Modeling monthly streamflow of intermittent rivers in semi-arid regions using a hybrid forecasting model

Abstract The process of forecasting the monthly flows of intermittent rivers is one of the important issues in hydrology as well as in managing water resources and designing hydraulic structures for those rivers. In this research, Support Vector Machine (SVM) model was used to predict the monthly flows of the Adhaim River in Iraq, and the results showed that the application of SVM model failed to model intermittent monthly flows, as the value of the determination coefficient did not exceed 0.23 in the validation stage of the model. A hybrid model consisting of three machine learning models, K-means clustering, Bayesian networks, and Support vector machine, was applied to predict intermittent monthly flows. The results proved the efficiency of the proposed hybrid model in predicting monthly streamflow. The value of the coefficient of determination was 0.84 at the model validation stage. The application of the proposed hybrid model improved the results by 72% compared to the implementation of the SVM model.