Scholarlink

<p>In the context of rapid urbanization and global climate change, understanding the spatiotemporal heterogeneity of land surface temperature (LST) is crucial for effective urban thermal environment management. However, most existing research has focused on single cities, lacking comprehensive analyses across multiple city clusters and providing limited insight into the nonlinear threshold effects and interactions among driving factors. This study investigates 19 representative city clusters across China, covering diverse climatic zones and topographic gradients. By integrating the XGBoost model with the SHAP interpretability framework, we systematically analyze the spatiotemporal evolution of LST at seasonal and diurnal scales and clarify its driving mechanisms. The results reveal that: (1) LST demonstrates significant seasonal and diurnal variability, with the North Tianshan Urban Agglomeration (NTSUA) showing the largest summer daytime temperature range (20–62&#xa0;°C), and the Harbin–Changchun Urban Agglomeration (HCUA) exhibiting the largest winter nighttime range (− 31.25 to − 14&#xa0;°C). (2) Elevation and potential evapotranspiration (PET) are the dominant driving factors in most clusters, with their importance varying seasonally. In NTSUA, nighttime elevation in summer (|mean| SHAP = 5.4) and daytime PET in winter (|mean| SHAP = 3.4) are particularly influential. (3) Both elevation and PET show significant nonlinear threshold effects on LST, with elevation contributing positively below a threshold and negatively above it, and PET showing opposite trends, especially with higher winter thresholds in southern cities (30–65&#xa0;mm) compared to northern cities (0.2–8&#xa0;mm). (4) Interaction analysis indicates that in summer, elevation–EVI interactions dominate, while in winter, PET–elevation interactions dominate. This study quantitatively reveals, for the first time at a macro scale, the nonlinear driving mechanisms of LST in multi-climatic city clusters in China, providing a scientific basis for region-specific, climate-adaptive urban planning and thermal environment management.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Nonlinear thresholds and multi-factor interactions of land surface temperature in 19 Chinese urban agglomerations

  • 期刊论文

  • 发表日期：

    2026年4月7日

查看全文
  • Weidong Chen,
  • Lei Shi,
  • Qiyu Li,
  • Lingxu Meng

摘要

In the context of rapid urbanization and global climate change, understanding the spatiotemporal heterogeneity of land surface temperature (LST) is crucial for effective urban thermal environment management. However, most existing research has focused on single cities, lacking comprehensive analyses across multiple city clusters and providing limited insight into the nonlinear threshold effects and interactions among driving factors. This study investigates 19 representative city clusters across China, covering diverse climatic zones and topographic gradients. By integrating the XGBoost model with the SHAP interpretability framework, we systematically analyze the spatiotemporal evolution of LST at seasonal and diurnal scales and clarify its driving mechanisms. The results reveal that: (1) LST demonstrates significant seasonal and diurnal variability, with the North Tianshan Urban Agglomeration (NTSUA) showing the largest summer daytime temperature range (20–62 °C), and the Harbin–Changchun Urban Agglomeration (HCUA) exhibiting the largest winter nighttime range (− 31.25 to − 14 °C). (2) Elevation and potential evapotranspiration (PET) are the dominant driving factors in most clusters, with their importance varying seasonally. In NTSUA, nighttime elevation in summer (|mean| SHAP = 5.4) and daytime PET in winter (|mean| SHAP = 3.4) are particularly influential. (3) Both elevation and PET show significant nonlinear threshold effects on LST, with elevation contributing positively below a threshold and negatively above it, and PET showing opposite trends, especially with higher winter thresholds in southern cities (30–65 mm) compared to northern cities (0.2–8 mm). (4) Interaction analysis indicates that in summer, elevation–EVI interactions dominate, while in winter, PET–elevation interactions dominate. This study quantitatively reveals, for the first time at a macro scale, the nonlinear driving mechanisms of LST in multi-climatic city clusters in China, providing a scientific basis for region-specific, climate-adaptive urban planning and thermal environment management.