HR: 0800h
AN: H41E-0835    [Abstracts]
TI: Subsurface Characterization using Sensitivity based Pilot Point Methods
AU: * Jung, Y
EM: yjung@ncsu.edu
AF: Department of Civil, Construction, and Environmental Engineering, North Carolina State University, Campus Box 7908, Raleigh, NC 27695, United States
AU: Mahinthakumar, K
EM: gmkumar@ncsu.edu
AF: Department of Civil, Construction, and Environmental Engineering, North Carolina State University, Campus Box 7908, Raleigh, NC 27695, United States
AU: Ranjithan, R
EM: ranji@ncsu.edu
AF: Department of Civil, Construction, and Environmental Engineering, North Carolina State University, Campus Box 7908, Raleigh, NC 27695, United States
AB: A large number of contaminated groundwater sites suffer from insufficient hydraulic property information leading to inefficient cleanup strategies and inaccurate prediction of contaminant fate and transport. Thus, significant researches have focused on developing efficient parameter estimation methods to obtain accurate subsurface hydraulic properties (i.e., hydraulic conductivity or permeability) from secondary measurements such as hydraulic head and/or tracer/contaminant concentrations. Due to the limited amount of information and their associated uncertainty, these problems are often ill posed and non-unique. To minimize these shortcomings, many methods have been developed including a class of methods called the pilot point methods (PPMs). The physical meaning of pilot points is that they are not direct measurement points; rather these are strategically selected points in the domain and are added to the parameter searching procedure to reduce ill-posedness. In the context of subsurface characterization, this method can provide improved estimates of hydraulic conductivity distribution by combining direct measurements of hydraulic conductivity with secondary measurements. In this study, an evolutionary algorithm based PPM using the D-Optimality sensitivity criterion is developed. This method is based on searching for a collection of pilot points that have the greatest influence on secondary information with slight perturbation of hydraulic conductivities. Thus, this criterion maximizes the dataworth of the secondary measurement (e.g., hydraulic head) while minimizing the impact to the original covariance structure of measured hydraulic conductivities. Four different synthetic scenarios were built and used to explore the effectiveness of this method in finding suitable pilot points for hydraulic conductivity estimation. Results show that pilot point selection using the D-optimality criterion can lead to improved hydraulic conductivity characterization than either a random pilot point selection method or a sequential sensitivity based PPM developed previously. Final hydraulic conductivity distributions obtained using our method are shown to be very close to the synthetically generated unknown reality.
DE: 1829 Groundwater hydrology
DE: 1832 Groundwater transport
DE: 1847 Modeling
DE: 1849 Numerical approximations and analysis
SC: Hydrology [H]
MN: 2007 Fall Meeting