Su, Yuan, LeXua, LeXuan, Han, Yu, Yan, Qiao, Meng, Qinglin, Luo, Zhiwen  ORCID: https://orcid.org/0000-0002-2082-3958 and Wu, Tianji
2026.
Nonlinear impact of urban morphological characteristics on air pollutants in dry-hot regions from the perspective of spatial type differences.
Sustainable Cities and Society
144
, 107415.
10.1016/j.scs.2026.107415
|
Preview |
PDF
- Accepted Post-Print Version
Available under License Creative Commons Attribution. Download (2MB) | Preview |
Abstract
Rapid urbanization has intensified air pollution, quantitative understanding of how multidimensional urban morphology drives pollutant variability across seasons and spatial regimes remains limited. Study integrates K-means clustering, a CatBoost-based explainable machine learning framework, and SHAP analysis to quantify nonlinear relationships between 27 urban morphology indicators and six major air pollutants across four seasons at a 1000 m resolution. Model achieved robust predictive performance R² exceeding 0.80 in winter. Global SHAP analysis revealed a dominant contribution from Natural Environment and Meteorological Environment factors, jointly accounting for 60 to 85% of total feature importance. Elevation, precipitation, wind speed, and solar radiation are the features with the highest explanatory power. Socioeconomic and Activities factors contributed 3 to 10%, and Land-Cover and Architectural Form generally remained below 15%. Spatial clustering identified six urban morphology types with distinct pollution regimes and driver sensitivities. Global and clustering perspectives of nonlinear SHAP dependence revealed threshold, including DEM is 500m or 1000 m, PREC is 13mm and 15 mm, and WS thresholds is 2 or 2.75 m/s, beyond which pollutant responses shift markedly. LISA-based spatial autocorrelation analysis demonstrated strong hotspot clustering in urban cores. Peripheral regions were dominated by low-value clusters and non-significant patterns, indicating pronounced urban–rural gradients in pollution mechanisms. Overall, our study provides an interpretable framework for disentangling the nonlinear, spatially heterogeneous drivers of urban air pollution. The results highlight the dominant role of topography–meteorology interactions and reveal threshold effects, offering insights for different urban space types of pollution mitigation and climate-adaptive urban planning.
| Item Type: | Article |
|---|---|
| Date Type: | Publication |
| Status: | Published |
| Schools: | Schools > Architecture |
| Additional Information: | RRS applied. License information from Publisher: LICENSE 1: Title: This article is under embargo with an end date yet to be finalised. |
| Publisher: | Elsevier |
| ISSN: | 2210-6707 |
| Date of First Compliant Deposit: | 30 April 2026 |
| Date of Acceptance: | 18 April 2026 |
| Last Modified: | 30 Apr 2026 14:30 |
| URI: | https://orca.cardiff.ac.uk/id/eprint/186703 |
Actions (repository staff only)
 |
Edit Item |
Altmetric
Altmetric