Urban heat islands (UHI) significantly affect urban sustainability, especially in arid cities facing rapid urbanization and climate change. This study aims to quantify the spatiotemporal variations in land surface temperature (LST) and identify key biophysical drivers affecting UHI patterns in Dammam, Saudi Arabia-an arid city experiencing rapid urbanization. We analyzed LST dynamics and their relationship with biophysical parameters over three decades (1993-2023) using Landsat imagery and advanced machine learning models, including Decision Tree (DT) and Random Forest (RF). Principal Component Analysis was employed to reduce data dimensionality and identify dominant thermal drivers. NDWI (r = -0.652), NDBI (r = 0.590), and NDVI (r = -0.259) emerged as the most significant predictors of LST, collectively explaining 78.6% of the thermal variance. The RF model significantly outperformed the DT model, explaining 88% of LST variance (95% CI: 0.862-0.906) with substantially lower mean squared error (0.151 vs. 0.274) and superior cross-validation performance across five folds (mean R-2 = 0.884 +/- 0.025). Results reveal significant thermal intensification over the study period, with mean LST rising from 29-32 degrees C in 1993 to 44-47 degrees C in 2023, indicating a warming rate of 0.485 degrees C/year (p < 0.001). High-LST zones (> 44 degrees C) expanded exponentially from 15 to 42% of Dammam's urban area, with built-up regions experiencing the most severe temperature increases (0.627 degrees C/year). Water-related indices demonstrated the strongest cooling potential, with NDWI contributing 22.95% to model accuracy, followed by NDBI (19.06%) and NDVI (16.12%). Vegetation and moisture indices showed strong negative correlations with LST (NDVI: r = -0.259, 95% CI: -0.298 to -0.218; NDWI: r = -0.652, 95% CI: -0.689 to -0.612), while urbanization indices exhibited positive relationships. Bootstrap uncertainty analysis (n = 1000) showed prediction intervals of +/- 2.1 degrees C in vegetated areas to +/- 3.4 degrees C in heterogeneous zones. These findings emphasize the urgent need for climate-sensitive urban planning in arid environments, specifically recommending the integration of water features in high-NDBI zones (> 0.3) and mandatory 20% green space coverage in areas exceeding 44 degrees C. This research provides a robust methodological framework for assessing thermal environments in similar arid regions and offers quantitative evidence for sustainable urban development strategies.