This study involves stepwise application of Unconstrained Linear Mixer Model (ULMM) for sub-pixel classification of residential areas using Land sat 7 TM image. The image was geometrically and radiometrically corrected and spectral enhancement and classifications were done to determine the possible number of target classes. In the first step, five end-members were used as inputs and the pixels which were considered as well fit to ULMM were identified as outputs. The unidentified pixels were separated and taken to the second step with new end members. This method identified 52% of the mixed pixels were identified in the first phase and 6% in the second phase. 42% of the pixels were left as unidentified after the two steps. The pixels identified by ULMM were grouped into high and low density residential subclasses. The resulting image indicated very low RMS errors. However the percentages of pixels unidentified were high. The independent accuracy test carried out using census population density and the resulting image indicated a low relationship. A hyper-spectral imagery with finer spatial resolution may provide a better sub pixel classification. / Department of Geography

Urbanization is one of the most significant phenomena of the anthropogenic influence on the Earth’s environment. One of the principal results of the urbanization is the creation of megacities, with their local climate and high impact on the surrounding area. The design and evolution of an urban area leads to higher absorption of solar radiation and heat storage in which is the foundation of the urban heat island phenomenon. Remote sensing data is a valuable source of information for urban climatology studies. The main objective of this thesis research is to examine the relationship between land use and land cover types and corresponding land surface temperature, as well as the urban heat island effect and changes in these factors over a 10 year period. 10 megacities around the world where included in this study namely Beijing (China), Delhi (India), Dhaka (Bangladesh), Los Angeles (USA), London (UK), Mexico City (Mexico), Moscow (Russia), New York City (USA), Sao Paulo (Brazil) and Tokyo (Japan). Landsat satellite data were used to extract land use/land cover information and their changes for the abovementioned cities. Land surface temperature was retrieved from Landsat thermal images. The relationship between land surface temperature and landuse/land-cover classes, as well as the normalized vegetation index (NDVI) was analyzed. The results indicate that land surface temperature can be related to land use/land cover classes in most cases. Vegetated and undisturbed natural areas enjoy lower surface temperature, than developed urban areas with little vegetation. However, the cities show different trends, both in terms of the size and spatial distribution of urban heat island. Also, megacities from developed countries tend to grow at a slower pace and thus face less urban heat island effects than megacities in developing countries.

A urbanização modifica a superfície, promovendo a troca da cobertura natural por materiais de construção. As áreas urbanas, além do solo impermeável, têm a presença de edifícios, que alteram a rugosidade da superfície, a velocidade e a direção dos ventos e provocam o sombreamento da superfície, bloqueando a incidência da energia solar. Assim, analisar a geometria de ocupação da cidade é importante para o entendimento do clima urbano e para o planejamento da cidade. O sensoriamento remoto é uma importante forma de obtenção de informações das áreas urbanas, contudo é preciso considerar a heterogeneidade deste ambiente e a mistura espectral existente nos dados satelitais. Dessa forma, o modelo linear de mistura espectral apresenta-se como importante método de extração de informações dos ambientes urbanos. O objetivo deste trabalho é identificar áreas com padrão horizontal e com padrão vertical de ocupação urbana, em Porto Alegre, e relacionar essa característica geométrica com valores de temperatura de superfície. Para tanto, imagens do sensor TM do satélite Landsat 5, adquiridas entre 1984 e 2009 foram utilizadas, bem como dados censitários, dados meteorológicos e modelos gerados por varredura laser. A partir das imagens TM foram geradas três imagens fração: solo, sombra e vegetação. A fração solo foi utilizada na identificação de áreas de ocupação horizontal e de expansão urbana, e a fração sombra foi utilizada na identificação de áreas verticalizadas. Utilizando as mesmas imagens, obtiveram-se os valores de temperatura de superfície. As áreas com ocupação horizontal, caracterizadas por moradias em casas, apresentaram baixos valores de sombra e altos valores de solo. As áreas verticalizadas apresentaram altos valores de sombra e baixos valores de solo. Os resultados extraídos das imagens fração têm similaridade com dados de artigos científicos e com os dados da varredura laser. A temperatura de superfície, em Porto Alegre, mostrou forte correlação com dados meteorológicos, e se caracteriza por valores mais altos nas áreas urbanizadas e mais baixos onde a ocupação é rarefeita. Nas áreas urbanizadas, maiores valores de temperatura de superfície são encontrados nas regiões com padrão de ocupação horizontal, enquanto os menores valores são encontrados nas regiões verticalizadas. A metodologia escolhida gerou resultados compatíveis com outros dados de uso e ocupação do solo, provenientes de diferentes fontes, e contribui com características da área urbana e do clima urbano da cidade de Porto Alegre, informações essas escassas nos principais bancos de dados acadêmicos. / The urbanization modifies the landscape, promoting changes from natural to man-made environment. Besides the impermeable soil, the urban areas have a lot of buildings, that changes the surface roughness, the wind speed and direction and also are responsible for shading the surface, blocking the incidence of solar energy. Analysing the city occupation geometry is important to understanding of the urban climate behaviour, and naturally the city planning. Remote sensing is a very important tool to get information about the urban areas, but is necessary to consider the heterogeneity of this environment and the existing spectral mixing in satellite data. Based on this, the linear model of spectral mixing can be classified as an important method of information extraction from urban environments. The goal of this research is to identify areas with horizontal and vertical patterns of urban occupation in the city of Porto Alegre – Brazil and relate this geometric characteristic with values of surface temperature. Therefore, images of the TM sensor of the Landsat 5 satellite were used (during the 1984-2009 period) and also the census data, meteorological data and models generated by laser scanning. Three fraction images were generated based on TM images: soil, shade and vegetation. The soil fraction was used for the identification of the areas with horizontal occupation and urban expansion, and the shadow fraction was used to identify verticalized areas. Based on the same images the surface temperature was obtained. Areas with horizontal occupation, mostly represented by houses, presented low shading values and high soil values. On the other hand, verticalized areas presented high shading values and low soil values. These results, obtained from the images fraction, are similar with the results from scientific papers and data from laser scanning. In Porto Alegre, the surface temperature indicated strong correlation with meteorological data, and was characterized by higher values in urbanized areas and lower values where the occupation is least intense. In urban areas, higher values of temperature are found in areas with horizontal occupation pattern, while the lowest values are found in verticalized regions. Furthermore, it is possible to suggest that the chosen methodology lead to conclusions that are consistent with other data of land use and occupation from different sources. Contributing with some information about characteristics of the urban area and urban climate of the city of Porto Alegre, which are usually not well documented in academic databases.

A urbanização modifica a superfície, promovendo a troca da cobertura natural por materiais de construção. As áreas urbanas, além do solo impermeável, têm a presença de edifícios, que alteram a rugosidade da superfície, a velocidade e a direção dos ventos e provocam o sombreamento da superfície, bloqueando a incidência da energia solar. Assim, analisar a geometria de ocupação da cidade é importante para o entendimento do clima urbano e para o planejamento da cidade. O sensoriamento remoto é uma importante forma de obtenção de informações das áreas urbanas, contudo é preciso considerar a heterogeneidade deste ambiente e a mistura espectral existente nos dados satelitais. Dessa forma, o modelo linear de mistura espectral apresenta-se como importante método de extração de informações dos ambientes urbanos. O objetivo deste trabalho é identificar áreas com padrão horizontal e com padrão vertical de ocupação urbana, em Porto Alegre, e relacionar essa característica geométrica com valores de temperatura de superfície. Para tanto, imagens do sensor TM do satélite Landsat 5, adquiridas entre 1984 e 2009 foram utilizadas, bem como dados censitários, dados meteorológicos e modelos gerados por varredura laser. A partir das imagens TM foram geradas três imagens fração: solo, sombra e vegetação. A fração solo foi utilizada na identificação de áreas de ocupação horizontal e de expansão urbana, e a fração sombra foi utilizada na identificação de áreas verticalizadas. Utilizando as mesmas imagens, obtiveram-se os valores de temperatura de superfície. As áreas com ocupação horizontal, caracterizadas por moradias em casas, apresentaram baixos valores de sombra e altos valores de solo. As áreas verticalizadas apresentaram altos valores de sombra e baixos valores de solo. Os resultados extraídos das imagens fração têm similaridade com dados de artigos científicos e com os dados da varredura laser. A temperatura de superfície, em Porto Alegre, mostrou forte correlação com dados meteorológicos, e se caracteriza por valores mais altos nas áreas urbanizadas e mais baixos onde a ocupação é rarefeita. Nas áreas urbanizadas, maiores valores de temperatura de superfície são encontrados nas regiões com padrão de ocupação horizontal, enquanto os menores valores são encontrados nas regiões verticalizadas. A metodologia escolhida gerou resultados compatíveis com outros dados de uso e ocupação do solo, provenientes de diferentes fontes, e contribui com características da área urbana e do clima urbano da cidade de Porto Alegre, informações essas escassas nos principais bancos de dados acadêmicos. / The urbanization modifies the landscape, promoting changes from natural to man-made environment. Besides the impermeable soil, the urban areas have a lot of buildings, that changes the surface roughness, the wind speed and direction and also are responsible for shading the surface, blocking the incidence of solar energy. Analysing the city occupation geometry is important to understanding of the urban climate behaviour, and naturally the city planning. Remote sensing is a very important tool to get information about the urban areas, but is necessary to consider the heterogeneity of this environment and the existing spectral mixing in satellite data. Based on this, the linear model of spectral mixing can be classified as an important method of information extraction from urban environments. The goal of this research is to identify areas with horizontal and vertical patterns of urban occupation in the city of Porto Alegre – Brazil and relate this geometric characteristic with values of surface temperature. Therefore, images of the TM sensor of the Landsat 5 satellite were used (during the 1984-2009 period) and also the census data, meteorological data and models generated by laser scanning. Three fraction images were generated based on TM images: soil, shade and vegetation. The soil fraction was used for the identification of the areas with horizontal occupation and urban expansion, and the shadow fraction was used to identify verticalized areas. Based on the same images the surface temperature was obtained. Areas with horizontal occupation, mostly represented by houses, presented low shading values and high soil values. On the other hand, verticalized areas presented high shading values and low soil values. These results, obtained from the images fraction, are similar with the results from scientific papers and data from laser scanning. In Porto Alegre, the surface temperature indicated strong correlation with meteorological data, and was characterized by higher values in urbanized areas and lower values where the occupation is least intense. In urban areas, higher values of temperature are found in areas with horizontal occupation pattern, while the lowest values are found in verticalized regions. Furthermore, it is possible to suggest that the chosen methodology lead to conclusions that are consistent with other data of land use and occupation from different sources. Contributing with some information about characteristics of the urban area and urban climate of the city of Porto Alegre, which are usually not well documented in academic databases.

A urbanização modifica a superfície, promovendo a troca da cobertura natural por materiais de construção. As áreas urbanas, além do solo impermeável, têm a presença de edifícios, que alteram a rugosidade da superfície, a velocidade e a direção dos ventos e provocam o sombreamento da superfície, bloqueando a incidência da energia solar. Assim, analisar a geometria de ocupação da cidade é importante para o entendimento do clima urbano e para o planejamento da cidade. O sensoriamento remoto é uma importante forma de obtenção de informações das áreas urbanas, contudo é preciso considerar a heterogeneidade deste ambiente e a mistura espectral existente nos dados satelitais. Dessa forma, o modelo linear de mistura espectral apresenta-se como importante método de extração de informações dos ambientes urbanos. O objetivo deste trabalho é identificar áreas com padrão horizontal e com padrão vertical de ocupação urbana, em Porto Alegre, e relacionar essa característica geométrica com valores de temperatura de superfície. Para tanto, imagens do sensor TM do satélite Landsat 5, adquiridas entre 1984 e 2009 foram utilizadas, bem como dados censitários, dados meteorológicos e modelos gerados por varredura laser. A partir das imagens TM foram geradas três imagens fração: solo, sombra e vegetação. A fração solo foi utilizada na identificação de áreas de ocupação horizontal e de expansão urbana, e a fração sombra foi utilizada na identificação de áreas verticalizadas. Utilizando as mesmas imagens, obtiveram-se os valores de temperatura de superfície. As áreas com ocupação horizontal, caracterizadas por moradias em casas, apresentaram baixos valores de sombra e altos valores de solo. As áreas verticalizadas apresentaram altos valores de sombra e baixos valores de solo. Os resultados extraídos das imagens fração têm similaridade com dados de artigos científicos e com os dados da varredura laser. A temperatura de superfície, em Porto Alegre, mostrou forte correlação com dados meteorológicos, e se caracteriza por valores mais altos nas áreas urbanizadas e mais baixos onde a ocupação é rarefeita. Nas áreas urbanizadas, maiores valores de temperatura de superfície são encontrados nas regiões com padrão de ocupação horizontal, enquanto os menores valores são encontrados nas regiões verticalizadas. A metodologia escolhida gerou resultados compatíveis com outros dados de uso e ocupação do solo, provenientes de diferentes fontes, e contribui com características da área urbana e do clima urbano da cidade de Porto Alegre, informações essas escassas nos principais bancos de dados acadêmicos. / The urbanization modifies the landscape, promoting changes from natural to man-made environment. Besides the impermeable soil, the urban areas have a lot of buildings, that changes the surface roughness, the wind speed and direction and also are responsible for shading the surface, blocking the incidence of solar energy. Analysing the city occupation geometry is important to understanding of the urban climate behaviour, and naturally the city planning. Remote sensing is a very important tool to get information about the urban areas, but is necessary to consider the heterogeneity of this environment and the existing spectral mixing in satellite data. Based on this, the linear model of spectral mixing can be classified as an important method of information extraction from urban environments. The goal of this research is to identify areas with horizontal and vertical patterns of urban occupation in the city of Porto Alegre – Brazil and relate this geometric characteristic with values of surface temperature. Therefore, images of the TM sensor of the Landsat 5 satellite were used (during the 1984-2009 period) and also the census data, meteorological data and models generated by laser scanning. Three fraction images were generated based on TM images: soil, shade and vegetation. The soil fraction was used for the identification of the areas with horizontal occupation and urban expansion, and the shadow fraction was used to identify verticalized areas. Based on the same images the surface temperature was obtained. Areas with horizontal occupation, mostly represented by houses, presented low shading values and high soil values. On the other hand, verticalized areas presented high shading values and low soil values. These results, obtained from the images fraction, are similar with the results from scientific papers and data from laser scanning. In Porto Alegre, the surface temperature indicated strong correlation with meteorological data, and was characterized by higher values in urbanized areas and lower values where the occupation is least intense. In urban areas, higher values of temperature are found in areas with horizontal occupation pattern, while the lowest values are found in verticalized regions. Furthermore, it is possible to suggest that the chosen methodology lead to conclusions that are consistent with other data of land use and occupation from different sources. Contributing with some information about characteristics of the urban area and urban climate of the city of Porto Alegre, which are usually not well documented in academic databases.

This doctoral thesis investigates the potential of classification methods based on spatial context to infer specific forms of land use from remote sensing data. The problem is that some types of land use are characterized by a complex configuration of land covers that traditional per-pixel based methods have problems classifying due to spectral heterogeneity. The problem of spectral heterogeneity is also present in classification of high resolution imagery. Two novel methods based on contextual information are evaluated, Spatial Relational Post-Classification (SRPC) and Window Independent Context Segmentation (WICS). The thesis includes six case studies in rural and urban areas focusing on the classification of: agricultural systems, urban characteristics, and dead wood areas. In the rural case studies specific types of agricultural systems associated with different household strategies are mapped by inferring the physical expression of land use using the SRPC method. The urban remote sensing studies demonstrate how the WICS method is able to extract information corresponding to different phases of development. Additionally, different urban classes are shown to correspond to different socioeconomic profiles, demonstrating how urban remote sensing can be used to make a connection between the physical environment and the social lives of residents. Finally, in one study the WICS method is used to successfully classify dead trees from high resolution imagery. Taken together these studies demonstrate how approaches based on spatial context can be used to extract information on land use in rural and urban environments where land use manifests itself in the form of complex spectral class and land cover patterns. The thesis, thus, contributes to the research field by showing that contextual methods can capture multifaceted patterns that can be linked to land use. This, in turn, enables an increased use of remote sensing data, particularly in the social sciences. / <p>At the time of the doctoral defense, the following papers were unpublished and had a status as follows: Paper 3: Manuscript. Paper 4: Manuscript. Paper 5: Manuscript. Paper 6: Manuscript.</p>

劇烈的城市化過程已經在世界上多個地區發生並產生了許多的城市群，珠江三角洲正是這樣的一個城市群。目前，珠江三角洲上的城市土地利用和土地覆蓋已經發生了巨大的變化。而其中最重要的一個結果就是大量城市不透水層的出現，並已經極大地影響著當地的城市環境，如城市洪水、城市氣候、水污染和大氣污染等。因此，城市不透水層及其分佈的估算對於監測和管理城市化進程及其對環境的影響有著重要的意義。然而，由於城市土地覆蓋類型的多樣性，精確的城市不透水層的估算（ISE）仍然是一個極具挑戰性的課題。本論文旨在通過融合光學遙感和合成孔徑雷達（SAR）遙感技術來提高亞熱帶濕潤區城市不透水層估算的精度。此外，論文還將探索亞熱帶濕潤區土地覆蓋類型分類的季節性變化及其對城市不透水層提取的影響。本論文的研究結果主要包括以下幾個部分。 / 首先，本研究發現亞熱帶濕潤區不透水層提取的季節性效應與中緯度地區截然不同。在亞熱帶濕潤地區，冬季是最適合用遙感影像進行不透水層提取的。原因是由於冬季是旱季，雲量少，許多可變來源區域（VSA）沒有水體覆蓋，而在遙感影像中，水體容易和暗不透水層混淆。另一方面，秋季最不適合不透水層提取，因為存在大量的雲層，並且，大量的降水導致VSA區域充滿水，從而增加了與暗不透水層混淆的區域面積。此外，大量的雲層在影像中也是呈現高反射特徵的，因此極容易和亮不透水層混淆，這是秋季不適合用於提取不透水層的另一重要原因。 / 其次，提出了一種新的基於形狀自我調整鄰域（SAN）的特徵提取演算法。該特徵提取演算法類比人類視覺對圖像感知的強大能力，進行遙感影像低層特徵的提取。實驗結果表明，SAN特徵提取方法對非監督分類有顯著的提高，其中總體分類精度從0.58提高到0.86，而Kappa係數從0.45提高到0.80。SAN特徵對於監督分類的精度也有一定的提高，這些都表明，與傳統的特徵提取方法相比，SAN特徵對遙感影像分類具有重要的作用。 / 再次，通過對比分析光學遙感影像和SAR影像發現，單獨採用光學遙感影像進行不透水層提取比單獨採用SAR影像取得更好的結果。同時，單獨採用光學遙感（Landsat ETM+）時，支援向量機（SVM）比人工神經網路（ANN）取得更好的結果，這是因為ANN對於亮不透水層與幹裸土之間，以及暗不透水層與陰影之間的光譜混淆更加敏感。然而，當單獨採用SAR遙感（ENVISAT ASAR）時，ANN則取得更好的結果，這是由於SVM分析SAR影像時更容易產生雜訊，並具有明顯的邊緣效應。因此，融合光學遙感和SAR遙感具有重要的意義。通過比較不同圖像融合層次發現，像元級融合（Pixel Level Fusion）會降低單獨採用光學遙感提取不透水層的精度，因而不適合光學和SAR影像的融合。而特徵級融合（Feature Level）決策級融合（Decision Level）可以更好的把不透水層從陰影區域和裸土中區分出來，因為更加適合光學與SAR的融合。 / 最後，三組不同的光學遙感和SAR遙感影像被用於評估本論文提出的光學和SAR融合方法，包括Landsat ETM+與ASAR影像，SPOT-5與ASAR影像，以及SPOT-5與TerraSAR-X影像。此外，還比較了不同的融合方法（人工神經網路、支援向量機和隨機森林）對融合結果的影響。結果表明，用光學和SAR遙感影像融合提取不透水層有利於減少在光學遙感影像中容易出現的光譜混淆現象，從而提高不透水層提取的精度。另外，隨機森林在融合光學和SAR影像中效果較其它兩種方法，因為隨機森林對兩種不同的資料來源區別對待，而這正是符合光學遙感與SAR遙感截然不同的工作方式的特點，從而能更好的融合光學遙感和SAR遙感。 / 本論文的研究成果有助於探索亞熱帶濕潤區中物候特點和氣候特點對城市不透水層提取所產生的季節性效應；同時也為融合光學遙感和SAR遙感影像提取城市不透水層提供了一個技術框架。由於珠江三角洲是一個亞熱帶濕潤區一個典型的快速城市化的城市群區域，本文所提出的方法框架和所得到研究結論可為世界上其它亞熱帶濕潤區的城市遙感研究提供一定的參考價值。 / Dramatic urbanization processes have happened in many regions and thus created a number of metropolises in the world. The Pearl River Delta (PRD) is one of such typical areas, where the urban land use/land cover has been significantly changing in the recent past. As one of the most important implications, a large increment of impervious surface (IS) turned out to be one of the features of fast urbanization process and has been influencing the urban environment significantly, including urban flooding, urban climate, water pollutions, and air pollutions. Therefore, the estimation of IS would be very helpful to monitor and manage the urbanization process and its impacts on the environment. However, accurate estimation of urban IS remains challenging due to the diversity of land covers. This dissertation attempts to fuse optical and SAR remote sensing data to improve the accuracy of urban impervious surface estimation (ISE) in humid subtropical regions (HSR). The seasonal characteristics of land covers and its impacts on ISE in HSR are all investigated. Some interesting findings are summarized as follows. / Firstly, the study demonstrates quite a special pattern of the seasonal effects of ISE in humid subtropical areas that is different from that in mid-latitude areas. According to the results, in subtropical monsoon regions, winter is the best season to estimate IS from satellite images. There are little clouds, and most of the Variable Source Areas (VSA) is not filled with water. On the other hand, autumn images obtained the lowest accuracy of IS due to the clouds coverage and the water in VSA. Autumn is a rainy season in a subtropical monsoon region, for which clouds occur very often and VSA areas are always filled with water. Consequently, clouds are confused with bright IS due to their similarly high reflectance, and more water in VSA is confused with dark IS due to their similarly low reflectance. / Secondly, a novel feature extraction technique, based on the shape-adaptive neighborhood (SAN), is proposed to incorporate the advantages of human vision into the process of remote sensing images. Quantitative results showed that improvement of SAN features is particularly significant improvement for the unsupervised classifier, for which the overall accuracy increased from 0.58 to 0.86, and the Kappa coefficient increased from 0.45 to 0.80, indicating promising applications of SAN features in the unsupervised processing of remote sensing images. / Thirdly, a comparison study of ISE between optical and SAR image demonstrates that single optical image provides better results than using single SAR image. In addition, results indicate that support vector machine (SVM) is a better choice for ISE using Landsat ETM+ (optical) images, while artificial neural network (ANN) turns out to be more sensitive to the confusion between dry soils and bright IS, and between shades and dark IS. However, ANN gets a better result using ASAR (SAR) image with higher accuracy, while the SVM classifier produces more noises and has some edge effects. Considering both the merits and demerits of optical and SAR images, synergistically fusing the two data sources should be a promising solution. Comparison of three different levels of fusion shows that pixel level fusion seems not appropriate for optical-SAR fusion, as it reduces the accuracy compared to the single use of optical data. Meanwhile, feature level fusion and decision level fusion obtained better accuracy, since they improves the identification of IS from shaded areas and bare soils. / Fourthly, a methodological framework of fusing the optical and SAR images is proposed. Three different data sets are used to assess the effectiveness of this methodological framework, including the Landsat TM and ASAR images, the SPOT-5 and ASAR images, and the SPOT-5 and TerraSAR-X images. In addition, different methods (e.g. ANN, SVM and Random Forest) are employed and compared to fusion the two data sources at a mixed level fusion of pixel and feature levels. Experimental results showed that the combined use of optical and SAR image is able to effectively improve the accuracy of ISE by reducing the spectral confusions that happen easily in optical images. Moreover, Random Forest (RF) demonstrated a promising performance for fusing optical and SAR images as it treats the two data sources differently through a random selection procedure of variables from different data sources. / The major outcome of this research provides evidence of the seasonal effects on IS assessment due to phenological and climatic characteristics, as well as provides an applicable framework of methodology for the synergistic use of optical and SAR images to improve the ISE. Since the PRD region is highly typical of many fast growing areas, the methodology and conclusions of this research would serve as a useful reference for other subtropical, humid regions of the world. / Detailed summary in vernacular field only. / Detailed summary in vernacular field only. / Detailed summary in vernacular field only. / Detailed summary in vernacular field only. / Detailed summary in vernacular field only. / Detailed summary in vernacular field only. / Zhang, Hongsheng. / Thesis (Ph.D.)--Chinese University of Hong Kong, 2013. / Includes bibliographical references (leaves 169-185). / Abstract also in Chinese. / ACKNOWLEDGEMENTS --- p.i / ABSTRACT --- p.iii / 論文摘要 --- p.vii / Table of Contents --- p.xi / List of Abbreviations --- p.xv / List of Tables --- p.xvii / List of Figures --- p.xviii / Chapter CHAPTER 1 --- INTRODUCTION --- p.1 / Chapter 1.1 --- Research background --- p.1 / Chapter 1.2 --- Research questions and hypotheses --- p.4 / Chapter 1.3 --- Objectives and significance --- p.6 / Chapter 1.4 --- Organization of the thesis --- p.7 / Chapter CHAPTER 2 --- LITERATURE REVIEW --- p.11 / Chapter 2.1 --- Introduction --- p.11 / Chapter 2.2 --- Significance of Impervious Surface --- p.12 / Chapter 2.2.1 --- Environmental significance --- p.12 / Chapter 2.2.2 --- Socio-economic significance --- p.16 / Chapter 2.3 --- Climatology and Phenology in HSR --- p.18 / Chapter 2.3.1 --- Characteristics of the climate and phenology --- p.18 / Chapter 2.3.2 --- Seasonal effects from Climatology and Phenology --- p.20 / Chapter 2.4 --- Land-cover complexity in rapid urbanized region --- p.21 / Chapter 2.5 --- Approaches of ISE --- p.22 / Chapter 2.5.1 --- Sub-pixel approaches --- p.22 / Chapter 2.5.2 --- Per-pixel approaches --- p.23 / Chapter 2.5.3 --- Synergistic use of optical and SAR data for ISE --- p.27 / Chapter 2.6 --- Summary --- p.28 / Chapter CHAPTER 3 --- STUDY AREA AND DATA SETS --- p.31 / Chapter 3.1 --- Study area --- p.31 / Chapter 3.1.1 --- Site A: Guangzhou --- p.32 / Chapter 3.1.2 --- Site B: Shenzhen --- p.33 / Chapter 3.1.3 --- Site C: Hong Kong --- p.34 / Chapter 3.2 --- Satellite data --- p.35 / Chapter 3.2.1 --- Landsat ETM+ --- p.35 / Chapter 3.2.2 --- SPOT-5 --- p.36 / Chapter 3.2.3 --- ENVISAT ASAR --- p.36 / Chapter 3.2.4 --- TerraSAR-X --- p.37 / Chapter 3.3 --- Digital Orthophoto data --- p.38 / Chapter 3.4 --- In-situ data --- p.39 / Chapter 3.5 --- Summary --- p.40 / Chapter CHAPTER 4 --- METHODOLOGY --- p.43 / Chapter 4.1 --- Framework --- p.43 / Chapter 4.2 --- Per-pixel modeling of ISE --- p.45 / Chapter 4.3 --- Investigation of the seasonal effects --- p.46 / Chapter 4.4 --- Feature extraction --- p.47 / Chapter 4.4.1 --- Conventional approach --- p.48 / Chapter 4.4.2 --- Novel approach based on shape-adaptive neighborhood --- p.48 / Chapter 4.4.2.1 --- The concept of shape-adaptive neighborhood --- p.49 / Chapter 4.4.2.2 --- The determination of a shape-adaptive neighborhood --- p.51 / Chapter 4.4.2.3 --- Extracting spatial features --- p.54 / Chapter 4.5 --- Fusing the optical and SAR data --- p.58 / Chapter 4.5.1 --- Multi-source image co-registration --- p.60 / Chapter 4.5.2 --- Compare the single use of optical and SAR image --- p.61 / Chapter 4.5.3 --- Compare different levels of fusion --- p.62 / Chapter 4.5.4 --- Fusion with supervised classifiers --- p.65 / Chapter 4.5.4.1 --- Artificial neural network --- p.66 / Chapter 4.5.4.2 --- Support vector machine --- p.68 / Chapter 4.5.4.3 --- Random Forest --- p.71 / Chapter 4.6 --- Results validation and accuracy assessment --- p.75 / Chapter 4.6.1 --- Training and testing data sampling --- p.75 / Chapter 4.6.2 --- Accuracy assessment --- p.76 / Chapter 4.7 --- Summary --- p.77 / Chapter CHAPTER 5 --- RESULTS AND DISCUSSION (I) - ASSESSMENT OF SAN FEATURES --- p.79 / Chapter 5.1. --- Analysis of threshold to determine the SAN --- p.79 / Chapter 5.2. --- Feature extraction from SAN --- p.80 / Chapter 5.3. --- Assessment of the SAN features with classification --- p.82 / Chapter 5.3.1 --- Training samples and classification --- p.82 / Chapter 5.3.2 --- Testing samples and accuracy --- p.84 / Chapter 5.3.3 --- Assess the effectiveness of the SAN based features --- p.85 / Chapter 5.4 --- Summary --- p.87 / Chapter CHAPTER 6 --- RESULTS AND DISCUSSION (II) - SEASONAL EFFECTS OF ISE --- p.89 / Chapter 6.1 --- Seasonal effects of ISE --- p.89 / Chapter 6.2 --- Analyzing the seasonal changes of typical targets --- p.92 / Chapter 6.3 --- Comparing the seasonal sensitivity of methods --- p.96 / Chapter 6.4 --- Summary --- p.97 / Chapter CHAPTER 7 --- RESULTS AND DISCUSSION (III) - URBAN LAND COVER DIVERSITY --- p.101 / Chapter 7.1 --- Introduction --- p.101 / Chapter 7.2 --- Urban LC classification Using RF --- p.102 / Chapter 7.2.1 --- Optimization of RF --- p.102 / Chapter 7.2.2 --- Land covers classification with optimized RF --- p.104 / Chapter 7.2.3 --- Compare RF with other decision tree-based methods --- p.107 / Chapter 7.3 --- Summary --- p.108 / Chapter CHAPTER 8 --- RESULTS AND DISCUSSION (IV) - FUSING OPTICAL&SAR DATA --- p.111 / Chapter 8.1 --- Introduction --- p.111 / Chapter 8.2 --- Comparison of ISE with single optical and SAR data --- p.111 / Chapter 8.2.1 --- ISE with ETM+ data --- p.112 / Chapter 8.2.1.1 --- Mapping the IS --- p.112 / Chapter 8.2.1.2 --- Effects of the parameter configurations of the methods --- p.114 / Chapter 8.2.2 --- ISE with ASAR data --- p.115 / Chapter 8.2.2.1 --- Mapping the IS --- p.115 / Chapter 8.2.2.2 --- Effects of the parameter configurations of the methods --- p.117 / Chapter 8.2.3 --- Comparisons over the data and methods --- p.119 / Chapter 8.2.4 --- Discussion and implications --- p.121 / Chapter 8.3 --- Comparison of different levels of fusion method --- p.122 / Chapter 8.3.1 --- Fusion strategies at different levels --- p.122 / Chapter 8.3.2 --- Results of feature extractions --- p.124 / Chapter 8.3.3 --- Fusion results on different levels --- p.126 / Chapter 8.3.4 --- Comparisons --- p.128 / Chapter 8.3.5 --- Discussion and implications --- p.129 / Chapter 8.4 --- Synergizing optical and SAR data with RF --- p.130 / Chapter 8.4.1 --- Feature extraction from ASAR data --- p.130 / Chapter 8.4.2 --- Determine the optimal number of features in each decision tree --- p.132 / Chapter 8.4.3 --- Determine the optimal numbers of decision trees in the RF --- p.134 / Chapter 8.4.4 --- ISE with optimized RF --- p.135 / Chapter 8.4.5 --- Discussion and implications --- p.140 / Chapter 8.5 --- A comprehensive study: ISE using SPOT-5 and TerraSAR-X data --- p.142 / Chapter 8.5.1 --- Data set and experiment design --- p.142 / Chapter 8.5.2 --- Feature extraction of SPOT-5 data --- p.145 / Chapter 8.5.3 --- Feature extraction of TerraSAR-X data --- p.148 / Chapter 8.5.4 --- LULC classification with optimized models --- p.149 / Chapter 8.5.5 --- ISE with optimized models --- p.152 / Chapter 8.5.6 --- Discussion and implications --- p.155 / Chapter 8.6 --- Summary --- p.156 / Chapter CHAPTER 9 --- CONCLUSIONS --- p.159 / Chapter 9.1 --- Findings and conclusions --- p.159 / Chapter 9.1.1 --- Seasonal effects of ISE in HSR --- p.159 / Chapter 9.1.2 --- Feature extraction methods --- p.160 / Chapter 9.1.3 --- Comparison between optical and SAR data --- p.161 / Chapter 9.1.4 --- Fusion level and fusion methods --- p.162 / Chapter 9.2 --- Recommendations for future research --- p.163 / Chapter 9.2.1 --- Feature extraction --- p.163 / Chapter 9.2.2 --- Study areas selection and design --- p.163 / Chapter 9.2.3 --- Validation with in-situ data --- p.164 / Chapter 9.2.4 --- Fusion level and strategy --- p.164 / Chapter 9.2.5 --- Fusion methods --- p.165 / References --- p.169 / Chapter Appendix I --- Codes for Determining Shape-adaptive Neighborhood --- p.186 / Chapter Appendix II --- Publication list related to this thesis research --- p.188

Das weltweite Ausmaß der Urbanisierung zählt zu den großen ökologischen Herausforderungen des 21. Jahrhunderts. Die Fernerkundung bietet die Möglichkeit das Verständnis dieses Prozesses und seiner Auswirkungen zu erweitern. Der Fokus dieser Arbeit lag in der Quantifizierung der städtischen Landbedeckung mittels Maschinellen Lernens und räumlich unterschiedlich aufgelöster Hyperspektraldaten. Untersuchungen berücksichtigten innovative methodische Entwicklungen und neue Möglichkeiten, die durch die bevorstehende Satellitenmission EnMAP geschaffen werden. Auf Basis von Bilder des flugzeugestützten HyMap Sensors mit Auflösungen von 3,6 m und 9 m sowie simulierten EnMAP-Daten mit einer Auflösung von 30 m wurde eine Kartierung entlang des Stadt-Umland-Gradienten Berlins durchgeführt. Im ersten Teil der Arbeit wurde die Kombination von Support Vektor Regression mit synthetischen Trainingsdaten für die Subpixelkartierung eingeführt. Ergebnisse zeigen, dass sich der Ansatz gut zur Quantifizierung thematisch relevanter und spektral komplexer Oberflächenarten eignet, dass er verbesserte Ergebnisse gegenüber weiteren Subpixelverfahren erzielt, und sich als universell einsetzbar hinsichtlich der räumlichen Auflösung erweist. Im zweiten Teil der Arbeit wurde der Wert zukünftiger EnMAP-Daten für die städtische Fernerkundung abgeschätzt. Detaillierte Untersuchungen unterstreichen deren Eignung für eine verbesserte und erweiterte Beschreibung der Stadt nach dem bewährten Vegetation-Impervious-Soil-Schema. Analysen der Möglichkeiten und Grenzen zeigen sowohl Nachteile durch die höhere Anzahl von Mischpixel im Vergleich zu hyperspektralen Flugzeugdaten als auch Vorteile aufgrund der verbesserten Differenzierung städtischer Materialien im Vergleich zu multispektralen Daten. Insgesamt veranschaulicht diese Arbeit, dass die Kombination von hyperspektraler Satellitenbildfernerkundung mit Methoden des Maschinellen Lernens eine neue Qualität in die städtische Fernerkundung bringen kann. / The global dimension of urbanization constitutes a great environmental challenge for the 21st century. Remote sensing is a valuable Earth observation tool, which helps to better understand this process and its ecological implications. The focus of this work was to quantify urban land cover by means of machine learning and imaging spectrometer data at multiple spatial scales. Experiments considered innovative methodological developments and novel opportunities in urban research that will be created by the upcoming hyperspectral satellite mission EnMAP. Airborne HyMap data at 3.6 m and 9 m resolution and simulated EnMAP data at 30 m resolution were used to map land cover along an urban-rural gradient of Berlin. In the first part of this work, the combination of support vector regression with synthetically mixed training data was introduced as sub-pixel mapping technique. Results demonstrate that the approach performs well in quantifying thematically meaningful yet spectrally challenging surface types. The method proves to be both superior to other sub-pixel mapping approaches and universally applicable with respect to changes in spatial scales. In the second part of this work, the value of future EnMAP data for urban remote sensing was evaluated. Detailed explorations on simulated data demonstrate their suitability for improving and extending the approved vegetation-impervious-soil mapping scheme. Comprehensive analyses of benefits and limitations of EnMAP data reveal both challenges caused by the high numbers of mixed pixels, when compared to hyperspectral airborne imagery, and improvements due to the greater material discrimination capability when compared to multispectral spaceborne imagery. In summary, findings demonstrate how combining spaceborne imaging spectrometry and machine learning techniques could introduce a new quality to the field of urban remote sensing.