Accurate Land Use and Land Cover (LULC) mapping is critical for environmental monitoring, resource management, and sustainable planning. This study evaluates the performance of four machine learning algorithms Random Forest (RF), Gradient Boosted Trees (GTB), Classification and Regression Trees (CART), and K-Nearest Neighbors (KNN) for LULC classification in Vazhikkadavu, Kerala, India, using combined Sentinel-1 Synthetic Aperture Radar (SAR) and Sentinel-2 multispectral imagery. Pre-processed datasets included spectral indices (NDVI, NDWI, NDSI) and were trained on 791 field-verified samples across seven LULC classes. Accuracy assessment employed Overall Accuracy (OA), Kappa Coefficient, Producer’s Accuracy (PA), and User’s Accuracy (UA). RF achieved the highest performance (OA = 0.96, Kappa = 0.93), followed by GTB (OA = 0.94, Kappa = 0.90), CART (OA = 0.93, Kappa = 0.88), and KNN (OA = 0.87, Kappa = 0.79). RF consistently outperformed others in PA and UA for major classes such as forest, built-up, and water, demonstrating robustness to spectral and textural variability. GTB showed strong but conservative classification, while CART exhibited slight overfitting and KNN underperformed in heterogeneous landscapes. Area quantification indicated RF and GTB yielded more balanced class distributions. The results highlight the superiority of ensemble-based approaches, particularly RF, for operational LULC mapping in complex tropical environments.