摘要
Abstract
With the acceleration of the urbanization,the spatial layout and surface materials of the city have changed significantly,increasing high-temperature risks.Effectively quantifying these risks is essential for the sustainable development of the society.This paper,based on the Local Climate Zones(LCZ)framework,takes Guangzhou Yuexiu District as the research area,integrates multi-source data,including Landsat land surface temperature inversion,semantic segmentation of the street-view images,multi-dimensional architecture/landscape shape parameters,and socioeconomic indicators,and XGBoost-SHAP machine learning model,and analyzes the multi-scale impacts of building form,landscape pattern,street-view visibility,and other factors on the urban thermal environment in subtropical high-density cities.It provides a multi-dimensional theoretical basis for the targeted regulation of urban heat island effect in subtropical cities.关键词
局地气候分区/地表温度/城市热岛效应/XGBoost-SHAP/亚热带高密度城市Key words
local climate zones/land surface temperature/urban heat island effect/XGBoost-SHAP/subtropical high-density city分类
土木建筑
