水稻遥感监测预测技术体系

Remote Sensing-Based Monitoring and Prediction Technology System for Rice

关键词:  水稻  遥感制图  生育期提取  长势监测  产量预测  农业气象灾害监测

Key words:  Key words: rice  remote sensing mapping  phenological stage extraction  crop growth monitoring  yield prediction  agro-meteorological disaster monitoring

摘要:

【目的】水稻是全球最重要的粮食作物之一，水稻生产对保障全球及我国粮食安全具有关键意义。针对水稻遥感制图、生育期提取、长势监测、产量预测及农业气象灾害监测五个核心环节，对水稻遥感监测预测技术体系进行系统梳理，以期为大尺度水稻精准管理与粮食安全决策提供理论与方法支撑。【方法】采用文献综述方法，重点分析基于物候、机器学习和时序匹配的水稻遥感制图方法；基于经验阈值、特征点提取、形状匹配和机器学习的水稻生育期提取方法；基于植被指数对比的水稻定性长势监测方法和基于经验模型、机理模型、遥感数据与作物生长模型同化的水稻定量长势监测方法；基于经验模型、光能利用率模型和遥感数据与作物生长模型同化的水稻产量预测方法；以及基于距平、监测范式和遥感数据与作物生长模型同化的水稻农业气象灾害监测方法。【结果】（1）水稻遥感制图与生育期提取正由单一阈值法向多源数据协同、机器学习和深度学习方法发展，显著提升了复杂种植条件下的识别精度。（2）水稻长势监测与产量预测正由经验统计模型向数据驱动和遥感—作物模型同化方法演进，提高了大尺度长势表征与估产可靠性。（3）水稻农业气象灾害监测正由单变量距平分析向多源数据与机理模型耦合评估转变，增强了灾害识别与产量损失评估能力。【结论】现阶段已初步形成覆盖水稻全生命周期的遥感监测预测技术体系，并在不断完善。未来需重点加强高质量样本数据自动化获取、多源异构数据深度协同、模型跨区域泛化能力与稳定性验证，以满足国家及区域尺度水稻全生命周期精准监测与智慧预测的业务化需求。

Abstract:

Abstract: [Purpose] Rice is one of the most important staple food crops worldwide, and rice production plays a critical role in ensuring both global and national food security. Focusing on five key components, namely rice remote sensing mapping, phenological stage extraction, crop growth monitoring, yield prediction, and agro-meteorological disaster monitoring, this study systematically reviews the technical framework of remote sensing-based rice monitoring and prediction, with the aim of providing theoretical and methodological support for large-scale precision rice management and food security decision-making. [Method] A literature review approach was employed to comprehensively analyze rice remote sensing mapping methods based on phenology, machine learning, and time-series matching; rice phenological stage extraction methods based on empirical thresholds, feature-point extraction, shape matching, and machine learning; qualitative rice growth monitoring based on vegetation index comparison, as well as quantitative rice growth monitoring methods based on empirical models, mechanistic models, and the assimilation of remote sensing data with crop growth models; rice yield prediction methods based on empirical models, light use efficiency models, and the assimilation of remote sensing data with crop growth models; and rice agro-meteorological disaster monitoring methods based on anomaly analysis, monitoring paradigms, and the assimilation of remote sensing data with crop growth models. [Result] (1) Rice remote sensing mapping and phenological stage extraction shifted from single-threshold methods toward multi-source data synergy, machine learning, and deep learning, which significantly improved identification accuracy under complex planting conditions. (2) Rice growth monitoring and yield prediction evolved from empirical statistical models toward data-driven methods and remote sensing–crop model data assimilation, thereby improving the reliability of large-scale growth characterization and yield estimation. (3) Rice agro-meteorological disaster monitoring changed from single-variable anomaly analysis to integrated assessment based on multi-source data and mechanistic models, enhancing the ability to identify disaster processes and estimate yield losses. [Conclusion] A remote sensing-based monitoring and prediction framework covering the entire rice life cycle is preliminarily established and continues to be refined. Future research should place greater emphasis on the automated acquisition of high-quality sample data, the deep integration of multi-source heterogeneous data, and the validation of cross-regional model generalization capability and stability, in order to meet the operational demands of precise full-life-cycle rice monitoring and intelligent prediction at both national and regional scales.