基于地表参量的污水灌溉区农田生态安全遥感监测研究

For the environmental problem caused by sewage irrigation is worldwide, it is a research subject paid attention to how to monitor it effectively and timely at home and abroad. At present, the domestic and international academic circles mostly do the microscopic research on environmental effects of sewage irrigation. However，the new research method is the use of the combination of microcosm and macrocosm. This project selects the plain in the north of Longkou city which is located in Shandong province and which lies at longitude between 120 13′ and 120 44′ east and at latitude between 37 27′ and 37 47′ north. It is set as an example to do a research on farmland ecological safety problems caused by sewage irrigation. Based on the receiving the data of biochemistry of crop leaves in the experimental area and soil as well as the data of spectrum synchronously, it systematically analyzes reflectivity of different deleterious element in different spectrum, absorption valley and reflectance difference, and it does a reserch on original spectrum data mentioned above and spectrum transformation. In addition, the project finds out the spectrum features of contents of different deleterious elements in crops and soil, and sensitive spectral band is filtered with the variation in the content of harmful elements, and then various types of spectral characteristic parameters are constructed to establish the quantitative relationship model based on the various content of harmful elements of the hyperspectral data; Extracting crops growing parameters by building remote sensing evaluation model and diagnosis model, such as individual characteristic parameters(root, stem, leaf, spike), group characteristics parameters (density, distribution, dynamic), integrated characteristic parameter (leaf area index and so on). Analyzing the influences of different harmful elements content on the physical, chemical and biological processes of farmland ecological system, as well as the changes of resulting farmland surface parameters amount; studying these changes in the hyperspectral satellite remote sensing data that has the independent geoscience variables; to match farmland surface parameters and satellite remote sensing data in order to achieve the remote sensing information to contact state variables. The multi-source remote sensing data cooperating with inversion are used to monitor the sewage irrigation on farmland ecological security purposes to provide fast, low-cost, lossless means for monitoring and forecasting farmland ecological security.

由污水灌溉产生的环境问题是世界性的，如何有效地对其进行监测一直是学术界普遍重视的研究课题。传统的研究无法做到对大面积污水灌溉区进行适时、动态、同步监测与分析。本项目在同步获取试验区作物叶片、土壤生化数据和光谱数据的基础上，系统分析不同有害元素含量在不同光谱波段的反射率、吸收谷和反射差异，研究其原始光谱数据及其光谱变换形式，摸清作物、土壤中不同有害元素含量的光谱特征，筛选随有害元素含量变化而变异的敏感光谱波段，构建各类光谱特征参数，建立基于高光谱数据的不同有害元素含量定量关系模型，分析不同有害元素含量对农田生态系统的物理、化学和生物过程影响所造成的地表参量的变化，以及这些变化在卫星遥感数据中的独立的地学变量，将农田地表参量与卫星遥感数据相匹配，实现以遥感信息为联系状态变量，将多源遥感数据协同反演用来监测污水灌溉区农田生态安全的目的，为农田生态安全的监测和预报提供快捷、低廉、无损的监测手段。

项目以山东省龙口市北部平原为研究区，以揭示污水灌溉区农田生态系统的物、化性状与地表参量之间的关系、地表参量变化引起的光谱特征变异及其在遥感影像中所具有的独立的地学变量为主要研究目标，根据农田灌溉、播种的实际情况，在充分考虑了空间异质性对地表参量调查影响的基础上，选择典型样区，设计和架构了不同时间尺度和空间尺度的观测平台，进行反射的精确控制测量和详细的地面调查，获取了大量实测地表参量、地物光谱以及作物、土壤等生化指标化验数据；通过分析不同有害元素含量在不同光谱段的反射率、吸收谷和反射差异，提取有害元素胁迫下的土壤、小麦和玉米光谱特征参数；采用相关分析法，计算不同采样时期内农田地表参量与作物叶片、土壤光谱数据之间的相关程度，探究土壤、作物叶片光谱对农田地表参量的敏感性，分析农田生态系统的物理、化学和生物过程与地表参量之间的关系；采用光谱变换技术增强不同有害元素胁迫下的光谱弱信息及其在遥感图像中所具有的独立的地学变量，揭示污水灌溉区不同有害元素含量对光谱参量的影响；采用逐步多元回归、主成分回归、支持向量机（SVM）和人工神经网络（ANN）等方法，建立了不同有害元素含量的高光谱定量关系模型和不同有害元素含量与农田地表参量之间的定量关系模型，对污水灌溉区农田有害元素含量进行估算；利用内梅罗综合污染指数法、污染负荷指数法、潜在生态风险危害指数法等分析了研究区土壤重金属生态风险；通过解析地理空间与影像空间的映射关系，同化地面状态参数与卫星遥感参数，建立基于像元尺度的遥感反演模型，实现大尺度农田生态安全遥感定量监测研究，为定量遥感技术积累必要的理论、方法和试验基础。