ABSTRACT
In this study, an inverse solution approach is proposed for simultaneous identification of unknown groundwater pollution sources and hydraulic conductivity zone structures. Many studies have provided methodologies about aquifer parameter estimation or groundwater pollution source identification separately, but only few studies deal with the simultaneous solution of both problems. Here, the proposed approach uses a linked simulation-optimization method to solve the problem. MODFLOW and MT3DMS models are used to simulate groundwater flow and contaminant transport processes. The hydraulic conductivity zone structure is identified using the fuzzy c-means clustering (FCM) approach. The association of hydraulic conductivity zone structures with the unknown pollution sources is accomplished through a linked simulation-optimization approach. The main objective of the linked approach is to determine the release histories of the potential pollution sources together with the hydraulic conductivity zone structures and homogeneous conductivities within the zones by minimizing the error value between simulated and measured hydraulic head values and pollution concentrations at available observation locations. In the optimization model, Particle Swarm Optimization (PSO) algorithm is implemented due to its efficiency in finding the global or near global optimum solutions. Since the solution is based on a heuristic approach, there is no need to define any initial values for the optimization procedure which is an important advantage of the proposed approach. The applicability of the identification procedure is demonstrated on a hypothetical problem setting. The performance of the proposed solution approach is first compared with previously developed Artificial Neural Networks (ANN), Genetic Algorithm (GA), and Harmony Search (HS) based solution models for the known hydraulic conductivity field conditions. Then, model performance is evaluated by solving the same problem for the unknown hydraulic conductivity field conditions. With this purpose, the problem is solved for different number of zone structures until reaching the best solution in terms of the final objective function value. Further, the model performance is also evaluated by considering measurement errors. The identified results indicate that the proposed solution approach provides identical or better results than ANN, GA and HS based solution models for the known hydraulic conductivity field and can also be used for the cases where any information regarding hydraulic conductivity field does not exist.
