<![CDATA[This study aims to predict the Ciliwung River water level in DKI Jakarta using an Long-Short Term Memory (LSTM) model with a multi-head attention mechanism. Increasing flood frequency due to climate change necessitates an effective early warning system. Utilizing historical water level data and related meteorological variables, the LSTM model with multi-head attention demonstrated superior performance, with Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), and Symmetric Mean Absolute Percentage Error (SMAPE) of 31.74, 3.3%, and 3.3%, respectively. Predictions for the next 72 hours indicate safe water levels between 450 cm and 500 cm, suggesting no flooding. In conclusion, the LSTM model with multi-head attention enhances water level forecasting accuracy and serves as a useful flood risk mitigation tool in Jakarta. This research significantly contributes to the development of flood early warning systems and the application of machine learning in disaster mitigation.]]>
Copyrights © 2024
