Abstract:

Different land use types have different effects on urban surface temperatures. At present, studies on the spatial autocorrelation and the bivariate spatial correlation between surface temperature and land use types lacked systematic examination of scale effects. Taking Chengdu City as the study area, the remote sensing and GIS methods were used to obtain the surface temperature and NDBI, NDVI, and NDWI data based on a total of six remote sensing images in this area during winter and summer between 2010 and 2021. The semi-variance function was applied to identify their spatial correlation analysis scales in different years. The spatial correlation between surface temperature and each land use type was combined with Moran’s I value to analyse the spatial correlation between surface temperature and each land use type to reveal the spatial scale dependence of the correlations between the two and their spatial variability characteristics, which provided an important reference for rational planning and construction of urban blue-green systems, reducing urban heat island effects, and improving the urban ecological environment. The study found that (1) the study area exhibited a significant heat island effect in both winter and summer, with NDBI and NDWI values showing a high centre and low surroundings, while NDVI values showing a low centre and high surroundings. (2) The univariate spatial autocorrelations of LST, NDBI, NDVI and NDWI had significant scale effects, with the scale effect of autocorrelation varying significantly in the 300m scale range and weakening in the 300-600 m scale range. The univariate spatial autocorrelation of LST was the strongest in summer, and NDVI and NDWI as a whole showed a larger scale of autocorrelation in summer than in winter. (3) The scale of bivariate spatial correlation between LST and each local surface parameter was around 300 m, and the scale effect was significant in the 300 m scale range, and the overall correlation was greater in summer than in winter. (4) There was a close relationship between LST and the three land use types. The results of this study provided a better understanding of the scale effect of the spatial correlation between LST and land use types.

Key words: spatial correlation, land surface temperature, Moran’s I, semi-variogram, Chengdu City

摘要：

不同土地利用类型对城市地表温度具有不同的影响。目前关于地表温度与土地利用类型间的空间自相关性、双变量空间相关性的研究缺乏系统的尺度效应探讨。以成都市为例,利用遥感与GIS方法获取研究区2010—2021年间冬夏两季共6期遥感影像的地表温度、NDBI、NDVI和NDWI数据,运用半变异函数识别其在不同年份下的空间相关性分析尺度,并结合Moran’s I值分析地表温度与各土地利用类型间的空间相关性,试图揭示两者之间的相关性的空间尺度依赖规律,及其空间变异特点,为合理规划与建设城市蓝绿系统,减弱城市热岛效应,改善城市生态环境提供重要参考。研究发现,（1）研究区在冬夏两季均表现出明显的热岛效应,NDBI、NDWI值均表现为中心高四周低,NDVI值表现为中心低四周高。（2）LST、NDBI、NDVI、NDWI的单变量空间自相关性存在明显的尺度效应,300 m尺度范围内自相关性尺度效应变化明显,300—600 m尺度范围内自相关性尺度效应减弱。夏季的LST单变量空间自相关性最强,NDVI与NDWI整体上表现出夏季自相关性尺度大于冬季。（3）LST与各地表参数间双变量空间相关性尺度均在300 m左右,300 m尺度范围内,尺度效应明显,整体表现为夏季的相关性大于冬季。（4）LST与3种土地利用类型存在密切关系。该研究结果对于进一步认识LST与土地利用类型间的空间相关性的尺度效应具有推动作用。

关键词: 空间关联性, 地表温度, Moran's I, 半变异函数, 成都市