Hourly impact of urban forests on land surface temperature based on machine learning

Machine learning reveals that urban forests in Rome, particularly those with midday canopy heights under 6 meters and afternoon cohesion around 70–95%, provide the most effective cooling against urban heat island (UHI) effects. Using ECOSTRESS thermal imagery from 12 clear-sky summer scenes, researchers analyzed vegetation-structure impacts on land surface temperature (LST) with random forest models, identifying nonlinear thresholds for optimal cooling. These findings offer structure-informed greening strategies tailored to heritage cities with limited space, bridging the gap in hourly and morphology-dependent UHI mitigation research.

Fan, P., Wang, H., & Imbroglini, C. (2025). Hourly impact of urban forests on land surface temperature based on machine learning. Results in Engineering, 28, 107584. https://doi.org/10.1016/j.rineng.2025.107584