Long-Term Thermal Performance of Urban Green Roofs: Insights from Multi-Year Monitoring and Microclimate Mitigation

Green roofs are widely recognized as a critical nature-based solution for mitigating the urban heat island (UHI) effect. However, prior research has predominantly focused on small-scale experimental plots or idealized simulations, with limited long-term observational data from full-scale implementations. Additionally, systematic investigations into their thermal regulation mechanisms across spatial and temporal scales remain scarce, impeding evidence-based design and broader adoption. This study addresses these gaps through a comprehensive analysis of green roofs in Nanjing, China—a city with a subtropical monsoon climate. Leveraging multi-year observational data from extensive and intensive green roofs, we assess their outdoor cooling efficacy across diurnal, seasonal, and annual cycles. Using correlation analysis and multivariate regression modeling, we elucidate how environmental drivers shape thermal performance variability. Furthermore, we integrate the ENVI-met microclimate model and the Weather Research and Forecasting (WRF) model to evaluate how vegetation type, coverage rates, and irrigation practices influence cooling efficiency. Our findings reveal that green roofs exhibit distinct diurnal and seasonal variations in cooling effects while sustaining annual thermal benefits. They significantly moderate neighborhood-scale microclimates, with cooling impacts on medium- and low-rise buildings extending to pedestrian levels. Scenario analyses demonstrate that varying vegetation types, coverage rates, and irrigation strategies reduce near-surface air temperature, though their cooling performance differs markedly. Crucially, urban-scale green roof implementation can substantially mitigate UHI intensity. These insights offer actionable strategies for optimizing green roof designs to enhance thermal resilience in subtropical cities, bridging critical gaps between research and practical application.