Design of optimal groundwater-quality monitoring networks via computer assisted analysis

摘要

Long-term groundwater quality monitoring can be achieved at lowest cost when the sampling strategy is determined using an optimal design approach. The design variables are the locations and times for water-quality sample selection. The analysis tools consist of a stochastic model of groundwater flow and transport, a Latin-hypercube sampling strategy for the creation of random-field hydraulic conductivity realizations, a Kalman filter, and an optimization algorithm. The goal of the monitoring network is to determine with a specified level of confidence the concentration history at selected locations which, in general, are not the measurement locations, It is through the use of a mathematical model surrogate for the groundwater system that new information is provided to augment the collected data such that the number, and therefore the cost of, of sampling is reduced.