【作者】 修鹏；

【导师】 刘玉光；

【作者基本信息】 中国海洋大学 ， 物理海洋学， 2008， 博士

【摘要】 本文对渤海海域的水色遥感,包括对近岸水体光学性质、叶绿素浓度垂向分布和水色遥感算法进行了深入的研究。渤海是我国的一个内陆海,深入了解其光学性质对我国水色遥感的发展有极其重要的作用。然而,由于其水深较浅以及大型河流泥沙、营养盐和可溶有机物的输入等原因,渤海海域的光学性质变得非常复杂,国内外的很多遥感模式在渤海海域都不适用,模式输出与现场数据之间存在较大的误差。针对这个问题,本文根据辐射传输方程和实测数据建立了相应数值模型,全面分析了渤海海域特别是渤海近岸的光学性质,并发展了叶绿素浓度垂向最大值遥感反演算法。依据2005年的实测数据,本文分析了渤海近岸水体的光学性质,包括水体的吸收系数,散射系数以及衰减系数等,建立了相应的叶绿素吸收光谱模型,黄色物质吸收光谱模型以及颗粒物后向散射系数模型等。通过对比现行的业务化海表层叶绿素浓度反演模式及人工神经网络的结果,本文深入分析了这些模式在渤海的适用性,通过相互比较找出了在渤海各个海区内反演结果精确度相对较高的模式。然而总的来说,将这些模式完全照搬到渤海海海域是会产生较大反演误差的。因此,针对渤海必须要发展其自身独特的遥感算法。根据辐射传输理论,本文建立了一个适用于一类水体辐射传输的理想模型,用来研究叶绿素垂向分布结构对水色遥感信息所产生的影响。该模型的三个主要遥感参数包括海表面遥感反射率、海水穿透深度以及垂向积分的叶绿素浓度值。模型的输出结果显示,当海表层叶绿素浓度不变而叶绿素垂向分布结构变化的时候,上述三个参数在可见光波段范围内都受到很大的影响,特别是在蓝绿光波段受到的影响最大。经验蓝绿波段比值法是反演海表层叶绿素浓度的有效方法,例如SeaWiFS的OC4V4算法是一个最有代表性的蓝绿波段比值法。因此为研究叶绿素垂向结构对叶绿素浓度反演所产生的影响,本文以OC2V4算法为例进行了讨论。结果显示,当海表层叶绿素浓度不变的时候,OC2V4算法的输出数据是随着叶绿素垂向分布结构的变化而变化的,这对于一个反演海表层叶绿素浓度的算法来说显然是不能接受的。通过分析模式结果,本文利用了一个特殊的遥感波段,建立了一种新型的叶绿素反演算法,该算法可以不受叶绿素垂向分布结构的影响,较准确地反演出海表层叶绿素浓度。目前,还没有人专门针对叶绿素垂向分布结构不一致的海域发展反演海表层叶绿素浓度的算法,而现行的大部分经验算法都是依靠蓝绿波段比值法来反演海表层叶绿素浓度的,这就不可避免的会产生相应的误差。因此可以说本文提出的新算法对以后的研究可以起到指导作用。根据实测数据,文中首先分析了渤海海域的叶绿素垂向分布结构特征并且将其加以分类。针对每一类的垂直方向分布特征,在前人研究的基础上建立了相应的适用于渤海海域的数学模式,该模型的建立可以使人们更准确的研究渤海海域的水体性质以及生态特征。另外,采用蓝绿波段比值法,文中建立一个反演叶绿素浓度最大值的遥感模式,并且利用实测数据对其进行检验,取得了较好的结果。根据Lee等人的研究结果,作者进一步建立了一个近岸二类水体水色遥感算法的“半分析模型”,它是一个适合渤海海域的参数化模型。通过根据渤海海域的光学性质将模式加以修改,本文利用该模式成功的反演出水体中的一些光学参数如叶绿素吸收系数,黄色物质吸收系数,颗粒物的后向散射系数,以及叶绿素垂向浓度最大值对应的深度等。至此,渤海海域的叶绿素垂向最大值的遥感模式已经基本建立,经验模式用来反演最大值浓度,半分析模型用来反演最大值的对应深度。本文的研究为将来拓展海洋水色遥感从海表层进入水体内部的应用提供了一个新的途径。更多还原

【Abstract】 Ocean color remote sensing has become important in the ocean sciences, especially for the coastal oceans. Bohai Sea is a semi-enclosed inland sea with a typical case-2 water environment located at the northernmost end of eastern Chinese mainland. Understandings of its bio-optical properties will be quite important for the improvements of ocean color applications in China. Due to the shallow bottom depths and large amounts of suspended matters transported from some big rivers, however, the optical properties of Bohai Sea become so complicated that the operational chlorophyll algorithms of the world often fail in this area.Based upon the measured data in the coastal areas of Bohai Sea, preliminary results are obtained about the optical properties including the absorption, scattering, and attenuation properties of the water body. Also, this study builds some remote sensing models to predict the inherent optical properties from above surface remote sensing reflectance.Empirical band-ratio algorithms and artificial neural network techniques to retrieve sea surface chlorophyll concentrations are evaluated in the Bohai Sea by using an extensive field observation data set. The comparison results show that these empirical algorithms developed for case-1 and case-2 waters can not be applied directly to the Bohai Sea, because of significant biases. For example, the mean normalized bias (MNB) for OC4V4 product is 1.85 and the root mean square (RMS) error is 2.26. It appears that the Bohai Sea requires new approaches and new parameterizations for both empirical and semi-analytical pigment algorithms.According to the radiative transfer theory, an idealized ocean color model is used to study the effect of nonuniform chlorophyll profile on the ocean color parameters such as penetration depth, the above-surface spectral remote-sensing reflectance, and the optically weighted chlorophyll concentration. The simulations for a vertically nonuniform chlorophyll concentration are compared with reference simulations for a homogeneous ocean whose chlorophyll concentration is identical to the surface chlorophyll concentration of inhomogeneous cases. Due to the influence of the nonuniformity, the maximum relative error at 445 nm for penetration depth is up to 60%, spectral remote-sensing reflectance is about 40% and optically weighted chlorophyll concentration is about 40% within the range of our simulations.Model results show that there is always a spectral band where the value of above-surface remote-sensing reflectance is not influenced by the nonuniformity. Depending on this band, a new model for retrieving sea surface chlorophyll concentration is designed by adding a compensation term into the variable in SeaWiFS OC2V4 algorithm. By using an iterative method with this new model, sea surface chlorophyll concentration can be well retrieved even in the area where the vertical chlorophyll distribution is unknown. This is a new method which is not ever proposed before, and by using this model, people don’t need to consider the vertical distributions of chlorophyll.Special characteristics of deep chlorophyll maximum in Bohai Sea of China are examined in this study with four cruises data measured in June and August of 2003 and 2005 respectively. Based on what we measured in 2003, a new blue-to-green ratio method ocean color model to retrieve concentration of deep chlorophyll maximum from the remote sensing reflectance above sea surface is proposed. This model is able to be used in case-2 water areas with depth of deep chlorophyll maximum shallower than 7 meters. This model confirms that satellite sensors can detect deep chlorophyll maximum in most areas of Bohai Sea and the existence of deep chlorophyll maximum also can bring big retrieval errors to ocean color models. The maximum depth from which the radiometer receives significant signal varies as a function of wavelength and of the clarity of the water. Blue or green light can penetrate deeper than red light in sea waters, so satellite sensors in blue and green bands are able to receive the signal of deep chlorophyll maximum. In the shallow seas of case-2 water with shallow depth and high concentration of deep chlorophyll maximum, contribution of deep chlorophyll maximum to remote sensing signals is much larger than that from the sea surface chlorophyll. Therefore, we suggest that we should not focus our attention only in sea surface for study of the ocean color retrieval models, but also should broaden the field of vision.According to Lee’s study, a hyperspectral remote-sensing reflectance inversion model is parameterized by using measured data in coastal areas of Bohai Sea. The model only uses remote-sensing reflectance to derive a set of values of absorption, backscattering, water bottom albedo, and bottom depth. The model shows good performances to retrieve phytoplankton absorption and colored detrital matter (detritus plus gelbstoff) absorption. More important, statistical analysis result shows that model derived depths agree well with measured depths of deep chlorophyll maximum, where no significant bias is found, which is primarily due to the existence of water stratification. Therefore, together with the empirical blue-to-green ratio method to retrieve concentration of deep chlorophyll maximum, this study succeed in retrieving the properties of deep chlorophyll maximum (the concentration and the depth) by using remote sensing methods. This work broadens the view of applications of ocean color remote sensing not only just from the surface but also into the oceanic interior, which provides a new method to use ocean color data.更多还原