Abstract: Heat waves are unevenly distributed across the city, which leads to geographical differences in land use, albedo, and building form within the city. Previous studies of heat waves and urban heat islands in Seoul have used temperature data from the Automated Weather System (AWS) or satellite imagery to select vulnerable areas. However, AWS has only one or two stations per district(gu), so it is not possible to check temperature changes at the micro-spatial scale. Satellite imagery has a high spatial resolution of 30-250 meters, but it can be difficult to analyze due to cloud cover during monsoon season (Jangma) in Korea. In addition, satellite imagery captures the roofs of buildings, which can be different from the actual temperature experienced by humans. Therefore, this study aims to explore heat wave vulnerable areas at the micro-scale by utilizing temperature data from the Smart Seoul Data of Things (S-DoT) sensors installed over 1,000 in Seoul from 2020 to 2023 heatwaves. Both ordinary least squares and random forest models were adopted to explore the relationship between diurnal temperature and spatial variables.

Figure 1. 2024 AAG Annual Meeting

Recommended citation: Cho, S., Hwang, T., Lee, Y., Lee, S., & Hwang, C. S. (2024, April). Nonlinear analysis of high-resolution urban thermal environment using random forest method: Seoul, Korea [Poster presentation]. 2024 AAG Annual Meeting Honolulu, HI, United States.