Application of stochastic methods to the simulaton of large-scale unsaturated flow and transport

Abstract

Numerical simulations are used to demonstrate features of large-scale flow and transport in heterogeneous unsaturated soils using effective expressions derived from a stochastic theory developed at MIT. The case of stratified soil is examined for one- and two-dimensional (2D) flow and 2D transport. Finite-difference methods are used in the flow analysis and a random-walk algorithm is used for transport. Input parameters are derived from statistics of soil samples collected at the site of a large-scale tracer experiment conducted by New Mexico State University (NMSU). This study examines spatial variability of saturated and unsaturated hydraulic conductivity regarding effects on the bulk character of moisture flow and solute transport (directional rates of movement and spreading of the plume). The numerical simulations demonstrate tension-dependent anisotropy, hysteresis and macrodispersion derived in the stochastic theory based on this spatial variability. These features cannot be explained using conventional deterministic models. Preliminary results suggest that the stochastic theory is better able to simulate the bulk character of the NMSU flow experiments compared to a deterministic model. Sensitivity analyses identify major factors of soil variability which control moisture flow and spreading. Results of this study indicate the need for careful experimental design and soils data collection. 88 refs.,more » 42 figs., 4 tabs.« less