Automated Mangrove Detection Method using Combined Machine Learning and Mangrove Index over Indonesia

Indonesia possesses approximately 22.6% of the world’s mangrove ecosystem. The country’s mangroves exhibit diverse characteristics, both in terms of their growing environment and species variety. This ecosystem serves as a vital habitat for numerous species of birds, fish, crabs, and other organisms, significantly contributing to global biodiversity. However, mangrove mapping in Indonesia is still primarily conducted on a local scale within individual regions or provinces. Consequently, there is a need for a comprehensive mangrove detection method that can facilitate national-scale mapping, making the process more effective and efficient. Integration of satellite data which covers a wide area and machine learning can obtain mangrove detection to be faster and more accurate. The study proposes an automated mangrove detection method using a combined mangrove index and machine learning over Indonesia. Normalized Difference Moisture Index and Mangrove Vegetation Index from Landsat 8/9 images were used to obtain training datasets for the Random Forest process. The results show that the overall accuracy is 0.93, and the kappa accuracy is 0.91. It proves that the proposed method can be used for mapping mangroves and non-mangroves accurately.