Spatial Dynamics of Mangrove Community Ecosystem in Muaragembong, Indonesia: Integrating Drone Data, NDVI, and Support Vector Machine (SVM)

Abstract

The mangrove forest in Muaragembong has experienced fluctuations in area over time, prompting this study to analyze the dynamics of mangrove community changes from 2019 to 2023. The study focuses on two main areas in Muaragembong: Muara Nawan and Muara Kuntul. The approach integrates remote sensing, including Planetscope satellite imagery and aerial photographs from DJI Phantom 3 Pro taken directly in 2019. The methodology combines remote sensing technology, Normalized Difference Vegetation Index (NDVI) analysis, Support Vector Machine (SVM) algorithm, and drone validation. The classification results obtained two dominant mangrove genera: Avicennia sp. and Rhizophora sp. Spectrally, Avicennia sp. reflects a greater amount of near-infrared than Rhizophora sp. with a difference range of 10–15%. Validation and accuracy assessment showed a high level of accuracy with an Overall Accuracy value of 89% and a Kappa Coefficient value of 0.83. There was an increase in Muara Nawan of 7.1 hectares, while Muara Kuntul experienced a reduction of 2 hectares. The changes in mangrove communities occur due to rehabilitation, land-use conversion, and local utilization of mangrove resources. Overall, this study shows the effectiveness of integrating the SVM algorithm, drone data, and NDVI analysis to monitor mangrove community dynamics over 5 years.