Comparing Different Supervised Classifications Methods for Land Use and Land Cover Mapping with Sentinel-2 Data in Google Earth Engine

Environmentalists increasingly recognize the importance of land use and land cover (LULC) changes, making accurate LULC mapping crucial. This study classifies LULC in Yuanling County using supervised classification methods on Sentinel-2 satellite images via the Google Earth Engine platform. Four machine learning algorithms—support vector machine (SVM), random forest (RF), classification and regression trees (CART), and minimum distance (MD)—were employed to evaluate their performance. Results showed overall accuracy and kappa coefficients for SVM, RF, CART, and MD as 86.9% (0.83), 96.4% (0.94), 91.8% (0.89), and 72.3% (0.65), respectively. The RF classifier outperformed others, indicating superior accuracy, while the MD classifier is deemed unsuitable for LULC classification due to its low accuracy metrics.