【作者】 崔廷伟；

【导师】 郑荣儿； 张杰；

【作者基本信息】 中国海洋大学 ， 光学工程， 2003， 硕士

【摘要】 我国拥有1.8万公里的漫长海岸线，海洋资源丰富。随着沿海城市工业化的发展，海洋生态环境受到严重破坏；赤潮灾害频发，已经成为我国的一大海洋灾害和三大近海污染问题之一。 赤潮是海水中的浮游生物在一定条件下过度繁殖或聚集致使海水变色的一种生态异常现象，因其具有随机性、突发性、持续时间短、影响范围大等特点，客观上要求利用航空高光谱遥感技术对其进行监测；但首要的前提与基础性工作是大量获取赤潮水体的现场光谱数据，通过处理与分析，掌握其光谱特性，不断充实赤潮优势种类的光谱数据库。本文的工作正是在此背景下层开的。 论文由五部分组成。 论文首先在第一章中对于海洋光谱数据获取、处理与分析等方面的国内外研究进展与动态进行了综述与分析，为相应研究工作的开展奠定了较完整的方法基础，具体包括：(ⅰ)海洋光谱数据获取方法；(ⅱ)海洋光谱数据处理方法；(ⅲ)海洋光谱数据分析方法。 我们采用围隔实验培养赤潮为主，现场捕获赤潮为辅的光谱数据获取策略：依据围隔实验以及赤潮自身的特殊性，选用了海面以上光谱数据获取法(above water method)来完成赤潮光谱数据的获取工作；具体的实验器材选择、操作规程确定、注意事项等内容在第二章中进行了详细阐述。 第三章是本论文的核心部分。对所获光谱数据进行处理，得出了丹麦细柱藻、海洋褐胞藻、中肋骨条藻、红色中缢虫等不同优势种类赤潮水体的遥感反射率光谱曲线；上述数据处理步骤、数据处理结果以及不同赤潮种类的生物特征在本章的3.1节给出。 为了定量表达赤潮水体的光谱吸收特征，3.2节进行了光谱吸收深度(H)、吸收宽度(W)、吸收面积(A)、吸收峰对称性(S)等特征吸收参数的提取方法研究。对中肋骨条藻三条参考光谱的吸收特征参数与叶绿素a浓度进行了相关分析，发现450～520nm区间光谱吸收深度与叶绿素a浓度有较好的正相关关系，625～700nm区间吸收峰的对称性与叶绿素a浓度呈正相关关系，吸收宽度与叶绿素a浓度呈负相关关系。由于可供比对分析的叶绿素a浓度数据数量有限，本节的研究还只是对相应内容的初步探讨，结论的普适性有待进一步检验。 光谱微分方法是一种简单而有效的确定光谱曲线极值点波长位置的方法。3.3节运用该方法得出了实验赤潮水体光谱曲线的两个反射峰和一个吸收峰的波长范围，并对其产生机理进行了分析;更为重要的是，利用光谱曲线第二反射峰的波长位置实现了赤潮与非赤潮水体的光谱辨别，以及红色中绕虫、海洋褐胞藻等不同优势种类赤潮的识别。 光谱匹配是一种基于整体波形相似性的光谱识别方法。3.4节将光谱角度制图(SPectral Angl“MapPing，SAM)这一光谱匹配方法引入赤潮光谱识别研究之中，所进行的赤潮优势种类识别取得了令人较为满意的结果:红色中绕虫、丹麦细柱藻赤潮的光谱识别取得了成功;海洋褐胞藻赤潮有误识别的情况发生;只有中肋骨条藻赤潮的识别失败。文中对误识别产生的可能原因进行了分析，并指出了相应的改进方法。 3.5节对互相关光谱匹配(eross eorrelogam speetral Matching，eesM)方法用于赤潮优势种类识别研究进行了尝试，红色中绕虫、丹麦细柱藻、海洋褐胞藻赤潮的识别取得成功，中肋骨条藻赤潮的识别失败。总体来看，CCSM识别结果较SAM更佳，但也存在误识别的可能。 在上述工作的基础上，论文的第四章提出了《赤潮现场光谱调查规范》，力图对赤潮现场光谱数据获取、处理与分析工作起指导作用，内容包括:赤潮现场光谱调查前的准备工作;(ii)赤潮现场光谱调查操作规程;(iii)其他注意事项;(iv)光谱数据处理。 第五章对本论文所取得的研究成果和结论进行了总结与讨论，在此基础上指出了论文工作中尚存在的不足与改进措施以及下一步的工作方向。更多还原

【Abstract】 China has 18,000-kilometer coastline and abundant marine resources. With the industrialization of littoral cities, marine entironment has been severely destroyed. Red tide occurs more and more frequently in China’s coastal sea area, and has been one of China’s ocean calamities and one of the three severe pollution problems.Red tide is ecological abnormal phenomenon caused by phytoplankton, which reproduce in excess and assemble, resulting in ocean color changed. Red tide occurs randomly, with short duration and large incidence. Airborne hyperspectral remote sensing technique can play an important role in red tide’s monitoring. And the precondition is to acquire in situ spectral data of red tide, to grasp red tide’s spectral characteristics by data processing and analysis and to set up spectral library of different dominant species of red tide. The thesis is developed in this background.This thesis comprises 5 chapters.In the first chapter, the development and status in quo of ocean spectral data acquisition, processing and analysis are summarized, in order to establish a foundation for corresponding work in this thesis. Three parts of the chapter are: (i) acquisition method of ocean spectral data; (ii) processing method of ocean spectral data; (iii) analysis method of ocean spectral data.Red tide is cultivated by mesocosm experiment with seawater enclosed and captured near the Bayuquan port in the Liaodong Bay. In situ spectral data of red tide are acquired by above water method, according to the condition of mesocosm experiment. Chapter 2 deals with the experiment equipments selection, spectral data acquirement regulations and proceedings that should be noticed in experiment.Chapter 3 is the core of this thesis. After spectral data processing, remote sensing reflectance data of red tide are got, whose dominant species are Leptocylindrus danicus, Chattonella marina, Skeletonema costatum and Mesodinium rubrum respectively. Section 3.1 gives the data processing steps, results and biological details of the dominant species described above.Method of extract spectrum absorption parameters such as absorption depth, width, symmetry and area is given in section 3.2. In the research of correlation between chlorophyll a concentration and absorption parameters of Skeletonema costatum spectra, we find that there is a pretty good positive correlation between absorption depth and chlorophyll a concentration in the spectral range between 450nm and 520nm. Yet in the spectral range between 625nm and 700nm, we find a positive correlation between absorption symmetry and chlorophyll a concentration and a negative correlation between absorption width and chlorophyll a concentration. As the chlorophyll a concentration data that can be used in the correlation analysis are limited quantitatively, these conclusions are preliminary and subjected to be tested.Spectral derivative is a simple and effective method to acquire the wavelengths of extremum points. The wavelengths of reflectance peaks and vales are got by spectral derivative method in section 3.3. Mechanism of them is analyzed. Red tide and normal seawater, as well as some different dominant species of red tide, such as Chattonella marina and Mesodinium rubrum, can be discriminated by using the wavelength information of second reflectance peak.Spectral matching algorithm is applied, such as Spectral Angle Mapping (SAM) in section 3.4 and Cross Correlogram Spectral Matching (CCSM) in section 3.5 in order to discriminate different dominant species of red tide. The results of SAM spectral matching are satisfying: identifications of Mesodinium rubrum and Leptocylindrus danicus succeed, identification of Chattonella marina has wrong possibility and that of Skeletonema costatum fails. The probable reasons for wrong identification and approach to get better results are also pointed out in section 3.4.The results of CCSM spectral matching are even better than those of SAM. Identifications of Mesodinium rubrum, Leptocylindrus danicus and Chattonella marina su更多还原