Multi-Objective Optimization Groundwater Network Using Genetic Algorithm (NSGA-II) and Empirical Bayesian Kriging (EBK) Method (Case Study: Silakhor plain)

Groundwater resource management depended on data obtained from the aquifer. Groundwater monitoring network can provide groundwater levels, but sometimes this information so much and not useful. This study develops a new multi-objective simulation-optimization model involving two objectives minimization of the total sampling cost for monitoring and maximization the spatially monitoring accuracy. Optimal design of groundwater network was considered in the Silakhor plain, regions of Lorestan, Iran. As the first step, a database includes of groundwater elevation in potential wells with use empirical Bayesian kriging (EBK) method in ArcGIS was produced. Inverse distance weighting (IDW) method used as simulation model and objective functions written in MATLAB software. Optimal groundwater monitoring network determines with preset conventions and finds by the non-dominated sorting genetic algorithm (NSGA-II). At final, a network with twelve observation stations that shown root mean square error (RMSE) value 0.605 meter. The optimal monitoring network, in comparison with the existing observation network, has been able to reduce the number of monitoring stations by 60%, improve the spatial distribution of stations, and predict appropriate zoning for unpredictable points.