Application of Extreme Learning Machine (ELM) for Water Level Prediction in Krueng Peusangan River Basin (2014–2023)

The Krueng Peusangan Watershed in Aceh Province is highly vulnerable to flooding, with 20.39% of its area classified as flood-prone, particularly in Bireuen Regency. This study aims to develop a water level prediction model using the Extreme Learning Machine (ELM), a type of Artificial Neural Network known for its computational efficiency and ability to handle uncertainty in hydrological data. The model was trained using water level data from the Krueng Peusangan River from January 2014 to June 2023. The results show a Mean Squared Error (MSE) of 0.063, indicating high predictive accuracy. Compared to conventional methods, ELM delivers faster computation and better precision. This research contributes to the development of data-driven flood early warning systems, supports adaptive and sustainable water resource management, and offers potential for replication in other watersheds with similar characteristics. Furthermore, the model provides a scientific basis for formulating disaster risk reduction policies leveraging artificial intelligence technologies. The promising accuracy of ELM supports its potential integration into real-time flood early warning systems and long-term adaptive water resource management in vulnerable river basins.