Stochastic modeling of flow and solute transport processes in unsaturated zone
Abstract
A systematic approach is presented in this study to model stochastically unsaturated flow and solute transport. The approach combines a one-dimensional Galerkin-type finite element analysis with a special Monte Carlo technique. Flow and solute transport processes in the unsaturated zone were modeled by coupling both the unsaturated flow model and the solute transport model, and then solving the system using Fortran program, HYDRUS, developed by U.S. Salinity Laboratory. System uncertainty was analyzed by introducing a special Monte Carlo algorithm that accounts for marginal distributions and correlations of flow and solute transport parameters. Unnecessary restrictions commonly encountered in stochastic analysis; such as the independence of all random variables, or imposing multivariate normal or lognormal distributions, were removed by using this algorithm. A user-friendly computer program, RELAN, was developed to implement the introduced Monte Carlo technique by using Visual Basic. Three simulation experiments, including steady flow transport, transient flow involving hysteresis, and a simplified spatial variability problem on stratified soil, were conducted to illustrate the use of the proposed approach. Results showed that (1) the correlation between dispersion coefficient, D, and pore water velocity, v, has a moderate effect on the adjusted arrival time of the peak concentration, (t$\sb{\rm max}$-t$\sb0$). The coefficient of variation of (t$\sb{\rm max}$-t$\sb0$) increases from 73% to 91% for correlation coefficient, $\rho\sb{\rm D,v},$ varying from 0 to 1; (2) hysteresis has only a minimal effect on the total amount of drainage water collected at the bottom drainage layer in the case of unsaturated transient flow, regardless of the soil boundary conditions and the soil type. It appears that, if a simulation involves a "total" or "average" parameter related to the unsaturated flow phenomenon, hysteretic effects may be neglected without causing significant error; (3) autocorrelation plays an important role in calculating the equivalent vertical hydraulic conductivity. The difference in the equivalent vertical hydraulic conductivity due to autocorrelations can be as large as an order of magnitude.
Degree
Ph.D.
Advisors
Harr, Purdue University.
Subject Area
Civil engineering|Mechanical engineering|Geotechnology
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.