Optimization Model of Cirata Reservoir Management Using Discrete Markov and ARIMA Discharge Forecasting Methods

Abstract

Even though the management operation guidelines have been determined, the optimal management of the Citarum Cascade Reservoir has not yet been achieved. Optimal management of cascade reservoirs in an integrated manner is difficult to be carried out due to complex operating procedures and high uncertainty of hydrological components. So, research on the optimal management of each reservoir considering the discharge forecasting model as input discharge (Qin) was proposed. The objective of this study was to optimize Cirata Reservoir's operational management. Hence all raw water demands are met without any water wasted through the spillway, as indicated by the correlation between inflow discharge (historical and model) and the correlation between trajectory (guideline and actual). In this case, discharge forecasting methods used were Discrete Markov and ARIMA. The continuous with five years return period (R5 continuous), continuous with ten years return period (R10 continuous), 3-classes Discrete Markov, and 5-classes Discrete Markov methods were applied for guideline trajectories. Based on the results, the correlation coefficients between inflow discharge were 0.63 (3-Classes Discrete Markov); 0.78 (5-Classes Discrete Markov); and 0.68 (ARIMA (1,0,0)(1,0,1) & ARIMA (0,0,2)(1,0,1)). Optimization simulations were carried out for 16 scenarios with combinations of four discharge models and four guideline trajectories during the 2016-2020 period. Based on the research, the scenario by the 5-Classes Discrete Markov discharge model and the R10 continuous guideline trajectory is the most optimal management of the Cirata Reservoir, with a correlation coefficient between the trajectory was 0.91.