Exploring the cooling potential of green roofs for mitigating diurnal heat island intensity by utilizing Lidar and Artificial Neural Network

Urban areas frequently exhibit higher elevated temperatures than their rural counterparts due to the prevalence of structures over natural resources, a phenomenon known as daytime surface urban heat island (DSUHI). This study simulates the cooling effects of green roofs (GR) for mitigating DSUHI by utilizing 2D and 3D urban morphological parameters over downtown Austin, Texas, USA. We estimated spectral indices using Landsat 8, Sentinel-2A, and Lidar data to include built-up, vegetation, waterbodies, daytime land surface temperature (DLST), buildings (height volume and density), sky view factor (SVF), and solar radiation (SR). Finally, we integrated eleven different neural network algorithms for GR simulation, validation, and correlation between DLST and the above urban features- the strongest model generated an R² of 0.783 and an RMSE of 0.925°F. We found converting 4.2% of the total rooftop area to GR resulted in an average DLST decrease of 2.80°F. The most significant cooling effects occurred with buildings heights between 13 and 28 m, high SVFs, SR, and closer proximity to water bodies. Our findings amplify the strategic importance of GRs in urban morphology and planning, guiding green infrastructure development to mitigate and foster urban environment sustainability.