Article Per Year (5 Year)
p-Index From 2021 - 2026
0.23
P-Index
This Author published in this journals
All Journal Jurnal Kelautan : Indonesian Journal of Marine Science and Technology
Adhitya Nugroho
Pattimura University

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

 1 
Comparative Performance Evaluation Of Machine Learning Algorithms For Sentinel-2 Benthic Habitat Classification Using Google Earth Engine Adhitya Nugroho; Muhammad Abdul Ghofur Al Hakim
Jurnal Kelautan Vol 18, No 3: Desember (2025)
Publisher : Department of Marine Sciences, Trunojoyo University of Madura, Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.21107/jk.v18i3.32389

Abstract

This study evaluates the optimal classification methodology and analyzes temporal changes over five years across four benthic habitat classes (Seagrass, Coral Reef, Rubble, and Sand) in the shallow waters of Ohoidertawun, Southeast Maluku, using Sentinel-2 imagery and the Google Earth Engine (GEE) platform. A comparative assessment of Machine Learning (ML) algorithms revealed that Random Forest (RF) demonstrated the best classification performance compared to Support Vector Machine (SVM), Classification and Regression Tree (CART), K-Nearest Neighbors (KNN), and Minimum Distance (MD) in benthic habitat mapping, achieving an Overall Accuracy of 0.856 and a Kappa Coefficient of 0.870. The classification results and accuracy assessment using the best-performing ML model from the 2025 Sentinel-2 imagery were used to analyze temporal changes relative to the 2020 Sentinel-2 data. Temporal analysis indicated a significant ecosystem shift, marked by a 52.41% increase in seagrass cover and a 31.46% decrease in coral reef area. These findings can serve as a recommendation for conservation site selection and urge stakeholders to help mitigate coral reef loss by utilizing the results of this research. The resulting benthic habitat map can serve as a reference for effective coastal resource management and blue carbon initiatives. Based on these findings, the Random Forest ML algorithm can be considered an optimal methodology for tropical benthic habitat mapping in the study area.Keywords: Benthic Habitat, Sentinel-2, Machine Learning, Google Earth Engine
Co-Authors Muhammad Abdul Ghofur Al Hakim