Expert system methodology for evaluating reductive dechlorination at TCE sites

An expert knowledge site-screening methodology has been developed to evaluate naturally occurring reductive dechlorination as a remedial option for sites with TCE-contaminated groundwater. This methodology combines a causative model for the reductive dechlorination of TCE and expert knowledge within a Bayesian Belief Network. The knowledge base for this expert system was obtained from 22 experts via an expert elicitation protocol. The resulting expert system can be used to aid environmental decision making by evaluating the adequacy of reductive dechlorination at TCE-contaminated sites. Comparisons between this expert system and a commonly used screening tool show that this expert system produces predictive models that may better discriminate between locations that were sampled. The 22 elicitations revealed different beliefs and assumptions among experts about the biochemical processes involved in reductive dechlorination. The decisionmaking value of some types of evidence is a matter of dispute; however, findings about biodegradation daughter and/or end products have high decision-making value for all of the experts. The methodology demonstrated herein can provide insights for other environmental decision-making challenges.