Seagrass Percent Cover Mapping Around Teluk Terima Using Machine Learning for Blue Carbon Stock Estimation
Alfian1, M. Yozar Amrozi1, Puji Rahmaini1, and Pramaditya Wicaksono.2

1Departement of Geographic Information Science: Postgraduate Student, Faculty of Geography, Universitas Gadjah Mada, Indonesia

2Departement of Geographic Information Science: Lecturer, Faculty of Geography, Universitas Gadjah Mada, Indonesia

Abstract

Seagrass ecosystems are important for absorbing blue carbon and providing ecosystem services in coastal areas. But, there is not a lot of spatial data on seagrass distribution and above-ground carbon (AGC) estimates in Indonesia especially Teluk Terima. This study aims to map seagrass coverage percentages and estimate AGC using PlanetScope SuperDove imagery with a Random Forest (RF) machine learning approach. Field data was collected using the photo transect method to generate training and validation data. Image processing included sunglint correction and pixel-based classification using the RF algorithm. Classification results showed an overall accuracy of 59.1% with a seagrass class user accuracy of 59.4%. Seagrass cover distribution was identified as dominant in the Tanjung Kotal and Labuan Lalang areas with density variations ranging from 0% to 82.6%. The estimation results indicate that AGC values vary from low to high levels, with the model demonstrating a moderate correlation between seagrass cover percentage and AGC. These findings indicate that RF is capable of identifying the spatial distribution of seagrass and predicting carbon stocks with sufficient accuracy, supported by representative field data. The results of this study are expected to serve as a reference for spatial data-based conservation management and climate change mitigation planning through the preservation of seagrass ecosystems in coastal areas.

Keywords: Blue Carbon, Climate Change Mitigation, Machine Learning, Seagrass

Topic: Topic E: Sustainable Development Goals