Estimation of the surface urban heat island intensity across 1031 global cities using the regression-modification-estimation (RME) method

Abstract

The urban heat island (UHI) effect profoundly impacts both humans and the environment. Satellite data have been widely employed to investigate the surface UHI (SUHI) effect. The SUHI intensity (SUHII) is a key indicator of SUHIs, and its estimation necessitates the acquisition of land surface temperature (LST) data for urban areas. However, in many previous studies, urban areas were simply defined as those with the proportion of impervious surfaces exceeding a certain threshold. Therefore, the delineated urban areas often included natural surfaces such as croplands and water bodies, leading to uncertainties in SUHI studies. In this study, the regression-modification-estimation (RME) method was proposed to quantify the SUHII and analyze the relationships between the SUHII and its driving factors across 1031 global cities. The results can be summarized as follows: (1) the RME method mitigates the uncertainties associated with the definitions of urban and rural areas, providing accurate estimation of the impact of urbanization-induced land cover changes on the LST. (2) The annual daytime and nighttime average SUHII values for the 1031 cities derived from the RME method were 2.05 °C and 1.79 °C, respectively. The SUHII values derived from the RME method were 24.8%–53.3% higher than those obtained from previous methods. (3) Vegetation and precipitation emerged as important drivers of the daytime SUHII, while the nighttime SUHII exhibited a significant relationship with the albedo and precipitation. An interesting result was that the correlations between SUHIs and some important drivers derived from the RME method were stronger than those derived from previous methods.