Pemetaan Sebaran Terumbu Karang di Perairan Pulau Kalimantung Gadang Menggunakan Citra Satelit dengan Algoritma Lyzenga dan Optimisasi Adaptive Moment Estimation

Abstract

Kalimantung Gadang Island, located in Central Tapanuli Regency, boasts a rich and diverse coral reef ecosystem. However, this ecosystem is under pressure due to various human activities, such as pollution and the use of dynamite for fishing, which threaten the sustainability of coral reefs in the area. The use of satellite imagery is one of the innovative approaches to map the distribution and condition of coral reefs effectively. With its wide range, high resolution, and integration capabilities with technology, satellite imagery provides an opportunity to conduct in-depth analysis that supports ecosystem conservation efforts. The research methods used are the Lyzenga algorithm, Adaptive Moment Estimation optimization, and the Support Vector Machine model that can overcome spectral variations in satellite image data. The data used is Sentinel-2A satellite imagery which includes Band 1, Band 2, Band 3, Band 4, and Band 8. The dataset used in this study includes 212 multipoints. This data is then divided into 70% train data, 20% test data, and 10% validation data. The Support Vector Machine model is trained and evaluated to classify aquatic areas: sand, coral reefs, vegetation, and the sea. The Support Vector Machine model has an overall accuracy of 92%, according to the final results.