3D Dynamic Flood Visualization System with AI-based Prediction Using LSTM, ANUGA and Leafmap for Jakarta Urban Area

This study presents the development of a 3D dynamic flood visualization system that integrates artificial intelligence-based rainfall prediction for the urban region of Jakarta, with a particular focus on North Jakarta. The system incorporates a Long Short-Term Memory (LSTM) model to forecast rainfall, combines it with hydrodynamic simulations using ANUGA, and displays the outcomes through interactive 3D visualizations using Leafmap. Historical rainfall records from BMKG (2015–2024), along with digital elevation and building data, serve as the foundation for model training and simulation. The LSTM model’s performance was assessed using several evaluation metrics including MAAPE, MAE, MSE, RMSE, and R², showing strong predictive capability (e.g., MAAPE: 0.58, RMSE: 31.74, MAPE: 3.3%). These rainfall predictions are converted into discharge inputs for the ANUGA simulation, whose outputs are then visualized in 3D using Leafmap to provide an intuitive and spatial understanding of flood dynamics and impacted areas. The system is aimed at supporting stakeholder decision-making in flood mitigation and management, with improved prediction accuracy and better spatial understanding