- We measured tree canopy temperature (Tc) and below-canopy surface temperature (Ts).
- Mean Tc (24.1 ± 2.9 °C) did not significantly differ from mean Ts (24.5 ± 1.8 °C).
- Tc was significantly lower than below-canopy air temperature (Ta) (25.4 ± 2.3 °C).
- Tc was highly and significantly correlated with Ts (r = 0.81) and Ta (r = 0.92).
- The differences between Tc, Ts and Ta were dependent on weather conditions.
Thermal infrared remote sensing is a popular tool to measure the impacts of land cover on land surface temperature (LST) in cities. Tree-covered areas usually exhibit a lower LST than other urban surfaces such as concrete, suggesting that urban trees can improve human thermal comfort. However, the remotely-sensed LST of trees is only representative of their canopy temperature. The linkage between the LST of trees and the below-canopy thermal conditions has not been investigated scientifically. This study aimed to measure the differences and determine the correlations between tree canopy temperature (Tc), below-canopy ground surface temperature (Ts) and below-canopy air temperature (Ta) from 42 trees of 16 species in a golf course in Hong Kong. Under the meteorological conditions of the study period (air temperature: ~26 °C; mean radiant temperature: ~30 °C), t-tests showed that mean ± SD Tc (24.1 ± 2.9 °C) did not significantly differ from Ts (24.5 ± 1.8 °C, p = 0.158), but it was significantly lower than Ta (25.4 ± 2.3 °C, p < 0.001). Tc was highly and significantly correlated with Ts (r = 0.81, p < 0.001) and Ta (r = 0.92, p < 0.001). Further analyses showed that the differences between Tc and Ts (ΔTcs) and between Tc and Ta (ΔTca) were significantly dependent on background air temperature (ΔTcs only), relative humidity, mean radiant temperature and incoming shortwave radiation. This study serves as a preliminary investigation on an important knowledge gap in urban climate research. The findings are particularly relevant to researchers who use remotely-sensed LST data to study human thermal comfort.