城市不透水层的大规模扩展在不同尺度上导致了显著的环境效应。开展城市不透水层格局变化、驱动力和环境影响的多尺度研究，不仅可以从理论和方法上促进城市不透水层综合研究的发展，还可以在实践上为优化城市用地布局、提升城市环境质量和促进城市可持续发展提供科学依据。因此，本论文基于多尺度的景观视角和分析手段，按照“量化格局变化-揭示驱动力-评估环境影响”的城市景观生态学概念框架，开展中国近二十年城市不透水层格局变化、驱动力和地表温度效应的综合研究。主要工作和结论如下。（1）发展了基于植被校正的夜间灯光城市指数测量城市不透水层信息的新方法，提高了大尺度城市不透水层信息监测精度。新方法能有效降低城市不透水层测量误差，克服以往测量中城市不透水层空间分布饱和现象，有效识别城市不透水层空间分布差异和动态变化过程。与当前常用的测量方法相比，基于新方法估测的城市不透水层与真实城市不透水层之间的均方根误差、平均绝对误差和系统误差分别可降低0.188、0.153和0.247，相关系数可提高0.384。1992-2009年，基于新方法估测的中国城市不透水层平均均方根误差仅为0.136，平均相关系数达0.852。（2）量化了中国1992-2009年城市不透水层格局变化。中国城市不透水层扩展迅速，存在明显的区域差异。城市不透水层总面积自1992年的10,614.23 km2增加到2009年的31,147.63 km2，年均增长率为6.54%。全国共有珠江三角洲地区、长江三角洲地区和北京-天津-唐山地区等六个扩展热点区。这六个热点区面积仅占全国陆地总面积的0.87%，但其城市不透水层扩展面积却占全国总扩展面积的37.66%。此外，热点区新增城市不透水层的类型存在明显差异。珠江三角洲地区以中、高密度类型为主，北京-天津-唐山地区、长江三角洲地区、青岛地区和成都地区以中、低密度类型为主，长沙-湘潭地区以低密度类型为主。（3）识别了城市不透水层尺度效应，明确了城市不透水层随空间幅度、人口数量和城市面积变化的尺度推绎关系及该关系随尺度变化的特征。改变空间幅度会对城市不透水层百分比和总面积产生显著影响。城市不透水层百分比随空间幅度的增大而减小，而城市不透水层总面积随空间幅度的增大而增大。在三个城市群中，长江三角洲城市群城市不透水层百分比的减小量和总面积的增加量最大，分别为42.49%和6485.13 km2。城市不透水层百分比和总面积随空间幅度变化的尺度推绎关系总体上可以概括成3种类型：简单易预测的数学函数关系（如指数、对数和线性函数关系）、阶梯状变化和无规则变化。尺度推绎关系在不同尺度上存在明显差异。在城市核心区尺度上，尺度推绎关系以简单易预测的数学函数关系为主，其回归模型的决定系数均大于0.98，而在城市群尺度上，尺度推绎关系则以阶梯状和无规则状变化为主。此外，城市不透水层总面积与人口数量和城市面积之间均存在幂函数的尺度推绎关系，但其幂函数指数存在明显差异。城市不透水层总面积与人口数量之间的幂函数指数小于1，与城市面积之间的幂函数指数大于1。（4）揭示了不同尺度上影响城市不透水层的主导社会-经济因子及其随尺度变化的特征。不同尺度上影响城市不透水层的主导社会-经济因子存在明显差异。在省级尺度上，经济因子（如GDP、地方财政预算收入和固定资产投资）是影响城市不透水层的主导因子，其回归模型的决定系数最大为0.873。在市级尺度上，经济、人口和交通因子（如总人口、GDP、地方财政预算收入和民用汽车数量等）共同影响城市不透水层，其回归模型的决定系数最大为0.867。在县级尺度上，人口因子（如总人口、城市人口和非农人口）是影响城市不透水层的主导因子，其回归模型的决定系数最大为0.814。（5）评估了城市不透水层对地表温度的影响，揭示了城市不透水层与地表温度之间相关关系的空间尺度依赖性、昼夜和季节变化特征以及气候和生态背景对其的调控作用。城市不透水层与地表温度之间的相关关系具有明显的空间尺度依赖性。两者之间的相关性随尺度的下降而逐渐增大。其中，夏季夜晚两者之间的决定系数从生态区尺度到城市核心区尺度共增加了0.231。城市不透水层对地表温度的影响还存在明显的昼夜和季节差异。在夏季和冬季夜晚，城市不透水层与地表温度之间均存在显著的正相关性，而在冬季白天，两者之间往往呈现显著的负相关性。其中，在城市群尺度上，各城市群的平均相关系数在夏季白天、夏季夜晚和冬季夜晚分别是0.35、0.44和0.32，而在冬季白天却小于0。同时，两者之间的相关性普遍在夏季高于冬季，在夜晚高于白天。其中，在城市核心区尺度上，夏季白天的决定系数是冬季白天的3.24倍，冬季夜晚的决定系数是冬季白天的4.10倍。此外，气候和生态背景（即降水、温度和植被）会影响生态区尺度上夏季白天城市不透水层与地表温度之间的相关关系。其中，以植被的影响最大，决定系数为0.633。降水、温度和植被的增多会减弱城市不透水层对地表温度的影响。

Understanding the causes, processes, and consequences of urban impervious surfaces (UIS) is crucial for urban sustainability on multiple scales. China provides an ideal laboratory for UIS-related research due to its massive expansion of UIS and myriad environmental problems. This dissertation research aimed to conduct comprehensive studies of UIS on its spatiotemporal patterns, socio-economic drivers, and the effects on land surface temperatures (LST) in China during 1992-2009, from a multiscale landscape perspective. The main findings are shown as follows:(1) A new method using the Vegetation Adjusted NTL Urban Index (VANUI) was developed to improve the accuracy of estimating UIS on broad scales with nighttime light (NTL) data. Key to this improvement was VANUI’s ability to alleviate the problem of saturation inherently associated with NTL data. The average RMSE, MAE, and SE based on VANUI in 2009 were 0.128, 0.105, and -0.008, respectively, which were obviously less than the corresponding values based on the other two commonly used methods. The average R based on VANUI in 2009 amounted to 0.846, while the average R based on the other two methods were all less than 0.664. The estimated UIS using VANUI accurately identified not only the regions with high percent UIS but also those with low percent UIS, and also corresponded well with the characteristics of percent UIS dynamics presented in the corresponding high-resolution images. The average RMSE for mainland China from 1992 to 2009 was 0.136, with MAE of 0.108, SE of -0.018, and R of 0.852.(2) The spatiotemporal patterns of UIS in China from 1992 to 2009 were quantified using the new method. The total amount of UIS in China increased substantially from 10,614.23 km2 in 1992 to 31,147.63 km2 in 2009, at an annual increase rate of 6.54%. China’s UIS expansion exhibited pronounced regional differences, with six large “hotspot” areas where UIS expanded most substantially. These hotspot regions accounted for 0.87% of China’s total land area, but 37.66% of the total area of UIS expansion during 1992-2009. The six large hotspot areas exhibited different ways of adding new urban impervious surfaces. The main type of urban impervious surface expansion was medium- and high-intensity (50-90%) for Pearl River Delta, low-intensity (1-40%) for Changsha-Xiangtan area, and medium- and low-intensity (20-60%) for Chengdu area, Yangtze River Delta, Beijing-Tianjin-Tangshan area, and Qingdao area.(3) The scaling relations of UIS with respect to spatial extent, population size, and urban area were explored at different scales in China’s three key urban agglomerations. Changing spatial extent had significant effects on the percentage and total area of UIS. The percentage of UIS decreased and the total area of UIS increased with increasing spatial extent. The responses of the percentage and total area of UIS to changing spatial extent fell into three general types: simple scaling equations (i.e., exponential, logarithmic, or linear functions), staircase-like scaling behavior, and unpredictable behavior. The scaling relations differed with spatial scales. At the city proper scale, most of scaling relations belonged to the simple scaling functions with the coefficient of determination (R2) values larger than 0.98. In contrast, most of scaling relations at the urban cluster scale were staircase-like or unpredictable scaling behaviors. The total area of UIS showed power-law scaling relations with respect to population size and urban area, with different scaling exponents. The total area of UIS increased more rapidly than urban area with the exponents larger than 1, but did not grow as rapidly as population size grew with the exponents smaller than 1.(4) The key socio-economic drivers of UIS dynamics were determined on multiple scales. The key influencing factors of UIS varied substantially across hierarchical administrative levels: economic factors dominated the provincial level with the largest R2 value of 0.873, demographic factors dominated the county level with the largest R2 value of 0.814, and a mixed group of economic, demographic and traffic factors complicated the prefectural level with largest R2 value of 0.867. This seems to suggest that, to control rampant expansion of UIS, more attention needs to be paid to economic factors at the provincial level, to demographic factors at the county level, and to both kinds of factors as well as traffic factors at the prefectural level.(5) The spatial scale dependence, temporal variations, and bioclimatic modulation of the UIS-LST relationship were examined. In general, UIS and LST were positively correlated in summer and winter nighttime, but negatively in winter daytime. The strength of correlation increased from broad to fine scales. For example, the mean R2 for winter nights was 3 times higher at the urban core scale than at the ecoregion scale. The relationship showed large seasonal and diurnal variations: generally stronger in summer than in winter and stronger in nighttime than in daytime. At the urban core scale, for instance, the mean R2 was 2.2 times higher in summer daytime than in winter daytime, and 3.1 times higher in winter nighttime than in winter daytime. Vegetation and climate modified the relationship during summer daytime on the ecoregion scale. UIS has substantial influences on LST, and these effects vary greatly with spatial scales, diurnal/seasonal cycles, and bioclimatic context. Our study reveals several trends on the scale multiplicity, temporal variations, and context dependence of the UIS-LST relationship, which deserve further confirmation. Importantly, high mean R2 values with large variations on the local urban scale suggest that a great potential exists for mitigating urban heat island effects via urban landscape planning.