How does urban heat island differ across urban functional zones? Insights from 2D/3D urban morphology using geospatial big data

Abstract

The deteriorating urban thermal environment poses a huge impediment to sustainable urban development, which is closely related to the urban morphology under different urban functional zones (UFZs). By integrating remote sensing and geospatial big data, this work aims to reveal the divergent mechanism behind Urban Heat Island (UHI) across UFZs taking insights from 2D/3D urban morphology. The Minimum Spanning Tree (MST) indicator depicting 3D building distribution was introduced and integrated with the classic indicators. Their impacts on UHI were measured by ensemble learning and SHapley Additive exPlanations (SHAP) model. Taking Wuhan as the study area, the eighteen 2D/3D urban morphology indicators affecting UHI in different locations and UFZs were extensively examined and compared. The results reveal that: 1) The impact of 2D/3D urban morphology indicators on UHI significantly varies across different UFZs, with dominance in transportation zones exhibiting opposite polarity compared to the other types; 2) positive impacts have a decreasing trend from the urban center to the edge, while negative impacts exhibit opposite trend; 3) XGBoost outperforms other classic methods in interpreting the impact of urban morphology on UHI for all UFZ types. The findings improving knowledge of UHI across UFZs will provide a practical guide for urban planning.