Groundwater Vulnerability Zonation, Using Nero Fuzzy Algorithm (The Case Study: Zanjan Province)

Assessing the vulnerability of an aquifer is essential for development, management and the ways to qualitatively control ground water resources as well as prevention of these waters’ contamination. The objective of this study has been the usage of a new intelligent Adaptive Nero Fuzzy Inference System (ANFI) to predict groundwater vulnerability in Zanjan province, Iran. In this study ANFIS model has been developed as a new way to predict the mentioned vulnerability. Data used in this study were as follows: the depth of water table, net recharge, aquifer circumference, soil environment, topography, the effect of non-saturation zone and electric conductivity. Several statistical and graphical criterions have been used to check the accuracy of the model. The model results validated using obtains data from groundwater vulnerability isolate. Results showed that the developed model had the proficient capability to successful prediction of groundwater vulnerability. It was found that the ANFIS model with R2=o.98 and RMSE=101/18 in training stage and R2=0.94 and RMSE=173.8 in testing stage was the best predictor among the other ones. The presented methodology in the current study was a new approach in estimating groundwater vulnerability and could be combined with other mathematical models for model updating. Also, in analysis of the ANFIS model, maximum percentage of contamination potential of the field has been shown to be related to low class 4 percent and the minimum was related to very high class 1 percent.