INTEGRATED ASSESSMENT OF URBAN GROWTH, LAND USE/ LAND COVER DYNAMICS, AND URBAN HEAT ISLAND VARIATIONS USING RANDOM FOREST CLASSIFIER

Abstract

Urbanization significantly leads to changes in Land Use Land Cover (LULC) patterns, LST and Urban Heat Island intensity. This study analyzes the spatio temporal dynamics of LULC, LST and urban growth in Pune-Pimpri-Hinjewadi area over 25 year period (2000 to 2025) using multi temporal Landsat imagery. LULC classification is performed using Random Forest algorithm over four LULC classes (Water, Built-up, Barren, Vegetation). The classification accuracy achieved is in between 84% to 87% with strong Kappa coefficients between 0.78 and 0.83. The built-up increased from 158.88 km² in 2000 to 201.94 km² in 2025. In addition, Land Surface Temperature (LST) analysis indicates increasing temperature with mean temperatures ranging from 34.36°C (2000) to 45.21°C (2020), while minimum and maximum LST values varied between 18.90–28.53°C and 50.94–74.51°C, respectively. The linear regression relationship between the Normalized Difference Built-up Index (NDBI) and LST showed strong positive correlation ranging from (R² = 0.446–0.605). This confirms the impact of urbanization on thermal characteristics. Directional analysis reveals uneven urban growth the northwest (NW) and west (W) zones experiencing the highest growth, while the southeast (SE) zone remained nearly saturated.