Machine Learning Algorithms for Land Cover Classification Using Hybrid Remote Sensing and Spectral Indices

Accurate land cover classification is essential for effective environmental monitoring and resource management. This study aims to evaluate the performance of different machine learning algorithms for land cover classification using hybrid remote sensing data and spectral indices. By integrating data from Landsat 8 and Sentinel-2 satellites, this study utilizes the advantages of both sensors to improve classification accuracy. Spectral indices such as NDVI, MNDWI, NDMI, NDBI, and BSI were computed to enrich the feature set used for classification. A comprehensive dataset of training samples representing different land cover types - water, vegetation, built-up area, and bare land - was used to train and validate the machine learning model. Random Forest, Decision Tree, and Support Vector Machine were applied, and their performance was evaluated using metrics such as Overall Accuracy (OA), Kappa coefficient, and F1-score. The results show that integrating hybrid remote sensing data significantly improves the classification performance compared to single source data. Among the evaluated algorithms, Random Forest showed superior accuracy and robustness across different land cover types. This study highlights the importance of hybrid data and the combination of spectral indices in achieving high accuracy land cover classification and provides insights into the use of Random Forest for land cover classification.