Spatiotemporal LULC transformation and its association with urban heat island amplification in Hyderabad city using machine learning

Hyderabad has experienced rapid urban expansion during the last two decades, resulting in substantial land use/land cover (LULC) transformation and increasing urban heat island intensity. However, comprehensive assessments integrating long-term predictive modelling with high-confidence hotspot identification remain scarce for rapidly growing tropical cities. Therefore, this study introduces a novel multi-method approach combining machine learning algorithm (Support Vector Machine), geospatial statistics (Getis-Ord GI hotspot analysis at 99% confidence level), and predictive modelling (ANN-MLP using MOLUSCE plugin) to assess and forecast LULC and land surface temperature (LST) dynamics in Hyderabad city. Landsat 5 TM and Landsat 8 OLI/TIRS data of 2001, 2011, and 2021 were used to quantify LST variations and identify persistent hotspots. The results indicate that the highest temperature is observed in urban built-up areas (35.93ºC) and open spaces (37.25ºC) and the lowest temperature is observed in water bodies (17.99ºC). The multiple linear regression analysis shows that urban built-up fraction (β = 0.58, p < 0.001) is the strongest positive driver of LST, while water bodies (β = − 0.44, p < 0.001) and NDVI (β = − 0.53, p < 0.001) have the most significant cooling effects. The predicted modelling for 2030 showed that the temperature of between 37–40°C covered the largest area (10,065 hectares) followed by 34–37 °C (6,994 hectares) influenced by urbanization (7,876 hectares of built-up area). Also, open space (1,959 hectares), water bodies (0.72 hectares) and aquatic vegetation (2.34 hectares) were associated with lower LST values than built-up areas in Hyderabad City. This study is the first to combine hotspot analysis with long-term predictive modelling in the city of Hyderabad, thus offering actionable insights to urban planners through a scenario-based assessment of future thermal conditions. Findings such as these highlight the need to incorporate water bodies, green spaces, and sustainable land-use planning in mitigating UHI effects and improving urban climate resilience.