Model Peramalan Debit Aliran Sungai Menggunakan Metode Gabungan Self Organizing Maps - Artifical Neural Network (Studi Kasus: Sungai Tapung Kiri)

Siak river basin now in criticial condition. This condition is due to land use changes in the upper body and cause silting in the river. The silting decreasing the capacity of the river so when heavy rain occurs, tha rivers could not accomodate it and it will cause flooding. The flood disaster which is very difficult to predict and lack of water resources at dry season made the citizens lack of preparation to deal with. A solution to anticipate that is to provide an early warning system. In order for the system to work well, it need a prediction method that could provide good quality of data, the method called SOM-ANN. SOM-ANN method consisted by two different types of algorithms, there are Self Organizing Maps (SOM) and Backpropagation. The purpose of this study was to test the reliability of SOM-ANN method in discharge predicting in Tapung Kanan River. The weighting result from SOM's learning applied to Backpropagation's learning, so that the pattern recognizing becomes faster and gaining more accuracy. By a comparison with ANN method, SOM-ANN method can improve a better performance and accuracy of predicting results with 1099% increase in performance and 44,74% increase in accuracy with an error value MSE = 0,001725, so that the discharge prediction modelling can be used to predict the discharge in the future