Generation of Long-term Monthly Stream flow Data (Case Study: Dez River)

An estimation of reservoir capacity substantially depends on river discharge. Either an overest imation or an underestimation of this parameter can bring about negative economical as well as social consequences. Generally, stream flow data generation is accomplished, using the stochastic methods.
Number of parameter in these methods, especially in monthly scale, will cause a considerable error of estimation. In this study, a nonparametric method called kernel density function was employed to generate the monthly stream flow for Dez River. Method for choosing the best simulation, among 50 monthly generated series, is based on an evaluation of mean absolute error (MAE) factor. A comparison between the most proper series and observed values of stream flow indicate a maximum difference of 11.6 % that can be acceptable to be observed in a problem. This approach can be proposed for generation of long-term monthly stream flow data for any river. This method especially allows, in cases of unavailability of long-term data, for reasonable decision-making in surface water
planning as well as in future management of water resources systems