The operation optimization of multipurpose reservoir between ARIMA, continuous, and Markov Chain models on Jatigede reservoir, Indonesia

The critical challenge in reservoir management is optimizing reservoir volume by balancing inflow and outflow discharges. Achieving this balance is essential not only to meet downstream water demands but also to preserve the reservoir's ecological function and structural integrity. Accurate inflow estimation plays an important role in this process, as it directly impacts the calculation of outflow discharges and the stability of reservoir volume. This study aims to compare the performance of the ARIMA ( Autoregressive Integrated Moving Average ), Continuous, and Markov Chain models in the estimating inflow discharge of Jatigede reservoir in West Java Province, Indonesia as a case study. Many studies conducted for inflow discharge estimation, from the most sophisticated to the simplest. However, the performance of model depends on its accuracy to the observation data. In this study, the results underscore the significance of inflow estimation, showing that the Continuous model yields the highest correlation (0.944), the lowest RMSE (0.408), and MAE (0.115). In contrast, the Markov Chain model exhibits a correlation of 0.923, RMSE of 0.443 and MAE of 0.116, while ARIMA model reports a correlation of 0.792, RMSE of 0.621, and MAE of 0.278. The findings of study indicated that the Continuous model has given better accuracy in inflow estimation, so it the most suitable approach for optimizing reservoir management. • ARIMA model improves discharge estimation accuracy. • Continuous model captures discharge correlations and distribution. • Markov model estimates transition probabilities between release states. • Discharge modeling reflects reservoir dynamics and water availability. • Optimizing models enhances sustainable reservoir management and adaptation.