LAND COVER CLASSIFICATION OF KRUENG LANGSA WATERSHED USING MAXIMUM LIKELIHOOD METHOD

Abstract

Land cover classification plays an important role in supporting environmental management and spatial planning, particularly in watershed areas. However, recent studies have predominantly emphasized advanced machine learning approaches, with limited attention given to the applicability of conventional methods such as Maximum Likelihood Classification (MLC) in specific local contexts. This study aims to evaluate the performance of the MLC method in classifying land cover in the Krueng Langsa watershed using high-resolution RapidEye imagery. This research was conducted from July to October 2024 in the Krueng Langsa watershed, covering East Aceh, Aceh Tamiang, and Langsa City. Land cover classification was performed using supervised Maximum Likelihood Classification in ArcGIS 10.8.2, followed by accuracy assessment using a confusion matrix and kappa coefficient based on stratified random sampling. The results identified ten land cover classes, with dryland forest dominating the area at 38.01%, while open land had the smallest proportion at 0.80%. The classification achieved an overall accuracy of 89% and a kappa coefficient of 0.88, indicating an almost perfect level of agreement. These findings demonstrate that the MLC method remains effective and reliable for land cover classification in tropical watershed environments. This study highlights the continued relevance of conventional classification approaches and provides valuable baseline information for sustainable watershed management and regional planning in the Krueng Langsa area.