Stream Flow Forecasting using Least Square Support Vector Regression

Abstract

Effective stream flow forecast for different lead-times is useful in almost all water resources related issues. The Support Vector Machines (SVMs) are learning systems that use a hypothetical space of linear functions in a kernel induced higher dimensional feature space, and are trained with a learning algorithm from optimization theory. The support vector regression attempts to fit a curve with respect to the kernel used in SVM on data points such that the points lie between two marginal hyper planes which helps in minimizing the regression error. The current paper presents least square support vector regression (LS-SVR) to predict one day ahead stream flow using past values of the rainfall and stream flow at three stations in India, namely Nighoje and Budhwad in Krishna river basin and Mandaleshwar in Narmada river basin. The relevant inputs are fixed on the basis of autocorrelation, Cross-correlation and trial and error. The model results are reasonable as can be seen from low value of Root Mean Square Error (RMSE), Coefficient of Efficiency (CE) and Mean Absolute Relative Error (MARE) accompanied by scatter plots and hydrographs.