【作者】 郑宗生；

【导师】 周云轩；

【作者基本信息】 华东师范大学 ， 自然地理学， 2007， 博士

【摘要】 淤泥质潮滩作为陆海相互作用的敏感地带，滩面泥泞、潮沟密布、变化频繁，常规地形测量难度较大。由于淤泥质潮滩具有一些能被可见光和近红外传感器探测到的特征，所以遥感技术为其地形信息提取和定量反演提供了广阔的前景。本论文首先利用多时相卫星影像资料及海图资料，结合实地调查完成了上海市不同时期的滩涂资源解译工作，统计结果为探明上海市滩涂资源总量及其变化规律提供了科学依据。利用遥感水边线方法和数值模型建立淤泥质潮滩的数字高程模型(DEM)。作者在分析长江口区不同浓度水体与背景地物光谱特征的基础上，利用多时相卫星遥感影像，采用决策树方法及区域增长算法提取水边线信息，提高了水边线提取效率和精度。利用国际上成熟的水动力数值模型(Delft-3D)模拟卫星过境时刻的潮位。最后，对具有高程值的水边线系列利用不规则三角网(TIN)完成插值，生成潮滩的数字高程模型。将得到的初始高程模型输入水动力模型，细化原来的地形条件重复运行模型，并将模型结果与水边线提取结果对比，进一步微调潮滩地形，直到模型模拟的水边线与卫星影像提取的水边线满足精度要求为止。作者以九段沙为主要研究对象，为消除潮滩冲淤变化的影响，选取相近年份的遥感数据为数据源，利用上述方法建立了不同时间段内的潮滩高程模型，并通过对比分析研究了长江口深水航道工程对九段沙冲淤演变的影响。以多时相高分辨率航空影像为数据源，在分析潮滩的动力沉积、动力地貌和光谱信息特征的基础上，进行了崇明东滩潮沟信息的提取。根据上述的提取结果研究了Horton定律在崇明东滩潮沟系统中的适用条件，并利用Horton定律及分形分维理论从定量角度分析潮沟形态变化。利用水边线高程反演技术，结合实测潮沟宽深比资料实现了潮沟地形反演，使潮滩地形得到更精细的刻画。利用大量的实测植被光谱及生态调查数据，利用主成份分析方法(PCA)分析了潮滩植被光谱信息与生态环境因子的关系，并以此为基础将植被覆盖度指数(FVC)、潮滩高程、潮沟等信息作为植被分类的辅助信息。在植被初次分类的基础上，构造模糊矩阵，根据辅助信息对不同植被类型的隶属关系对误分的像元进行二次分类，从而提高了潮滩植被分类的精度。为了得到潮滩的沉积速率，本文利用不同年份的水边线位置和实测的高程剖面，计算了潮滩不同部位的多年平均沉积速率，并分析了潮滩冲淤的空间差异及影响因素。结合上述高程反演及平均冲淤速率计算结果，探讨崇明东滩高程及沉积速率之间的相互关系。根据植被信息提取结果，研究了崇明东滩植被对潮滩沉积速率的影响，同时根据野外实测光滩区及植被区的流速、流向及悬沙浓度特征进一步研究了植被对潮滩地貌演化的影响。论文最后探讨了空间可视化技术，利用地理信息系统三维可视化功能对潮滩地形及近岸潮位、流场模拟结果进行了虚拟表达，为海洋科学研究人员进行深入、综合分析提供了技术支持。更多还原

【Abstract】 As a sensitive belt of sea and land interaction, silt tidal flat is characterized as muddy surfaces, densely covered and frequently varied tidal creeks. Therefore, the traditional field observation has been restricted by the difficulty of access. Remote sensing, combined with in situ surveying, is an effective tool for monitoring the tidal flats which can be detected by visible and near infrared sensors.On the basis of field survey, multi-temporal remote sensing data and navigational maps were used to interpret wetland in Shanghai, which provide timely data (i.e. amounts and changes of wetland resource) for scientific management on coastal regions.Waterline method and hydrodynamic model were investigated to construct Digital Elevation Model (DEM). According to spectral characteristics of tidal flat and nearshore water body, a decision tree model and region growing method were applied in detecting waterline with improved accuracy. A series of waterlines were extracted from multi-temporal remotely sensing images by above approach. The assignment of an elevation to the waterline was performed according to hydrodynamic model(Delft3D) at the satellite overpass. The waterlines labeled elevation were used to construct DEMs (Digital Elevation Model). This initial DEM was then used to refine the topography in the intertidal zone, and the model was re-run to produce improved water levels and a new DEM. This procedure was iterated by comparing modeled and actual waterlines until no further improvement occurred. Jiuduansha Island with rapid evolution was selected as study area, and multi-temporal remote sensing images ware divided into different groups which spanned a short period to eliminate the error of the dynamics of mudflat. Then the generated DEMs, built by above method, were compared to evaluate the effects of the construction of Deep Water Channel Project (DWCP) at the north of Jiuduansha shoals.Based on the dynamic nature and spectral information of tidal flat, tidal creeks were extracted from high resolution aerial images, on which whether Horton law has general applicability was analyzed. Then Horton law and fractal dimension were applied in quantifying geomorphic changes. The topography of tidal creeks was also acquired using remote sensing waterline method and measured width-depth ratios.Paired measurements of saltmarsh community and spectral characteristics were carried out and were then ordinated using Principal Component Analysis (PCA). According to the relationship between vegetation spectrum and environmental factors, vegetation cover, elevation of tidal flat and tidal creeks were selected as additional information for vegetation classification. With the fuzzy matrix and previous classification, the second classification was made on the error pixels in terms of fuzzy membership function of different vegetation and additional information. The results indicated this approach obviously improved the accuracy of vegetation classification.Heighted waterlines and corresponding field measurements were employed to acquire the mean annual accretion/erosion rate between satellite and survey time. We also studied the accretion/erosion pattern and factors which influenced the evolution of tidal flat. The interactions of between vegetation and tidal elevation were also investigated according to the extracted vegetation and inversed DEM at Dongtan. To analyze the impacts of vegetation on hydrodynamics, tidal current and suspended sediment concentration were measured and compared at vegetation and unvegetation fields respectively. The interaction of the accretion/erosion rate and extracted vegetation using remote sensing method showed the same conclusion with which was made by field survey.At last, 3D visualization of tidal flat, tidal level and nearshore flow field were implemented by the field visualization technique of GIS, which made marine scientist more intuitionistic understanding the hydrodynamic process at coastal region.更多还原