Application of Support Vector Regression for Modeling Low Flow Time Series

摘要: Hydrologic time series modeling using historical records plays a crucial role in forecasting different hydrological processes. The objective of this study is to analyze the suitability Support Vector Regression (SVR) for monthly low flows three stations Mahanadi river basin, India. ‘low flow’ threshold was taken as Q75 discharge, i.e., flow equal or surpassed duration 75% observation period which obtained from daily discharge data. potential applicability SVR model assessed with two framework models (ANN-ELM, GPR) based on various statistical measures (r2, RMSE, MAE, Nash-Sutcliffe coefficient, function (OBJ), Scatter Index (SI) and BIAS). selection lowest OBJ value each station amongst (SVR, ANN-ELM, GPR). trained Radial Basis Function (RBF). Using same inputs, other (ANN-ELM also tested. From results, among all stations, outperformed GPR ANN-ELM (1.378, 1.202, 1.570). In addition, accuracy were checked mean error (0.474, 0.421, 0.509) SVR, (0.507, 0.489 0.500) (0.564, 0.603, 0.772) stations. results confirm that can be used satisfactorily Hence, could employed new reliable accurate data intelligent approach predicting (Q75 discharge) precedent water resources its allied field.