Data-driven analysis of urban heat island phenomenon based on street typology

Abstract

This study explores the intricate relationship between diverse street types and the urban heat island (UHI) phenomenon - a major urban issue where urban regions are warmer than their rural counterparts due to anthropogenic heat release and absorption by urban structures. UHI leads to increased energy consumption, diminished air quality, and potential health hazards. This research posits that a sample of representative streets (i.e., a few streets from each type of street) will be sufficient to capture and model the UHI in an urban context, accurately reflecting the behavior of other streets. To do so, streets were classified into unique typologies based on (1) socio-economic and morphological attributes and (2) temperature profiles, utilizing two clustering methodologies. The first approach employed K-Prototypes to categorize streets according to their socio-economic and morphological similarities. The second approach utilized Time Series Clustering K-Means, focusing on temperature profiles. The findings indicate that models retain strong performance levels, with R-Squared values of 0,85 and 0,80 and MAE ranging from 0,22 to 0,84 °C for CUHI and SUHI respectively, while data collection efforts can be reduced by 50 to 70 %. This highlights the value of the street typology in interpreting UHI mechanisms. The study also stresses the need to consider the unique aspects of UHI and the temporal variations in its drivers when formulating mitigation strategies, thereby providing new insights into understanding and alleviating UHI effects at a local scale. © 2023