The future of China's urban heat island effects: A machine learning based scenario analysis on climatic-socioeconomic policies

Abstract

Risk assessment and adaptation have become key foci in the examination of urban heat island (UHI) effects. Land use change and population growth are known to impact UHI effects. Taking the Changsha-Zhuzhou-Xiangtan urban agglomeration in China as the study context, a rapid and simple method based on artificial neural networks was employed to spatially estimate UHI effects from 2040 to 2100. Considering carbon neutrality and pro-natalist policies, Representative Concentration Pathways (RCPs) and Shared Socioeconomic Pathways (SSPs) were integrated into combination scenarios: RCP2.6-SSP1, RCP4.5-SSP2, and RCP4.5-SSP3. The risk of UHI effects was assessed through UHI hazard and population exposure. Results showed that compared with 2020, heat islands would decrease by 30% - 34% in area but increase by 29% - 37% in mean intensity by 2100. The population impacted by UHIs would increase by >10 million under the third scenario. Areas at high risk of UHI effects are mainly located in the urban core, with the total value of risk across the study area by 2100 being about 2 - 4 times higher than that in 2020. Our findings provide a forward-looking perspective to identify overall risk and critical areas in relation to UHI effects in the future through which clearer climatic-socioeconomic targets can be empirically grounded to minimize damage to public health.