Predicting long-term urban overheating and their Mitigations from nature based solutions using Machine learning and field measurements

Abstract

Urban overheating has become a global issue, exacerbated by climate change and leading to serious risks for public health and urban sustainability. Traditional methods, such as numerical simulations and field measurements, often face challenges due to uncertainties in input data. This study predicts the longevity and severity of future urban overheating by integrating field measurements with machine learning (ML) models, focusing on the impact of urban greening under different global warming (GW) scenarios. Field measurements were conducted from June 15 to September 14, 2024, at an office campus in Ottawa (a cold climate zone). Microclimate data was collected at four locations featuring distinct vegetation coverage: a large lawn area without trees (Lawn), a parking plot with no greening (Parking), a greenery area with sparsely distributed trees (Tree), and a forested area with 100 % tree coverage (Forest). Models—including Artificial Neural Networks (ANN), Recurrent Neural Networks (RNN), and Long Short-Term Memory (LSTM) networks—were trained on local climate data, with LSTM demonstrating superior accuracy. Four GW scenarios aligned with Shared Socioeconomic Pathways for 2050 and 2090 were examined. Results show that the Universal Thermal Climate Index (UTCI) at the Parking plot could increase from about 27 °C under GW1.0 to 31 °C under GW3.5. Low health risk (UTCI > 26 °C) is projected to rise at all sites, while dense tree coverage effectively prevents extremely high-risk conditions (UTCI > 38.9 °C). These findings underscore the importance of urban greening in mitigating severe thermal stress and enhancing outdoor comfort under future climates.