Assessment of the Impact of the Iype of Water Resource (surface and ground) on the Performance of Intelligent Models in Simulation of Water Quality Parameters

The main purpose of this study is to evaluate the effect of type of water resource in simulating water quality parameters in Savel plain by the intelligent models. In this investigation, the performance of ANN and CANFIS models in simulation of TDS, SAR, EC, and TH were compared with each other. For this aim, considering the correlation of the parameters, the Ph, Ca, Mg, Cl, and SO4 from 1966 to 2016 were selected as the input variables. The results indicated that the minimum error by the ANN in groundwater and surface water resources were related to the TH and the EC with 1.29 and 5.3 % respectively. The maximum error in both resources corresponded to the SAR. The same results were obtained by the CANFIS. The consequences of the T-Test showed that there was a significant difference between the ANN and the CANFIS in estimation of the SAR in the groundwater. So that the value of NRMSE decreased from 32.9% in the ANN to 22.82% in the CANFIS.  Concerning the other parameters, there was not any significant distinction. In case of the surface water, the CANFIS led to increase of proficiency of simulation of the SAR and the TDS. So thar, the value of the NRMSE declined from 60.35 and 29.27% in the ANN to 28.72 and 16.85 % in the CANFIS. In general, the impact of the type of the water resource in the accuracy of intelligent models in simulation of water quality parameters was approved.