AI systems can provide tailored solutions for different application scenarios. The ideal model for integrated agricultural non-point source pollution (ANPSP) monitoring and prevention-control from the perspective of temporal-spatial-process is presented in cover figure. This model comprehensively implements the 4R strategy to achieve the reduction, interception, removal, tracing and recycling of pollutants throughout the entire process. AI employs remote sensing, sensor technology and image recognition to capture terrestrial pollutant data. After preprocessing these multisource data sets, it extracts critical features for model training, using root mean square error as a validation metric. Real-time monitoring via Internet of Things supported by algorithms like support vector machine feeds measured data into trained models for analysis. Through this closed-loop workflow, AI enables precise monitoring, scientific assessment, and intelligent ANPSP early warnings. This technology provides crucial technical support for watershed pollution control and the reduction of fertilizer/pesticide inputs. Developing portable and rapid pollution source detection devices could also help to lower the adoption threshold for small- and medium-sized farming operations.
(Jing TAO, Haibing XIAO, Feng LIU, Yanfang FENG, Shunchang YANG, Lei ZHOU, Yonghong WU. Agricultural non-point source pollution and agricultural green development: status, challenges and prospects. Front. Agr. Sci. Eng., 2026, 13(2): 25650 DOI: 10.15302/J-FASE-2025650)
Download cover
