Improved measurement of conductivity on swelling clay soils using a modified disc permeameter method.
Disc permeameters (Perroux and White 1988) have been used extensively to assess the impact of land management on soil hydraulic conductivity (K) and macroporosity in the field (Cook 1994; Connolly et al. 1997; Turpin et al. 1999). They have also been used with mixed success to provide estimates of saturated and near-saturated hydraulic properties for modelling soil water movement, and in developing pedotransfer functions (McKenzie and Jacquier 1997; Cormolly et al. 2002; Vervoort et al. 2003).
Saturated ([K.sub.SAT]) and unsaturated ([K.sub.[psi]]) conductivity are essential parameters in soil water modelling as they describe the soil's intrinsic ability to transmit water. K can be quantified at tensions close to saturation with disc permeameters. The permeameter is placed on a prepared soil surface, and water, usually under negative tension, moves unconfined through the soil in 3 dimensions. Flow rates are analysed to determine [K.sub.[psi]] using methods that assume the soil is homogeneous and the wetting front is even (3-D bulb shaped) (Reynolds and Elrick 1991; McKenzie et al. 2002). [K.sub.SAT] is commonly extrapolated from the [K.sub.[psi]] data. The method was originally developed for use on rigid, non-swelling, homogeneous soils (White et al. 1992), but has also been used to characterise K on swelling clays (Lin and McInnes 1995). Although this method is only suitable on swelling soils if the soil is fully wet (McKenzie and Cresswell 2002), it has been used extensively on dry soils (Connolly 2000; Vervoort et al. 2003). As might be expected, flow rates are significantly higher on dry, cracked soils.
Two major difficulties arise on swelling soils which can inflate conductivities by several orders of magnitude compared with values measured by other techniques (such as Talsma and van der Lelij 1976; Forrest et al. 1985; Bird et al. 1996). Firstly, water can move rapidly in preferential flow paths created by cracks, bypassing most of the clay matrix (Lin et al. 1997). Such water flow violates the assumption that the wetting front is even. Secondly, difficulties arise on initially dry soil due to swelling during the measurement period. This creates moisture (and porosity) heterogeneity at the wetting front and violates a key assumption in using this method. The resulting measurements of K are therefore transient and moisture/time dependent. This is exacerbated when short measurement times are used and infiltration does not reach steady-state. In theses situations the use of 3-D steady-state analysis is inappropriate.
The strong dependence of K on initial soil moisture (Lin et al. 1998; Bell et al. 2005), coupled with the frequent calculation of either inflated or negative conductivities, has led to skepticism about the usefulness and reliability of disc permeameters as a tool for measuring hydraulic properties of swelling clay soils. For example, Connolly et al. (2002) found that for Vertosols, Chromosols, and Sodosols, K was over-estimated when permeameter [K.sub.[psi]] values were applied to the APSIM-SWIM model (Huth et al. 1996; Verburg et al. 1996). This resulted in an under-prediction of runoff. [K.sub.[psi]] was generally 2 orders of magnitude higher than data collected using the single-ponded ring technique. This was attributed to non-ideal 3-D flow under permeameters that was not adequately accounted for in the analysis, and the use of short measurement times.
As disc permeameters allow for large numbers of in situ [K.sub.[psi]] measurements to be made rapidly and at low cost (McKenzie et al. 2001), it would be useful to find practical solutions to these concerns. In this paper, we identify mechanisms leading to the common failure of disc permeameters on swelling clay soils, and report the benefits of simple modifications to the method to enable more realistic measurements of K.
Three experiments were conducted on a range of clay soils displaying mild to strong shrink-swell properties (Table 1). In the first experiment, a commonly used method of determining [K.sub.[psi]] with permeameters using 3-D measurement and analysis (Reynolds and Elrick 1991; Connolly 2000; McKenzie et al. 2002) was used to assess the dependence of flow rates on soil moisture content, for a swelling clay. Flow rates through either pre-wet soil or soil at in situ moisture content were compared with [K.sub.SAT] measured by the single-ponded ring method (Reynolds and Elrick 1990). In the second experiment, some permeameters were confined within rings driven into the soil to restrict flow to 1D and prevent lateral anisotropy. These flow rates were compared with those obtained from unconfined permeameters and from ponded rings, on both field-dry and pre-wet soil. In the third experiment, the confining method was used at 2 sites to obtain [K.sub.[psi]] for a range of cropping treatments. Details of the experimental methods are given below.
Experiment 1: Effect of soil moisture content and measurement duration on K
A Black Vertosol (Isbell 1996) (Craigmore soil type) at Kingsthorpe, 20km west of Toowoomba, Queensland, was used. This soil is a strongly self-mulching, medium to heavy cracking clay, with >30 years cultivation history and a well-established subsurface restrictive plough pan layer (Powell et al. 1988). The following treatments were applied in shallow trenches randomly located across the site: (i) soil at in situ field moisture content/short measurement times; (ii) soil at in situ field moisture content/long measurement times; (iii) soil pre-wet/short measurement times; (iv) soil pre-wet/long measurement times.
Three trenches 2.5 m by 0.5 m by 0.1 m deep (bottom of the cultivated layer) were prepared for each treatment. Trenches used for treatments (iii) and (iv) were pre-wet with rainwater (15 mm applied at 11 mm/h using a grid of 2 L/h flow irrigation drippers placed over hessian), and then drained for 30 min prior to measurement. Soil was reasonably moist after pre-wetting (0.43 g/g at 0-20mm, 0.42g/g at 20--40 mm, and 0.39 g/g at 40-60 mm from the floor of trenches, compared with 0.35 g/g prior to pre-wetting and 0.45 g/g at saturation).
Disc permeameters (0.2 m diameter) were set 0.4-0.5 m apart on levelled pads (approximately 3 mm thick) of fine bedding sand (mean diameter <0.001 m). Four permeameters were placed in each of the 3 treatment trenches (12 duplicates per treatment). Permeameters were operated with sequential supply tensions of -10, -4, -3, -2, and -lcm, for 9, 4, 4, 4, and 4 min, respectively, for short measurement times, or 30, 19, 9, 9, and 9 min, respectively, for longer measurement times. The supply tensions were measured in ascending order (soil wetting up) rather than descending order (soil draining) because only wetting-up occurs during infiltration, the process measured by the permeameters (Elrick and Reynolds 1992).
Twelve ponded rings (0.3 m diameter) were installed 0.1 m into the plough pan layer (from 0.1 to 0.2 m) and flow rates measured over a 24-h period. Six rings were run at in situ soil moisture and 6 were pre-wet. Infiltration rates (Qs) and field-saturated conductivity ([K.sub.FS]) were compared to measurements obtained from permeameters.
Experiment 2: Confining permeameters to prevent excessive lateral flow
During Expt 1, excessive lateral flow of water was observed in depressions around permeameters at less negative tensions, invalidating the determination of K. To prevent excessive lateral flow, we confined the permeameters in rings and compared results with unconfined permeameter and ponded ring methods. The effects of pre-wetting were further investigated.
The permeameter treatments were: (i) soil at in situ field moisture content/permeameters unconfined; (ii) soil at in situ field moisture content/permeameters confined; (iii) soil pre-wet/permeameters unconfined; (iv) soil pre-wet/permeameters confined.
For the confining treatments, permeameters were placed within metal rings (0.24 m diameter by 0.2 m depth) inserted 0.1 m into the plough pan layer (after pre-wetting, for treatment (iv)), similar to the original techniques used by White et al. (1992). Steel rings with bevelled edges were oiled and driven smoothly and levelly into the soil, to minimise soil disturbance and fracturing. Any cracks or gaps between the soil surface and the inner ring wall were gently finger-sealed to prevent edge-flow.
The above permeameter treatments were applied to cropped and grass pasture on a Black Vertosol 25 km NW of Dalby, Queensland (Bongeen soil type), and a non-cracking structured Red Ferrosol at Toowoomba, Queensland (Ruthven soil type). The Bongeen soil had been under cultivation for >50 years, with reduced/no till strips of cropping and pasture for the last 8 years. The Ruthven soil had an area under dryland cropping, and an adjoining area under native and improved grass pasture (for the last 15 years after prior cultivation). Both cropped soils had well established plough pans. Soil was removed to 0.15 m to expose the plough pan layer in cropped plots, and to 0.03 m in pasture plots.
Measurements were taken in 2 trenches on each treatment, with 6 permeameters in each trench (12 duplicates per treatment). Pre-wet treatments were wet overnight, with 30mm applied to the Bongeen soil (this site was already quite wet from recent rain), and 50 mm applied to the Ruthven soil.
Additional trenches were prepared on cropped treatments to enable measurements over longer durations (3-6 h) to determine the time necessary to reach steady-state infiltration rates ([t.sub.q]). Measurement times were extended at either -10 or -1 cm tension, with other tensions measured for standard times. Results of these long runs determined the measurement times used during experiments on the Bongeen soil (60, 19, 19, and 19 min for -10, -4, -2, and -1 cm tensions, respectively). Measurements had been taken on the Ruthven soil prior to determining [t.sub.q], using 15.5, 6.25, 6.25, and 6.25 min times for sequential tensions. Moisture content was sampled to 0.35 m (0.05-m depth increments) under permeameters after measurements ceased, and adjacent to the confining rings. Twelve ponded rings were also run on each soil and management treatment (pre-wet overnight prior to ring insertion) to obtain comparative values of [K.sub.FS].
Experiment 3: Comparing [K.sub.[psi]] across sites and cropping systems using the modified method
[K.sub.[psi]] was measured using the confining method on 2 Black Vertosols with contrasting soil properties but similar crop management treatments (Dalgliesh et al. 2001). The Bongeen site was again used, together with a nearby site on a Waco soil type, 25 km NNW of Dalby, Queensland. The Waco had a higher clay content and finer structure than the Bongeen. [K.sub.[psi]] was measured at both sites on lucerne leys after 1 year, 2 years, and 5 years of lucerne after cropping, as well as an annual summer/winter crop treatment.
Initial moisture content samples (to 0.6 m) indicated considerable difference in soil moisture between management treatments at both sites, and therefore a more extensive pre-wetting regime was undertaken. Treatments were pre-wet over 3 weeks with 70 mm of water. Samples taken after pre-wetting indicated the soil was approximately at drained upper limit (DUL) to a depth of 0.45 m, although variability still existed between treatments.
Two trenches were prepared on each treatment (6 permeameters in each trench), located in non-wheel traffic areas. The trenches were located at 2 pre-existing datum points (1 and 4) spaced 600 m apart along the management strips for Bongeen and 1200m apart for Waco. All permeameters were confined in rings and the longer measurement times were used, as for the Bongeen soil in Expt 2. After measurement, soil moisture was sampled at depth increments 0.05 or 0.1 m under selected permeameters (through the axis of the confining ring). Antecedent soil moisture content was also sampled after pre-wetting.
For 3-D flow under permeameters, [K.sub.[psi]] (specific tensions denoted as [K.sub.-10cm] to [K.sub.-1cm]) and an estimate of steady-state flow ([Q.sub.INF]) were measured at each tension and extrapolated to saturation using the methods of Reynolds and Eirick (1991). As steady-state was not reached during the first experiment, calculation of [K.sub.[psi]] using steady-state analytical methods would have led to errors in [K.sub.[psi]] values. However, they are still reported here as an example of the commonly used method being assessed. [Q.sub.INF] values are also provided for comparison, and are not subject to 'choice of analysis' errors. For I-D flow under permeameters, [Q.sub.INF] was assumed to approximate [K.sub.[psi]], provided steady-state was reached. For initially wet soils, the proportion of flow due to capillarity is small and errors in assuming I-D flow are not large (i.e. sorptivity = 0). However, this assumption is not entirely accurate, as flow caused by capillarity still occurs and this condition may be more accurately referred to as quasi-steady-state (Cook 2002). To determine the required measurement times, time to steady-state ([t.sub.q]) was calculated from the 3-6-h run data (using methods based on Philip 1986; from Cook 1994).
Single-ring ponded infiltration rates and [K.sub.FS] were calculated using the methods of Reynolds and Elrick (1990). Data were analysed using a general analysis of variance (ANOVA using GENSTAT Release 4.2 Fifth Edition). Individual measurements were treated as duplicates within a trench stratum, with blocking for trenches. A Greenhouse-Geisserepsilon box test for symmetry indicated that permeameter flow data was dependent on the order of tension sequencing, and so degrees of freedom were adjusted (reduced) for general comparisons.
Results and discussion
Flow rates from the permeameters (Q) were 2-4 times lower through the wettest soil (pre-wet, longer measurements) than through the driest soil (in situ moisture content, short measurements), at all tensions (Table 2). Q and [K.sub.[psi]] mostly decreased as soil swelled due to increased water added from both pre-wetting and longer measurements. Differences between the treatments were not statistically significant; however, when data were pooled for wetting treatments. [Q.sub.-1cm] and [K.sub.-1] cm from longer run times were significantly lower than those from shorter runs. This agrees with Lin et al. (1998) who found swelling clay soils had very high infiltration rates when dry and cracked, in contrast to low infiltration rates when wet. Drier soil has more macropores and large lateral capillary forces dominate flow, creating considerable hydraulic gradients at the wetting front to drive flow rates.
Moisture sampling indicated the amount of water applied during pre-wetting did not uniformly wet the entire depth of soil in which flow was measured (50-80 mm). This moisture content heterogeneity may have confounded the results.
A comparison of [K.sub.[psi]] from permeameters and KFS from ponded rings
[K.sub.[psi]] and Q under permeameters (after 25 min) were 2 orders of magnitude higher than in ponded tings (after 24h) (Tables 2 and 3). Flow under the permeameters at 25 min was driven by hydraulic gradients at the wetting front, whereas flow had slowed to steady-state in ponded rings after 24 h. Connolly et al. (2002) observed a similar discrepancy between the methods and suggested it was due to the difference in measurement times. However, in this study, even very early (e.g. 2-9 min) flow rates in ponded rings (Table 3) were an order of magnitude lower than flow rates under permeameters (Table 2).
Ponding around permeameters caused inflated and negative K values
Small surface irregularities and depressions in the soil surrounding permeameters were observed to fill with water as tension was reduced, particularly on treatments with a lower soil K. Water filled depressions in preference to moving into dense or compacted soil. This created firstly generally inflated flow rates and a breakdown in the assumptions of 3-D flow (due to lateral anisotropy), and secondly surges in flow rates when bubble tube tensions were altered. Surface irregularities and depressions filled until the difference in height between the top of the ponded water and the base of the membrane was equal to the current bubble tube tension. Water from the permeameters ran into any space (soil or air) below this height. For example, if a 30-mm-deep depression occurred within the wetting zone, it would begin filling at the start of the -2 cm tension sequence. A surge in flow would occur until the height of water in the depression reached 20 mm below the membrane. When the tension was adjusted to -1 cm, the depression would fill to 10mm below the membrane, and so on. If most of the total surrounding depression volume was filled at -2 cm tension, [K.sub.-2 cm] would be inflated relative to [K.sub.-1] cm and a negative [K.sub.-1 cm] would be calculated (K being calculated by difference). However, if the surface was reasonably smooth, most of the depression volume would be filled during the -1 cm tension sequence, and [K.sub.-1 cm] would simply inflate to well above the 'real' rate of water percolating through the soil.
Inflated flow rates are often unknowingly measured, or negative values of [K.sub.[psi]] (knowingly) calculated, from permeameter data (D. McGarry, B. Bridge, R. Connolly, pers. comm.). It is not associated with profile backfilling, but rather appears to be related to the design and use of the permeameter. When water moves under tension from the permeameter into the soil, the sizes of the conducting pores are directly related to the potential at which the water is supplied. However, as the wetting front advances, the influence of the supply potential weakens. Water then indiscriminately fills both soil and depressional voids to the equivalent height below the membrane, of the current bubble tube setting. We observed depressions filling with water within 0.1 m of the permeameter base. This phenomenon can be severe on soils or layers with very low K, as it is far easier for water to move into air filled depressions than through these layers. It occurs to a far lesser extent on more freely infiltrating soils.
To overcome the distortions created by excessive lateral flow and flow surges, physically confining permeameters within tings was trialled. On both soils and management treatments, lowest [K.sub.-1 cm] values were derived from the pre-wet confined method, and conversely, highest values were mostly derived from the in situ moisture content unconfined method. Confining reduced [K.sub.-1cm] by 1-2 orders of magnitude on the highly swelling Bongeen soil (Table 4). Effects of confining were not as strong on the slightly swelling Ruthven soil, due to higher infiltration rates and probably less lateral distortion of 3-D flow in this soil, but were still significant. Confining had a much smaller effect on K-10cm values, indicating that there were less lateral distortions at this tension.
K decreased from [K.sub.-10cm] to [K.sub.-1cm] on the Bongeen cultivated treatment, due to increased wetting and swelling causing a reduction in both medium and large micropores and macropores. While total porosity increased, the effective porosity and macropore volume decreased. Steady-state was not reached in all treatments. Calculating K from transient flow data using steady-state methods, although commonly used, and presented here for comparative purposes, was not appropriate. For this reason, Q data is also presented (Table 4).
Conductivity of plough pans and pastured soils
At the experimental sites, soil under grass pastures contained extensive root channels and biopores and [K.sub.[psi]] was greater than under cropping (Table 4). [K.sub.-1 cm] (pre-wet, confined) values under grass were nearly 2 orders of magnitude higher (25 mm/h) than through the plough pan layer (0.3 mm/h) on the Bongeen soil, and about 4 times greater for the Ruthven soil. [K.sub.-10cm] (pre-wet, confined) values were similar for crop and pasture on the Bongeen, indicating similarity of matrix pore size distributions. Thus, differences between crop and pasture soils were due to macropore distributions (>0.3 mm diameter). This highlights both the benefits of rooting channels and biopores in maintaining high soil conductivity, and the severe restriction to water flow created by a dense plough pan layer.
The relative difference between agronomic treatments was not as strong for the Ruthven soil, probably due to the existence of a compacted layer in the pasture from periodic heavy traffic. However, the absolute differences in [K.sub.1cm] are quite large.
Comparison of [K.sub.[psi]] (confined permeameters) and [K.sub.FS] (ponded rings)
Across both soils and treatments, there was excellent agreement between [K.sub.-1 cm] under permeameters (confined, pre-wet) and [K.sub.FS] in ponded rings, at comparable early measurement times (16 v. 15, 60 v. 56, 0.3 v. 1, and 25 v. 24 mm/h, for permeameters and ponded rings, respectively; from treatments and soils presented in Tables 4 and 5). The (approximate) order of magnitude difference previously measured by the 2 techniques (Expt 1) was removed when permeameters were confined and K was measured using comparable 1-D methods.
K in ponded rings declined during the 24-h measurement period to approximately one-third of the early (25 min) values (Table 5). Hence, capillary forces at the wetting front apparently still had a significant impact on flow rates during the early stage measurements.
Measurement time necessary to attain steady-state infiltration ([t.sub.q])
Data from the extended runs on cultivated treatments indicated measurement time was important to accurately determine K (Fig. 1). On the cultivated Ruthven soil, gravity-dominated flow only commenced after the permeameters had run for 2h at -10cm. Moreover, [t.sub.q] was still not reached after 4 h (Fig. 1a), although for practical purposes a reasonable estimate of [K-10.sub.cm] was reached by 120min, as measurements did not alter significantly after this time. During the -1 cm tension sequence, [t.sub.q] was reached after 35 min (Fig. 1b). The shorter times used in Expt 2 (data shown as points in Fig. 1a, b, from Table 4) were insufficient to attain steady-state, over predicting [K.sub.-1cm] by a factor of 2 for pre-wet soil and by a factor of 9 for soil initially at in situ moisture content. Furthermore, increasing the application of water at higher tensions did not reduce the time necessary to reach [t.sub.q] at tensions closer to zero.
[FIGURE 1 OMITTED]
For the cultivated Bongeen soil, water applied by either pre-wetting or permeameters was important to obtain quasi-steady-state K. There was a positive cumulative effect of running permeameters longer at smaller tensions, as it decreased the time necessary to reach steady-state at tensions closer to zero. The best measure of steady-state [K.sub.-1cm] (and a good approximation of [K.sub.SAT as few/no macropores >3 mm diameter remained) was achieved after the soil had reached near-saturation during longer measurements (100min) at more negative tensions, even when using shorter measurement times (20 min) at tensions close to zero (Fig. 1c, d).
K for these soils was obviously highly influenced by the length of measurement time, contrary to many reports where flow rates reached within minutes of adjusting the supply tension were used as a good estimate of steady infiltration rates (Lin and McInnes 1995).
[K.sub.[psi]] values were small at both the Bongeen and Waco sites, indicating few macropores remained in the soil after extensive pre-wetting and measurement. [K.sub.[psi]] was similar (n.s.d.) for the 4 agronomic treatments at each site (0-, 1-, 2-, 5-year lucerne leys after cropping) (Fig. 2). The Bongeen soil had significantly lower [K.sub-4cm] to [K.sub.-1cm] than the Waco soil ([K.sub.-1cm] shown in Fig. 2), although K values for both soils are low in absolute terms. [K.sub.-1cm] was 0.8mm/h for Bongeen and 2.0mm/h for Waco. This is consistent with the higher bulk density (1.11 kg/[m.sup.3]) for Bongeen compared with Waco (1.03kg/[m.sup.3]) and therefore a smaller total pore volume.
[FIGURE 2 OMITTED]
[K.sub.[psi]] was consistent (n.s.d.) between datums 1 and 4 (600 m apart at Bongeen and 1200 m apart at Waco), demonstrating hydraulic uniformity over considerable distance.
Effectiveness of a long-term pre-wetting regime in unifying soil moisture
Extensive pre-wetting standardised the moisture contents at both datum locations and for all treatments on the Waco soil (0.51 and 0.50g/g for datum points 1 and 4, [F.sub.pr] = 0.29). However, on the Bongeen soil, datum 1 was significantly wetter than datum 4 after pre-wetting (0.44 and 0.36g/g, respectively, [F.sub.pr] = 0.041) and differences in moisture content also remained between the various management treatments. Regression analysis indicated final [K.sub.[psi]] was not affected by these differences in initial moisture content. However, sorptivity (S) at -10cm tension (run for 1 h) was slightly influenced by antecedent moistures ([R.sup.2] =0.27 and 0.26, [F.sub.pr] = 0.023 and 0.025, for 0.15-0.25 and 0.25-0.35 m depths, respectively). As sorptivity is more sensitive than [K.sub.[psi]] to initial moisture contents, this was not unexpected.
Water movement below confining rings during measurement runs
At both sites, approximately one-third of the water applied during measurements moved below the confining rings. As a result, soil was wetter below the rings than in the surrounding soil (Table 6). This difference was significant for the Bongeen 2-year and 1-year lucerne Icy plots only ([F.sub.pr]= 0.05), as these plots had been drier than other treatments prior to measurement.
Although water movement below the tings indicates 1-D flow analysis may have been inappropriate for part of the run time, no increases in the flow data were discernable. The 0.1 m of soil in the confining ring appeared to act as a highly restrictive 'plug' preventing an observable increase in flow once the wetting front moved below the ring (F. Cook, pers. comm.). However, increased flow has been observed on some soils once the wetting front dropped below the confining ring (D. McGarry, pers. comm.). The tings could be inserted to greater depth to reduce this problem.
The following mechanisms appear to be responsible for inflating flow rates under unconfined (3-D) permeameters:
(i) Water moves rapidly, and predominately laterally, from unconfined permeameters into the soil where any form of heterogeneity occurs. This includes variations in soil water content, as it determines the degree of swelling and porosity/pore sizes in the soil. Removal of this anisotropy is difficult, even with pre-wetting. This source of distortion is not limited to highly swelling soils. It was also evident in the slightly swelling Red Ferrosol.
(ii) Water moving out of the permeameters at tensions close to zero may fill surrounding depressions, and this is mistakenly measured as flow into the soil. This is a common occurrence on highly-swelling/low-conductivity soils. It can also negate the benefits of running unconfined permeameters until steady-state, as much of the 'measured' water may in fact be held in surface depressions, not the soil.
(iii) The absence of 'matching' swelling in the drier soil surrounding the wetting front provides lateral 'stress relief' from swelling pressures normally experienced when the entire soil surface is wetted. Flow rates are larger when there are no restrictions to swelling, either by soil at the wetting front or by the overburden (Kutilek 1996). However, this does not occur in the field under rainfall, where there is no laterally spreading wetting front. This distortion only occurs on swelling soils, and can be minimised by pre-wetting soil prior to measurement, and preferably confining flow to 1-D.
(iv) Flow rates are much higher for short run times where steady-state is not attained, and in dry soil, as discussed previously.
Comparison with other data
From the literature, it appears that K measured on similar (or even the same) swelling clay soils has also depended on whether 1-D or 3-D measurement methods were used (Table 7). For example, the smaller K values for Vertosols in Table 7, that is < 10 mm/h (Talsma and van der Lelij 1976; Forrest et al. 1985; Bird et al. 1996; Schaap et al. 1998), were all measured by laboratory cores, well permeameters, or in situ ponding. In all these methods, flow was confined to 1 dimension or the soil extensively wet, removing some or all of the mechanisms responsible for inflated flow. In contrast, values up to 2 orders of magnitude larger (Turpin et al. 1999; Connolly 2000; Vervoort et al. 2003) were measured with unconfined (3-D) disc permeameters, usually on dry soils using insufficient measurement times, activating all of the mechanisms responsible for inflated flow rates. Lower K values from permeameters that fall somewhere between these extremes were usually obtained from studies where the soil had been sufficiently pre-wet or long measurement times were used (Lin and McInnes 1995; Ringrose-Voase et al. 2003), thus removing many causes of inflated flow.
[K.sub.SAT] values reported for Red Ferrosols (Table 7) are quite high, around 100 mm/h, compared with those measured in this study (3.8 and 21.9mm/h for plough pan and pasture treatments, respectively; Table 5). Conductivity has been found to decline considerably with increasing soil moisture content, and with structural degradation and organic matter decline in Red Ferrosols (Bell et al. 2005). Similarly, in this study, flow took several hours to reach steady-state (i.e. for soil to fully wet) and degraded plough pans considerably restricted flow. Bell et al. (2005) measured a [K.sub.SAT] of 16.4 mm/h for degraded soil and 88 mm/h for rehabilitated soil. Because K varies considerably depending on land management history for these soils, generalized soil type comparisons are difficult. For example, McKenzie et al. (2001) used soil from the B22 horizon, probably without a restrictive layer, and [K.sub.SAT] was 4 times higher than through our pasture soil.
The results of this study indicate various problems with using 3-D unconfined permeameters to derive [K.sub.[psi]] for modelling internal redistribution of soil water on both Vertosols and Ferrosols. To derive estimates of [K.sub.[psi]] that reflect natural flow during internal drainage, we recommend (i) pre-wetting the soil and extending measurement time to attain steady-state flow, and (ii) confining permeameters within rings to restrict flow to one dimension. These methods result in [K.sub.[psi]] values several times lower than with 3-D unconfined permeameters, closer to [K.sub.SAT] observed with ponded rings.
Hydraulic properties were determined for 2 Vertosols and a Red Ferrosol in southern Queensland, using the modified permeameter method for [K.sub.[psi]] and ponded rings for [K.sub.SAT]. These indicate a near-saturated K on the Black Vertosols of 0.3-2 mm/h (7-48 mm/day) for shallow plough pans (annual cropping and lucerne leys, controlled traffic, minimum tillage), and 8-25 mm/h (192-600 mm/day) under grass pasture. Near-saturated K on the Red Ferrosol was 3.8 mm/h (91 mm/day) for shallow plough pans (trafficked, regular tillage), and 22-60 mm/h (528-1440 mm/day) under grass pasture. These K values are considerably lower than previously reported disc permeameter results, particularly for Vertosols, and are expected to be more suitable for modelling internal drainage in such soils.
This project was funded by GRDC projects DNR3 and DNR15, NAPWQS-SIP project Ag07 and Department of Natural Resources, Mines and Water, in collaboration with the GRDC Eastern Farming Systems Project DAQ 0050.
Thanks go to farm owners Rob Taylor and Jamie Grant for providing field sites, also to Leslie Research Centre for the use of Kingsthorpe Research Station. A special thank you to the dedicated CSIRO team led by Neal Dalgliesh and Brett Cocks, for conducting the third experiment, and to Kerry Bell for advice on statistical methods. Special thanks also go to Dr Bryan Bridge and Dr Steven Raine for reviewing this manuscript and providing invaluable comments.
Manuscript received 12 December 2005, accepted 10 August 2006
Bell M J, Bridge B J, Hareh GR, Orange DN (2005) Rapid internal drainage rates in Ferrosols. Australian Journal of Soil Research 43, 443-455. doi: 10.1071/SR04063
Bird TL, Willis TM, Melville GJ (1996) Subsoil hydraulic conductivity estimates for the lower Macquarie Valley. Australian Journal of Soil Research 34, 213-228. doi: 10.1071/SR9960213
Connolly RD (2000) Improved methodology for simulating infiltration in soil-crop systems. PhD thesis, University of Queensland, Australia.
Connolly RD, Bell M, Huth N, Freebairn DM, Thomas G (2002) Simulating infiltration and the water-balance in cropping systems with APSIM-SWIM. Australian Journal of Soil Research 40, 221-242. doi: 10.1071/SR01007
Connolly RD, Freebairn DM, Bridge BJ (1997) Change in infiltration characteristics associated with cultivation history of soils in south-east Queensland. Australian Journal of Soil Research 35, 1341-1358. doi: 10.1071/S97032
Cook FJ (1994) Effects of gypsum slotting on the hydraulic properties of a red-brown earth. Soil and Tillage Research 28, 239-252. doi: 10.1016/0167-1987(94)90133-3
Cook FJ (2002) The twin-ring method for measuring saturated hydraulic conductivity and sorptivity in the field. In 'Soil physical measurement and interpretation for land evaluation'. Australian Soil and Land Survey Handbook, Vol. 5. (Eds NJ McKenzie, KJ Coughlan, HP Cresswell) pp. 108-118. (CSIRO Publishing: Melbourne, Vie.)
Dalgliesh N, Tolmie P, Probert M, Robertson MJ, Brennan L, Connolly R, Sutton G, Hole S, Taylor R, Grant J (2001) Benefits and economics of a lucerne ley on the heavy clay soils of the Darling Downs. In 'Proceedings of the 10th Australian Agronomy Conference'. Hobart, Tasmania.
Elrick DE, Reynolds WD (1992) Infiltration from constant-head well permeameters and infiltrometers. In 'Advances in measurement of soil physical properties: Bringing theory into practice'. Soil Science Society of America Special Publication No. 30, pp. 1-24.
Forrest JA, Beatty J, Hignett CT, Picketing JH, Williams RGP (1985) A survey of the physical properties of wheatland soils in Eastern Australia. CSIRO Division of Soils, Divisional Report No. 78, Canberra, Australia.
Harris P, Biggs AJW, Stone BJ, Crane LN, Douglas NJ (1999) Resource information book. In 'Central Darling Downs land management manual'. DNRQ990102. (Department of Natural Resources: Brisbane, Qld)
Huth NI, Keating BA, Bristow KL, Verburg K, Ross PJ (1996) SWIMV2 in APSIM: An integrated plant, soil water and solute modelling framework. In 'Proceedings of the 8th Australian Agronomy Conference'. Toowoomba, Queensland.
Isbell RF (1996) The Australian Soil Classification. In 'Australian soil and land survey handbook'. Vol. 4. (CSIRO Publishing: Mellbourne, Vic.)
Kutilek M (1996) Water relations and water management of Vertisols. In 'Vertisols and technologies for their management'. Developments in Soil Science 24. (Eds N Ahmad, A Mermut) pp. 201-230. (Elsevier)
Lin HS, McInnes KJ (1995) Water flow in clay soil beneath a tension infiltrometer. Soil Science 159, 375-382.
Lin HS, McInnes KJ, Wilding LP, Hallmark CT (1997) Low tension water flow in structured soils. Canadian Journal of Soil Science 77, 649-54.
Lin HS, McInnes KJ, Wilding LP, Hallmark CT (1998) Macroporosity and initial moisture effects on infiltration rates in Vertisols and Vertic Intergrades. Soil Science 163, 2-8. doi: 10.1097/00010694-199801000-00002
McKenzie N, Jacquier D (1997) Improving the field estimation of saturated hydraulic conductivity in soil survey. Australian Journal of Soil Research 35, 803-825. doi: 10.1071/S96093
McKenzie NJ, Cresswell HP (2002) Selecting a method for hydraulic conductivity. In 'Soil physical measurement and interpretation for land evaluation'. Australian Soil and Land Survey Handbook, Vol. 5. (Eds NJ McKenzie, KJ Coughlan, HP Cresswell) pp. 90-107. (CSIRO Publishing: Melbourne, Vic.)
McKenzie NJ, Cresswell HP, Green TW (2002) Field measurement of unsaturated hydraulic conductivity using tension infiltrometers. In 'Soil physical measurement and interpretation for land evaluation'. Australian Soil and Land Survey Handbook, Vol. 5. (Eds NJ McKenzie, KJ Coughlan, HP Cresswell) pp. 119-130. (CSIRO Publishing: Melbourne, Vic.)
McKenzie NJ, Cresswell HP, Rath H, Jacquier D (2001) Measurement of unsaturated hydraulic conductivity using tension and drip infiltrometers. Australian Journal of Soil Research 39, 823-836. doi: 10.1071/SR99136
Perroux KM, White 1 (1988) Designs for disc permeameters. Soil Science Society of America Journal 52, 1205-1215.
Philip JR (1986) Linearized unsteady multidimensional infiltration. Water Resources Research 22, 1717-1727.
Powell B, Baker DE, Christianos NG (1988) Soils of the Kingsthorpe Field Station eastern Darling Downs. Queensland Department of Primary Industries, Research Establishments Publication QR88003.
Reynolds WD, Elrick DE (1990) Ponded infiltration from a single ring: I. Analysis of steady flow. Soil Science Society of America Journal 54, 1233-1241.
Reynolds WD, Elrick DE (1991) Determination of hydraulic conductivity using a tension infiltrometer. Soil Science Society of America Journal 55, 633-39.
Ringrose-Voase A J, Young RR, Paydar Z, Huth NI, Bernardi AL et al. (2003) Deep drainage under different land uses in the Liverpool Plains Catchment. Report 3, Agricultural Resource Management Report Series, NSW Agriculture.
Schaap MG, Leij FJ, van Genuchten MTh (1998) Neural network analysis for hierarchical prediction of soil water retention and saturated hydraulic conductivity. Soil Science Society of America Journal 62, 847-855.
Talsma T, van der Lelij AJ (1976) Infiltration and water movement in an in situ swelling soil during prolonged ponding. Australian Journal of Soil Research 14, 337-349. doi: 10.1071/SR9760337
Turpin JE, Bridge BJ, Orange D, Thompson JP (1999) Water and bromide movement in a Vertosol under four fallow management systems. Australian Journal of Soil Research 37, 75-89. doi: 10.1071/S97108
Verburg K, Ross PJ, Bristow KL (1996) SWIM v2. I user manual. CSIRO Division of Soils, Divisional Report No. 130, Canberra, Australia.
Vervoort RW, Cattle SR, Minasny B (2003) The hydrology of Vertosols used for cotton production: I. Hydraulic, structural and fundamental soil properties. Australian Journal of Soil Research 41, 1255-1272. doi:10.1071/SR02154
White I, Sully M J, Perroux KM (1992) Measurement of surface-soil hydraulic properties: disk permeameters, tension infiltrometers and other techniques. In 'Advances in measurement of soil physical properties: Bringing theory into practice'. SSSA Special Publication No. 30. (Eds GD Topp, D Reynolds, RF Green) (Soil Science Society of America: Madison, WI)
J. L. Foley (A,C), P. E. Tolmie (A), and D. M. Silburn (A,B)
(A) Agricultural Production Systems Research Unit, Queensland Department of Natural Resources and Water, PO Box 318, Toowoomba, Qld 4350, Australia.
(B) Cooperative Research Centre for Catchment Hydrology.
(C) Corresponding author. Email: Jenny.Foley@nrm.qld.gov.au
Table 1. Soil properties for the experimental sites (0-0.1 m) Coarse sand corresponds to particles of diameter 2-0.2 mm; fine sand, 0.2-0.02 mm; silt, 0.02-0.002 mm; clay, <0.002 mm Australian Coarse Fine Silt Soil name (A) classification sand (B) sand (%) Craigmore Black Vertosol 1 9 14 Ruthven Red Ferrosol 3 20 24 Bongeen Black Vertosol 7 17 12 Waco Black Vertosol 3 13 15 Clay Organic C CEC (B) Soil name (A) (cmol/kg) Craigmore 76 1.30 86 Ruthven 53 3.3 41 Bongeen 64 1.28 67 Waco 69 1.33 74 (A) Harris et al. (1999). (B) pH 8.5. Table 2. Flux (Q) and conductivity (K[psi]) of the plough pan layer for Craigmore soil, measured with disc permeameters No letter indicates that there are no significant differences. Values in parentheses are standard deviations of means Wetting Measurement Tension (cm) treatment time -10 -4 -3 Q (mm/h) Pre-wet Long 1.5 (1.6) 16.0 (4.2) 17.1 (5.1) Short 6.0 (4.8) 17.8 (10) 19.4 (11.9) In situ Long 5.3 (1.6) 15.5 (7.2) 16.7 (8.6) Short 6.2 (3.0) 24.6 (8.3) 27.4 (9.4) K [psi] (mm/h) Pre-wet Long 1.0 12.3 3.3 Short 2.5 7.5 5.0 In situ Long 4.0 20.5 16.7 Short 4.0 16.3 12.2 Wetting Measurement Tension (cm) treatment time -2 -1 Q (mm/h) Pre-wet Long 18.3 (5.1) 23.3 (7.2) Short 22.1 (12.3) 36.8 (27.5) In situ Long 19.0 (11.2) 36.0 (21.8) Short 31.3 (10.9) 48.6 (18.7) K [psi] (mm/h) Pre-wet Long 5.3 14.4 Short 11.1 33.0 In situ Long 16.4 35.9 Short 15.5 37.5 Table 3. Flow rates ([Q.sub.S]) and field-saturated conductivities ([K.sub.FS]) of the plough pan layer for Craigmore soil, measured by single ponded ring method The 2-9 min and 3-24 h times are determined from increments of the cumulative flow v. time curves. Values in parentheses are standard deviations of means Hydraulic Pre-wet In situ parameter 2-9 min 3-24 h 2-9 min 3-24 h [Q.sub.S] (mm/h) 4.5 (5.6) 0.6 (0.4) 9.7 (5.8) 0.8 (1.5) [K.sub.FS] (mm/h) 2.0 (2.5) 0.3 (0.2) 4.4 (2.6) 0.3 (0.7) Table 4. Flux (Q) and conductivity (K [psi]) in mm/h at -10 and -1 cm tensions measured with disc permeameters, for cropped and grass pasture treatments on Ruthven and Bongeen soils Within column treatments and tensions, means followed by different letters are significantly different (P < 0.05). No letter indicates no significant differences. Q measured on unconfined treatments is not included in the statistical analysis. For confined treatments, Q is assumed to approximate quasi steady-state K Treatment Ruthven Cropped Grass pasture -10cm -1cm -10cm -1cm Unconfined Pre-wet 8.1 71.9 3.7 190.4 (Q) In situ 12.6 128.6 6.4 209.3 Unconfined Pre-wet 6.4b 51.3b 2.4a 165.6b (K) In situ 8.3c 98.9c 3.9b 179.8b Confined Pre-wet 3.7a 16.4a 1.9a 60.3a (Q or K) In situ 7.1bc 52.3b 5.2b 79.1a 1.s.d. (P = 0.05) 1.8 12.1 1.4 47.1 Confined 3-5 h runs 1.6 6.2 pre-wet Treatment Bongeen Cropped Grass pasture -10cm -1cm -10cm -1cm Unconfined Pre-wet 1.1 52.3 1.5 312.7 (Q) In situ 1.2 47.8 4.5 433.4 Unconfined Pre-wet 0.7 49.3b 1.0a 262.0b (K) In situ 0.8 38.7b 3.4b 362.3b Confined Pre-wet 1.0 0.3a 1.6a 25.3a (Q or K) In situ 1.7 0.5a 2.9b 34.6a 1.s.d. (P = 0.05) 1.0 28.2 1.3 124.7 Confined 3-5 h runs 0.7 1.7 pre-wet (A) Surface ponding occurred in rings. Table 5. Conductivity ([K.sub.FS], mm/h) for Ruthven and Bongeen soils measured by single ponded ring method Early and final time values are calculated from increments on cumulative flow v time curves. Values in parentheses are standard deviations of means Ruthven Cropped Grass pasture [K.sub.FS] (9-25 min) 14.9 (7.8) 56.2 (42.6) [K.sub.FS] (24 h) 3.8 (1.4) 21.9 (9.5) Bongeen Cropped Grass pasture [K.sub.FS] (9-25 min) 0.9 (1.6) 24.2 (45.5) [K.sub.FS] (24 h) 0.3 (0.5) 8.0 (23.2) Table 6. Soil moisture contents (g/g) for Bongeen and Waco, at 0.25-0.35 m depth below disc permeameters and in surrounding soil Treatment Bongeen Permeameter Surrounding soil 1-year lucerne ley 0.454 0.435 2-year lucerne ley 0.387 0.357 5-year lucerne ley 0.429 0.418 Annual crop 0.409 0.398 Average 0.42 0.40 Treatment Waco Permeameter Surrounding soil 1-year lucerne ley 0.539 0.509 2-year lucerne ley 0.518 0.504 5-year lucerne ley 0.529 0.493 Annual crop 0.512 0.497 Average 0.52 0.50 Table 7. A comparison of [K.sub.SAT] and K [psi] derived from different measurement methods, for Vertosols and Red Ferrosols Soil type (A) Location Method of collection Vertisol 8 sites around Small cores, lab. Dalby, Qld Calcic Vertisol 4 sites, Lower Well permeameter Macquarie Valley, (0.6 and 1.5 m depths) NSW Black Vertisol Coleambally, Large cores, in situ NSW infiltration study Burleson clay Texas, USA Tension infiltrometer (B) Ships clay (0.5 m horizons) (50,70% clay) Black Vertosol, Blackville, Disc permeameter (C) Black Vertosol, Liverpool Plains, ~0.35 and 1 m horizons Grey Vertosol NSW Black Vertosol Dalby, Qld Disc permeameter (B) (no pre-wetting), ponded ring (0.1 m) Black Vertosol Darling Downs, Qld Disc permeameter (DE) (no pre-wetting) surface, below crumb layer Black Vertosol Warwick, Qld Disc permeameter (B) (no pre-wetting) surface, subsurface dry, subsurface wet Clay soil texture 84 soils, USA ROSETTA (PTF) (F) Red Ferrosol Kingaroy, Qld Ponded ring (0.1 m), disc permeameter (B) (no pre-wetting) Red Ferrosol Wollongbar, NSW Large cores, lab.; tension infiltrometer (B), lab. Soil type (A) [K.sub.SAT] K [psi] (mm/h) Vertisol Av. 0.15 (0.02-0.34) -- Calcic Vertisol 0.08-0.36 (0.6 m), -- 0.13-0.17 (1.5 m) Black Vertisol Av. 0.09 (0.02-0.4) -- Burleson clay -- 14 (Bc) 5.4 (Sc) Ships clay (50,70% clay) Black Vertosol, -- 10 Black Vertosol, 18 Grey Vertosol 8 Black Vertosol 0.8 98 Black Vertosol 90.8 36.8 Black Vertosol -- 152-645, 81-330, 41-271 Clay soil texture 6.15 [+ or -] 3.47 -- Red Ferrosol 100 96 Red Ferrosol 99 91 (mean) Soil type (A) Author Vertisol Forrest et al. (1985) Calcic Vertisol Bird et al. (1996) Black Vertisol Talsma and van der Lelij (1976) Burleson clay Lin and McInnes (1995) Ships clay (50,70% clay) Black Vertosol, Ringrose-Voase et al. Black Vertosol, (2003) Grey Vertosol Black Vertosol Connolly (2000) Black Vertosol Vervoort et al. (2003) Black Vertosol Turpin et al. (1999) Clay soil texture Schaap et al. (1998) Red Ferrosol Connolly (2000) Red Ferrosol McKenzie et al. (2001) (A) Classification as originally reported. (B) -1 cm tension. (C) -1.5 cm tension. (D) [K.sub.SAT] extrapolated from -2 cm tension data. (E) -2 cm tension. (F) PTF, pedotransfer function.
|Printer friendly Cite/link Email Feedback|
|Author:||Foley, J.L.; Tolmie, P.E.; Silburn, D.M.|
|Publication:||Australian Journal of Soil Research|
|Date:||Nov 1, 2006|
|Previous Article:||Limitations in the use of electrical conductivity to monitor the behaviour of soil solution.|
|Next Article:||Sampling considerations for surveying copper concentrations in Australian vineyard soils.|