城市高速发展和城市人口急剧膨胀等快速城市化进程，加剧了城市土地利用/覆盖的变化，也产生了一系列环境问题，城市热岛效应是其中最典型的问题之一。城市热岛效应影响因子复杂且众多，土地利用/覆盖变化是各因子相互影响的桥梁和纽带，因此，开展城市土地利用/覆盖变化对城市热岛效应的影响研究具有重要意义。本论文以北京市为例，基于城市土地利用/覆盖变化及其对地表城市热岛效应影响这一主题，利用多源遥感数据，从城市土地利用/覆盖变化的准确获取和时空格局分析，地表城市热岛效应遥感监测和时空格局变化分析，城市土地利用/覆盖变化对地表城市热岛效应的定量影响三大方面开展模型方法、时空格局分析以及定量影响研究，主要成果与结论如下:（1）提出了多源遥感数据相对辐射归一化方法MIPIF（Multiple Images with Pseudo-Invariant Features，MIPIF），该方法基于地表反射物理特性，自动选择包括暗集和亮集的伪不变特征点PIFs （Pseudo-Invariant Features，PIFs），并基于统计规则和最小绝对偏差法自动、逐步地迭代优化PIFs，能有效剔除PIFs中地类发生变化的像元点。四组不同景观类型的实验验证及基于定性分析、统计参数值和PIFs差异直方图的精度评价结果充分说明了MIPIF方法的有效性.（2)基于利用MIPIF方法归一化后的多期Landsat系列数据和北京二号卫星高分辨率数据，构建了北京市城市土地利用/覆盖三级结构分类体系，提出了融合面向对象、多指数分层分类、机器学习优势的多模型融合提取方法：采用目视解译及“流方式”勾画法得到1900-2015共6期的北京建成区边界；利用基于对象的多指数分层分类提取建成区二级土地利用/覆盖，并用J-M距离（Jeffreys-Matusita Distance，简称J-M距离）决定分层分类顺序；利用基于对象的随机森林分类方法提取首都功能核心区三级土地利用/覆盖，经验证，分类精度均达85%以上，较好满足城市土地利用/覆盖格局分析。（3）从总体格局及变化、城市扩展、城市内部优化三个维度对北京市1990-2015年间城市土地利用/覆盖时空格局进行深入分析，结果表明：北京市一级土地利用/覆盖中，耕地面积占比下降10%以上，城镇用地扩张最大，面积占比从3.0%扩张到12.3%。北京市建成区二级土地利用/覆盖中，不透水地表面积占比先增加后下降，城市绿地面积占比先下降后增加。北京市建成区扩展呈“跳跃式”增长，面积净增约3.0倍；城市扩展以占用大量耕地为主，25年占用耕地量达到城市总扩展量的53.5%，各扩展阶段耕地占用呈减少趋势。建成区内部各阶段变化特征不同。其中1990-2000年，以“向城市不透水地表转变”为主；2000-2010年，以“城市绿地转换与城市不透水地表互换”为主，2005-2010年互换区域南北对称分布；2010年后，以“向城市绿地转变”为主。（4）利用MODIS温度产品交叉验证了基于1990-2015年Landsat卫星数据反演的北京市地表温度的精度，继而采用城市热岛效应强度、城市热岛效应足迹以及城市热岛比例指数开展了北京市地表城市热岛效应时空格局分析，结果表明：地表温度反演结果与验证数据的决定系数在0.9以上，满足分析要求。年际变化上，整体表现出热岛效应范围加大，中心城区由于旧城改造、河流两岸绿化表现出热岛效应缓解；城市热岛比例指数由1990年的0.450上升到2010年的0.637，又缓降到2015年的0.507；热岛效应足迹由90年代的0.5倍城市面积距离处上升到2000年4-5倍距离又缓升到2015年的5-5.5倍距离。季节变化上，5月底到9月上旬热岛效应强度较高且范围覆盖六环以内城区；12月、1月有“冷岛效应”；夏季热岛效应足迹最大，达城市面积3-4.5倍距离处。（5）比较了城市不同土地利用/覆盖的热岛效应强度随年际、季节变化特征，从土地利用/覆盖组分、结构和物理性质3个维度构建了17个影响分析因子，采用随机森林模型开展了北京市土地利用/覆盖对热岛效应的定量影响分析，并探索其时空变化特征，结果表明：年际变化上，城市水体的热岛效应强度最低，城市不透水地表最高，城市绿地时高时低；季节变化上，冬季和初春各种土地利用/覆盖热岛效应强度差异较小，夏秋季差异加大。城市湖泊热岛效应缓解作用最强，设施用地热岛效应最强。2010年，17个因子对热岛效应影响按重要性排序前6个既有组分因子，又有物理性质因子和结构因子；各因子与城市热岛效应的偏依赖性关系各有不同；2010-2015年，建成区物理性质因子对热岛效应影响增大，结构因子影响略有下降；从建成区到首都功能核心区，结构因子对热岛效应的影响大幅增强，物理性质因子有所下降。无论建成区还是首都功能核心区，城市不透水地表面积和城市绿地面积比例都是地表城市热岛效应影响的主导因子。（6）提出加强城市热环境生态调控的途径：增加城市面状水域；控制城市不透水地表面积比例及空间结构，建议其加权形状因子大于2.6；增加城市绿地面积，避免空间破碎，尽量保持其加权形状因子不大于1.0；在城市土地利用/覆盖要素、结构确定的基础上，要考虑材质的物理属性。

Rapid urban development and expansion of urban population intensify urban land-use and land-cover change and produce a series of environmental problems. Urban Heat Island effect (UHI) is one of the most typical problems. The influencing factors of UHI are complex and numerous, and the land-use and land-cover change is the bridge and link between them. Therefore, it is of great significance to study the impact of urban land-use and land-cover change on UHI. This paper takes Beijing as an example and focuses on urban land-use and land-cover change and its impact on Surface Urban Heat Island effect (SUHI). This paper uses multi-source remotely sensed data to precisely extract urban land-use and land-cover change, analyze the temporal and spatial pattern, monitor SUHI and its temporal and spatial pattern, and analyze the quantitative impact of urban land-use and land-cover change on SUHI. The main achievements and conclusions are as follows:

Firstly, a new method for normalizing multiple images with pseudo-invariant features (PIFs) (MIPIF) was proposed. Based on the physical characteristics of surface reflection, MIPIF method automatically selects PIFs including dark set and bright sets, iteratively optimizes automatically and step by step based on statistical rules and least absolute deviation method. MIPIF method can effectively eliminate the pixel points in PIFs where land type changes. Four groups of experiments based on different landscape types were carried out to get qualitative analysis, statistical parameter values and PIFs difference histogram. All the results proved the MIPIF’s validity.

Secondly, three-level classification system of urban land-use and land-cover in Beijing was established based on normalized Landsat series images and Beijing-2 constellation high-resolution data. A multi-model fusion extraction method is also established which integrated object-oriented, multi-index hierarchical classification and machine learning. This method is described as follows. First, visual interpretation combined with "flow mode" sketch method was used to obtain Beijing built-up boundary in six periods from 1900 to 2015. Then, object-based segmentation with multi-index hierarchical classification method was proposed to extract land-use and land-cover of secondary level in Beijing built-up area, and J-M Distance method (Jeffreys-Matusita Distance, simplified J-M Distance) was used to determine the hierarchical classification order. Finally, object-based segmentation with random forest classification method was used to extract land-use and land-cover of the third level of capital functional core area. Validation results show that classification accuracy is over 85%, which satisfies the temporal and spatial analysis of urban land-use and land-cover.

Tirdly, the temporal and spatial pattern of urban land-use and land-cover in Beijing from 1990-2015 was analyzed from three dimensions: overall pattern and change, urban expansion and urban internal optimization. The results show that in the first-class land-use and land-cover in Beijing, the proportion of cultivated land decreased by more than 10%, and the urban land expanded the most, from 3.0% to 12.3%. In the second-class land-use and land-cover of Beijing built-up area, the proportion of impervious surface area increased first and then decreased, and the proportion of urban green space area decreased fist and then increased. The expansion of built-up area in Beijing shows a "leap" growth, with a net increase of about 3.0 times. The urban expansion is mainly to occupy a large amount of cultivated land, which has reached 53.5% of the total urban expansion in 25 years. The cultivated land occupation in each expansion stage shows a decreasing trend. In built-up area internal optimization, the change characteristics of each stage are different. Among them, 1990-2000 is dominated by "transformation to urban impervious surface"; 2000-2010 is dominated by "transformation of urban green space and urban impervious surface", and 2005-2010 is characterized by symmetric distribution of the exchange areas from north to south; after 2010, it is dominated by "transformation to urban green space".

       Fourthly, the accuracy of Beijing surface temperature retrieved from Landsat data from 1990-2015 was corss validated by MODIS temperature products. Then, using Urban Heat Island Intensity (UHII), Urban Heat Island Footprint (UHIFP) and Urban-Heat-Island Ratio Index (URI) to carry out the spatial and temporal pattern analysis of SUHI. The results show that: the determination coefficient between the inversion results and the validation data is above 0.90, which meets the analysis requirements. In terms of inter-annual changes from 1990-2015, the scope of SUHI has been enlarged as a whole, while that in the central urban area has been alleviated due to the old city reconstruction and the greening on both sides of the rivers; the URI has increased from 0.450 in 1990 to 0.637 in 2010, and then slowly decreased to 0.507 in 2015; the UHIFP has increased from 0.5 times of the distance between urban areas in the 1990s to 4-5 times of the distance between urban areas in the 2000 and then increased to 5-5.5 times the distance in 2015. In terms of seasonal changes, from the end of May to the first ten days of September, the UHII is higher and covers the urban area within the Sixth Ring Road; in December and January, there is "cold island effect"; in summer, the UHIFP is the largest, reaching 3-4.5 times of the distance between urban areas.

Fifthly, the characteristics of UHII in different urban land-use and land-cover was compared with each other in different years and seasons. Seventeen factors were constructed from three dimensions of land-use and land-cover composition, structure and physical properties. The quantitative analysis of the impact of urban land-use and land-cover on SUHI in Beijing is carried out by using Random Forest Model (RFM), and the temporal and spatial change characteristics are explored. The results show that: in the aspect of interannual change, the UHII is the lowest in urban water body, the highest in urban impervious surface and high and low in urban green space; in the aspect of seasonal change, the UHII of different land-use and land-cover in winter and early spring has little difference, and the difference increases in summer and autumn. The UHII mitigation of urban lake is the strongest, and the UHII of facility land is the strongest. In 2010, the influence of 17 factors on SUHI ranked in order of importance, and the first six factors are not only component factors, but also physical and structural factors. The partial correlation of each factor on SUHI is different. From 2010 to 2015, the impact of physical property factors on SUHI in the built-up area increased, while that of structural factors decreased slightly. From the built-up area to the functional core area of the capital, the impact of structural factors on SUHI is greatly enhanced, and that of the physical factors is decreased. No matter the built-up area or the functional core area of the capital, the area ration of urban impervious surface and urban green space are the leading factors influencing on SUHI.

In summary, the ways to strengthen the ecological regulation and control of urban thermal environment are put forward: increase urban planar water bodies; control the proportion and spatial structure of urban impervious surface area and suggest that its weighted shape factor is bigger than 2.6; increase the urban green space, avoid its fragmentation, and try to keep its weighted shape factor not greater than 1.0; on the basis of determination of urban land-use and land-cover composition and structure, the material physical properties should be considered.