From near real-time urban data to an explainable city-scale model to help reduce the Urban Heat Island (UHI) effect

Abstract

Abstract Urbanization is associated with increasing the temperature of urban areas and phenomena like the Urban Heat Island (UHI). The latter produces a sensible impact on people’s health, quality of life, urban liveability, and even the mortality rate. As a result, reducing high temperatures in cities will improve the quality of life in cities. Overall, the understanding of the relationship between UHI and urban features is not well captured due to a lack of accessibility to data and cost implications related to long-term monitoring and simulation processes. Machine Learning (ML) offers significant opportunities to develop approaches that lead to high accuracy in discovering the significant features of buildings and urban areas, improving the conditions of the inhabitants. The paper aims to describe how the actual spatial data of Tallinn, Estonia, were assessed and applied to build an Explainable ML-based model. The study outcomes include the ML-based models that help understand the importance of urban features and their value in developing strategies and solutions to mitigate the UHI effect. The study’s findings show that changes in the most important features of urban areas and buildings, such as the built area in the neighborhood zone and the area and orientation of buildings, play the most significant role in transforming an urban area from one with no UHI effect to one that experiences heat waves. The results provide essential information for urban planners who implement built environments.