Air-quality-sensitive urban design on NO2 Mitigation in the high-rise high-density City of London: A machine-learning land-use regression analysis

Kazakos, Vasilis, Luo, Zhiwen ORCID: https://orcid.org/0000-0002-2082-3958, Munir, Said, Futcher, Julie, Davis, Juliet ORCID: https://orcid.org/0000-0002-2056-5792 and Deng, Qihong 2026. Air-quality-sensitive urban design on NO2 Mitigation in the high-rise high-density City of London: A machine-learning land-use regression analysis. Sustainable Cities and Society , 108008. 10.1016/j.scs.2026.108008

PDF - Accepted Post-Print Version Available under License Creative Commons Attribution. Download (2MB)

Abstract

Elevated nitrogen dioxide (NO₂) remains a persistent urban air-quality and public-health concern. In the UK, despite progressive emissions-control policies, exceedances have historically occurred in central London. This study quantifies neighbourhood-scale NO₂ variability in the high-rise, high-density City of London and evaluates how air-quality-sensitive urban design, alongside traffic measures, influences concentrations in a morphology-constrained street network. We developed machine-learning land-use regression (LUR) models to predict NO₂ distribution: the Random Forest (RF) model performed best (adjusted R² = 0.88; RMSE = 7.6 μg m⁻³; MAE = 6.25 μg m⁻³). The RF-LUR model was then applied to assess air-quality-sensitive urban design scenarios. Increased built fraction associated with population growth was associated with widespread NO₂ increases (up to 5.5 µg m⁻³), consistent with reduced dispersion in dense street-canyons. A 50% reduction in heavy-vehicle activity was associated with corridor-focused NO₂ decreases of 1–6 µg m⁻³. Introducing additional high-rise buildings increased height variability and was associated with NO₂ reductions up to ∼7 µg m⁻³ in the eastern cluster. Targeted tree-canopy expansion was associated with hotspot reductions of up to ∼4 µg m⁻³ but produced small local increases in some areas. The combined intervention package was associated with the largest overall improvement, with NO2 reductions up to ∼7 µg m⁻³. The results show that integrating urban morphology within machine-learning LUR models can provide a practical, fine-scale evidence base for spatial targeting and comparative appraisal of NO₂ mitigation options in high-density urban districts.

Item Type:	Article
Date Type:	Published Online
Status:	In Press
Schools:	Schools > Architecture
Publisher:	Elsevier
ISSN:	2210-6707
Date of First Compliant Deposit:	19 October 2026
Date of Acceptance:	10 October 2026
Last Modified:	19 Oct 2026 11:30
URI:	https://orca.cardiff.ac.uk/id/eprint/190295

Actions (repository staff only)

Edit Item