Two approaches involving source identification using dating and artificial neural network (ANN) are demonstrated within a conceptual modeling framework. The first approach helps in reconstructing historical contaminant loading at a given location, using concepts of dating or age of the contaminant material moving with groundwater at an observation well. The second approach using ANN is superior as it helps identifying both source strength and its explicit location in a finite difference grid system. Both the cases are demonstrated using synthetic data by assuming contaminant sources at random locations and by generating concentration profiles at observation wells with the help of existing groundwater flow and transport simulators.

Identification of unknown contaminant source(s) in space and time in terms of location and strength for remediation purposes have been central issues in environmental engineering. The source often occurs in the form of a spillage on the surface or as a leakage from underground tanks and pipelines. Generally the only input data that could possibly be obtained for such problems is concentration profile or release histories at one or more observation wells downstream along the direction of groundwater flow. The problem essentially involves solution of an inverse groundwater flow and transport problem through modelling.

artificial neural network, source identification, conceptual, modeling, framework, contaminant, observation, simulators, remediation, environmental, engineering, concentration