Natural-anthropogenic environment interactively causes the surface urban heat island intensity variations in global climate zones

Abstract

The inconstant climate change and rapid urbanization substantially disturb the global thermal balance and induce severe urban heat island (UHI) effect, adversely impacting human development and health. Existing literature has revealed the UHI characteristics and driving factors at an urban scale, but interactions between the main factors of a global grid scale assessment on the context of climate zones remain unclear. Therefore, based on the multidimensional climatic and socio-economic statistical datasets, the multi-time scale of surface urban heat island intensity (SUHI) characteristics was investigated in this study to analyze how natural-anthropogenic drivers affect the variance of SUHI and vary in their importance for the changes of other interaction factors. The results show that the mean value of SUHI in summer is higher than in winter, and in daytime is higher than in nighttime on a seasonal and daily scale. SUHIs in different global climate zones have significant differences. When analyzing drivers’ contributions and interactions with LightGBM model and SHAP algorithm, we know that monthly precipitation (PREC), the estimated population (POP) and surface pressure (PRES) are the three major drivers of daytime SUHI. The nighttime SUHI is mainly PREC, POP and anthropogenic heat emission (AHE), the influence rules of the natural drivers are mostly opposite to that of daytime. This study highlights the fundamental role of background climate for designing strategies. Irrigation or artificial rainfall will be effective to mitigate SUHI in low rainfall areas, while it is more effective to reduce AHE in high rainfall areas. In where greening can be difficult in the most developed cities, reducing AHE, increasing per capita GDP and controlling the population scale may also contribute to alleviating the SUHI. This study provides ideas for developing responsive urban heat island mitigation policies in a more realistic setting.

Keywords

• Surface urban heat island
• Climate zones
• Machine learning
• SHAP value
• Driving factor analysis
• Interaction effects