地球信息科学学报 ›› 2019, Vol. 21 ›› Issue (7): 1051-1060.doi: 10.12082/dqxxkx.2019.180611

• 地理空间分析综合应用 • 上一篇    下一篇

基于半变异函数的重庆市地表温度空间异质性建模及多尺度特征分析

陈昭1(), 罗小波1,2, 高阳华2,*(), 叶勤玉2, 王书敏1   

  1. 1. 重庆邮电大学 计算机科学与技术学院,重庆 400065
    2. 重庆市气象科学研究所,重庆 401147
  • 收稿日期:2018-09-05 修回日期:2019-03-19 出版日期:2019-07-25 发布日期:2019-07-31
  • 通讯作者: 高阳华 E-mail:506556163@qq.com;gaoyanghua@sina.com
  • 作者简介:

    作者简介:陈 昭(1993-),男,湖北仙桃人,硕士生,主要从事遥感技术应用及城市热岛等研究。E-mail: 506556163@qq.com

  • 基金资助:
    国家自然科学基金项目(41871226);重庆市博士后特别资助项目(Xm2016081);重庆市气象局开放基金项目(KFJJ201602);重庆市应用开发计划重点项目(cstc2014yykfB30003);中国气象局省所科技创新发展专项(SSCX201917)

Modeling and Multi-Scale Analysis of the Spatial Heterogeneity of Land Surface Temperature in Chongqing based on Semi-Variogram

Zhao CHEN1(), Xiaobo LUO1,2, Yanghua GAO2,*(), Qinyu YE2, Shumin WANG1   

  1. 1. College of Computer Science and Technology, Chongqing University of Posts and Telecommunications, Chongqing 400065, China
    2. Chongqing Institute of Meteorological Sciences, Chongqing 401147, China
  • Received:2018-09-05 Revised:2019-03-19 Online:2019-07-25 Published:2019-07-31
  • Contact: Yanghua GAO E-mail:506556163@qq.com;gaoyanghua@sina.com
  • Supported by:
    National Natural Science Foundation of China, No.41871226;Chongqing Postdoctoral Special Funding Project, No.Xm2016081;Chongqing Meteorological Bureau Open Fund Project, No.KFJJ201602;Chongqing Application Development Plan Key Project, No.cstc2014yykfB30003;China Meteorological Administration Provincial Science and Technology Innovation Development Project., No.SSCX201917

摘要:

城市地表温度空间异质性的研究对理解城市地表温度空间结构有重要意义。本文利用大气校正法反演地表温度,基于半变异函数构建城市地表温度空间异质性模型,并进一步分析不同空间尺度下地表温度空间异质性结构参数的变化规律。以2013年6月16日的Landsat 8为数据源,以重庆为研究区开展实验,研究结果表明:① 不同空间尺度下重庆地表温度空间异质性均呈现指数模型分布特征;② 在30 m空间尺度下,地表温度空间异质性主要是由空间结构引起,但随机因素引起的空间变异占比为0.45,呈现出明显的块金效应,表明该尺度下随机因素引起的空间变异不可忽略;③ 从空间尺度(30~1500 m)整体变化上看,地表温度空间异质性主要由空间结构引起,同时表现出明显的尺度效应;随着空间尺度增大,块金值(C0)、偏基台值(C)、基台值(C0+C)以及块基比(C0/(C0+C))逐渐减小,表明地表温度空间异质性逐渐减弱但空间自相关性逐渐增强。变程(A)逐渐增大,表示空间自相关性范围逐渐扩大;④ 随机因素引起的空间变异占比为0.23~0.46,呈现出波动变化,这是因为地表温度在像元内部也存在空间异质性。空间结构引起的空间变异相对平缓,这是因为空间尺度的变化不会改变地形结构;⑤ 从尺度域来看,基台值与块金值在尺度域(690 m,1500 m)内呈现出较大幅度波动变化状态,且变化趋势相似,表明地表温度空间异质性的变化与随机因素有较大关联。综上所述,分析地表温度空间结构需要选取合适的空间尺度,尺度较小时,容易受到随机因素干扰,从而影响地表温度在空间结构上的空间变异性;尺度较大时,地表温度空间异质性较弱且变化不稳定。

关键词: 地表温度, 半变异函数, 空间异质性, 空间结构, 尺度效应, 重庆市

Abstract:

Analyzing the spatial heterogeneity of land surface temperature (LST) is important for understanding the spatial structure of LST. This study retrieved LST by the atmospheric correction method, and constructed a spatial heterogeneity model of LST by using the semi-variogram function. It then took a multi-scale perspective to discuss LST’s spatial heterogeneity in the study area of Chongqing. A Landsat 8 OLI imagery in June 16, 2013 was the primary data source. Results show that: ① The LST’s spatial heterogeneity was exponentially distributed at different spatial scales. ② At the 30 m spatial scale, the spatial heterogeneity was mainly caused by spatial structure, though the proportion of spatial variation caused by random factors accounted for 0.45, showing obvious nugget effect; thus, random factors cannot be ignored at this scale. ③ On the whole spatial scale (30~1500 m), the spatial heterogeneity was mainly caused by spatial structure, and showed obvious spatial scale effect. As the spatial scale increases, the nugget (C0), the partial sill (C), the sill (C0+C), and the nugget-sill ratios (C0/(C0+C)) gradually decreased, indicating that the spatial heterogeneity declined and the spatial autocorrelation gradually increased. Meanwhile, the range (A) gradually increased, indicating that spatially autocorrelated regions gradually enlarged. ④ On one hand, the proportion of spatial variation caused by random factors ranged from 0.23 to 0.46, showing obvious volatility, because the LST also had spatial heterogeneity within each pixel. On the other hand, the spatial variability caused by spatial structure was relatively flat, because the change of spatial scale did not affect the topographic structure. ⑤ From the scale effect perspective, both sill and nugget showed large fluctuations, and the trend was similar from 690 m to 1500 m, indicating that the change of the LST's spatial heterogeneity was related to random factors. In summary, choosing the appropriate spatial scale is very important for analyzing the spatial structure of LST. When the scale is small, the spatial distribution of LST is easily disturbed by random factors, which affects the variability in spatial structure. When the scale is large, the spatial heterogeneity of LST is weak and unstable.

Key words: land surface temperature, semi-variogram, spatial heterogeneity, spatial structure, scale effect, Chongqing