Integrating Sentinel-2 and ESA world cover for effective land use and land cover assessment using machine learning

Muhammad Iqbal Habibie, Nety Nurda, Doni Fernando, Robby Arifandri, Prabu Kresna Putra, Hari Prayogi, Dionysius Bryan Sencaki,
Integrating Sentinel-2 and ESA world cover for effective land use and land cover assessment using machine learning,
Advances in Space Research,
Volume 76, Issue 9,
2025,
Pages 4925-4958,
ISSN 0273-1177,
https://doi.org/10.1016/j.asr.2025.07.083.
(https://www.sciencedirect.com/science/article/pii/S0273117725008385)
Abstract: This study investigates Land Use and Land Cover (LULC) changes in East Kalimantan, focusing on the largest rainforest in Kalimantan. As urbanization intensifies in this region, understanding LULC dynamics is crucial for sustainable development and environmental conservation, particularly with the notable shift from forested areas to grasslands. Utilizing multi-spectral remote sensing techniques and machine learning models, this research employs satellite datasets, including Sentinel-2 and European Space Agency (ESA) World Cover, to analyze LULC changes. Various machine learning models, such as Random Forest, K-Nearest Neighbors (KNN), XGBoost, Support Vector Machine (SVM), and Artificial Neural Networks (ANN), were applied to assess their effectiveness in classification tasks. Among these, Random Forest emerged as the most effective model, achieving an impressive accuracy of 97.59 %, attributed to its ability to robustly utilize key features for classification. The findings highlight the importance of feature selection, as different models prioritize various environmental indicators. The research highlights those geospatial technologies, including remote sensing and GIS, have their fair share of advantages and disadvantages in LULC change monitoring. Satellite imagery and machine learning permit accurate monitoring of land cover changes to render it useful to policymakers and urban planners for evidence-based decisions. In addition, the findings recommend avenue for further work in the areas of multi-temporal analysis, advanced feature engineering techniques, and ensemble-learning techniques to improve predictive abilities and model robustness for future applications. This study, therefore, highlights the need for developing sustainable urban design strategies, especially for Indonesia’s new capital in East Kalimantan, to complement economic development with environmental conservation.
Keywords: Sentinel-2; European Space Agency (ESA) WorldCover; Hyperparameter tuning; Feature selection; IKN (Ibu Kota Nusantara); Land Use Land Cover (LULC) classification