Monitoring networks are designed based on a newly developed methodology linking contaminant transport simulations and optimization models. Advection and dispersion simulation generates plume realizations, while the use of the Latin Hypercube Sampling accounts for uncertainty in transport parameters and in contaminant source characteristics. The optimization model using a Genetic Algorithm adequately designs a given number of wells in order to maximize the detection probability and to cover the vacant area where plumes can pass through.


Contamination in geoenvironmetal system is most serious when leachates enter a hydrogeological environment that allows rapid transport of the contaminants through the groundwater system. Therefore, common goals of monitoring groundwater quality include detecting and mapping contaminants migrating from landfills, hazardous waste sites, or agricultural fields. Accurate and timely information on the spatial distribution of the contaminant is essential in the formulation of corrective action plan and environmental management strategies for aquifers. Whether or not the appropriate management of aquifer is accomplished is highly dependent on the groundwater monitoring configuration from which samples are collected.

In order to detect a contamination, a sufficient number of observation wells in the underlying aquifer is required to intercept potential pathways of contaminant migration. However, in practice, the number and location of observation wells are subjectively determined because of physical and economical constraints. In addition, difficulty in locating these wells is due to the uncertainty that is characteristic of mass transport phenomena including the hydrogeological uncertainty that governs groundwater flow and contaminant transport and the uncertainty about the exact location of the contaminant leak in an unknown source.

Monitoring network design is often formulated mathematically as an optimization problem in which an objective function is to be minimized or maximized over a given search space subject to a set of constraints (Loaiciga et al., 1992).

This content is only available via PDF.
You can access this article if you purchase or spend a download.