Zahra Sharifi; Raoof Mostafazadeh; Abazar Esmali Ouri; Zeinab Hazbavi; Mohammad Golshan
Abstract
Daily flow data are a prerequisite for water resources management, but it is not possible to measure it in many upstream watersheds. In this study, different optimization algorithms have been used to evaluate the efficiency of the SIMHYD model. Therefore, the discharge data of Kouzetopraghi rive gauge ...
Read More
Daily flow data are a prerequisite for water resources management, but it is not possible to measure it in many upstream watersheds. In this study, different optimization algorithms have been used to evaluate the efficiency of the SIMHYD model. Therefore, the discharge data of Kouzetopraghi rive gauge station was selected as the study data (805 km2) located in Ardabil province. The daily data of rainfall, evapotranspiration of the meteorological stations in the study area were used to simulate the daily river flow data. Optimization methods including genetic algorithm, comprehensive competitive evolution, search pattern, multi-start search pattern, uniform random sampling, Rosenbrook, multi-start Rosenbrook optimization were evaluated based on statistical efficiency criteria. The mean value of discharge values by genetic algorithms, multi-year pattern search, uniform random sampling, multi-start Rosenbark, Rosenbork, comprehensive competitive evolution, search pattern were 0.031, 0.023, 0.085, 0.032, 0.024, 0.032, 0.031, respectively. The results showed that the change of optimization algorithms has a significant effect on the calibration accuracy of the model, so that the values of the Nash-Sutcliffe efficiency criteria for the employed algorithms were 0.42, 0.31, -8.55, 0.38, 0.56, 0.023, and 0.24, respectively. The Rosenbrook algorithm had higher accuracy in calibrating the SIMHYD hydrological model compared to other algorithms used. A part of the modeling error can be related to the inconsistency of precipitation and runoff data due to the multiplicity of stations.
