Cooling Cities Through Data-Driven Insights: Distinguishing and Weighting Driving Climates, Urban Attributes, and Local Climates

Urban overheating poses a significant risk to most cities worldwide. To support effective heat mitigation decision-making, we propose a mitigation-centered machine learning model. A feature classification framework based on the scale at which physical processes or urban characteristics influence urban climate is introduced. Specifically, physcial processes primarily governed by background climate are classified as driving climates (DC), while those exhibiting two-way couplings with urban areas are categorized as local climates (LC). Meanwhile, urban morphologies, materials, and landscaping that not only have strong heat mitigation potential but are also manageable are classified as urban attributes (UA). This model enables the assessment of the most effective mitigation strategies because the effectiveness of multiple potential mitigation measures-i.e. the impact of urban attributes-can be evaluated through their respective weights. This is crucial, as some high-ranking urban climate drivers in conventional models, such as urban boundary layer height, may be less practical for direct mitigation efforts, as discussed in our perspective article [1].

As an initial case, we applied our model to Zurich, Switzerland, using Weather Research and Forecasting (WRF) simulations for the summers of 2017 and 2019, covering both heatwave and non-heatwave periods. Our model outperformed a reference model that lacked the classification of influencing factors. Additionally, incorporating historical heatwave data further improved model performance. Our findings suggest that increasing leaf area intensity (LAI) is the most effective strategy for reducing pedestrian-level air temperatures in Zurich, outperforming other heat mitigation interventions. Our model offers a practical framework to support urban heat mitigation, particularly in scenarios where multiple measures need to be assessed, prioritized, and implemented.

Reference

[1] Prioritizing Nature-Based Solutions and Technological Innovations to Accelerate Urban Heat Mitigation Pathways. Y Zhao, J Carmeliet, R Hamdi, C Yuan, X Ding, D Derome, Y Fan, S Jiang, J Peng. Annual Review Environment Resources, 2025.