Decision-Tree Analysis to Assist in Developing a Remediation Strategy
Decision-tree analysis is a method that uses probabilistic analysis to help in the selection of remedial actions. The method, which includes an adapted software package, is used to analyze large quantities of data and provide a systematic evaluation of the effects and costs of various remedial alternatives.
This was used, for example, at a site where the concentration of contaminants in the soil, although too low to pose a significant risk to human health by soil exposure, was a potential source of groundwater contamination. A decision-tree analysis was employed to help with the decision whether: a) additional soils should be remediated to prevent continuing contamination of the groundwater; or b) the groundwater treatment system should address the existing groundwater contamination, plus the additional contamination released from the unremediated soils. In other words, one option would treat additional soil; another would treat groundwater for a longer period of time. Both options were assumed to achieve the same goal for groundwater, albeit through different methods, in different time periods, and with different costs.
Limitations and Concerns
Probabilistic analysis may help assist risk managers make better remediation decisions, as long as cleanup objectives remain constant. However, the trade-off of cost vs. cleanup is not a technical decision. It should not be addressed with this tool.
The effectiveness of remediation of soil and groundwater is not always certain, and there are a great number of variables and assumptions that affect the outcome. If decision-tree analysis is going to be used, the variables and assumptions must be clearly explained to the community.
Decision-tree analysis can be used to formulate remediation strategies, especially where there is a complex interaction between contaminants in the soil and the groundwater.
Technology Development Status
Decision-tree analysis is refined and well developed. There are many commercial applications of this analysis, and there is existing software to support it.
Other Resources and Demonstrations
See example of decision tree at http://www.clu-in.org/download/stats/decisiontree.pdf.
Also see http://www.clu-in.org/download/partner/2000specialreport.pdf for reference to decision support tools.