Evaporation is one of the main processes in the hydrological cycle and one of the most important factors in agriculture, hydrology, aerology, reservoir operation, designing of irrigation systems and irrigation scheduling. Management and operation of reservoirs, especially in arid and semi-arid regions, require accurate evaporation estimates. As Zayandeh-Rud reservoir is a strategic source of supply Isfahan drinking water, precise estimation of evaporation is required for managing. In this regard, this study has investigated uncertainty analysis of estimation models from the Zayandeh-Rud reservoir using Bayesian approach. For this aim, Jensen-Haise, Stephen-Stewart, Makkink, Hargreaves and Doorenbos-pruitt models from the radiation-temperature group and Blany-Criddle and Hamon models from the temperature-day length group have been used to predict monthly evaporation of Zayandeh-Rud reservoir during the 18-year period (2002-2019). To improve the accuracy of the results and reduce the uncertainty of evaporation estimation models, Bayesian Model Averaging (BMA) was applied to merge simulation results of evaporation estimation models. Then to identify suitable models for estimating monthly evaporation, the results of all methods were compared with Class A pan evaporation. According to our finding, among the used models, The BMA has significantly increased the simulation accuracy and exhibited results approximately close to the evaporation pan data. Sequentially, the Hamon model by presenting the meaningful values is in the second priority in terms of estimated accuracy. In the next step, 90% confidence intervals were derived for results of BMA methods as well as other models using Monte Carlo sampling. Based on the result, has been discovered except for Hamon model, the uncertainty of estimation evaporation methods is relatively high and the smaller percentage of observed data (especially in the hot months) fall into the 90% uncertainty bounds of models, while the BMA method covers more observed point and significantly reduce the uncertainty evaporation estimation. Therefore, BMA and Hamon methods were used to estimate evaporation at monthly time step to simulate reservoir storage, water shortage, and release with standard operation policy (SOP) during the period 2002-2019. To eva‎luate the methods performance, efficiency criteria of the reservoir in terms of reliability, resiliency, vulnerability and root mean square error (RMSE) were used. The results of reliability, resiliency, and vulnerability show that the BMA method is closer to pan evaporation method. Also, this method with RMSE equal to 24.71 million cubic meters has exhibited more acceptable performance than the Hamon model with RMSE equal to 35.33 million cubic meters in the estimation of reservoir storage of Zayandeh-Rud dam during the study period.

جهانگير عابدي كوپايي

عليرضا گوهري

سعيد سلطاني كوپايي، محمد شايان نژاد