HR: 0800h
AN: H11D-1297 [Abstracts]
TI: An Evaluation of Polynomial Chaos Methods for Approximate Solution of Stochastic Groundwater
Flow
AU: * Rupert, C P
EM: crupert@email.unc.edu
AF: Department of Environmental Science and Engineering, UNC - Chapel Hill
Rosenau Hall CB# 7431, Chapel Hill, NC 27599-7431
United States
AU: Miller, C T
EM: casey_miller@unc.edu
AF: Department of Environmental Science and Engineering, UNC - Chapel Hill
Rosenau Hall CB# 7431, Chapel Hill, NC 27599-7431
United States
AB:
Our inability to fully characterize heterogeneous subsurface systems at appropriate scales motivates continuing interest in
the careful stochastic treatment of subsurface fluid flow and contaminant transport. Unfortunately, existing methods often
appear to require restriction to low variability regimes, and significant challenges remain as heterogeneity increases, In
the last decade, models based on Wiener polynomial chaos and Karhunen-Loeve expansion have been developed for a range of
scientific fields. A careful assessment of these techniques for application to subsurface systems is needed. We examine chaos
˙approximation of a transient, two-dimensional groundwater flow problem for a range of variances and˙correlation models with
a random log-normally distributed˙hydraulic conductivity field., with attention to convergence and computational efficiency.
The errors in moment ˙approximations are investigated, by comparison to Monte Carlo methods, and as a function of log
conductivity variance, of covariance model, of domain size, of the number of eigenfunctions, and of the generalized Hermite
polynomials used.˙ Each of these factors affects the computational performance of the chaos method.˙ Finally, we consider
challenges that should be addressed to extend the range of applicability of chaos methods.
DE: 1800 HYDROLOGY
DE: 1828 Groundwater hydraulics
DE: 1869 Stochastic hydrology
SC: Hydrology [H]
MN: Fall Meeting 2005