Water Level Elevation Variations Modeling Using Support Vector Machine and Neural Network

This study aimed at analyzing the hydrological changes in the Lake Urmiabasin with focus on the response of the lake water level to meteorological factorsby means of two models was applied. For this, Support Vector Machines (SVM) and MLP-Artificial Neural Network (ANN) models developed for simulating the Urmia Lake water level variations. The yearly historical data of rainfall, temperature and discharge of the Urmia Lake basin and lake water level fluctuation were used. The outcome of the SVM based models are compared with the ANN.The root mean squareerrors (RMSE), sum square errors (SSE) and determination coefficient statistics (R 2) are used as comparison criteria. Analysis results showed that the (RMSEs) of 0.23and 0.5 m obtained by SVM and ANN respectively and SSEs of 0.43 , 2.01 and R 2 of 0.97, 0.93 obtained by SVM and ANN respectively. The results of SVM model show better accuracy in comparison with the ANN models.