Optimizing cultivated land productivity through integrated fertilization strategies and UAV-assisted management

This study investigates the impact of long-term fertilization modes on crop health and productivity using unmanned aerial vehicle (UAV)-assisted monitoring and a Q-learning-based approach for optimized fertilization strategies. An approximate state-matching Q-learning algorithm was developed to improve the efficiency of agricultural UAVs in fertilization tasks. The spatiotemporal patterns of cultivated land productivity under different fertilization regimes were analyzed in a long-term experiment. An evaluation system for cultivated land health productivity was established. The results indicated that the Q-learning algorithm enabled UAVs to achieve efficient task completion. QLASM outperformed traditional methods on all metrics. QLASM achieved a 92% task completion rate within the target time frame, outperforming deep Q-networks (78%) and rule-based algorithms (65%). Long-term fertilization, especially with organic fertilizer, had a positive impact on soil health and cultivated land productivity. Total irrigation plus organic fertilizer plus green manure reduced irrigation needs by 8–12% by increasing soil organic matter and water holding capacity. The average healthy production capacity was significantly lower in the control group compared to other fertilizer models. Balanced fertilizer application and the long-term use of organic fertilizers are recommended to improve yield and maintain the healthy productivity of cultivated land, which is the basis for sustainable agricultural management.