On Bayesian Decision Analysis for Evaluating Alternative Actions at Contaminated Sites
Today, contaminated land is a widespread infrastructural problem and it is widely recognised that returning all contaminated sites to background levels, or even to levels suitable for the most sensitive land use, is not technically or financially feasible. The large number of contaminated sites and the high costs of remediation, are strong incentives for applying cost-efficient investigation and remediation strategies that consider the inherent uncertainties.
This thesis presents an approach based on Bayesian decision analysis for handling uncertainties and evaluating alternative actions at contaminated sites. These actions include remediation, investigation and protection strategies for contaminated soil and groundwater. The expected utility decision criterion for individual decision-makers is used, where utilities are expressed in monetary terms. The main idea of the working approach is to focus on decision-making and risk valuation at a much earlier stage of the project than in contemporary practice.
The evaluated approach allows for: explicit economic valuation and comparison of alternatives, identifying factors that are important for the optimal decision, data worth analysis, including model uncertainty and alternative hypotheses, and, it requires the use of expert judgement. This thesis has resulted in a decision framework, which was developed from applying the approach in a number of case studies including remediation of a landfill, design of reclaimed asphalt storage, design of a road stretch situated on mine tailings, and remediation of a part of an industrial site. The framework is believed to provide a logical structure for the proposed working approach, provide a structure for documentation such that the work process becomes traceable, and thereby provide a basis for communication between project participants. The major tasks in applying the approach are delimiting the decision problem, finding a reasonable level of complexity in the analysis, and effectively communicating the results.
Bayesian decision analysis
data worth analysis