This paper describes a series of soil water tracer experiments and approaches taken to numerically model the flow and transport behavior observed in the field experiments. These experimental and numerical results strongly suggest that current widely held views and commonly applied modeling approaches are flawed in many cases for unsaturated flow, and provide strong supporting evidence for a variable, state‐dependent anisotropy in the hydraulic conductivity of an unsaturated medium. This phenomenon has been previously postulated in a number of independent theoretical and experimental investigations. In general, the previous studies identify layered heterogeneity as a primary cause of the macroscopic anisotropy. In addition, we show how hysteresis in the soil moisture characteristics (θ‐ψ relationship) can cause a lexturally homogeneous porous media profile to behave anisotropically under transient unsaturaied conditions. Recognizing that both of these factors (layered heterogeneity and capillary hysteresis) contribute the anisotropic behavior observed in the tracer experiments, we attempt to quantify the relative magnitude of their contributions in a numerical modeling investigation. For the numerical modeling study we use a finite element flow and transport code, and introduce a simple procedure for incorporating variable anisotropy into a predictive numerical model. To determine the relative magnitude of textural heterogeneity and capillary hysteresis as causes of the observed macroscopic anisotropy, we employ a diagnostic modeling approach. The results of the diagnostic modeling study indicate that textural heterogeneity is by far the most important contributor to the variable macroscopic anisotropy observed at the field site. The diagnostic simulations further show that the variable anisotropy approach is well suited to modeling field‐scale problems. Subsequently, a sensitivity analysis was performed to determine how climate, geologic and topographic structure, and media lithology affect flow and transport behavior when soils were specified to have a variable macroscopic anisotropy. The results of this study clearly indicate that variable state‐dependent anisotropy is a real and significant process at the field site and that modeling with consideration of variable anisotropy strongly affects model predictions.

An untilled medium‐grained fluvial sand near Socorro, N. Mex., was sampled in horizontal and vertical transects to study the statistical and stochastic properties of particle‐size parameters and saturated hydraulic conductivity. Hydraulic conductivity is log‐normally distributed, whereas the 10% finer, median, and geometric mean particle sizes are normally distributed. The strongest correlation between hydraulic conductivity and particle‐size distribution parameters is that of the log of hydraulic conductivity with the 10% finer particle size. Stochastic analyses using the autocorrelation function and spectrum indicate that the log of hydraulic conductivity and particle size are characterized by dissimilar spatial correlation structures in the vertical directions. In general, particle size is more structured, regular, and predictable and shows a close similarity with the stratigraphy as observed in the field. On the other hand, hydraulic conductivity in the vertical direction may best be modeled as a simple random variable. Variogram and kriging analyses indicate that both hydraulic conductivity and particle size are relatively isotropic in the horizontal plane and that marked similarities in spatial structure exist in this plane. The spatial distribution of saturated hydraulic conductivity in the horizontal plane is estimated reasonably well using the empirical relationship between particle size and conductivity along with the kriged estimates of the 10% finer particle size.

No abstract

Part of a small drainage basin on the Sevilleta National Wildlife Refuge (about 25 km north of Socorro, NM) was intensively instrumented with soil monitoring equipment to estimate natural ground-water recharge. Soil-moisture data were analysed with special attention to characterizing the influence of topography on the direction of vadose water flow paths in fine to medium aeolian sand. Moisture content data were obtained by the neutron scattering technique, and hydraulic head data were obtained using tensiometers. In addition, tracer experiments were conducted on a sandy hillslope to delineate the flow paths of vadose water. The results indicate that there is a strong lateral component to unsaturated flow on a hillslope, even in the absence of apparent sublayers of much lower permeability. Darcian calculations estimate the long-term, steady, deep flux beneath a concave location to be about 4 per cent of an assumed mean annual precipitation of 20cm. The deep soil water flux downward varied by several orders of magnitude during the 17 month period of record.

A laboratory experiment was performed to investigate the movement of water infiltrating from a point source into a porous medium comprised of alternating layers of fine and coarse sand packed in a clear plastic tank. The purpose of the experiment is to demonstrate whether the effective anisotropy of stratified materials is dependent upon saturation. The noncolinearity of the specific discharge and hydraulic head gradient directions is used as a measure of anisotropy. Dyes were used to map the specific discharge direction and tensiometers were used to characterize the hydraulic gradient field under unsaturated, steady state conditions. The results indicate a divergence between the direction of the specific discharge and gradient which increases as the pressure head decreases. Anisotropy determined from the experiment exceeds that calculated from the hydraulic conductivities of the individual materials.