Green space pattern and thermal comfort simulation based on high-resolution remote sensing images

The green space pattern has obvious impacts on thermal comfort simulation, and clarifying the correlation between the two is crucial for optimizing urban green space planning. To explore the correlation between green space pattern and thermal comfort simulation, a green space pattern extraction method based on high-resolution remote sensing images is designed (Method 1), and anti-interference green space recognition features and object level shadow recognition features are introduced. The study also develops a supplementary method for extracting green space patterns based on InternImage (Method 2) to remove noise from the green space dataset generated by Method 1. The study uses ENVI-met software to simulate the thermal environment and selects different thermal comfort indicators. The average accuracy and time consumption of Method 1 were 95.15% and 51.97ms, respectively, which were significantly better than those of comparison methods. The minimum accuracy of Method 2 was 94.28%, while the maximum accuracy of the comparative models was below 92%. From 2010 to 2024, the proportion of green space in Hangzhou city increased, with the most significant growth in area F, which had the best thermal comfort and a maximum growth rate of 3.93%. The green space extraction method has good performance, and areas with large green space areas and a high proportion can improve the thermal comfort. The research provides scientific methods and data support for urban green space planning and thermal environment optimization.