【作者】 况润元；

【导师】 周云轩；

【作者基本信息】 华东师范大学 ， 自然地理学， 2010， 博士

【摘要】 长江口水色定量遥感对于监测长江口泥沙入海通量、时空分布有着重要的意义。当前水色遥感常用经验关系模式反演成分浓度,算法简单、稳定性差、信息机理不明确。为更加准确的反演长江口水色浓度需要从深层次地了解长江口水体成分的光学特性和光在水中的辐射传输问题,构建水体光谱辐射传输模型,建立一个符合长江口区域的水色反演模型。本文基于此,选择长江口为研究区,通过大量的实地水体光学调查工作和室内实验分析,进一步探讨水色反演问题。根据现场实测数据和室内实验数据,分析了长江口水色组分的吸收、散射特性,建立了长江口泥沙吸收光谱、(后向)散射光谱参数化模型,叶绿素吸收光谱参数化模型,黄色物质吸收光谱参数化模型。根据长江口实测资料,分析了长江口水体反射光谱特征,长江口水体反射光谱可以分为8种类型；根据各种类型的水体反射光谱特征,提出相邻或相间波段的反射光谱斜率算法应用于MODIS中高分辨率影像进行水体分类,分类结果与实际基本符合,分类结果可用来监测不同水体的空间分布情况。借鉴前人研究成果,建立长江口水体辐射传输模型,模拟了水体表面遥感反射率光谱。模拟表明,建立的辐射传输模型和参数化模型在泥沙中低浓度范围内适应性比较好,而在高浓度范围(0.2kg/m3以上),因超过光学参数化模型的范围,结果不理想：泥沙浓度垂向不同分布结构对反射光谱有明显的影响,与泥沙垂向分布均一型相比,近表层水域泥沙浓度梯度变化大的水面反射光谱比梯度变化小的反射光谱偏差大；黄色物质的主要效应是降低短波长反射率值,而反射光谱形态特征不会有明显变化；口外泥沙低浓度区,叶绿素的吸收效应明显大于叶绿素的散射效应,叶绿素对反射光谱的贡献以吸收为主：各种类型水体透光层深度模拟表明,长江口透光层深度小于实际水深,说明水面的离水辐射主要来自一定深度内水体本身的辐射贡献,水底反射贡献可忽略不计；最后提出透光层深度的泥沙垂向加权积分浓度,传统经验方法中,与反射光谱建立算法关系的表层浓度建议用此代替,以真实反映反射光谱对应的浓度信息。耦合非线性优化方法Levenberg-Marquaudt和水体辐射传输模式建立水色多组分反演模型。根据实测的反射光谱反演水色组分浓度,模式导出的遥感反射率与实测非常接近。反演浓度与实测水体组分浓度对比表明,泥沙和叶绿素浓度反演精度较高,黄色物质反演精度偏低。泥沙浓度平均误差25.3%,叶绿素浓度平均误差31.5%,黄色物质平均误差59.4%。更多还原

【Abstract】 Ocean color remote sensing is a great significance in monitoring sediment flux and spatial and temporal distribution of suspended sediment in Yangtze estuary. The empirical approach often applies to retrieval of water compositions concentration. However, the disadvantages of empirical algorithms are regional or seasonal effects, fuzzy of information mechanism. In order to accurately retrieve concentration of water compositions, we need understand deep optics properties of ocean color in Yangtze estuary and radiative transfer of light in water, and establish the radiative transfer model and the retrieval model in Yangtze estuary. Based on these reasons, we discuss about inverse problem by in site optics investigation and experimental analysis indoor.Based on the measured data in Yangtze estuary and experimental data indoor, we analyze absorption and scattering properties of water body. Furthermore, the study builds parameterization model about suspended sediment absorption and scattering, chlorophyll absorption, and yellow matter absorption.According to the measured data in Yangtze estuary, the characters of reflectance spectrum were analyzed. There are eight types reflectance spectrum. We put forward an algorithm of spectrum slope in order to distinguish different spectrum types. The algorithms applied to water body classification of MODIS images. The results are consistent with the facts and monitor dynamic distributions of different water body.The study builds water radiative transfer model by using previous research results for reference, and model remote sensing reflectance on water surface. The result shows the radiative transfer model and parameterization model is effective in the range of low-middle concentration of suspended sediment. However, in the range of high concentration (>0.2km/m3), the result is inconsistent with fact. The model indicates the depth of euphotic zone is less than water depth in the different type water body. This show the water-leaving radiances on water surface mainly come from radiance contribution of water body itself, and bottom effects are negligible. Based on the depth of euphotic zone, the paper suggested the optically weighted suspended sediment concentration instead of surface concentration when building traditional empirical algorithms in order to reflect truly suspended sediment concentration correspondence with remote sensing reflectance.Coupling non-linear optimization techniques and water body radiative transfer, the article set up ocean color retrieval model. The retrieval results show the average error of suspended sediment is 25.3%, and the average error of chlorophyll concentration is 31.5%, and the average error of yellow matters absorption is 59.4%. The retrieval precision of suspended sediment concentration and chlorophyll concentration is high. However, the retrieval precision of yellow matter absorption is low.更多还原