Calibration of a Water Resource Planning Model using Many-Objective Optimization

Water resource planning models traditionally incorporate hydrologic, water allocation and economic modules whose parameters should be properly estimated. This paper focuses on calibration of a water resource planning model. Our methodology includes three main parts: 1) development of a water resource planning model using WEAP software for Zarineh and Simineh River Basins located in Urmia Lake Basin in Iran, 2) application of VARS algorithm for sensitivity analysis of parameters of the developed model, and 3) usage of Many Objective Particle Swarm Optimization (MaOPSO) algorithm for model calibration and parameter estimation. It is worth mentioning that many objective optimization algorithms are utilized for problems with more than 3 objectives and here in this study we intend calibration of 7 objective functions defined based on Nash-Sutcliffe efficiency indices calculated in terms of residuals of time series of simulated and observed values of river discharge at stream gauges’ locations, dam reservoir’s storage volume, and aquifers’ storage volume. Furthermore the model’s parameters includes allocation priority, consumption rate, maximum withdrawal, and parameters of elevation-volume curves of aquifers. Sensitivity analysis using VARS algorithm shows that there are 17 influential parameters from total 27 parameters which are considered in calibration phase. To evaluate performance of MaOPSO, its results are compared to results derived using multi-objective PSO, which shows that MaOPSO has better performance dealing with such a complex problem.