Abstract

The continuous advancement of urbanization and the acceleration of global climate warming have severely aggravated the heatwave vulnerability of the urban complex human–land system. Therefore, the scientific assessment of urban heatwave vulnerability (UHV) is particularly critical. We used Xiamen City, one of the representative heatwave disaster-prone cities, as a case study area. We then constructed a UHV index system that coupled adaptability and selected 12 indicators from the three dimensions: exposure, susceptibility, and adaptability. The back propagation neural network (BPNN) model was used to composite each indicator layer and produce UHV results. Finally, we analyzed the spatial distribution characteristics of UHV. We found that the BPNN model had good training performance, with an overall accuracy of 0.92986. The value of UHV ranged from 0 to 1 and was divided into five grades, from low to high were 18.45%, 18.72%, 17.16%, 28.76%, and 16.91. In terms of spatial characteristics, high adaptability significantly improved UHV. The high value of UHV presented specific agglomeration characteristics, and the extremely vulnerable and disordered areas were mainly located in Huli District and the junction of Siming District and Huli District. The research will provide a new theoretical perspective and framework for urban heatwave assessment and help for disaster management and sustainable development in a high-risk area.