عنوان مقاله [English]
نویسنده [English]چکیده [English]
Background and Objectives: One of the ways to predict and estimate the amount of runoff from rainfall is the use of hydrological models. Calibration of effective parameters in hydrological models is one of the basic steps in using these models. However, this process is a critical step which should be carried out carefully to optimize the model parameters. Multi-objective optimization algorithms as one of the most important and practical topics in various fields of study could be employed to achieve a reasonable calibration. The purpose of these algorithms is to determine the values of model parameters to find the best possible solution and achieve different goals.
Materials and Methods: In this study، Multi-objective optimization algorithm (AMALGAM) used to calibrate conceptual daily hydrologic model (MILC). AMALGAM method combines two new concepts and takes place in two modes of evolution. In the first case, the evolutionary capabilities of four multi-objective algorithms NSGA-II, PSO, DE and AMS are used simultaneously to evolve the population, and in the latter case, the AMALGAM algorithm itself is used to evolve the population. This algorithm is called a multi-objective hybrid algorithm due to the simultaneous use of multiple multithreading algorithms. The reason for choosing the AMALGAM algorithm is the superiority of this algorithm in achieving a fast and accurate access to the sum total of Pareto's optimal responses to other multi-objective algorithms such as MOPSO, SPEA2 and NSGA-II. MILC model employs the Soil Conservation Service—Curve Number method for abstraction (SCS-CN) for estimation of losses، the geomorphological Instantaneous Unit Hydrograph (GIUH) for routing of rainfall excess of catchment. This paper applies a four-objective calibration strategy focusing on peak flows (NSE)، low flows (TRMSE)، water balance (ROCE)، and flashiness (SFDCE) to parameter estimation of MILC model. After calibration process، a trade-off point extracted from Pareto- front was selected to include the appropriate values of all four objectives simultaneously. This point is applied to verify the validation period.
Results: The obtained values during the validation period (0.71 ≤ NSE ≤ 0.78) indicate that the MILC model has Good performance to simulate the amount of peak flows but according to ROCE and SFDCE values it has weak performances to simulate the balance water and median flow respectively.
Conclusion: It’s recommended to use multi-objective optimization algorithm rather than one objective optimization for calibration of hydrological models because this optimization covers the all hydrograph flows. Selecting objective functions for calibrate the rainfall-runoff model is the key to recognizing the model as much as possible.