Simultaneous investigation of surface and canopy urban heat islands over global cities

Interpreting the similarities and dissimilarities in spatiotemporal variations and various controls between surface and canopy urban heat islands (UHIs) is critical for a better understanding of their vertical structure. Preceding comparisons of the surface UHI (SUHI) and canopy UHI (CUHI), however, remain mostly restricted either in a single city or over a few cities within limited background climates; therefore, the associated similarities and dissimilarities between the SUHI and CUHI under different climates, especially at a global scale, remain largely unknown. Based on both satellite and in situ data, we simultaneously investigated the spatiotemporal patterns of the SUHI intensity (SUHII) and CUHI intensity (SUHII) of 366 global cities within various background climates. We further investigated the different impacts of several controls (e.g., vegetation coverage, population size, precipitation) on SUHII and CUHII. Our results indicate the following: (1) For the selected 366 cities, the annual mean SUHII is higher than CUHII by 1.1 ± 1.9 °C (mean ± Std) during the day and 0.3 ± 1.5 °C (mean ± Std) at night. The SUHII and CUHII in the equatorial, warm temperate, and snow climates are generally consistent with the above characteristics (i.e., SUHII > CUHII), however, in arid regions SUHII is lower than CUHII by 0.8 °C during the day. (2) The annual mean day–night difference in SUHII is positive (i.e., 0.6 ± 1.8 °C (mean ± Std)), while the difference in CUHII becomes negative (i.e., −0.2 ± 1.6 °C (mean ± Std)), indicating that urbanization increases the diurnal temperature range (DTR) based on land surface temperature, but it decreases the DTR based on surface air temperature. (3) Despite the high correlation between vegetation coverage and impervious surface percentage (ISP), their impacts on SUHII and CUHII were not consistent. The urban–rural difference in ISP exerts an insignificant impact on both SUHII and CUHII during the day and a greater impact on CUHII than on SUHII at night, whereas the urban–rural difference in vegetation coverage has a greater impact on SUHII than on CUHII during the day, while the opposite occurs at night. The impacts of population size on SUHII and CUHII are much greater during the night than on the day in which their impacts can be minimal. The relationship between annual mean precipitation and SUHII is positive during the day but negative at night, while for CUHII, their relationship is insignificantly negative both during the day and at night. These results can improve our understanding of the spatiotemporal patterns and controls of these two types of UHIs under various climates.