Combination of SWEG and SSP Models for Monthly Stream Flow Forecasting; TOROGH Dam in North Khorasan province as a Case Study

Stream flow forecasting is one the effective mean for reservoir operation and hydropower rule curve optimizing. Using multivariable liner regression model is one of the conventional approaches for stream flow forecasting. High sensitivity of regression model to independency of independent variable (predictors) and ratio of statistical period to the number of predictors are challengeable problems for these models. Inter-correlation between predictors causes wrong estimation of predictor. At this paper for reduction of predictors for increase ratio of statistical period to the number of predictors and elimination of Inter-correlation between predictors, SSP model was prepared. SSP model use a strong searching algorithm to select effective initial predictors and principal component for seasonal and monthly forecasting. Deficiency of snow data such as snow water equivalent (SWE) is a problem of spring and summer stream-flow forecasting in cold catchment. SWE is measured twice a year in IRAN. There for, it is a prepared physical model SWEG to simulate daily SWE according other parameter such as precipitation, temperature, wind velocity and etc. it is used of Genetic algorithm to calibrate parameters of SWEG. In this paper output of SWEG is used as one variable of input variables of SSP for monthly stream-flow of TOROGH dam forecasting. The result of (RMSE) and coefficient of correlation shown that combination of these two models have enough accuracy for monthly stream-flow forecasting.