时空耦合的内陆水体叶绿素浓度多源数据协同反演研究

Chlorophyll concentration, which reflects the extent of water eutrophication and algae content, is one of the main parameters of water quality evaluation. The retrieval model of chlorophyll concentration has been the hot study topic since 1970s in water environment remote sensing. However, the experimental observation of the ground cannot be really fully synchronized, leading to the deviations between observed experimental data and image. And thereafter, it directly affects on the accuracy of the inversion model; on the other hand, the inversion from remote sensing image does not have continuity because of the temporal resolution and weather impact. In response to these two difficulties, this project intends to study on the mechanism and method for fusing the ground real-time observational data - bio-optical model, and water dynamic model - satellite image data to achieve the inversion by synergy of multi-source data. At the same time, we get the synchronize data by buoy observations, and fetch the spatial continuous data by satellite imagery, and then assimilate water dynamic model and bio-optical model. Thus, by transforming discrete sample observations to space continuous, and instantaneous to time continuous from space-time data coupling, we could assess the spatial and temporal distribution of chlorophyll concentration. The research not only can improve the accuracy of chlorophyll concentration remote sensing, but also provide the technical and methodological reference of multi-source data collaboration inversion for the other water color parameters. This is very important with practical significance for real-time and continuous monitoring of water environment.

叶绿素浓度反映了水体富营养化程度、藻类含量，是水环境质量评价的主要参数，对其遥感反演模型方法的研究，一直是水环境遥感研究的热点。但是，地面实验观测无法真正做到与卫星影像完全同步，导致实验观测数据与影像瞬时获取的数据存在偏差，直接影响了反演模型的精度；另一方面，遥感影像的时间分辨率及受天气影响等因素，使其对叶绿素浓度的反演不具有时间连续性。针对上述两个难点，本项目拟研究地面连续观测数据-生物光学反演模型、水动力模型-多源卫星影像数据协同的机理与方法，以浮标观测等获得时间连续数据及影像同步数据，以卫星影像获取空间全覆盖数据，同化水动力模型、生物光学反演模型实现离散样点-空间连续、瞬时观测-时间连续的时空数据耦合，进行叶绿素浓度时空分布的实时评估。项目的研究，不仅能够提高叶绿素遥感反演的精度，而且可为其他水色参数的多源数据协同反演提供技术和方法论借鉴，对水环境实时、连续监测具有重要的现实意义。

针对水环境监测中地面监测站点数据在空间上不连续，而遥感监测数据在时间上不连续的问题，以太湖为研究区，利用地面连续观测数据、多光谱卫星数据、高光谱卫星数据，研究多源数据时空耦合的机理，构建叶绿素浓度多源数据协同反演模型，实现对水体叶绿素浓度的时空连续估算和预测。项目开展了4次原位遥感实验，获得太湖累计360个样点的同步水质参数及光学参数，获取了太湖2个站点2013-2016年水质自动连续监测数据。取得了以下主要成果：（1）针对环境1号卫星CCD等多光谱数据，构建了基于绿光波段稳定性的大气校正方法，针对MERIS等高光谱数据，构建了基于氧气、水汽吸收波段的大气校正改进算法。有效提高了太湖区域水体大气校正精度，为内陆水色遥感大气校正提供了新的经验和借鉴，同时也为通过遥感影像反演其它水质参数等后续应用提供了数据保障。（2）构建了一种新的二类水体叶绿素a浓度遥感估算模型。该模型通过选取特定的4个波段位置，有效地消减了叶绿素a吸收峰附近非色素碎屑物质的干扰，提高了特定波段叶绿素吸收系数估算精度，进而在拓展了模型通用性的同时，提高了叶绿素浓度的遥感估算精度。（3）构建了基于风生流的太湖叶绿素a扩散模型。该模型充分考虑了水下地形、风速、风向、水位和流速等因素对水质参数输移扩散的影响，采用三角形网格对湖流方程进行离散求解，可以有效模拟叶绿素的三维输移扩散，实现了湖泊水质参数的连续动态模拟。（4）结合水色要素扩散模型，构建了太湖叶绿素浓度估算的数据同化系统，实现了地面实时观测—水色要素扩散模型、生物光学模型—多源卫星数据的协同反演。（5）完成了课题预定的研究目标。课题发表论文21篇，其中，SCI收录论文10篇，申请发明专利2项，其中1项获得授权。（6）培养毕业博士研究生3名，硕士研究生4名。