城市精细景观格局对热环境的影响
文献类型：学位论文
| 作者 | 钱雨果 |
| 学位类别 | 博士 |
| 答辩日期 | 2015-05 |
| 授予单位 | 中国科学院研究生院 |
| 授予地点 | 北京 |
| 导师 | 周伟奇 |
| 关键词 | 景观遥感分类 气温 地表温度 城市绿地 北京，landscape classification air temperature land surface temperature（LST） urban greenspace Beijing |
| 其他题名 | The relationship between landscape pattern at fine scale and urban thermal environment |
| 学位专业 | 生态学 |
| 中文摘要 | 全球范围内的快速城市化进程，使城市不透水地表增加，人为热排放加剧，导致了严重的城市热岛效应。与此同时，各个国家和城市也在采取各种政策和措施，如增加绿地，来改善城市热环境。本研究旨在回答城市精细景观格局，尤其是绿地景观格局的特征和变化，及其对城市热环境的影响。厘清上述关系，可为缓解城市热岛的规划、设计和评价提供科学依据和理论支撑。本研究发展了城市景观格局量化的技术和方法，定量分析了城市绿地的格局与变化，揭示了绿地格局对地温和气温的影响规律。 本研究以北京市为例，发展了基于高分辨率遥感影像的城市景观分类方法，定量研究了北京 5环内的景观格局及变化，探讨了中、高分辨率遥感数据对量化结果的影响。在此基础上，结合遥感反演的地表温度和地面实测的气温，分析了绿地对地温和气温的影响，及其尺度效应。主要结果如下： （1）建立了基于面向对象图像分析的景观制图分类器参数优化方案：对比分析了样本数量、参数方案设置对 4种常用分类器分类精度和效率的影响，发现支撑向量机（SVM）的分类精度最高，且所需的样本数量最少；并揭示了当SVM分类器的参数C设置在10(6至10(8之间，gamma在10(-5至10(-3之间时,SVM有稳定的高分类精度。 （2）发展了城市景观的多等级分类方法：应用面向对象的分类方法和自上而下的分类思路，将城市区域表征为景观类型和景观要素两个等级，对研究区域进行两级景观分类。首先将景观类型划分为城市景观、农田景观和森林景观3种景观类型，然后分别在各景观类型中进行植被、裸地、水体和不透水表面等 4类景观要素的提取，得到具有隶属关系的两级分类结果。其中景观类型的精度达到 93.36%，景观要素的精度达到 87.89%。相比单尺度的景观分类，多等级的分类不仅能够体现景观要素的组成，还能体现它在上一级景观中的分布，因而能更完整的描述景观格局，为认识城市生态系统结构和功能提供新视角。 （3）量化了北京 5环内绿地格局及动态。基于上述多等级分类方法和分类器参数优化方案，研究了北京 2005至2009年的绿地格局与动态变化，揭示了建成区中绿地景观的高度动态性。2005至2009年，在所有土地覆盖类型中，绿地的变化最为强烈，5环内新建绿地 69 km2，同时减少绿地32.7 km2，分别占整个研究区面积的 10.36%和4.91%。新增的城市绿地主要由不透水地表转变而来，2005年约 13%的不透水地表覆盖转变为 2009年的绿地，占新增绿地总量的 89.82%。通过对比分析发现，城市内部绿地斑块面积普遍较小，需要高分辨率影像数据才能准确揭示城市建成区内部的绿地格局及变化，常用的中等分辨率数据如 Landsat TM会严重低估城市绿地的覆盖比例及其动态变化。 （4）揭示了不同尺度的热环境时空特征。研究发现，城市地表热环境即使在很小的空间范围内也有显著差异，并随空间尺度增加，地温和气温的空间异质性明显增强；但在相同空间范围内，地温的异质性高于气温。在夏季，城郊的气温差异在 19点左右达到最大，且差异显著；而在城市的不同居民区中，14点左右气温差异最大，且差异不大，随时间变化不明显。 （5）量化了精细尺度绿地覆盖比例和空间配置对地表温度、气温的影响。对于地表温度，增加绿地的比例，增大绿地的斑块面积以及增大最大绿地斑块的面积能显著降低地表温度，该研究结果在北京、天津、南京和上海等 4个城市，以及 120m、600m和1080m等3个分析尺度下都基本一致。对于气温，不同季节影响局地气温的绿地空间范围不同，在冬季，200米或更大范围的绿地格局对监测点的气温影响显著，而在夏季，50至100米范围内的绿地格局显著影响监测点的气温，其中，绿地覆盖比例、最大绿地斑块面积、以及绿地边缘 密度的增加都会显著降低监测点的气温。 |
| 英文摘要 | The growth of urbanization has made the phenomenon of urban heat island （UHI） a crucial topic for urban ecological research. UHI results from the increased coverage of impervious surface, the reduction of transpiring and shading vegetation,and anthropogenic heat. Consequently, projects increasing the amount of urban greenspace have been undertaken to mitigate the UHI effect. Therefore, the aim of this study is to 1） quantify greenspace as a component of the urban landscape and its change, and 2） explore how the urban landscape, especially urban greenspace, affects the urban thermal environment. To achieve these two aims, first, we established an optimized object-based classification including two levels: landscape types and their constituent cover elements；second, we quantified greenspace as a component of the urban landscape and its dynamics from 2005 to 2009 within the 5th ring road of Beijing；third, we characterized the spatial-temporal thermal environment；finally, we statistically analyzed the relationship between the urban landscape classification and the thermal environment. The details of these procedures and their outcomes are described below. （1）Object-based urban land classification. We tested four frequently used classifiers, K nearest neighbor, normal Bayes,decision tree, and support vector machine（SVM）. The SVM had the best performance among these four. Even with a small number of training samples, SVM can achieve classification accuracy larger than 90%. In addition, we found the optimal range of the two parameters C and gamma to be between 10(6 and 10(8 , and between 10(-5 and 10(-3, respectively. （2）Two-level classification of urban region The SVM was used to generate a two level classification. The highest level was divided into landscape types of urban, forest, and agricultural types. Each of these general types was subdivided into impervious surface, water, greenspace, and bare soil. The two-level classification can reveal more information on the landscape pattern than single-level classification, and achieve high classification accuracy. The study area was dominated by urban landscapes, with proportional coverage of 43.54%. The proportions of agricultural and forest landscapes were 36.02% and The relationship between landscape pattern at fine scale and urban thermal environment 20.44%, respectively. As for landscape elements, the urban region was dominated by impervious surfaces, amounting to 45.08%. The proportion of impervious surface,however, varied greatly by different landscape types. The proportion of impervious surface was 70.95% in urban landscapes, but only 38.87% in agricultural landscapes and 12.34% in forest landscapes. Vegetation covered 37.42% of the study area but only 22.75% in urban landscapes, much less than that of 80.28% in forest landscapes. Percent cover of water was 1.99% in the study area, mostly occurring in the agricultural landscapes（56.68%） and urban landscapes（43.32%）. Bare soil took up 15.51% of the whole area, with 76.84% located in agricultural landscapes,only 14.28% in urban landscapes, and 8.88% in forest landscapes. In addition, the overall accuracy of classifications of landscape types was 93.36%, and was 87.89% for landscape elements. （3）Quantifying urban greenspace pattern and dynamics. Urban greenspace pattern and dynamics are examined at the lower level of the classification, that of landscape elements. The proportion of urban greenspace increased by 5.45%, or 36.3 km2 loss of greenspace were 69 km2 in size from 2005 to 2009. The amounts of gain and and 32.7 km2, respectively, or 10.36% and 4.91% of the entire study area. In fact, among all the four landscape elements, the change in percent cover of greenspace was the largest. “New” urban greenspace mostly occurred in lands previously covered by impervious surfaces. Approximately 13% of the impervious surfaces in 2005 were converted to greenspace by 2009, accounting for 89.82% of the new urban greenspace. In addition, we found the quantification of urban landscape needs high resolution imagery；the frequently used 30 m resolution TM imagery will greatly underestimate the percent cover of greenspace and its dynamics. （4）Characterizing Spatial-temporal temperature patterns. Land surface temperature（LST） and air temperature（AT） are spatially heterogeneous in urban areas, even in relatively small spatial extents. The larger the study area is, the higher the difference in LST and AT. However, even in areas as small as a few city blocks, large contrasts in LST and AT can exist. In the same study area, the maximum difference of LST is always larger than that of AT. In the summer,the magnitude of air temperature difference between urban and rural sites peaked at 7 pm；however, within the urban area the maximum magnitude mostly occurred at 2 pm, and the temperature difference among residential areas were much smaller. （5）Assessing relationships between temperature and urban greenspace. This relationship between greenspace and LST is consistent across different cities, namely Beijing, Tianjin, Nanjing and Shanghai, and across different analysis scales, namely 120 m, 600 m, and 1080 m. We found that the LST decreased with increase in the percent cover of greenspace, including the mean vegetated patch area, and the largest patch area. Therefore, to mitigate the LST, the cover and configuration of greenspace can be effective tools. The relationships of greenspace and AT were assessed only for Beijing due to the intensity of the measurements required on the ground. The pattern of greenspace affects the local AT. Mean AT decreased with increase the percent cover of greenspace, the area of the largest patch, and the edge density within a 100 m radius around the local site. The impact of greenspace on AT also varied by seasons；in the winter, increase in the percentage of greenspace（radius ≥200 m）affected the local AT. On the contrary, percent cover of greenspace and the size of the largest patch（50≤ radius ≤100 m） affected the AT in summer. Consequently, to mitigate the local high AT, it will be useful to manipulate the relevant greenspace characteristics identified here. Summary/Significance A significant methodological contribution of this dissertation is the discovery of the optimal parameters of C and gamma, especially in an urban environment. The two level classification is also a contribution that goes beyond the usual coarse land cover types of urban, agricultural, forest and bare ground. The refined elements of cover within these types better exposes urban structure that may have ecological significance. In addition, the relatively dense sampling scheme using HOBO thermometers reveals the local distribution of air temperature and allows its relationship with fine urban structure to be assessed. This contradicts the usual assumption of mesoscale homogeneity of urban air temperature. The results of this research can assist city planners and managers to understand whether their interventions work to reduce UHI effects. The methods described in this dissertation can be used to evaluate whether policies for urban greening are effective. The relationship between landscape pattern and the thermal environment can guide urban designers and planners to better exploit features that mitigate UHI The relationship between landscape pattern at fine scale and urban thermal environment effects. Furthermore, the two level classification has demonstrated the ranges of parameters that can be used to separate different categories of land cover, and hence generate high accuracy in urban land classification. |
| 源URL | [http://ir.rcees.ac.cn/handle/311016/34362]  |
| 专题 | 生态环境研究中心_城市与区域生态国家重点实验室 |
| 推荐引用方式 GB/T 7714 | 钱雨果. 城市精细景观格局对热环境的影响[D]. 北京. 中国科学院研究生院. 2015. |
入库方式： OAI收割
来源：生态环境研究中心
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。