The playa wetland was located near Shallowater, TX (33.656 Lat., −102.0358 Long.). This site was chosen based upon proximity, landholder permission, and because previous studies had been conducted at this location (e.g., [5][9]-[11]). This particular playa wetland was embedded within and proximal to an irrigated cotton cropping system surrounding the east and south and southwest portion of the playa (Figure 1). The west and north sides of the playa were surrounded by terraced Conservation Reserve Program (CRP) grassland. The playa basin is predominately a Randall Clay (very-fine, smectitic, thermic Ustic Epiaquerts), though soil composition was likely changing as a result of transport by surface runoff from land used for row cropping [12]. The soil proximal to the annulus is a Lofton clay loam (fine mixed Vertic Argusol). Farther upland the soil is an Amarillo fine sandy loam (fine-loamy, mixed, superactive, thermic Aridic Paleustalfs). Instrumentation was installed at the deepest point within the playa basin. The deepest part of the basin was approximated by referencing U.S.G.S. topographic data and precisely located with a high-resolution global positioning system (GPS) consisting of a Trimble R-6 GNSS base and a rover with a Pacific Crest HPD-450 radio link Trimble TSC2 survey controller (Sunnyvale, CA), mounted on an all-terrain vehicle to collect real-time kinematic data of continuous topographic points at fixed 4.5 m intervals. GPS data were analyzed using Golden Software Surfer 11 (Golden, CO) to locate the lowest region of the playa. Four meso-scale double ring infiltrometers were installed in a cluster to minimize potential effects of soil and vegetation variability but separated by what was subjectively judged to be a distance that would minimize if not to eliminate the effects of one installation upon another. The individual infiltrometers were separated by about 50 m.

Figure 1.Playa embedded within 1.6 km (1 mile) square watershed bounded by roadways. Irrigated crop land (dark tan), dryland crop land (light tan), C.R.P. grassland (green) and clay playa basin bottom (dark gray) identified from aerial photography. Elevation lines (red) and basin edge (blue) are from U.S.G.S. topographic data.

Rings (Figure 2) were fabricated from rolls of 0.04 cm (30 ga.) thickness × 91 cm (36 inches) width galvanized sheet steel. The metal was cut to length, the ends overlapped, and the overlapped portion fastened with screws and sealed with a silicone sealant. The resulting inner and outer ring diameters were 120 cm and 200 cm, respectively. The infiltrometer was set to a depth of 40 cm. In an effort not to affect the soil hydraulic properties care was taken to leave the soil undisturbed while setting the rings. Precautions included using plywood forms to stand upon and work from. The forms served to more precisely guide manual trenching operations. Infiltrometer rings were placed in the resulting trenches, and the plywood forms carefully removed. Small spaces remaining on the inner sides of the trenches were filled with a slurry of sifted soil to prevent water from escaping under the rings. The remaining trench space on the outer surface was backfilled, but additional soil was mounded a few centimeters high to allow for settling. After installation of the rings in January of 2012, the soil was left undisturbed for five months until early summer.

Water was hauled from an irrigation well and fed into large holding tanks at the experimental site, and hose-fed into the infiltrometer rings from a portable generator-driven electric pump. Water was allowed to circulate back into the holding tanks through valves adjusted such that system pressure was maintained at 138 kPa (20 psi). The water depth within each ring was controlled with rugged inexpensive float valves of the type commonly used to maintain livestock water trough levels. The float valves were manually adjusted over the course of several hours during filling, and periodically throughout experimentation, so that no difference could be measured between the top edge of the inner ring and the surfaces of the water. Infiltrometer flow was forced through either a paddle wheel flow meter or a more sensitive low-flow turbine type flow meter. During initial filling the higher flow required to maintain a 20 cm head was passed through the paddlewheel meter. As flow decreased, water to the inner ring was passed through a filter and then through the high-resolution turbine flow meter. Flow to the outer ring was similarly measured. Flow meter outputs were summed over 15-minute intervals and recorded for later analysis.

Figure 2.Cross-sectional diagram of meso-scale infiltrometer placed in playa bottom. Flow inlets to float valves are shown. Dimensions and head maintained throught experiment are indicated. Infiltration (I_t), Evaporation (H_evap) and total flow (H_ttl) are indicated. Soil infiltration is indicated by large blue arrows.

There were originally five functional infiltrometers installed. During the filling of one of the devices, the contents unexpectedly drained into what appeared to be an animal burrow. Here we report results taken from the remaining devices. During installation of another infiltrometer, a textile tube drain was encountered. These were installed as part of an engineering scheme to harvest playa water and increase aquifer recharge [10]. The drain tube was simply severed where the infiltrometer walls were to be installed leaving the remainder in place. Drain tubes external to the outer wall were torn from the ground and discarded.

Flow rates (cm³·min⁻¹) were converted to the apparent drop in head (H_ttl, cm·min⁻¹) from the inner ring (Figure 2). It should be borne in mind that the “drop in head” is a rate calculated from flow; and that actual head, the distance from the soil to the water surface, was kept constant. The total head drop values (H_ttl) were corrected for evaporation (H_evap) using values from a “near Class-A” pan located at the USDA-Plant Stress and Water Conservation Laboratory at a distance of 14.5 km ESE (33.593849 Lat., −101.900978 Long.) from the study site [13]. Head loss from infiltration at any time was calculated as: I_t = H_ttl − H_evap. Data were plotted, inspected, and then fitted according to three models to evaluate infiltration: the “Green-Ampt” model [14], “Horton’s” infiltration model [15], and a simple hyperbolic decay model of the form I = I_ss + (a × b)/(b + t); where, I_ss = steady state apparent head drop, t = time (minutes), and both a and b are site specific coefficients introduced by iterative curve fitting. Infiltration estimates were made by iterative curve fitting using SigmaPlot v. 11 (Copyright^© 2011, Systat Inc.). Infiltration through a single ring infiltrometer was numerically simulated by combining flows to both inner and outer rings and treating data as described above for inner rings of the double ring infiltrometers. For presentation, long term infiltration values, averages, and standard error were presented in tabular form.

Six infiltrometers were originally installed. One was sacrificed to investigate infiltration patterns beneath and around the device (not presented). Another unexpectedly failed when the contents suddenly drained into what appeared to be the remains of an animal burrow. The results of the four remaining infiltrometers are shown.

Long term infiltration estimates are shown in Table 1 and Table 2 (below). Table 1 shows estimates made by each numerical modelling routine and the standard error of the estimates, for Horton’s and the hyperbolic decay models. The Green-Ampt model was calculated using a spreadsheet and errors about the mean were not calculated.

Because the three models yielded slightly different results, values for each device across models are presented as the averaged results of all models. That is, the long-term estimates made using; Horton’s, Green-Ampt, and the simple hyperbolic regression models averaged devices A through D in Table 2. The numerically simulated “single ring”, or cylinder infiltrometer are presented likewise.

Table 1.Long term infiltration estimated by Horton’s, Green-Ampt, and Hyperbolic with double ring infiltrometers methods. Standard error (S.E.) of estimate is shown in parentheses. Estimate error was not calculated for the individual Green-Ampt estimates.

Table 2.Long term infiltration estimated with double and single ring infiltrometers. Results are averaged across modelling techniques.