Pemetaan Tutupan Lahan Wilayah Pesisir Aceh Barat menggunakan Citra Sentinel-2 dan Algoritma Random Forest
Year: 2026
Type: article
Abstract: Accurate and up-to-date land cover information plays a crucial role in spatial planning, natural resource management, and environmental monitoring, particularly in rapidly changing coastal regions such as West Aceh, Indonesia. One of the key challenges in land cover mapping is developing a classification approach that is efficient, accurate, and reproducible in distinguishing diverse surface features. This study aims to generate a detailed land cover map of Meureubo Subdistrict, West Aceh Regency, using Sentinel-2 imagery (2025) and the Random Forest (RF) machine learning algorithm implemented in R Studio. The workflow involved several stages, including image preprocessing, selection of key spectral bands (B2-Blue, B3-Green, B4-Red, and B8-Near-Infrared), preparation of training and validation samples, supervised classification using the “terra” and “randomForest” packages, and accuracy assessment through a confusion matrix. Five major land cover classes were identified: built-up areas, dense vegetation, sparse vegetation, bare land, and water bodies. The classification model achieved an Overall Accuracy of 94.68% and a Kappa Coefficient of 0.92. The dominant land cover was dense vegetation (45.31%), followed by sparse vegetation (23.56%), bare land (15.70%), built-up areas (14.09%) and water (1.32%). These findings demonstrate that the RF-based classification method implemented in R Studio provides an accurate, efficient, and reproducible solution for land cover mapping. This approach offers valuable insights for spatial planning and sustainable coastal management in West Aceh. Keywords: Land cover, Sentinel-2, randomForest, terra, R Studio, West Aceh, coastal.
Institutions: Universitas Teuku Umar, Universitas Siber Asia
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7166183418