Automated Calibration and Optimization of Water Planning and Allocation Models: Gilan Case Study

Abstract

Large-scale water-resource planning and allocation models with several interconnected components can have a significant number of unknown return-flow coefficients. This study presents an optimization-based approach for automatic calibration of a complex water planning and allocation model of the Gilan Province water-resource system (GWRS), in the heart of Iran's rice-production area. The frequent natural cycling of water among supply and demand sites is conceptualized by using flow links and their return-flow coefficients, which need to be optimally determined. A control module, consisting of a model-developer unit linked to optimization solvers in a software program solves the nonlinear program of the calibration problem by minimizing the differences between observed and predicted discharges. The simulated calibration results correspond well with observed records of river outflows at hydrometric stations, as well as with historical water allocations. The optimized parameter values are used subsequently to evaluate the GWRS performance against a scenario involving a reduced inflow to the Sefidrud Reservoir, Gilan, Iran. The derived systemwide operation policies for the reservoir and other components show that the GWRS has the potential to adapt to the planned 30% future reduction in water availability. (C) 2016 American Society of Civil Engineers.