A bicriteria knapsack program for planning remediation of contaminated lightstation sites

Abstract

The Coastguard manages over 40 lighthouse sites on the West Coast of Canada. All of these have some ground contamination from lighthouse activities or other earlier uses of the location, such as military fortification. Now the federal government is requiring its departments to identify all pollution on land that it owns or leases, and develop an environmental management plan for contaminated sites. The author has worked with the Canadian Coast Guard, Pacific Region (CCG-PR) to develop a set of environmental criteria for prioritizing management and remediation at the lighthouse sites. The next step is to apply these criteria to determine the best action to take once preliminary environmental assessments of the sites have been completed. Two criteria dominate the immediate next actions at a site. One is the Canadian Council of Ministers of the Environment (CCME) Score, which is a measure of estimated environmental risk, and would be reduced by some immediate remediation. The other is Uncertainty, which estimates the incompleteness of the preliminary assessment and would be reduced by further testing. Working with these two criteria and a limited budget, deciding the best next step can be formulated as a bicriterion 0–1 knapsack problem. A comprehensive solution of the problem would require a parametric analysis over all possible relative weights of the two criteria, and also a parametric analysis for the budget. An exact solution would require much computation, and such a solution process could not easily be handed over to CCG-PR personnel. However, if the integer requirement in the formulation is relaxed, the problem becomes simply a matter of ranking projects by their best benefit/cost ratio, and selecting projects down the list until all the budget is allocated. This is the solution approach used here, and which is being passed to CCG-PR headquarters for their continued use. The paper illustrates the method with some disguised data from the actual CCG-PR studies.