地球信息科学学报 ›› 2018, Vol. 20 ›› Issue (11): 1657-1666.doi: 10.12082/dqxxkx. 2018.180303

• 遥感科学与应用技术 • 上一篇    下一篇

不同下垫面遥感指数与地温关系的空间差异性研究

帅晨1(), 沙晋明1,*(), 林金煌1,2, 季建万1, 周正龙1, 高尚1   

  1. 1. 福建师范大学地理科学学院,福州350007
    2. 中国科学院地理科学与资源研究所 资源与环境信息系统国家重点实验室, 北京 100101
  • 收稿日期:2018-06-27 修回日期:2018-09-17 出版日期:2018-11-20 发布日期:2018-11-28
  • 通讯作者: 沙晋明 E-mail:1628639901@qq.com;jmsha@fjnu.edu.cn
  • 作者简介:

    作者简介:帅 晨,(1993-),男,硕士生,主要从事生态遥感与地理信息建模。E-mail: 1628639901@qq.com

  • 基金资助:
    欧盟ERAMUS+项目(586037-EPP-1-2017-1-HU-EPPKA2-CBHE-JP、551857-EPP-1-20L5-1-DE-EPPKA2-CBHE-JP);科技部专项(247608);福建省国际合作重点项目(2018I0005)

Spatial Difference of the Relationship Between Remote Sensing Index and Land Surface Temperature under Different Underlying Surfaces

SHUAI Chen1(), SHA Jinming1,*(), LIN Jinhuang1,2, JI Jianwan1, ZHOU Zhenglong1, GAO Shang1   

  1. 1. College of Geographical Science, Fujian Normal University, Fuzhou 350007, China
    2. State Key Laboratory of Resources and Environmental Information System, Institute of Geographical Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, China
  • Received:2018-06-27 Revised:2018-09-17 Online:2018-11-20 Published:2018-11-28
  • Contact: SHA Jinming E-mail:1628639901@qq.com;jmsha@fjnu.edu.cn
  • Supported by:
    International University Cooperation on Land Protection in European-Asiatic Countries IUCLAND, No.586037-EPP-1-2017-1-HU-EPPKA2-CBHE-JP, No.551857-EPP-1-20L5-1-DE-EPPKA2-CBHE-JP;Special Project under Ministry of Science and Technology, No.247608;Special Project of International Cooperation of Fujian Province, No.2018I0005.

摘要:

利用遥感指数解释城市地温(LST)时空特征是国内外城市热环境研究的热点。但是由于遥感指数针对不同下垫面条件下的差异性表达,造成其在不同研究当中对城市地温的解释存在区别。本文选取莆田市都市区为研究区,以2016年7月27日Landsat 8影像为数据源,基于单窗算法反演地温,提取归一化植被指数(NDVI)、归一化建筑指数(NDBI)、改进的归一化水体指数(MNDWI)3种遥感指数,并利用同期GF-2影像提取研究区内下垫面类型。本研究对比分析了下垫面与地温的关系、各遥感指数在不同下垫面与地温的关系;构建遥感指数-地温的最小二乘拟合(OLS)模型与地理加权回归(GWR)模型,并对两个模型的分析结果进行对比。分析结果表明:① 下垫面对地温有显著影响,受地理空间连续性影响,人工表面的聚热效应会在临近林地、水体的区域减缓,而林地与水体的降温效果也会在临近人工表面时下降,表明下垫面对地温的影响也具有空间差异。② NDVI与MNDWI因本身值域特征,在水体区域与非水体区域与地温关系呈现不同的正负相关,而NDBI在定量解释其与地温的关系时呈现稳定的正向相关,因此基于NDBI构建OLS模型,在解读NDBI与LST的关系上,效果会优于NDVI、 MNDWI。③ 相较于OLS模型,GWR因充分考虑遥感指数在定量描述与地温关系时存在的空间差异性,能够更好的解释遥感指数与地温的关系,对2个模型的模型残差进行空间自相关分析发现相较于GWR模型,利用OLS构建的遥感指数-地温关系模型存在一定空间局限。

关键词: 地表温度, 归一化植被指数, 归一化建筑指数, 改进的归一化水体指数, 空间异质性, 地理加权回归

Abstract:

Use of remote sensing indices to explain the changes and distribution in urban surface temperature has become a hot topic in the study of urban thermal environment. However, there are obvious differences in interpretation of surface temperature using the same remote sensing index in different studies. The reason for this is that when we use remote sensing index to explain the land surface temperature changes, the spatial differences caused by the interaction of different underlying surfaces are ignored. In this paper, we select Putian metropolitan area as study area to explain the spatial difference of LST using the 2016 Landsat 8 image as the data source. Three remote sensing indices were used for the explanation: normalized difference vegetation index (NDVI), normalized difference building index (NDBI), and modified normalized difference water index (MNDWI). Then we use GF2 image to extract land cover types during the same period. the LST and its relationship with remote sensing indices was anayzed for different underlying surfaces. Then we use the OLS and GWR methods to construct the remote sensing index-LST model, and analyzed the spatial autocorrelation for model residuals. The results show: (1)The underlying surface has a significant impact on LST. Due to the continuity of the geospatial space, LST appear abnormal changes between different underlying surfaces. For example, the high temperature area is mainly concentrated on the artificial surface, but LST drops near water or forest lands as compared to its values on pure artificial surface; (2) NDVI and MNDWI have different positive and negative correlations with LST in water area and non-water area due to their own value range characteristics, while NDBI shows a stable positive correlation when quantifying its relationship with LST. (3)Compared with the OLS model, GWR can better explain the relationship between the remote sensing index and LST by fully considering the spatial difference. Through the spatial autocorrelation analysis for model residuals, there is an aggregation of model residuals under OLS, while the model residual under GWR shows a random distribution. Compared to GWR, OLS has stronger spatial limitations in describing he relationship between remote sensing index and LST.

Key words: Land Surface Temperature (LST), Normalized Difference Vegetation index (NDVI), Normalized Difference Building Index(NDBI), Modified Normalized Difference Water Index (MNDWI), Spatial Heterogeneity, Geographically Weighted Regression (GWR)