You are here:
Cost Effective Long-Term Groundwater Monitoring DesignEPA Grant Number: U915546
Title: Cost Effective Long-Term Groundwater Monitoring Design
Investigators: Reed, Patrick M.
Institution: University of Illinois at Urbana-Champaign
EPA Project Officer: Jones, Brandon
Project Period: August 1, 1999 through August 1, 2002
Project Amount: $102,000
RFA: STAR Graduate Fellowships (1999) RFA Text | Recipients Lists
Research Category: Academic Fellowships , Engineering and Environmental Chemistry , Fellowship - Civil/Environmental Engineering
The objective of this research project is to apply stochastic data analysis (fate and transport simulation) and discrete optimization to quantify the cost-benefit tradeoffs inherent to long-term groundwater monitoring design. Explicit knowledge of these tradeoffs will aid stakeholders and regulators when negotiating the appropriate long-term management strategies for a site. This research will result in a reduction of the costs associated with long-term monitoring of sites with groundwater contamination.
This research will focus on developing a management model that has three primary components: (1) groundwater fate-and-transport simulation; (2) geostatistical estimation; and (3) optimization using a genetic algorithm (GA). The fate-and-transport model will be used to project the migration and mass of dissolved contaminants. The coordinates and predicted contaminant concentrations of every potential sampling location within the domain will be used to randomly generate a specified number of sampling plan designs. Each of these designs then is evaluated in terms of their cost, and a performance criterion is determined using geostatistics. Examples of performance criteria include global mass estimates attained using ordinary kriging, or uncertainty estimates attained using indicator kriging. The cost and performance criteria are used to compute a fitness (or measure of the quality) of each potential sampling design. The GA uses fitness values to determine which individual sampling designs are allowed to reproduce and evolve in later generations in a process analogous to Darwinian evolution. The management model undergoes several iterations of evolving new populations of potential sampling plans until the GA converges to several optimal or near-optimal sampling designs.