To give a preliminary indication of the soil infiltration characteristics to be expected, we conducted 20 small infiltration-ring (“bottomless bucket”) tests on soils of various ages at candidate experimental field sites in the Mojave National Preserve near Kelso, California. We obtained flow-rate and spatial-distribution data for use in site selection and formulation of the detailed design of full-scale tests. For some locations, we also obtained bulk samples of soils tested, and analyzed their particle-size distributions for correlation with hydraulic properties and observed surface characteristics. Our observations show a trend toward lower infiltration rates in older soils. This confirms and quantifies the expectation that the more distinctly stratified character of more highly developed soil impedes downward flow.
Field Experiments
To obtain measurements of hydraulic properties and quantitative observations of
soil-water flow behavior, we have completed two of three planned infiltration/redistribution experiments at field sites in the Mojave National Preserve.
Field test 1 (instrumented site pictured below, left) was conducted in an active wash with a wide (>12 m) channel, classified as map unit Qyw1. We applied 1450 liters of water in 140 min on October 29, 2004. Water was minimally ponded (depth from 0 to about 5 cm) during the infiltration. About 90% of soil surface within the ring had ponding depth greater than 0. We monitored nearby soil moisture conditions until mid January 2005.
Field test 2 (infiltration ring pictured in close-up below, right) was conducted in a soil of intermediate age, classified as map unit Qya3. We applied 810 liters of water in 140 min on March 3, 2005. Ponded water was maintained at 4.5-6.5 cm depth (downslope edge) 2.5-4.5 cm depth (upslope edge) during the infiltration. Monitoring of nearby soil moisture conditions continues beyond March 31.
Instrumentation for both tests included an electrical resistance tomography (ERT) array. The array of electrodes was symmetrical, with two lines centered on the infiltration ring in the pattern of a cross at right angles. Each line of the cross is 11.5 m long and has 24 electrodes equally spaced. Our collaborator Kamini Singha of Stanford University designed this aspect of the experiment, arranged for loan of the specialized apparatus from Stanford, and calculated initial results.
Other subsurface instruments included water content measurement by two-prong time-domain reflectometry (TDR) probes and flat-plate “ECH2O” probes, both based on the relationship between soil dielectric constant and water content. (Use of trade names does not imply endorsement by the USGS.) Measurements of soil water matric pressure (needed for calculation of driving force in hydraulic conductivity determinations) were made using tensiometers and heat-dissipation probes (HDP). Field test 1 also included thermocouples for temperature measurement, used to indicate the arrival time of newly infiltrated water at the points of installation. Pictured here is the instrument schematic (map view, below left) for infiltration test 2, with black dots indicating ERT electrodes and green dots indicating the other subsurface instruments, whose vertical positions are shown separately (cross-sectional view, below right).
In general, the observed soil-water behavior shows less contrast than expected between the Qyw1 and Qya3 sites; the more developed soil crust and horizons of the Qya3 site were only a modest impediment to flow. In both tests, the infiltration rate was nearly constant, decreasing slightly with time during field test 1 and increasing very slightly during field test 2 (see graph, below left). Preliminary estimates of infiltration capacity are 2.3 × 10-2 cm/s (83 cm/hr) for test 1 and 1.2 × 10-2 cm/s (42 cm/hr) for test 2.
The ERT results give a cross-sectional view of the relative water content at a sequence of times from the start of infiltration. In the picture below (right) are ERT results from field test 1, line 1 (across the channel; orientation here is left-to-right if looking upstream). In the site with the full ERT array pictured above, this is the line of electrodes going from upper left to extreme right. The results shown have not yet been calibrated for quantitative water content, but in general the wetter the soil, the further its color toward the blue end of the spectrum. These results show a relative dominance of vertical with respect to horizontal flow; there appears to be little impediment to vertical flow in the uppermost meter of soil, and very little water moves horizontally farther than about 1 m from the outer edge of the ring. In previous studies in Mojave Desert sediments, greater degrees of horizontal spreading were inferred for layers sharply contrasting in texture (Nimmo et al., 2002).
Pictures below show selected water-content, matric-pressure, and temperature results from various subsurface probes. In addition to the amplitude of changes in the moisture state, the timing of probe response is of interest. The delay between the start of infiltration and the initial or peak response (as for TDR results in the table) indicates the average speed of subsurface transport to the depth and radial position of the probe.
ECH2O probe data (below, left), from instruments in place following field test 1, shows responses to subsequent storms. Heat dissipation probe data from field test 2 are below, at right. These are not yet calibrated, but are generally indicative of matric pressure (greater “Delta T” implies the more strongly negative matric pressure of drier soil).
Thermocouple data (below, left) from field test 1 show wetting front arrival.
The main goal is to link hydraulic properties—unsaturated and saturated K, and soil water retention—to particle-size distribution (PSD) and other geomorphologic information. Many existing models can provide a link of this sort, but a combination or adaptation is necessary for several reasons: (1) We need a model applicable to Mojave Desert soils, for which existing models may be inappropriate because they were designed primarily for agricultural fields in non-desert regions. (2) Most models apply over a relatively wet range of soil water content, neglecting the drier range that is often of critical importance to desert plants. (3) We need a model tailored for optimal use of the particular types of geomorphologic information being obtained, mapped, and modeled by collaborators in the SDOILWR and RVDE investigations.
Until hydraulic property results from our field tests have been computed, we are testing and developing models using data from other sites, for example Oro Grande Wash (Winfield, 2000). We are testing (1) the widely used Arya-Paris (1981) model and other models based on capillary concepts and (2) more innovative models based on adsorption concepts, which may be more appropriate for dry-range data. The graph below (left) shows model predictions by several variations of the Arya-Paris model. Several of these appear to work reasonably for the wetter (right-hand-side) portions of the retention curve, but none of them work well for the drier portions. The graph below (right) shows predictions by one of the best-fitting of the capillary-based models (solid curve) and by an adsorption-based model using estimates of the soil’s specific surface area (dashed curve). The adsorption-based model works much better in the dry range.
Our current direction is to develop a model suitable for representing Mojave soils in general by combining a basic form of capillary model with a basic form of adsorption model. This combined model will be applied and tested using measured data from the field experiments now in progress.
Basic outline of approach: Develop a set of 1-D variably saturated soil water flow models over a wide area of the Mojave Desert. Applied simultaneously, each of these models in effect constitutes a grid-point of a large-scale three-dimensional soil moisture model. Divide the domain into polygons of geomorphologic similarity. Each grid point within one of these polygons has hydraulic properties of the geomorphologic class of the polygon, adjusted for known systematic trends, for example a correlation of characteristic particle size with position on a slope. Incorporate soil stratification as layers of distinct properties corresponding to the textural and other information, to the extent needed and justified on the basis of field data. Assume uniform effective hydraulic properties within each thin but broad layer. Apply equal-flux boundary condition between layers. Incorporate evapotranspiration as a term expressing the rate of water loss from the root zone, quantified from empirical data.
| National Research Program | USGS | Water Resources | UZ Flow Homepage |