Machine learning methods and approaches for Urban Heat Island (UHI) assessment: A comprehensive review

The Urban Heat Island (UHI) effect is a phenomenon in which urban areas experience significantly higher temperatures than their surrounding regions due to human activities and the built environment. This poses challenges to urban systems, living conditions, and exacerbate air pollution in urban areas. Over the years, the use of Machine Learning (ML) to study the UHI effect has increased, driven by advances in computational power, high predictive accuracy, and the growing availability of large-scale, environmental datasets. This study aims to comprehensively review the application and role of ML in addressing the UHI effect. The key findings of this review show that the application of ML can be categorized into: land surface temperature (LST) retrieval and prediction, land use land cover (LULC) classification and change detection, UHI intensity and pattern modelling, and mitigation strategy assessment. The most common application is the derivation and prediction of LST in conjunction with physical and climatic variables such as building energy consumption, air temperature, human thermal comfort, and wind movement. Our findings further show that ML is effectively utilized in multi-temporal investigations and predictive modelling, assessing energy demand surges, and validating the efficacy of cooling interventions. We also outline the key challenges faced in implementing ML in this realm, including aerial data accessibility, data pre-processing complexities, modelling sophistication or UHI, and lack of ground truth data for validation. This review sets the stage for future use of ML, including emerging Artificial Intelligence (AI) - driven methods that could change how we understand and mitigate UHI effects.