to the degradation of soil quality, nutrient imbalances, reduced biodiversity, and soil compaction. To address these challenges, hybrid soybeans with efficient biological nitrogen fixation capabilities and broad environmental adaptability were selected as green manure to reduce fertilizer application, thereby improving soil fertility and structure. This study utilized the varieties “Forage Soybean S001” (S001), “Neinong S002 Forage Soybean” (S002), “Mengnong S003 Forage Soybean” (S003), “Mengnong S004 Forage Soybean” (S004), “Mengnong S005 Forage Soybean” (S005), and “Mengnong S006 Forage Soybean” (S006) as green manure materials. The clean tillage (CK) treatment served as the control, ensuring a residue-free soil surface while maintaining consistent practices in soil preparation, irrigation, and field management across all treatments. Field planting of green manure and subsequent crops was conducted at the M-Grass Ecology and Environment (Group) Company’s experimental site in Hohhot in early May of 2023 and 2024. The plots each measured 20 m², with three replications arranged in a randomized block design. A combination of field experiments and laboratory analyses was utilized to investigate the effects of incorporating various hybrid soybean varieties as green manure on soil nutrient levels, soil enzyme activity, soil microbial communities, and the subsequent growth of oats. The results indicated that incorporating various hybrid soybean varieties as green manure into the soil significantly improved soil nutrient levels and enzyme activity. The diversity and richness of soil bacterial communities increased significantly, accompanied by alterations in community structure and composition. These changes enhanced soil fertility and optimized the microbial community structure, promoting the growth of subsequent crops. Among all the treatments, S001 and S004 were particularly effective in enhancing the soil environment, indicating their potential as superior green manure resources for broader application.

Accurate alfalfa biomass prediction is crucial for pasture management and sustainable livestock production. However, traditional methods often perform poorly under complex field conditions. To address the limited prediction accuracy of traditional methods under complex planting environments, this study proposes an alfalfa biomass prediction method combining multispectral and LiDAR data with ensemble learning model. Based on the multispectral images acquired by unmanned aerial vehicle (UAV) and airborne LiDAR data, the spectral features, three-dimensional structural features, and their interaction features are systematically extracted at the quadrat scale, and a high-quality modeling dataset is constructed by feature selection. Secondly, an ensemble model for alfalfa biomass prediction was constructed, which was composed of random forest, extra trees, and histogram gradient boosting. After model training, the coefficient of determination (R²) of the integrated model on the test set reached 0.813, and the root mean square error (RMSE) and mean absolute error (MAE) were 0.178 kg m⁻² and 0.146 kg m⁻², which were significantly better than those of similar single models. Under feature combinations, the fusion model was better than that of spectral indices only (R² = 0.773) and LiDAR traits only (R² = 0.576), and the model achieved the highest accuracy from bud emergence to early flowering (R² = 0.917). The overall prediction error of the model was approximately normal distribution, and the absolute error of more than 65% of the samples was less than 0.2. However, there was still a trend of underestimation in the high biomass interval. This research showed that the multimodal data fusion and ensemble learning method could achieve high-precision prediction of alfalfa biomass, which provided reliable technical support for pasture resources monitoring and precision agriculture management.