农田遥感监测机理与生态过程关键参数反演

Taking typical farmland soil-vegetation binary systems in China as research objects, remote sensing monitoring of farmland ecological process changes as mainline, this project mainly focuses on the key problems to be solved in farmland ecological process parameters inversion, and the contents of this project include: (1) through response characteristic analysis and field experiment of farmland soil-vegetation in different spectral bands of electromagnetic waves, study on mechanisms of interaction between electromagnetic waves and farmland soil-vegetation and explore the surface scattering mechanisms from rough surface of bare soil and volume scattering mechanisms of farmland covered by vegetation in microwave domain; (2) using multi-source and multi-temporal remote sensed data, study on parameters inversion methods, solve key technologies of the retrieval of land surface parameters which characterize condition of ecological processes in farmland, including soil temperature, soil moisture, leaf area, crop chlorophyll content, crop canopy temperature, and increase the accuracy of parameters inversion; (3) to achieve the dynamic monitoring requirement, carry out time series analysis and modeling of farmland ecological process parameters; taking remote sensing data products as assimilation variables, study on combining remote sensed data and ecological process models through data assimilation, propose and improve farmland crop monitoring models over the whole growing season; (4) through farmland in-site experiment and other means, validate and improve the methods and model proposed in this project and then conduct a field demonstration of application in chosen study area. The research results of the project have a theoretical significance and application value for the remote sensing crop yield estimation and crop disaster monitoring.

本项目以我国农田土壤-植被二元体系为研究对象，以农田生态过程变化遥感监测为主线，从农田生态过程参数反演亟待解决的关键问题入手，重点开展以下研究：①通过农田土壤-植被在不同谱段电磁波响应特性分析与观测试验，开展电磁波与农田土壤-植被相互作用机理研究，探索裸露地表（农田）微波面散射机理和植被覆盖农田微波体散射机理； ②综合利用多源多时相遥感数据，进行参数反演方法研究，解决土壤温度、土壤水分、作物叶面积、作物叶绿素含量、作物冠层温度等表征农田生态过程状态参数定量反演关键技术，提高关键参数反演精度；③针对动态监测要求，开展农田生态过程参数时间序列分析与建模，以遥感反演数据产品为同化变量，开展遥感信息与生态过程模型数据同化研究，提出并发展适应作物全生长期的农田监测模型；④通过农田试验等手段，进行方法与模型验证、完善与应用示范。上述项目成果对于农作物遥感估产与作物灾害监测具有理论意义与应用价值。

针对目前精准农业发展中对农田定量遥感应用的迫切需求，本项目以农田生态过程变化遥感监测为主线，从机理、方法、技术与应用四个方面深入探索与研究了农田遥感监测机理与生态过程关键参数反演涉及的关键问题。在机理研究方面，分别开展了农田植被-土壤二元体系在不同电磁波谱段的响应特性观测试验与土壤不同含水量的光谱响应观测试验，形成了涵盖可见光-近红外、热红外和微波波段的复杂农田地表与电磁波作用的遥感模型，包括植被-土壤混合光谱模型与全极化SAR数据的植被覆盖地表土壤水分反演模型。在总结与分析已有建模方法的基础上，提出了适用于不同入射光条件、不同生育期、不同作物类型的冠层二向性反射统一模型，实现了几何光学原理、聚集指数和再碰撞理论方法的结合；构建了垄行农田结构雷达后向散射模型，揭示了周期性地表散射机理，实现了农田特有的垄行结构的定量化表达。在农田参数反演技术方面，综合利用多源多时相遥感数据，开展了参数反演的研究，提出了土壤温度、土壤水分、作物叶面积、作物叶绿素含量、作物冠层温度等农田参数反演的新技术，提高了参数的反演精度。在农田生态过程的动态监测方面，提出了区域尺度遥感-作物生长模型同化框架和作物长势动态监测的数据同化的优化方案，将作物生长关键时相和面向研究对象的空间分辨率作为同化时空尺度，在保证同化精度的前提下提高了数据同化效率，实现了大区域农田作物长势的高效、动态监测；在应用示范方面，分别开发了农田生态系统遥感诊断软件平台与地表参数反演系统，以国产卫星遥感数据为主，开展了农田生态过程多参数反演，实现了作物长势、农田旱情与作物估产等多项农情信息的精准提取，并在河北、山东、陕西、甘肃等省区开展了推广示范应用，进一步促进了定量遥感在我国精准农业与粮食安全保障中的应用。.本项目发表论文42篇，其中SCI/SCIE收录26篇,获得国家发明专利2项，软件著作权2项，出版专著1部，圆满完成基金考核指标。