Prediction of Variations of Groundwater Levels for the North of Sahand with Neural-Fuzzy Inference System, Time Series and Regression

Unsustainable application of groundwater in the north of Sahand (East Azarbaijan, Iran) caused a decline more than 14m in water level from 1983 (1362) by now. Therefore, the optimum and sustainable utilization of this resource is a management necessity which is depending on modeling, trend analysis and future study of its how application. The present study was conducted with the aim of groundwater level analysis in the north of Sahand with time series, regression and neural-fuzzy inference system methods. The modeling, test and future studying were made for 50 years which 35 years (from 1983 to 2018) were applied to model and test; and 15 years (from 2019 to 2033) were applied to future study. Results showed that based on indices of performance evaluations, neural-fuzzy inference system produced more precise than other methods to analyze groundwater levels. The groundwater level declined more than 14m with an annually of 40cm and its will be 20.03m in 2034 with current conditions of applications of groundwater. Some conservational scenarios are recommended to improve consumption patterns for groundwater resource in this plain.