機械学習を用いた空間的都市化と都市ヒートアイランドの関係性の理解：景観ベースの空間的フレームワーク

The urban heat island (UHI) effect driven by urbanization induced land use and land cover (LULC) change has emerged as one of the serious environmental issues. Urbanization is one of the prime drivers of the UHI effect in cities. Therefore, it is essential to understand the impact of the urbanization on UHI. This study aims to understand the relationship between urbanization and surface urban heat island (SUHI) using machine learning approach in Kolkata Metropolitan Area (KMA), India. A spatial urbanization index (SUI) has been developed to assess the spatial urbanization and its relationship with UHI using spatial landscapes metrics. A spatial gradient approach was also applied to understand the changes of land surface temperature (LST) along urban-rural gradient (URG). A correlation and regression analysis were employed to understand the relationship between urbanization and SUHI. The findings showed that the impervious surface (IS) substantially increased from 2000 to 2020 with a net increase of 128%. The area under high and very high SUHI intensity increased by 287%, and 207% The IS and LST from the core decreased with increasing distance along URG. The RF model showed that the area mean (AREA_AM) (built-up) had the greatest influence on spatial urbanization, followed by class area (CA), Cohesion (COHE), and percentage of landscape (PLAND). Among the spatial landscape metrics, largest patch index (LPI) had the greatest influence on SUHI followed by patch density (PD), division (DIV), number of patches (NP), and Interspersion and Juxtaposition index (IJI). Lastly, the spatial urbanization had a positive correlation with SUHI intensity. The findings emphasize the pressing necessity of sustainable urban planning strategies to alleviate the detrimental consequences of urban expansion on thermal environments. The integration of green and blue infrastructure into urban planning policies is essential for the effective combat of UHI.