Uncertainty Analysis of River Water Quality Based on Stochastic Optimization of Waste Load Allocation Using the Generalized Likelihood Uncertainty Estimation Method

The aim of this study is to improve the water quality of rivers while satisfying the interests of pollution sources and environmental protection agencies (EPA). For this purpose, a stochastic integrated simulation–optimization approach is developed for waste load allocation (WLA) in a river system. The water quality simulation model (QULA2Kw) is coupled with an evolutionary optimization model (multi-objective imperialist competition algorithm (MOICA)) to minimize wastewater treatment costs and biochemical oxygen demand (BOD) violations of the standard level. The applicability of the approach is demonstrated by the case study of the Dez River in Iran. The stochastic model (ARIMA) is used to forecast the headwater from 2022 to 2025. The influence of the uncertainty of the stochastic parameters (headwater, oxidation rate, point source inflow, abstraction, and point source concentration) is evaluated by the Generalized Likelihood Uncertainty Estimation (GLUE) model. The results showed that the point source inflow uncertainty is higher than other parameters. The results of optimal WLA under the uncertainties showed that the dissolved oxygen (DO) uncertainty bound was narrower than the BOD. The solutions in Pareto fronts showed the contradiction between polluters and environmentalists' interests, and according to the waste load criterion, using this methodology not only improved the river water quality but also there were least violations of standards along the river.