Technological innovations in urbanization and land surface temperature analysis: A remote sensing and machine learning case study of Delhi

The rapid urbanization of Delhi has significantly transformed its Land Use and Land Cover (LULC), leading to profound alterations in Land Surface Temperature (LST) and exacerbating environmental challenges. Therefore, this study integrates multi-temporal Landsat imagery, advanced machine learning techniques, and landscape fragmentation indices to assess the impact of urban expansion on LST from 2001 to 2021. LULC classification was conducted using the Random Forest algorithm, achieving high classification accuracy. Landscape metrics such as Number of Patches (NP), Largest Patch Index (LPI), Mean Patch Area (MPA), Patch Density (PD), Landscape Shape Index (LSI), and Fractal Dimension (FD) were computed to quantify urbanization trends. The mono-window algorithm was used for LST retrieval, and a spatial regression framework incorporating Spearman’s correlation and multivariate regression models was applied to establish statistical relationships between urbanization metrics and temperature variations. Results indicate a 50 % increase in built-up area, accompanied by an 81 % reduction in open land and a 15 % decline in cropland over two decades. LST exhibited a substantial rise, with the mean temperature in built-up regions increasing from 36.58 °C in 2001 to 41.81 °C in 2021, and peak temperatures reaching 51.60 °C, highlighting the intensification of the urban heat island (UHI) effect. Regression analysis revealed a strong negative correlation between LPI and LST (R² = 0.99), indicating that fragmented urban structures contribute to higher temperatures. Conversely, LSI and FD showed positive correlations with LST, confirming that irregular and dispersed urban growth patterns exacerbate thermal stress. These findings underscore the need for sustainable urban planning, emphasizing compact city development, green infrastructure, and water body conservation to mitigate UHI effects.