PREDICTION OF FLOOD IN KARKHEH BASIN USING DATA-DRIVEN METHODS

Abstract. Flood causes several threats with outcomes which include peril to human and animal life, damage property, adversity agricultural fields. Hence, flood prediction is highly significant for the mitigating municipal environmental damage. The aim of this study was assessing performance different machine learning methods in predicting Karkheh basin. To this, we used Support Vector Machine (SVM), Least Square (LSSVM), Feed Forward Back Propagation Neural Network (FFBPNN), Radial Basis Function (RBFNN) simulate monthly streamflow area. Furthermore, models compared flood. All four indicated good simulating streamflow. However, LSSVM model had highest accuracy other R2 RMSE 85.89% 30.02 m3/s during testing periods, respectively. Similarly, performed better annual maximum comparison models.