Modeling land use dynamics and land surface temperature using GEE and CA-ANN: a case study of Pakistan's freshwater basins

This study analyses land use/land cover (LULC) changes using observations from 2000, 2010 and 2020, and projects future dynamics up to 2070. Land surface temperature (LST) and spectral indices, including the normalized difference vegetation index (NDVI), normalized difference built-up index (NDBI) and normalized difference water index (NDWI), were derived from MODIS Terra imagery to examine spatial–temporal interactions. The Classification and Regression Trees (CART) algorithm achieved high classification accuracies (81.6%–88.1%). Future LULC was simulated using a Cellular Automata–Artificial Neural Network (CA-ANN) model (Kappa = 0.83). Results indicate increasing built-up areas and declining vegetation and agricultural land. Mean LST increased by ~4 °C (2000–2020), strongly correlated with NDBI (R² = 0.82), while NDVI showed an inverse relationship. NDWI exhibited weak correlations, whereas NDVI–NDWI interactions revealed a vegetation–water trade-off. The findings highlight significant anthropogenic impacts and emphasize the need for sustainable land management and climate-resilient planning.