APLIKASI ALGORITMA BACKPROPAGATION FEEDFORWARD NEURAL NETWORKS UNTUK PREDIKSI ALIRAN SUNGAI (STUDI KASUS DAS GUMBASA, PALU SULTENG)

Rais Rais

JURNAL ILMIAH MATEMATIKA DAN TERAPAN
Vol. 6 No. 2 (2009)

Hydrological models are necessary in assessing water resources and valuable tool for water resources management. This paper describes applications of feedforward neural networks (FFNN) for Gumbasa watershed  in  Palu  Sulteng,  Indonesia.  Back-propagation  was  used  in  the  learning  rule  of  FFNN.  A series of daily rainfall, evapotranspiration and discharge data for 2 years (2006-2007) from Gumbasa  watershed  was  used.  The accuracy  is  evaluated  by  statistical  performance  index,  the shape  of hydrographs and the flood peaks. The results show that  FFNN is successful in predicting watershed discharge  in Cidanau  watershed.  These  hydrological  models  have  been developed  in  form  of application program Matlab 7.0.4 and applicable to use in other watershed.