<![CDATA[This study aims to develop an accurate monthly rainfall prediction model for Sabang City, Indonesia, to support agriculture, disaster mitigation, and water resource management in coastal regions with complex climatic conditions. An Artificial Neural Network (ANN) trained using the Levenberg–Marquardt (LM) algorithm was employed, combining the Gradient Descent and the Gauss–Newton methods to enhance convergence speed and training stability. Meteorological data from 2015–2024, including temperature, humidity, air pressure, sunshine duration, wind direction, wind speed, and rainfall, were obtained from the Maimun Saleh Meteorological Station. Model performance was assessed using Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and the coefficient of determination (R²). The optimal architecture consisted of a single hidden layer with 25 neurons, producing an MSE of 955.84 mm², an RMSE of 30.91 mm, an MAE of 23.06 mm, a MAPE of 34.8%, and an R² of 0.93. These results indicate that the ANN-LM model effectively captures nonlinear climatic relationships and seasonal rainfall variability. The MAPE value falls within the acceptable range reported in forecasting literature, demonstrating practical reliability. Overall, the ANN-LM approach outperformed conventional backpropagation in accuracy and training efficiency, indicating its suitability for rainfall prediction in coastal areas.]]>
Copyrights © 2026
