A comparative analysis of surface and canopy layer urban heat island at the micro level using a data-driven approach

The Urban Heat Island (UHI) phenomenon, which is defined as the temperature differential between inner cities and surrounding areas, has been extensively studied over the past few years to unravel its mechanism and develop mitigation strategies. Nonetheless, two distinct types of temperature can be used to measure UHI, namely, (1) air temperatures, which refer to air temperatures at the canopy layer (CUHI), and (2) surface temperature, which refers to the temperatures at the surface layer (SUHI). The two types of UHI can have different deriving mechanisms and any effective mitigation strategies should be able to mitigate both concurrently. While efforts have been made to compare these two types of UHI, the studies so far seldom used consistent data. This is because SUHI is mostly based on remote sensing data, whereas CUHI commonly requires field measurements. These data are often not consistent with respect to spatial resolution, frequency, and accuracy. To address this gap, the present research aims to perform a comprehensive comparative analysis of CUHI and SUHI using data collected by a tailor-made mobile data collection unit at a high frequency and resolution at the micro-level (i.e., street-level). Data was collected from a fixed 8 km-long route in the city of Apeldoorn, the Netherlands, for the period of one year. Two different machine learning techniques, i.e., random forest and neural network, were used to study SUHI and CUHI. The results indicated a considerable variability between air and surface temperatures during the data collection campaign. Air temperatures ranged from -0,3 to 35,3°C, while surface temperatures fluctuated over a wider range, from -12,0 to 48,4°C at a micro-level. This variability of temperatures translated into an average of 0,10°C for CUHI, and -0,48°C SUHI. More importantly, however, the results highlighted the importance of investigating simultaneously the two types of UHI. This is because while urban features do not change dramatically in short periods of time, the impact of these same features on the CUHI and SUHI is different, therefore urban-heat resilience strategies planned for one type of UHI alone could have a different impact on the other type.