【作者】 姜玲玲；

【导师】 赵冬至；

【作者基本信息】 大连海事大学 ， 环境科学， 2014， 博士

【摘要】 后向散射系数是水色遥感的一个重要基础光学参数,其大小与水体组分有关,是重要的固有光学量之一。目前关于水体后向散射特性的研究主要集中在一类水体和近岸二类浑浊水体中,对于赤潮水体后向散射特性的研究相对较少,这在一定程度上阻碍了水色遥感分析模型的发展。本文从微微型浮游植物、微型浮游植物和小型浮游植物中选取四种我国近岸代表性赤潮藻种(微微藻、中肋骨条藻、强壮强沟藻和海洋原甲藻)进行培养和测量,并获得生物-光学数据,分析了不同藻种后向散射光谱变化特性及其影响因素,同时首次建立了基于MODIS数据的典型赤潮藻种细胞数遥感反演算法。主要内容和成果如下：1)对不同类型的赤潮藻种的后向散射光谱变化特性进行研究。通过对现场数据的分析计算,本文获得了四个代表性藻种不同叶绿素浓度下的后向散射系数和后向散射比率光谱,结果表明,各藻种的后向散射系数和后向散射比率均随叶绿素浓度的升高而增大,但是每个藻种都具有自己独特的光谱形状,并且微微型浮游植物和微型浮游植物在蓝光波段420nm-488nm的后向散射系数光谱形状还会随叶绿素浓度改变而发生变化,而小型浮游植物海洋原甲藻的光谱形状则始终保持不变。此外,通过对各藻种单位后向散射系数、单位后向散射比率以及后向散射截面的综合分析得出大颗粒藻种在420-620nm之间具有相对比较平缓的后向散射光谱曲线,尽管不同粒径大小的藻细胞颗粒对水体后向散射的贡献规律很难确定,但后向散射截面与粒径则呈良好的幂函数关系,该结论为日后赤潮藻种粒径分布的遥感反演提供了理论依据。2)针对藻类颗粒物蓝光波段后向散射光谱形状的变化特点,探讨了颗粒物吸收对后向散射的影响,并针对各藻种建立了后向散射与吸收的特征响应关系模型,得出微微藻和强壮前沟藻的颗粒吸收对后向散射的影响较大,其决定系数R2分别为0.996和0.93；而对于中肋骨条藻和海洋原甲藻而言,其影响则相对较弱,R2仅为0.57和0.52。3)建立水体表征参数与后向散射系数相关关系模型。本研究基于实测数据,分别建立了后向散射系数与叶绿素浓度、细胞数之间的相关关系模型,分析得出线性回归模型和乘幂回归模型均可用来描述各藻种后向散射系数与叶绿素浓度之间的关系；而后向散射系数与细胞数只满足幂指数关系模型,并且微微藻各波段的拟合系数基本相同；强壮前沟藻两者的相关性随着波长的增加而逐渐增强,至红光波段700nm处,两者之间的相关系数R2则高达0.99；海洋原甲藻的最佳拟合波段则出现在蓝光波段和红光波段处,该结论为日后建立典型赤潮藻种的细胞数遥感反演算法奠定良好的理论基础。4)建立了基于MODIS卫星的细胞数遥感反演算法,实现了赤潮水体信息的有效提取。本论文利用实验获得的测量数据,建立表观量与固有量之间的关系,并选择特征波段针对海洋原甲藻和微微藻建立了基于MODIS卫星的细胞数遥感反演算法。根据赤潮监测记录,选取2012年6月16日微微藻赤潮的MODIS影像进行提取实验,分别采用多波段遥感反演算法和单波段遥感反演算法进行细胞数反演。结果表明,对于高叶绿素水体,多波段细胞数反演模型的反演精度明显高于单波段细胞数反演模型,但是对于混合像元信号,赤潮检测比较困难,容易导致误判,特别对于悬沙较大的水体,反演的结果也偏大；而对于高悬沙水体,基于488nm的单波段的细胞数反演模型的反演精度则相对更高,尽管监测点位的反演结果比实测细胞数偏低,但从总体对赤潮的识别效果而言,单波段的反演算法明显剔除了悬沙的影响,识别精度更高一些。更多还原

【Abstract】 Backscattering coefficient is an important parameter in ocean optics, and also is one of the important inherent optical properties, it is determined by the concentration of water constituents. Recently, while considerable research has been conducted on the backscattering properties of water with the development of optical instruments, most of this research is focused on the particle backscattering characteristics of case1water and coastal turbid water. The phytoplankton backscattering coefficient (bbP) and the causes of its variability are still poorly known. This hinders the expansion of the remote sensing model to a considerable extent. Variability of the backscattering characteristics about the alga Aureococcus anophagefferens、Skeletonema costatum、Amphidinium carerae hulburt and Prorocentrum micans which represent for the picophytoplankto、 nanophytoplankton and microphytoplankton respectively are examined. At the same time, the cell density remote sensing inversion model of the red tide alga is first builded. And the main researches contents and achievements were as follows:(1) Variability of backscattering properties about different red tide alga. Particulate backscattering coefficients and backscattering ratio are obtained for the experimental cultures by calculating the in-situ measured data, the results show that the backscattering coefficient and the backscattering ratio increase with an increase in the chlorophyll concentration, but every alga has its own spectral shape, and the shape of the particulate backscattering coefficient spectra about the picophytoplankton and nanophytoplankton is also changed with the variation of the chlorophyll concentration, especially at420nm-488nm. Otherwise, the bigger particulate has relatively flattened spectra, it is not regular for the contribution to the water backscattering of different sized alga particles, and a good relationship is observed between the backscattering cross-section and the ESD which provide a good theoretical basis for building the particle size distribution inversion model.(2) According to the variation of particulate backscattering spectra shape, the study investigates the influence of absorption on the backscattering signal, and establishes the characteristic response relationship model between backscattering and absorption on every selected alga. The result indicates that the slope of the backscattering spectrum in the blue for Aureococcus anophagefferens and Amphidinium carerae hulburt shows a strong relationship with absorption and the correlation coefficient R2are0.996and0.93respectively. The slope of the backscattering spectra between442nm and488nm for the Skeletonema costatum and Prorocentrum micans has a weaker, positive linear relationship with the relative absorption(R2is equal to0.57and0.52respectively).(3) Building correlation model between backscattering coefficients and the water quality characteristic parameters. The relationship between backscattering coefficients and chlorophyll concentration, cell desity is studied respctively. And find that the linear regression and nonlinear regression are both fit for showing the relation between the backscattering coefficients and chlorophyll concentration. While the relationship between backscattering coefficients and cell density only fit the nonlinear regression, the determination coefficient at all bands are the same for the Aureococcus anophagefferens; and the relationship of the Amphidinium carerae hulburt’s becomes stronger with the longer wavelengh, especially at the700nm, R2is as much as0.99; moreover, the best fit bands of Prorocentrum micans appears at blue and red, that also lays the good foundation for building cell density distribution inversion model.(4)It develops the cell density remote sensing reversion model from MODIS data and retrieves red tide distribution effectively. After analyzing the spectral data and establishing the relation between inherent optical properties and apparent optical properties, the author builds a cell density remote sensing inversion model from MODIS data in allusion to Aureococcus anophagefferens and Prorocentrum micans, then apply this method to MODIS data on June16,2005, of which the red tide alga is Aureococcus anophagefferens. The results shows that multi-bands inversion model is more accurate than the single band inversion model for the high chlorophyll water, but it is difficult to identify the red tide information from the mixed pixel signal, that leads to the inversion results is larger than the monitoring data especially for high suspended sediment water, while the single band inversion model is more useful for such turbid water, although inversion value is lower than the monitoring results, it is available to eliminating the influence of suspended sediments and the recognition accuracy is more higher.更多还原