Estimation of release history of groundwater pollution source using ANN model

Abstract

Estimating the release history of a groundwater pollutant source is an important environmental forensics problem. The knowledge of the release history of pollution source is critical in the prediction of the future trend of the pollutant movement. In addition, for identifying the responsible parties for allocating the remediation costs as well as in choosing an effective remediation strategy. Estimation of the release history with the help of concentration data is an ill-posed inverse problem. A novel approach based on ANN modeling has been developed in this study to estimate the release history of groundwater pollution source without using the prior knowledge of lag time. The required sampling duration of the breakthrough curve has been decreased in this study using the only upper half portion of the breakthrough curve which also reduces the uncertainties associated at the tail ends of the breakthrough curve. The previous studies in this area utilize the complete breakthrough curve whose lag time is completely known. The Levenberg–Marquardt algorithm has been used to train ANN model. The problems solved in this study address both two and three-dimensional flow fields with erroneous concentration data. The results indicate that the developed ANN model appears to be robust even for large measurement error level in concentration data up-to 10% and very effective in solving these problems.