Measuring ocean physical asset account using machine learning approaches

The blue economy concept has been adapted as a strategy in setting development programmes and public policies in managing Indonesia’s marine resources. As a supporting instrument, accurate field data is needed when compiling the ocean account. Meanwhile, the support of qualified resources is needed during the field data collection process. Research on mapping water areas using satellite technology and machine learning techniques in producing water maps, especially in coastal areas. The approach is suitable for arranging a physical asset account, which is a component of the ocean account framework. So far, no research has implemented these developments to produce ocean physical asset account. Therefore, this study will cover in arranging the account by utilising Sentinel-2 imagery and implementing Random Forest, Support Vector Machine, and Extreme Gradient Boosting (XGBoost) machine learning methods, which according to previous studies are superior methods for mapping water areas. The modelling results show that there isan extensive change in coral, seagrass, and mixed ecosystem types (a combination of coral, seagrass, and macroalgae ecosystems) between 2020 and 2023.