基于机器学习量化对城市热岛的形态学影响OA

Taking the Pearl River New Town in Guangzhou as an example,this paper uses XGBoost,Shapley Value(SHAP)and Partial Dependency Map(PDP)to evaluate the impact of urban morphology on urban heat island.The results emphasize the differential impact of urban form attributes on heat islands,and optimizing building configurations can reduce heat island risks.Urban layout factors have a threshold impact on heat islands,and it is necessary to control the surface fraction and height of buildings within a limited time to reduce negative impacts.This research model focuses on adapting to spatial planning,resisting extreme urban heat island storms,and providing recommendations for the relationship between urban form and heat island risk through interpretable machine learning.