عنوان مقاله [English]
Background and objectives: The hydrological simulation models represent a simplified representation of the real hydrologic system that helps to study the functioning of the basin in response to various inputs and better understanding of hydrological processes. These models are able to estimate the runoff values with the lowest possible time and costs using the simulation of rainfall-runoff process. Despite high efficiency, these models have uncertainty. One of the most important issues among researchers is the elimination of these uncertainties. Hence, the application of the combination technique as one of the most important approaches to improve results of simulations is main aim of this study. In this study, four models including the simple combination models (SMA), Weighted Average Method (WAM), Multi Model Super Ensemble (MMSE) and Modified Multi Model Super Ensemble (M3SE) were used analyze the hydrological simulations, in the Gharesou catchment, located in Kermanshah province.
Materials and methods: The Gharesou catchment, with nearly 5354 km2 area, is located in the northwestern parts of Karkheh basin and the western parts of Iran. In this study, the baseline daily data including observed temperature, rainfall and runoff during the period of 1997- 2008 were gathered from selected stations in the study area. 70 percent of the data was used for the calibration period (1997- 2005) and the remaining 30 percent for validation (2006- 2008). To this end, the models in the RRL package such as Simhyd, AWBM, Sacramento and TANK and the SCS-Milc and Hymod models, which are coded in the Matlab programming language, were used. Then, four combination methods including SMA, WAM, MMSE and M3SE were used to improve the results. Finally, the performance of each method was evaluated using normalized root mean square error (NRMSE) and Nash-Sutcliff (NS).
Results: In the present study, all simulated models provide acceptable results. The results of the combination methods showed that the application of these methods led to improve the simulation results. Also, the most improvement of results was achieved by M3SE and MMSE, respectively. For the M3SE method, the value of the NS and NRMSE evaluation criteria were 0.80 and 0.97 in the calibration period and 0.87 and 0.53 in the validation period, respectively.
Conclusion: As resultant it can be expressed that Multiple Combination techniques improved the results of simulated flow by each simulation model obviously. It also may be resulted that recent technique (M3SE) is more efficient than other due to incorporating the bias correction step. Finally it is observed that multimodel simulation generated by M3SE can be better at least comparable to the best-calibrated single-model simulations.