Groundwater quality assessment using fuzzy inference system for drinking purposes (Case study: Sardasht city, West Azerbaijan-province)

Document Type : Research Paper


1 M.Sc, Department of Geology, Hydrology, Urmia University

2 Department of Geology. Urmia University, Urmia

3 Assistant professor Department of Geology. Payame Noor Abhar University


Assessment of Water quality is an important step toward the optimal and appropriate use of water resources for drinking. Therefore, the necessity of studying water quality characteristics in water resources management programs has been strongly emphasized. The lack of inherent certainty in water resources in evaluating goals, criteria, and decision-making units on the one hand, and inconsistency in the opinions and judgments of decision-makers on the other, tend to fuzzy set theories and subsequently, fuzzy logic as an efficient and useful tool for Planning and decision making are in the water resources. In the present study, we try to compare two conventional deterministic methods and a fuzzy inference model, from four years of average groundwater quality parameters of 33 sources including 10 wells, 22 springs, and one Qanat in Sardasht city to determine groundwater suitability for drinking purposes, based on WHO standard for each threshold. The parameters were used as inputs in the two models. In the definitive evaluation method (Schoeller diagram), water quality characteristics and diagrams were determined using Chemistry software. In the fuzzy inference model, the four-year average of eight water quality parameters was classified into three groups, In the first group parameters, and, the second group parameters, and in the third group two parameters TH and TDS were placed. Then, by defining fuzzy rules, the results of three groups of one output number in MATLAB software in a fuzzy inference system were obtained and then the output number was evaluated. Based on the deterministic method, the water quality of all studied samples was classified as good to acceptable. The results of fuzzy results showed that two samples with a 50% confidence level were acceptable and the other samples with 83-87% confidence level were desirable for drinking.


Articles in Press, Accepted Manuscript
Available Online from 26 September 2022