HR: 0800h
AN: H11C-1280    [Abstracts]
TI: Multiobjective Ambient Groundwater Quality Monitoring Network Design
AU: * Ammar, K A
EM: khalil@cc.usu.edu
AF: Graduate Student, 6 Aggie Village c, Logan, UT 84341 United States
AU: Khalil, A F
EM: akhalil@cc.usu.edu
AF: Postdoctoral research Scientest, Columbia University, Earth and Environmental Engineering, New York, NY 10027 United States
AU: Mckee, M
EM: mmckee@cc.usu.edu
AF: Professor, Utah Water Research Laboratory, Logan, UT 84322 United States
AB: Abstract Traditional monitoring network design is generally performed without a systematic understanding of the multiobjective nature of the issue. This paper presents a new methodology for designing an optimal multiobjective ambient groundwater quality monitoring network. The framework used in the analysis is based on a sparse Bayesian learning approach called relevance vector machines (RVMs). RVMs adopt a fully probabilistic framework that has an inherent capability to simultaneously address model complexity and all sources of uncertainty. The design of a monitoring network obtained through the application of RVMs is proven to be more efficient in terms of the number of monitoring wells than a network acquired by application of support vector machines (SVMs). Preliminary results and comparisons between the two methodologies suggest that the proposed RVM method could provide a reliable and parsimonious network configuration that is more pertinent to the physics of the case study. The methodology has been employed to identify redundancy in the monitoring of wells for nitrate (NO3- ) in the West Bank aquifers in Palestine. Tradeoff analyses between reducing the sampling cost and minimizing uncertainty was carried out to devise the optimal monitoring network design. The results show that this methodology is accurate and efficient. The proof of correctness and design options from cost and accuracy perspectives are also presented. Key Words: groundwater monitoring, multiobjective optimization, relevance vector machines.
DE: 1831 Groundwater quality
DE: 1848 Monitoring networks
DE: 1869 Stochastic hydrology
DE: 1873 Uncertainty assessment (3275)
DE: 3275 Uncertainty quantification (1873)
SC: Hydrology [H]
MN: Fall Meeting 2005