三峡支流回水区藻类群落结构光谱指纹识别研究

World-wide occurrence of algal blooms is common problem in water quality management. It remains the hot issue for ecological monitoring and early risk forecasting in the Three Gorges Reservoir (TGR). The characteristics of algal blooms in the backwater area of the TGR are detailed as follows: High turbidity of water body background, the diversity of dominant species, and rapid changes in community composition. As a result, traditional methods of remote sensing monitoring are insufficient to discriminate spectral response of algal community structure. To explore the aforementioned issue, this project will study the description method of algal community structure from the time-space-spectrum perspective (i.e. spectral fingerprint identification model). Through in-situ monitoring, the responses of spectral characteristics of three bloom-forming populations to diffuse scattering effect of suspended particulate matter were analyzed, including cyanobacteria, green algae and diatom, respectively. Then, we screened spectral parameters of corrected turbidity factor to construct diagnostic spectral index based on photosynthetic pigment. According to mathematical model and data assimilation, we used quantitative indicators for describing spatiotemporal autocorrelation features, in order to establish the multi-dimensional feature space model based on diagnostic spectral index. Subsequently, we abstracted the multi-dimensional feature space model of algal community as the concept of the spectral fingerprint identification. By means of mixed spectral simulation experiments, the reliability of identification model of spectral fingerprint was comprehensively evaluated by cross-validation. This project provides a new research framework for depicting the evolution mechanism of algal community structure. It will be helpful for monitoring, early warning, and scientific management of algal blooms in large reservoirs.

水华暴发是一个全球性水环境问题，也是三峡蓄水成库后生态环境监测预警研究的关键问题。三峡支流回水区水华暴发表现出水体背景浊度高、优势藻类种群多与群落结构变化快等特征，致使传统遥感监测方法对藻类群落结构光谱响应的甄别能力不足。本项目以探索面向复杂生境驱动下藻类群落结构的时-空-谱特征描述方法（光谱指纹识别模式）为目标，通过原位布点监测，分析蓝藻、绿藻和硅藻光谱特征对悬浮颗粒物漫散射效应的响应机理，筛选浊度因子校正光谱参量，构建基于光合色素的藻类种群诊断光谱指数；依据藻类机理模型和数据同化方法，利用时空关联性描述的定量化指标，建立基于藻类种群诊断光谱指数的多维特征空间模型，将其抽象化为藻类群落结构光谱指纹识别模式；借助混合光谱模拟实验，研究交叉比对策略，综合评价藻类群落结构识别结果的可信性。本项目为揭示藻类群落结构演化机制提供一种新的研究思路，将有助于对大型水库水华暴发问题的监测预警与科学管理。

本项目聚焦于运用卫星和地基光谱数据精确识别藻类群落结构（蓝藻、绿藻、硅藻相对丰度比例）的生物光学机理研究，针对三峡水库典型支流小江回水区“深-浅水类湖泊”复杂多态交替的独特生境，构建了一种面向优势藻种水华精确识别的光谱表征方法，在高浊背景干扰下提取了藻华群落中蓝-绿-硅藻种群所对应的叶绿素浓度，为探测大型水库藻华的时空演化轨迹提供理论依据和方法支撑，基本达成了预期的研究目标。首先，利用密度峰值聚类算法对水体光学类型进行分类，由此建立适用于环境一号卫星多光谱影像长时序数据的表层水体浊度遥感定量反演算法，获得了小江回水区浊度的时空分布，为夏季洪水期的藻华监测预警提供了参考背景值；其次，运用自主研发的地基高光谱成像仪，在小江回水区构建了叶绿素浓度高光谱反演模型，刻画了时-天-月尺度的叶绿素浓度时序变化模式，揭示了在气候变化与水位调度的交叠影响下叶绿素浓度多尺度变化规律，为通过高光谱卫星影像获取藻华时空演化信息提供了技术支撑；最后，采用基于连续小波变换的峰度分析方法，借助环境一号卫星多光谱影像长时序数据，展现了三峡支流回水区藻华暴发提前起始和延迟终止的物候趋势，并证实藻类群落结构演替物候趋势在气候变化和水位调度的交互作用下发生了改变。研究成果可为大型水库生态调度运行和藻华监测预警提供相应的科学依据。