Keywords
artificial neural networks, counterpropagation, parameter estimation, kriging
Start Date
1-7-2006 12:00 AM
Abstract
Current subsurface site characterization, plume delineation, remediation designs and monitoring network designs that rely on a limited, albeit large, number of sparsely collected data, tend to be expensive, cumbersome and frequently inadequate for solving multi-objective, long-term environmental management problems. We present a subsurface characterization methodology that integrates multiple types of data using a modified counterpropagation artificial neural network (ANN) to provide parameter estimates and delineate groundwater contamination at a leaking landfill. Apparent conductivity survey data and hydrochemistry data (i.e. heavy metals, BOD5,20, chloride concentration, etc.) are used to estimate the extent of subsurface contamination at the Schuyler Falls Landfill, located in Clinton County NY. The results of this research illustrate the feasibility of combining principal component analysis (used to reduce data dimensionality) with the counterpropagation ANN and traditional geostatistical methods (kriging) to estimate subsurface contamination. The ANN methodology for obtaining parameter estimates is data-driven and can easily incorporate a large number of data types obtained from diverse measurement techniques. This technique is also flexible as it does not require the computation of large covariance matrices and, once the ANN is trained, can produce realizations for subsurface characterization and monitoring in real time.
Application of an Artificial Neural Network for Analysis of Subsurface Contamination at the Schuyler Falls Landfill, NY
Current subsurface site characterization, plume delineation, remediation designs and monitoring network designs that rely on a limited, albeit large, number of sparsely collected data, tend to be expensive, cumbersome and frequently inadequate for solving multi-objective, long-term environmental management problems. We present a subsurface characterization methodology that integrates multiple types of data using a modified counterpropagation artificial neural network (ANN) to provide parameter estimates and delineate groundwater contamination at a leaking landfill. Apparent conductivity survey data and hydrochemistry data (i.e. heavy metals, BOD5,20, chloride concentration, etc.) are used to estimate the extent of subsurface contamination at the Schuyler Falls Landfill, located in Clinton County NY. The results of this research illustrate the feasibility of combining principal component analysis (used to reduce data dimensionality) with the counterpropagation ANN and traditional geostatistical methods (kriging) to estimate subsurface contamination. The ANN methodology for obtaining parameter estimates is data-driven and can easily incorporate a large number of data types obtained from diverse measurement techniques. This technique is also flexible as it does not require the computation of large covariance matrices and, once the ANN is trained, can produce realizations for subsurface characterization and monitoring in real time.
