Assessing the effects of 2D/3D urban morphology on the 3D urban thermal environment by using multi-source remote sensing data and UAV measurements: A case study of the snow-climate city of Changchun, China

Abstract

The urban heat island (UHI) effect has many adverse impacts on natural and social environments. A better understanding of how complex urban biophysical and social structures affect UHI is crucial to improve urban thermal environments. However, few studies have investigated the effects of 2D/3D urban morphology on land surface temperature (LST) and vertical air temperature simultaneously. Using LST data retrieved from Landsat 8 images and vertical temperatures at different heights collected from unmanned aerial vehicle (UAV) measurements, this study investigated the effects of 2D and 3D urban morphology on the 3D urban thermal environment among different urban function zones (UFZs) in the snow-climate city of Changchun, China. Our main findings are: (1) the average difference in vertical temperature between heights of 1.5 and 100 m was 1.78 °C and the temperature decreasing rate was 0.027 °C/m when the height was lower than 30 m, compared to 0.014 °C/m for 30–100 m. (2) Although 3D metrics had significant impact on LST, the relationships between the 2D metrics and LST were much stronger, accounting for 61% of the variations in LST. Additionally, the 2D composition metrics had stronger correlations with LST than the configuration metrics. The influence of 3D building metrics on LST was context-dependent for different UFZs. Increasing building height could reduce LST in residential UFZs but increase LST in industrial UFZs. (3) The effect of 2D urban morphology on the vertical temperature was stronger than that on the 3D ones when the height was lower than 30 m; however, opposite results were obtained when the height was greater than 30 m. The effects of 2D or 3D metrics on vertical temperature vary depending on whether the height is larger or smaller than a threshold, which is related to the characteristics of the building height. In general, 30 m appeared to be the threshold in our study. (4) The performances of models that used both 2D and 3D metrics to explain both LST and 3D thermal environments were better than those of models that used either 2D or 3D metrics alone, especially in UFZs such as residential areas.