Utilization of MERRA-2 Data for Precipitation Forecasting in Indonesia, Ibu Kota Nusantara: A Comparative Study of Traditional Methods and the Proposed Hybrid Fourier Series-ARIMA

We introduced the Hybrid Fourier Series-ARIMA model to enhance precipitation forecasting accuracy in Ibu Kota Nusantara (IKN) region. This hybrid approach combines the Fourier Series ability to capture seasonal variations with ARIMA capability to model residual errors, enhancing predictive performance. The study utilizes monthly precipitation data from January 2015 to December 2024, divided into a 90:10 split for training and testing. Model evaluation demonstrated that the Hybrid Fourier Series-ARIMA (5,1,3) outperformed traditional methods such as Naive Trend (NT), Simple Exponential Smoothing (SES), ARIMA, and Fourier Series. In the training and testing data, the hybrid model achieved the lowest MSE, RMSE, and MAPE, indicating superior accuracy. The 12-period forecast for 2025 projects distinct seasonal patterns in IKN, with the rainy season occurring from January to March and November to December, while the summer season spans from April to October. These findings suggest that the Hybrid Fourier Series - ARIMA model is a promising tool for long-term precipitation forecasting, aiding in strategic planning for water resource management and sustainable development in IKN. The use of MERRA-2 satellite data enhances model robustness and precision.