Laboratory measurements of infiltration capacity by a double ringed infiltrometer and the Cornell Sprinkle Infiltrometer

A soil's infiltration capacity (IC) represents the maximum rate at which water will infiltrate through the surface of the soil under given conditions. Accurate estimation of IC is important for numerous situations, including determining appropriate irrigation rates: if the irrigation rate exceeds IC, the excess water is lost to surface runoff (Ali 2010; Fullen & Catt 2014). IC is also important in hydrologic modeling of runoff (Brath & Montanari 2000), and for characterizing the suitability or effectiveness of installations for reducing urban stormwater runoff by increasing infiltration. ‘Green infrastructure’ like rain gardens and bioretention basins are promoted by many agencies (e.g., NYSDEC 2014; USEPA 2014) and environmental groups (CWC 2014; Sierra Club 2014). Many design manuals require or advise a field measurement of IC to determine whether the soil is capable of sustaining an infiltration installation (IDNR 2008; NVSWCD 2014).

Double-ring infiltrometers (DRI), with concentric inner and outer rings, are the most commonly used method of measuring IC (Johnson 1963; American Society Testing Materials 2003). Both rings are filled to a known level with water and either (1) the decrease in water level in the inner ring is noted over time (falling head method) or (2) the water level is maintained in the inner ring by adding measured volumes over measured time periods either manually or using Mariotte Tubes (constant head method). IC is measured as change in water depth per unit time in the falling head method, and as the volume added divided by its area per unit time in the constant head method. Both methods give IC in units of depth per unit time, such as mm/h.

Although the CSI has been used for more than 15 years (Ogden et al. 1997), no study was found that compared IC measurements using it and a DRI. Accordingly, the objective of this study was to compare the CSI and DRI on sands in a laboratory, specifically assessing (1) how well measurements of IC by the different infiltrometers agreed with each other, and (2) whether measurements of IC were more reproducible by one type of infiltrometer than the other.

Three different sands were used in the study. ‘Play sand’ (Sand 1) and ‘Builders' sand’ (Sand 2) were purchased from a home supply store; Sand 3 was obtained from the soil mechanics laboratory of the Department of Civil and Environmental Engineering of Manhattan College, Bronx, NY, where all the work was performed. Sand was chosen because its high infiltration capacity relative to loams, silts and clays allowed each IC measurement to be performed quickly, which enabled multiple tests to be done on each sand.

To characterize the 3 sands, the particle size distributions were determined by sieving and weighing according to the procedures described by Bowles (1992).

Three plastic tubs (75 × 45 × 12 cm) were prepared to hold each sand separately. Holes were drilled in bottom of each tub to allow free drainage and the bottom covered with a screen to retain the sand. One sand type was added to each tub and mixed by hand to ensure homogeneity.

IC was measured 6 to 8 times for each sand with the two devices: a DRI (15 cm inner ring, 30 cm outer) manufactured by Turf-Tec (model IN7-W) and the CSI. The sand was wet prior to each test to try to ensure that the measurement was made at saturation and under similar conditions for each test, which should aid comparability.

The falling-head method was used with the DRI. The inner and outer rings were filled to 5 cm above the sand surface, and the time until the water level in the inner ring dropped to the soil surface was recorded, giving the infiltration capacity in cm/h. Eight repetitions were conducted with the DRI with each sand type. Each test lasted between 3 and 13 minutes.

The CSI and its operation are described briefly here; more detail can be found in the instrument's manual (Van Es & Schindelbeck 2003). The device (Figure 1) consists of a 46 cm high by 26 cm diameter Lucite cylinder sitting on a steel ring of the same diameter which sits, in turn, on the soil. The bottom of the Lucite cylinder is sealed with a Lucite plate which supports 129 coiled, narrow tubes of 0.76 mm internal diameter. The tubes penetrate the plate to allow water to sprinkle from cylinder onto the soil. The top of the cylinder is also sealed with a plate. The rate of water application is regulated by a Mariotte-type bubbling tube extending through the top plate. The lower steel ring has a 25 mm diameter hole in its side. The ring is pushed or hammered into the soil so that the bottom edge of the hole is flush with the soil surface.

Any applied water that does not infiltrate (the equivalent of excess rainfall) will run off the soil through this hole. A stopper with a tube running through it is inserted into the hole to direct the runoff into a graduated beaker sitting in a small hole dug for it. The distance that the water level drops in the cylinder is measured against time, to determine the application rate. The volume of runoff captured in the beaker is measured at the same time. The infiltration rate (which is equal to the IC) is determined in cm/h by subtracting the volume of runoff captured from the volume of water applied, and dividing the result by the area of the cylinder and the duration of the test. Six replicate tests were conducted on each sand type. Each test lasted between 9 and 13 minutes.

Table 1 presents the particle size distributions for the three sands and Table 2 presents the results of the infiltration experiments for both infiltrometers. While the IC values are very large compared with what is reported for soils in the field – e.g. <12 cm/hr (Johnson 1963), they are in the range expected for sands based on saturated hydraulic conductivity reported elsewhere in the literature (Freeze & Cherry 1979).

Regarding agreement between IC measurements using the DRI and CSI (measured as the percentage difference in mean IC for a particular sand), the results varied substantially for the three sands. Agreement was close for sand 1 (4% difference between means of ∼54 cm/h), but not close for sands 2 and 3 (48% and 32%, respectively) with the CSI giving larger values than the DRI in both cases. The P-values of a two-tailed t-test (assuming unequal variances) for each sand are reported in Table 2, which shows statistically significant differences in the means for sand 2 and sand 3 at the 0.01 and 0.05 significance (99% and 95% confidence) levels, respectively.

The reproducibility of IC measurements was assessed using the relative standard deviation (RSD, i.e., the standard deviation divided by the mean) for each sand. As shown in Table 2, the RSD for the CSI was much less than that for the DRI for sand 2 (7 vs 26%), but the RSDs were much the same for each of sands 1 and 3 (23% vs 18%, and 22% vs 19%, respectively). The relatively high RSDs (>= 17%) found under laboratory conditions in 5 out of 6 cases, indicate significant variability in IC measurements introduced by instrument itself.

Secondary results include the observation that the IC did not tend to change – neither increasing nor decreasing – as additional trials were conducted for either instrument. Finally, although the data are not shown here, the rate of water application using the CSI was highly consistent, with an RSD of only 4% relative to the mean application rate of 1.88 cm/min.

The results were inconclusive regarding the level of agreement of IC measurements between the two infiltrometers. The percentage difference in the mean IC was small for sand 1 (4%), but substantially larger for sands 2 and 3 (32% and 48%, respectively), with the CSI giving higher values than the DRI in both cases. Likewise, the results were inconclusive regarding reproducibility. The CSI showed much better reproducibility than the DRI for sand 1 (RSDs of 7% and 26%, respectively), but slightly worse for sands 2 and 3 (23% and 18%, and 22% and 19%, respectively).

The results were obtained in a laboratory using commercially available sands. Accordingly, the results should not be considered representative of other soil textures – e.g. loams, silts or clays – nor of soils in the field. However, they do provide a useful, first comparison of the two infiltrometers.

Moreover, despite their inconclusive nature, these results have important, practical implications for practitioners who need to measure IC in the field and regulators of such activities. We tested the infiltrometers under idealized conditions (homogenized sand without any vegetation or soil biota) that should foster good agreement between the two instruments. The fact that, under these idealized conditions, the two infiltrometers gave very different values of IC for two of three sands tested means that the type of instrument used in field conditions will likely affect the results obtained. Furthermore, in these idealized conditions, both instruments had 18 to 26% relative standard deviations in 5 of 6 cases. This indicates that practitioners and regulators should also be aware of the variability in IC measurements that is associated with instrument itself – as well as the considerable spatially variability of IC in soils.

Future work could compare results from the two instruments in lab with different soil textures or in the field on real soils.

The authors wish to thank Professor Kerryanne Donohue, Ms. Maame Boakye and Mr. Taymar Walters for their assistance in performing the experiments, and Drs. Scott Lowe and Joshua Galster for their helpful reviews. This material is based upon work supported by the National Science Foundation under Grant No. 1203210.