Li et al. (2025) Investigating the effect of urban form on land surface temperature at block and grid scales based on XGBoost-SHAP

Identification

• Journal: Environmental Modelling & Software
• Year: 2025
• Date: 2025-10-11
• Authors: Hong Li, Jun Yang, Jiaxing Xin, Wenbo Yu, Jiayi Ren, Huisheng Yu, Xiangming Xiao, Jianhong Xia
• DOI: 10.1016/j.envsoft.2025.106738

Research Groups

• JangHo Architecture College, Northeastern University, Shenyang, China
• Liaoning Key Laboratory of Urban and Architectural Digital Technology, JangHo Architecture College, Northeastern University, Shenyang, China
• Human Settlements Research Center, Liaoning Normal University, Dalian, China
• School of Humanities and Law, Northeastern University, Shenyang, China
• School of Management Engineering, Qingdao University of Technology, Qingdao, China
• School of Biological Sciences, University of Oklahoma, Norman, USA
• School of Earth and Planetary Sciences (EPS), Curtin University, Perth, Australia

Short Summary

This study integrates XGBoost-SHAP to investigate the effects of urban factor indexes (UFIs) on land surface temperature (LST) at block and grid scales, and across local climate zones (LCZs). It reveals seasonal and scale-dependent LST patterns, identifying key UFI contributors and specific thresholds for warming or cooling effects.

Objective

• To investigate the effects of various urban factor indexes (UFIs) on land surface temperature (LST) at both block and grid scales.
• To examine the differences in LST and its driving factors across local climate zones (LCZs) at the grid scale.

Study Configuration

• Spatial Scale: Urban areas, analyzed at block and grid scales, and categorized by local climate zones (LCZs).
• Temporal Scale: Seasonal analysis (spring, summer, autumn, winter).

Methodology and Data

• Models used: eXtreme Gradient Boosting (XGBoost) integrated with SHapley Additive exPlanation (SHAP) method.
• Data sources: Land surface temperature (LST) data, and various urban factor indexes (UFIs) including normalized difference built-up index (NDBI), normalized difference vegetation index (NDVI), Shannon’s diversity index (SHDI), sky view factor, and building density.

Main Results

• LST is higher in central urban areas than in peripheral ones during summer and autumn, with this pattern reversing in spring and winter.
• LST varies significantly across different Local Climate Zones (LCZs).
• The normalized difference built-up index (NDBI), normalized difference vegetation index (NDVI), and Shannon’s diversity index (SHDI) are identified as the main contributors to LST variations.
• The sky view factor inhibits LST at the block scale but promotes it at the grid scale.
• The impacts of UFIs on LST follow a seasonal trend in terms of intensity: summer > spring > autumn > winter.
• LST responses to UFIs exhibit similar trends across scales, showing specific warming or cooling thresholds:
  • A cooling effect is observed when SHDI exceeds 0.65.
  • A warming effect occurs when building density exceeds 20 % in summer and autumn, or 40 % in spring and winter.
  • Significant cooling is only achieved when NDVI exceeds 0.4, although NDVI generally remains low in all seasons except summer.
• High-contribution UFIs typically exhibit the strongest interaction effects with artificial factor indicators.

Contributions

• Novel application of the integrated XGBoost-SHAP method to analyze the complex relationship between urban form and LST.
• Comprehensive investigation of UFI effects on LST across multiple spatial scales (block and grid) and within different Local Climate Zones (LCZs).
• Identification and quantification of seasonal variations and specific thresholds for urban factor indexes that lead to LST warming or cooling effects.
• Provides insights into scale-dependent impacts of urban form indicators, such as the sky view factor, on LST.

Funding

Not specified in the provided text.

Citation

@article{Li2025Investigating,
  author = {Li, Hong and Yang, Jun and Xin, Jiaxing and Yu, Wenbo and Ren, Jiayi and Yu, Huisheng and Xiao, Xiangming and Xia, Jianhong},
  title = {Investigating the effect of urban form on land surface temperature at block and grid scales based on XGBoost-SHAP},
  journal = {Environmental Modelling & Software},
  year = {2025},
  doi = {10.1016/j.envsoft.2025.106738},
  url = {https://doi.org/10.1016/j.envsoft.2025.106738}
}