矿区环境重金属污染的高光谱遥感监测与分析方法研究

In the recent years, the serious incidents about heavy metal pollution caused by mining and industrial production had frequently happened in China. It made the life and health of the local residents some kinds of severe harm. Consequently, it is very necessary and has important significance that to carry out the research of heavy metal pollution. The hyperspectral remote sensing is a frontal area in the application research of remote sensing technology nowadays, and the analysis and evaluation of environment during the exploiting and mining of mineral resources are the hot and difficult issues of hyperspectral remote sensing applications. .The Research work will take the Zijin's gold-copper mining areas in Fujian province as a case study area, and applies space-borne hyperspectral remote sensing and surface spectral measuring to acquire the micro-pollution information of heavy metal such as copper (Cu) or lead (Pb) especially to soil and vegetation in the study area. The key theories and technologies include (1) how to define the heavy metal spectral response intervals effectively; (2) the algorithm of wavelet packet transformation of hyperspectral image data of heavy metal environment pollution; (3) the spectral decomposition and explanation of linear wavelet packet transformation in frequency domain of micro-pollution information; (4) the harmonic analysis of measuring spectral data and the discrimination of micro-pollution information within soil and vegetation; (5) the modeling of self-organizing feature maps (SOFM) based on some kinds of ground measurement spectral characteristic data and the map of micro-pollution information; (6) the verification and accuracy evaluation of study results. .We expect that several effective, update theories and technologies to acquire and analyze and assess the information of heavy metal pollution to soil and vegetation in mining areas in China could be explored, and service the social sustainable development.

近年来我国重特大重金属污染事故频发。这种污染主要由采矿和排污等原因所致，尤以采矿污染为主并悄无声息地转移着，给矿区与其周边及相关流域的居民生活和健康带来严量危害，开展这方面的研究十分必要，意义重大。高光谱遥感是当今遥感研究的前沿领域，矿产资源开发的生态环境演变与污染是高光谱遥感应用的热点和难点之一。本项目以福建紫金铜矿区为测试区，应用星载高光谱影像与地面测量光谱等数据，对铜、铅等矿产开发的环境土壤和植被重金属污染进行监测与分析评价。研究要点是：重金属污染光谱响应区间的有效表达；高光谱影像数据重金属浸染的小波包分解算法；微量污染信息的光谱维频域线性小波包分解与解释；光谱测量数据的谐波分析及土壤与植被微量污染信息甄别；多类型光谱特征SOFM建模与微量污染信息映射；成果验证与精度评定。力求探索出工矿区土壤与植被等重金属污染信息获取与分析评价的高新技术手段和基本理论方法，并服务于社会的可持续发展。

针对利用高光谱遥感卫星影像与地面实测光谱等监测与分析评价矿区土壤和植被中铜、铅重金属污染的研究目标，本研究通过多次盆栽玉米实验获取了不同梯度浓度Cu2+、Pb2+胁迫下植株叶片和冠层光谱数据及其重金属元素与叶绿素含量测定数据，分析了Cu2+、Pb2+污染后原始光谱、光谱微分、光谱分形处理和频率域分析等数据的污染效应与光谱有效表达位置，胁迫响应特征及其特征参量、污染监测方法；开展了光谱特征参数、生化参数与Cu2+、Pb2+污染之间的相关性研究，并结合光谱特征参数、主成分分析、小波包变换、BP与小波神经网络、遗传算法、经验模态分解和谐波分析构建和改进了叶绿素含量的BP、PCA-WNN、GA-BP、WPT-PCA-BP、SCP-PCA-BP和多元线性回归等反演模型；研究了植被重金属污染检测方法，提出了微分光谱角正切(DSAT) 重金属污染的光谱弱差信息甄别模型和分段微分光谱角正切(SDSAT)的重金属污染程度度量算法等；同时研究了基于谐波分析、最佳小波包分解层和小波包最佳基等的高光谱影像去噪方法、融合算法、光谱信息保真度评价和微量信息识别模型，并利用多种现化数学理论构建了影像端元数目估计方法、混合像元分解算法和小目标探测模型，以及矿区重金属蚀变信息端元提取和丰度反演算法，也建立了多种适合微量信息识别的新型高光谱影像分类器。研究成果可为将来更好地利用航空航天高光谱遥感影像监测环境重金属污染提供理论与方法支持。