How do land use changes affect temperature and groundwater in urban areas? An integrated remote sensing, and machine learning approach

Present study examines the complex interrelationship between land use/land cover (LULC) changes, land surface temperature (LST), and groundwater levels in Lahore, Pakistan, from 2013 to 2038, with a focus on rapid urbanization. This study aims to fill the gap in projections of future LULC and its impacts on water resources, utilizing advanced deep learning techniques to enhance the classification accuracy of land cover maps and the reliability of future land use simulations. Using Landsat imagery, LULC changes were classified, and LST values were derived for the 2013–2023 period, with accuracy validated through metrics such as the kappa coefficient. To predict future LULC changes (2023–2038), a Cellular Automata-Artificial Neural Network (CA-ANN) model was employed, achieving a prediction accuracy of 93 %. The analysis revealed significant increases in cropland (46.32 %) and built-up areas (23.11 %), while water bodies and vegetation cover decreased by 2.02 % and 10.14 %, respectively. LULC changes correlated with LST increases ranging from 19.91 °C to 46.52 °C. The CA-ANN model forecasts a continued rise in built-up areas (26.21 %) by 2038, alongside reductions in water bodies (1.21 %) and increases in tree cover (9.15 %). Additionally, groundwater depth data from monitoring wells showed a concerning increase in depth over the past decade, highlighting the influence of LULC changes on both LST and groundwater depletion, indicative of the urban heat island effect.