Relationship between groundwater quality and distance to fault using adaptive neuro fuzzy inference system (ANFIS) and geostatistical methods (case study: North of Fars Province)

Abstract

The aim of this paper is to use Kriging (spherical, exponential, and Guassian models) and Inverse distance weighted (IDW) methods to prepare the water quality map. In addition, the relationship between water quality and distance to fault is determined in northeast of Fars province, Iran. Adaptive neuro fuzzy inference system method is also used to predict groundwater quality. The measured Sodium adsorption ratio and electrical conductivity parameters that are obtained from 384 wells in 2005 to 2014 are utilized to determine groundwater quality. The results show that the Kriging method (spherical model) has a higher accuracy with lower RMSE value than IDW method. Thereafter, this model is used to prepare the interpolation maps. Moreover, the results indicate the hybrid model in terms of maximum R2 and the minimum error is suitable enough to predict water quality parameters. In addition, the results depict by increasing the number of fault, the groundwater quality is decreased and vice versa.