The dominant factors of building pattern for regulating urban thermal environment vary across urban functional zones

Abstract Context Thermal environment degradation has emerged as a serious issue constraining urban sustainable development. Identifying key factors affecting thermal environment is critical for heat mitigation and risk control. Though earlier investigations have shown that urban spatial patterns exert an obvious influence on land surface temperature (LST), research exploring the comprehensive impacts of multidimensional building patterns on LST from the perspective of urban functional zones (UFZs) and fine-scale analysis remains lacking. Objectives This study attempted to quantify the comprehensive effect of 2D/3D building configurations on summer daytime LST among different UFZs to provide more refined decision support for urban heat risk management and heat alleviation. Methods Taking Shanghai as a case study, the Pearson correlation analysis, boosting regression tree (BRT) and variance partitioning were employed to quantify the relationship between 2D/3D building configurations and LST from UFZs perspective. Results There were obvious differences in terms of significance, positive/negative correlation and thresholds between building metrics and LST across different UFZs. Dominant building metrics affecting LST varied across different UFZs. Building Coverage Ratio (BCR) emerged as the most critical metrics for residential, industrial, public service and traffic zones, and Building Patch Density (BPD) had the highest contribution in commercial zone, and both of them showed significant positive influences. In addition, Building Height Standard Deviation (BHSD), Mean Building Projection Area (MBPA), Floor Area Ratio (FAR), Building Volume Ratio (BVR), Landscape Shape Index (LSI) and High-rise Building Ratio (HBR) were also the crucial influence metrics of LST for various UFZs, respectively. The building metrics selected in this study had a strong overall explanatory power (R 2 range was 31.95–46.43%) for the LST variation of different UFZs. The 2D building metrics group outperformed 3D in LST modulation, but the joint effect between 2D and 3D building metrics group was more important, as it had a higher R 2 than the independent effect of 2D or 3D metrics group. Conclusions The dominant building pattern factors influencing LST and their thresholds varied obviously across different UFZs. The results can provide more targeted support and guidance for decision-makers and planners in improving urban thermal environment of different UFZs through scientific planning and architectural design.