Coastal areas are highly vulnerable to the impacts of climate change, such as sea-level rise and abrasion, thereby necessitating dynamic management based on actual data. This study proposes and tests a Digital Twin framework integrated with Machine Learning to support climate resilience and sustainable coastal ecosystem management. The case study was conducted empirically in the Tangerang Coastal Area using multi-source datasets (2022-2026) comprising Landsat-8 satellite imagery (USGS), tidal data (BIG), and climatological data (BMKG). A Long Short-Term Memory (LSTM) algorithm was applied in the predictive analytics layer to model the time series of sea-level anomalies, while a Random Forest classifier was utilized to detect changes in mangrove coverage. The experimental results demonstrate that the LSTM model can predict sea-level anomalies with a Root Mean Square Error (RMSE) of 0.038 meters, and the mangrove ecosystem classification achieved an F1-score of 93.2%. This approach asserts that a Digital Twin is not merely a spatial visualization tool but a solid predictive analytical instrument for identifying vulnerable areas and achieving the Sustainable Development Goals (SDGs 13 and 14).
Copyrights © 2027