An Improvement on the Estimation of River ECs Using ANN Models and ANFIS Involving PCA Analysis, Case Study; Nekarood River, IRAN

Estimation of changes in water quality parameters including electrical conductivity along a river is essential. In this paper, ANN and ANFIS-SC were used to estimate the ECs of the Nekarood River, North Iran, from 1992-2013. The study period was divided into two periods of dry and wet, based on the river flow rate. Then, Using the PCA, the effective parameters in EC estimation were determined and used for model inputs. The results showed that the least amounts of RMSE in dry and wet periods in the two methods was related to the first component. The value of RMSE in the ANN model, during the training and testing process of dry period, were estimated equal to 22.35 and 10.81µS/cm, respectively and as well for the wet period were equal to 6.92 and 8.49µS/cm, respectively. Also, for the ANFIS-SC method, the RMSE values in the training and testing process were equal to 21.86 and 13.14µS/cm for the dry period and were equal to 8.5 and 13.19µS/cm during the wet period. The first component parameters in the dry period include Ca, So4, Hco3, Mg, TDS, total anions and cations, and for wet period, its parameters included Ca, Hco3, TDS, total anions and cations.