VARIABILITY OF HYDRAULIC CONDUCTIVITY DUE TO MULTIPLE FACTORS

Soil properties are greatly influenced by intrinsic factors of soil formation as well as extrinsic fac tors associated with land use and management and vary bo th in time and space. Intrinsic variability is caus ed by the pedogenesis and usually takes place at large ti me scales. The variability caused by extrinsic fact ors could take effect relatively quickly and could not be treated as regionalized. Saturated hydraulic conductivity is one of the most important soil prop erties for soil-water-plant interactions, water and contaminant movement and retention through the soil profile. It is a critically important parameter fo r estimation of various other soil hydrological param eters necessary for modeling flow through the natur ally unsaturated vadose zone. Among different soil hydro logical properties, saturated hydraulic conductivit y is reported to have the greatest statistical variabili ty, which is associated with soil types, land uses, positions on landscape, depths, instruments and methods of me asurement and experimental errors. The variability of saturated hydraulic conductivity has a profound inf luence on the overall hydrology of the soil system. Therefore, focus of this review is centered on the variability of saturated/unsaturated hydraulic cond uctivity due to a large number of factors. This study review s recent experimental and field studies addressing the measurements and variability of hydraulic conductiv ity. A synthesis of a large amount of data availabl e in literature is presented and the possible sources of the variability and its implications are discussed . The variability of a soil hydraulic conductivity can be expressed by range, interquartile range, variance and standard deviation, coefficient of variation, skewn ss and kurtosis. The spatial and temporal variabil ity of hydraulic conductivity and the influences of sample support, measurement devices/methods, soils, land uses and agricultural management on hydraulic conductivi ty are evaluated. Methods of measurements strongly impact variability, for example, saturated hydrauli c conductivity measured using a single ring may pr oduce significantly different mean and standard errors th an those measured using a double ring. The sample support can also influence the variability, for exa mple, increasing or decreasing the size of the infi ltrometer rings can change the mean and variability of the sa turated hydraulic conductivity. Similarly, hydrauli c conductivity measured in the field could show a muc h larger variability than those measured in the laboratory. The spatial and temporal variations of hydraulic conductivity and interactions among soil characteristics, land uses, agricultural management , climatic and environmental conditions and measurement methods are rather complex, which shoul d take into account multiple factors discussed in t h s review. Decisions and choices made by investigators during sampling, sampling designs, availability of resources, number of investigators involved in samp ling and analysis, skill level of investigators, ty pe and quality of tools and equipments used to collect sam ples and analyses, scale of the domain, availabilit y of time, accessibility of sites, criteria of success a nd ssumptions made for the sampling and analysis h ave profound influence on the variability of hydraulic conductivity.


INTRODUCTION
Darcy's law describes the one dimensional flow of water through a saturated soil profile whereas Darcy-Buckingham law describes the one dimensional flow of water through an unsaturated soil profile.A more exact and generalized differential form of Darcy's law for three dimensional saturated porous media was proposed by Science Publications AJES Slichter (1899).The formulations essentially indicate that the flow through the saturated porous media is proportional to the hydraulic gradient that is the driving force causing flow.And the term, included as a constant of proportionality, is known as hydraulic conductivity (K s ) of the saturated porous media (Lal, 2004).In essence the K s is a constant for a given saturated porous media in any given direction and it can have a different constant value in each of the three dimensions.The K s is strongly influenced by the properties of a porous media such as structure, pore connectivity as well as the properties of the fluid such as viscosity and temperature.
Another similar term is permeability that is defined as the property of the porous medium controlled only by the pore geometry (Richards, 1952).Permeability and the hydraulic conductivity are very different parameters.It is believed that Soil Conservation Service handbook first defined qualitatively the water movement through soil as two distinct permeability classes of favorable and unfavorable (Norton, 1939).Subsequently using the percolation field experiments of Uhland and O'Neal (1951), seven permeability classes were proposed (SSDS, 1993).Based upon the work of Mason et al. (1957), National Soil Moisture Committee proposed a "choice schema" of five to seven classes Soil Survey Division, 1997.There were two popular methods at that time, one of them was the auger hole method that can measure water flow in multiple directions and the other was Uhland core method.There were concerns about the suitability of each method and whether these two can provide similar estimates of the permeability.By 1969, saturated hydraulic conductivity obtained from the Darcy's law was recommended as the correct term and permeability classes were renamed as hydraulic conductivity classes.
Saturated hydraulic conductivity is one of the most important parameters for soil-water-plant interactions, water and solute movement and retention through the soil profile.It is a critically important parameter for estimation of various other soil hydrological parameters necessary for modeling flow through the naturally unsaturated vadose zone.Among different soil hydrological properties, the K s is reported to have the greatest statistical variability by several authors (Biggar and Nielsen, 1976;Hern and Melancon, 1986;Webb et al., 2000).The variability of K s is associated with soil types, land uses, positions on landscape, depths, instruments and methods of measurement and experimental errors (Stockton and Warrick, 1971).It has been suggested that more studies are needed on the variability of K s across different landscapes.The variability of K s has a profound influence on the overall hydrology of the soil system.Therefore, focus of this review is centered on the variability of saturated/unsaturated hydraulic conductivity due to a large number of factors.A summary of methods of measurements, synthesis of large amount of data available in literature is presented and the possible sources of the variability and its implications are discussed.

Methods of Hydraulic Conductivity Determination
The various laboratory and field methods used in determining the hydraulic conductivity are summarized in Fig. 1.Readers are referred to Schilfgaarde (1974); Dane and Topp (2002); Stephens (1996); Reynolds et al. (2000) and Dane and Topp (2002) for details on the saturated (K s ) and unsaturated (K) hydraulic conductivity determination methods.

Saturated Hydraulic Conductivity
The K s of undisturbed cylindrical core samples can be measured by using a constant head or a falling head method in the laboratory.Measurements of field K s in the unsaturated or vadose zone (above the water table) can be obtained using various ring or cylinder infiltrometers (e.g., single-ring and double-or concentricring infiltrometers, pressure infiltrometers, twin-or dualring and multiple-ring infiltrometers) and constant head well or borehole permeameter methods.The correlation methods are based upon relationships between the K s value and one or more of the soil properties such as soil texture, pore size distribution of the soil, grain size distribution of the soil and soil mapping unit.In the saturated zone (below the water table), auger hole and piezometer methods are commonly used techniques.Other methods in the saturated zone include the two-well method, the four-well method, the multiple-well method, the pit bailing test and the slug test.

Unsaturated Hydraulic Conductivity
The laboratory methods are: steady state flow methods in horizontal or vertical column under constant head or flux conditions and transient flow methods.Measurements can also be made in laboratory long soil columns by inducing evaporation or infiltration.A variety of transient laboratory techniques can be used: the instantaneous profile method, the Bruce-Klute method, the pressure plate method, the one-step outflow method and the ultracentrifuge method.The field methods include the instantaneous profile method, the flux control method, the flow net method and the borehole point source method.The crust method can be applied while using double-ring infiltrometers, pressure infiltrometers, or disc permeameters.There are two approaches to estimate the K function of an unsaturated soil: empirical equations and statistical models (Table 1).Several measured conductivity data are required to use an empirical equation, while a statistical model can be used to predict the K function when the K s and the soil water retention curve are available.

Variability of Soil Hydraulic Conductivity
Soil hydraulic conductivity displays large statistical variability and both short and long range spatial and temporal variability.It is greatly influenced by extrinsic factors of land use and management, usually at short time scales and the intrinsic factors of soil formation, usually at large time scales (Nielsen et al., 1973).Intrinsic variability, due to pedogenesis, is also known as regionalized with nearby areas being more similar than the areas farther away (Dane and Topp, 2002;Nielsen and Wendroth, 2003).On the other hand, the extrinsic variability could take effect rather quickly and might not be treated as regionalized.In addition to these two sources of variability, measurement devices used, sample support areas, assumptions and choices made by the investigators, types of sampling strategy and designs also have profound influence on the variability.
Methods of measurements strongly impact variability, for example, long term infiltration tests conducted to determine field K s using a single ring may produce significantly different mean and standard errors than those measured using a double ring, Guelph permeameter and/or tension infiltrometer at the same location.The sample support or size of the infiltrometer rings or disks also influence the variability.
Similarly, hydraulic conductivity measured in the field could show a much different amounts of variability than those measured in the laboratory (Shukla, 2011).

Statistical Variability
The variability of soil hydraulic conductivity is usually expressed as Coefficient of Variation (CV), ratio of standard deviation and mean.There are several criteria based on CV, a measure of relative variability, available in the literature.Statistical variability can also be represented by variance or standard deviation, range including interquartile range, skewness and kurtosis.Variance is defined as the second moment about mean that is the average of the square of deviations of a value from its mean value or the first moment.Standard deviation is the measure of absolute variability.The difference between the largest and the smallest value is presented as range.If data contains outliers then range does not truly describe the characteristic of the datasets and in that case interquartile range that takes middle 50% of values is a somewhat better statistic to express variability.Skewness is expressed as the ratio of third moment about mean and third power of standard deviation and kurtosis is the ratio of fourth moment about mean and fourth power of standard deviation.Skewness is a measure of the asymmetry of the probability distribution of a random variable, while kurtosis is a measure of the peakedness of the probability distribution of a random variable (Isaaks and Srivastava, 1989;Nielsen and Wendroth, 2003).
Several studies undertaken during the last decades reported that in general K s displays the greatest variability expressed as CV across sites (Table 2), although some of the important on site hydraulic properties affecting K s such as total porosity and field capacity water content could have much smaller variability (data not reported) across these sites (Shukla, 2011).Such a behavior supported earlier observations that pore size, shape and connectivity are more important than total porosity.

Influence of Sample Support on Saturated Hydraulic Conductivity
Among various factors influencing variability of K s , the sample support or the area of cross section of the flow domain is an important factor (Bagarello, 1997).Gupta et al. (2006) reported only minor differences in K s when infiltration tests were performed in field near College Station, Texas using disk infiltrometer with disk sizes of 10, 15, 17, 20 and 24 cm.The CVs for the K s ranged from 0.3-0.5 for disk sizes ranging from 0.10-0.17m.However, further increase in the disk sizes to 0.20 and 0.25 m resulted in a much higher increases in CVs to 0.87 and 0.86, respectively (Table 3).Not much difference was observed in the mean value of the K s and no relationship was observed between disk size and K s .The larger variability for bigger disks could be due to the inclusion of larger sample volume with attendant increase in heterogeneity and macropore network.In another study, Bagarello and Provenzano (1996) studied the effect of size of the undisturbed soil core on the laboratory estimates of K s using Constant Head Permeameter (CHP) method in a sandy clay soil and in particular, large cores (0.085 m-diameter × 0.11 m-high stainless cylinders) produced lower and less variable estimates of K s than small cores (0.05 m-diameter × 0.05 m-high stainless cylinders), in which the preferential flow increased greatly.A comparison between results of the in-situ Guelph Permeameter (GP) method and CHP method showed that overall, K s values obtained using CHP method produced larger means and CV values than K s values obtained from the GP method (Table 4).

Influence of Measurement Devices/Methods on Saturated Hydraulic Conductivity
The K s can be measured using various devices and procedures as described earlier.The K s was determined by conducting long duration infiltration tests (3 h) using double ring infiltrometers and using constant head Science Publications AJES method on soil cores collected from the same experimental location near Ohio, USA (Shukla et al., 2003b).Table 5 presents the average and standard deviations of i c and K s determined in field and lab, respectively, at three slope positions, shoulder, middle and foot.The hypothesis was that since double ring has a sample support of 15 cm that is almost double the sample support for the core (7.8 cm), i c would be much greater than K s .However, the data in Table 5 does not validate the hypothesis except for the data from No-Till Corn-Soybean-Rotation (NTCSR) field.Earthworm activities were clearly noted in NTM and NTWM fields but sample support did not influence the values of K s , indicating that macropore channels or earthworm burrows were not open at the soil surface.
However, an exactly opposite result is also possible for a variety of reasons, such as, smaller sample support for laboratory (or core) than the field experiment, spaces between the core and the soil, hitting or missing macropores in the soils.In an experiment, Reynolds et al. (2000) determined K s using tension infiltrometer, pressure infiltrometer and soil cores (Table 6).The mean K s values were always greater for the soil core method for sand and clay loam soil but not for loam.The smaller K s for tension and pressure infiltrometer methods can be because of the fact that in-situ methods mostly have a disadvantage of soils not being fully saturated and measurements are actually under quasi steady state.
In general, looking at the data in Table 5 and 6, the variability expressed as CV of K s was large irrespective of the method of measurement and no definite trends were visible among these methods.The possible explanation for the differences among these methods could be the differences in flow domains or sample sizes and flow geometries.The surface area for the infiltration was much higher (491 cm 2 ) for tension infiltrometer than for other methods (79 cm 2 ).Flow was three dimensional from a tension infiltrometer and near one dimensional from a pressure infiltrometer or soil core method.In addition, likely blockage of macropores by core walls and experimental artifacts could also change K s .For a randomly distributed domain, measuring K s over a larger volume of soil can be equivalent to pooling the measurements from within the smaller volumes (Parkin and Robinson, 1992).In this case the small and large supports are centered on the same mean because they are sampling the same population.
In a study, the K s from Guelph permeameter, the velocity permeameter, a pumping test procedure and the auger hole method were compared for a Ravenna silt loam at Wooster and a Hoytville silty clay loam near Fermont, Ohio and evaluations were conducted during high water table conditions established by subirrigation (Dorsey et al., 1990).Authors reported that the pumping test, auger hole and velocity permeameter methods provided results within similar ranges whereas the Guelph permeameter provided significantly lower estimates of K s (Table 7).
The Guelph permeameter method gave the lowest K s values because of small sample size, whereas the disk permeameter and double-tube methods gave maximum values for K s with minimum variability, likely due to the large sample size (Table 7).Maximum variability in K s values for soil cores at shallow depths was attributed to the presence or absence of open-ended macropores.However, estimates of K s were most comparable for the velocity permeameter and the laboratory method using a constant-head permeameter.

Influence of Land Use on Saturated Hydraulic Conductivity
Several accounts are available using different measurement devices to demonstrate that land use has a strong influence on the variability of soil hydraulic conductivity.The steady state hydraulic conductivity was measured by a double ring infiltrometer and soil core method in the lab under three different land uses, i.e., annual tillage by chiseling or mold board plowing, no-tillage (6-15 years) and woodland (Reynolds et al., 2000; Table 8).October (e-h), respectively.The number in parenthesis is the averaged initial volumetric soil moisture at each site in May or October (Zhou et al., 2008).(Reprinted with permission from Elsevier) Both K s and steady state infiltration rate were higher from woodland than agricultural fields and steady state infiltration rate values varied in the order: woodland > no-tillage>annual tillage.Such a trend is not surprising because of the higher macroporosity of the soils of the natural woodland than soils under no-tillage or conventional tillage system.The tension infiltrometer seemed to underestimate K s values for sand under all three land management systems.A possible explanation could be the restriction to flow from tubes or air in the mariotte bottle used to supply water.Other possible reasons could be the arrangement of macropores, three dimensional infiltration and restrictions to flow by the membrane (Reynolds et al., 2000).
Science Publications AJES Fig. 3. Temporal variability of saturated hydraulic conductivity at two soil depths (0-5 cm and 5-10 cm) in (A) prairie and fields with (B) 3, (C) 14 and (D) 32-year of cropping (Scott et al., 1994).(Reprinted with permission from Soil Science Society of America) However, average K s values did not follow the conventional wisdom and were higher for fields under conventional tillage than no-tillage in the other study in Ohio (Table 9).This could be due to a number of factors including the larger sample size used for determining the K s from no-tillage fields than from fields under annual tillage, measurement errors in the field and laboratory while collecting and preparing the core samples, timing of tillage operations and errors during sample analyses.
Difference of hydraulic conductivity and their temporal dynamics were examined among four land uses (woodland, cropland, pasture and urban) and four soil series (Glenelg, Hagerstown, Joanna and Morrison) in Pennsylvania with contrasting textures, structures and parent materials (Zhou et al., 2008).At each of the 16 sites of soil series-land use combinations, Zhou et al. (2008) measured field-saturated and near-saturated hydraulic conductivities during May and October 2004 to 2006 using tension infiltrometers at water potentials of -0.12, -0.06, -0.03, -0.02, -0.01 and 0 m (Fig. 2).The measurement time had the greatest impact on measured hydraulic conductivities, followed by the land use and soil series.Compared to the cropland, pasture and urban land uses, woodland showed less temporal change because of less human-induced impacts and more consistent ground cover.

Temporal Variability of Saturated Hydraulic Conductivity
The temporal variability of K s was determined by several researchers.For example, Scott et al. (1994) studied short-and long-term variability of K s in four adjacent fields located in Arkansas, USA.One of the fields was under prairie while other three cropped fields had mostly been in a rice-soybean rotation for either 3, 14, or 32 years (Scott et al., 1994).The fields were sampled monthly from March 1989 to March 1990.The K s was measured on intact soil cores in the laboratory using constant head method and the temporal variability of K s is presented in Fig. 3. Scott et al. (1994) defined the variability among fields as long-term and the variability among sampling times within a field as short-term.

AJES
Table 10.Spatial variability of saturated hydraulic conductivity (K s * ) presented using semivariogram parameters for an experimental farm, Austria (Shukla et al., 2004), a 162-ha cotton field Perthshire, Mississippi, USA (Iqbal et al., 2005), a 12-ha field in Kingston, North Carolina, USA (Duffera et al., 2007) and a 40-ha field in Las Cruces, New Mexico, USA (Sharma et al., 2011 (Shukla et al., 2004;Duffera et al., 2007); and cm d -1 (Iqbal et al., 2005;Sharma et al., 2011); § NSR: Nugget to sill ratio (%); S is strong spatial dependence (NSR<25%), M is moderate spatial dependence (25%<NSR<75%) and W is weak spatial dependence (NSR>75%) using the criteria suggested by Cambardella et al. (1994) The short-term temporal variability of K s in the 3, 14 and 32-year fields resulted from seasonal changes and crop management practices (e.g., irrigation of soybeans, flooding of rice, tillage, disking), while the short-term variability of K s in the prairie resulted from climate influences on the biological activities of the grasses and microorganisms.Although not consistent, in general a pattern emerged and K s values started increasing from spring until early summer, remained similar until early winter and then decreased.During the years when the field was disked, the monthly trends of K s changed.Scott et al. (1994) reported that the relatively high K s during June and November 1989 was due to the unstable, loose aggregation of the soil due to tillage.
Another study on the temporal variability of unsaturated hydraulic conductivity, K (Ψ), was by Gupta et al. (2006).They applied six different matric potentials of -0.2, -0.15, -0.1, -0.05, -0.02 and 0 m using 0.2 and 0.24 m disks.The observations were made during a 21month period for May 2003 to January 2005 on an abandoned agricultural in Texas, USA.The temporal variability of K (Ψ) at varying matric potentials (-0.2, -0.15 and -0.1 m) is depicted in Fig. 4 as an example.The average values from these two disks showed remarkable temporal variation in K (Ψ).

Spatial Variability of Saturated Hydraulic Conductivity
It is commonly known that most soil hydrological properties exhibit both short and long range variability (Nielsen et al., 1973).It has been generally accepted that samples collected close to each other are more similar than those collected at greater distances.The similarity decreases as the separation distance between samples increases up to a certain separation distance beyond which samples are known as spatially uncorrelated or independent.Spatial dependence is reported to occur at scales ranging from a few meters to several kilometers (Trangmar et al., 1987;Ovalles and Collins, 1988;Gaston et al., 2001).Geostatistical analysis is usually carried out to understand the spatial structure and spatial variability of soil hydrological properties.Geospatial analysis can also provide more insight on spatial variability of a property whether it is structured, unstructured or directional.A detailed overview of these methods and their application on field datasets can be found in Hillel (1980); Webster (1985) and Nielsen and Wendroth (2003).
A study was conducted at the experimental farm of University of Natural Resources and Applied Sciences, Austria to determine the spatial variability of K s of the soil (Table 10; Shukla et al., 2004).In-situ K s was determined using Guelph permeameter and the variability was identified as moderate using the nugget ratio criteria of Cambardella et al. (1994).In a study on a 162-ha cotton field near Perthshire, Mississippi, spatial variability of K s was determined using falling head method (Iqbal et al., 2005).Table 10 shows that the K s had a nugget ratio ranging from 0.25-0.75 and using the criteria suggested by Cambardella et al. (1994) and Iqbal et al. (2005) classified them as moderate spatial dependent.In spite of the similarity of Nugget to Sill Ratio (NSR), various physical properties displayed wide variations in their range of spatial dependence.Although this research provided very useful information on the structure of the variability and spatial dependence of a Science Publications AJES soil property, the question, what could be the best sampling strategy for collecting samples for analyzing various soil properties that are spatially independent or uncorrelated, was not definitively answered.The differences in spatial class and range of spatial dependence among different horizons also indicated to the large inherent spatial variability of soil properties in general.
Similarly, spatial variability of K s was assessed for a 12-ha field in Kingston, North Carolina and spatial dependence was described using Cambardella et al. (1994) Classification (Table 10).The spatial dependence of K s has been also reported for surface soil (0-15 cm) in an agricultural field located in southern New Mexico (Sharma et al., 2011; Table 10).The important difference between different datasets presented in Table 10 is that spatial variability of K s was reported as moderate by Shukla et al. (2004) and Iqbal et al. (2005), weak by Duffera et al. (2007) and strong by Sharma et al. (2011).Looking at the CV (Table 2 for some of data), in these studies K s was reported always as highly variable.The data in several of the tables in this review showed no stochastic correlation between the CV and the size of the field or CV and the range of dependence.This could be due to the small sample size but could also be due to the multi-scale variability of K s across these domains.
Since K s is an important parameter for water and solute application efficiencies and triggering GHG (greenhouse gas) emissions from an ecosystem, knowledge of the spatial structure and spatial variability on a landscape scale is a prerequisite for designing site specific management.In order to conserve the water (surface and groundwater) resources and use the available water efficiently without polluting the water resources, as well as to prevent or minimize GHG emissions from the agricultural fields, there is a need to increase the overall on-farm water application and water use efficiency.Accordingly, an accurate knowledge of the variability of the K s is a prerequisite for initiating an efficient water management scheme.

CONCLUSION
In this review, the variability of hydraulic conductivity due to a large number of factors is presented.The spatial and temporal variability of hydraulic conductivity centered on the statistical variability of hydraulic conductivity and the influences of sample support, measurement devices/methods, soils, land uses and agricultural management on hydraulic conductivity are reviewed and discussed.The spatial and temporal variation of hydraulic conductivity and interactions among soil characteristics, land uses, agricultural management, climatic and environmental conditions and measurement methods are rather complex, which should take into account multiple factors discussed in this review and one must adequately assess a representative value.Notably, hydraulic conductivity measured in the field could show a much larger/smaller variability than those measured in the laboratory because of hitting or missing of macropore channels or error associated with the measurement.Since the design and functioning of the soil-water-plant hydrological systems depends to a great extent on the soil's hydraulic conductivity, decisions and choices made by investigators during sampling, sampling designs, availability of resources, number of investigators involved in sampling and analysis, skill level of investigators, type and quality of tools and equipments used to collect samples and analyses, scale of the domain, availability of time, accessibility of sites, criteria of success and assumptions made for the sampling and analysis have profound influence on the variability of hydraulic conductivity.

ACKNOWLEDGEMENT
Fig. 1.Overview of methods used to determine the hydraulic conductivity Fig. 2: Soil hydraulic conductivity as a function of water potential [K (Ψ)] among the four land uses (woodland, cropland, pasture and urban) within each of the four soil series (Glenelg, Hagerstown, Joanna and Morrison) measured in May (a-d) andOctober (e-h), respectively.The number in parenthesis is the averaged initial volumetric soil moisture at each site in May or October(Zhou et al., 2008).(Reprinted with permission from Elsevier)

Fig. 4 .
Fig. 4. Temporal variations of hydraulic conductivity [K (Ψ)] at varying matric potentials (Ψ = -0.2,-0.15 and -0.1 m).The average values for both 0.2 and 0.24 m disk diameters are presented, together with 10% error bars, because the means for both disk sizes were statistically similar.(Modified from Gupta et al., 2006).(Reprinted with permission from Soil Science Society of America).

Table 2 .
The statistical variability of saturated hydraulic conductivity (K s § , cm h -1 ) for some sites within the United States

Table 3 .
Gupta et al., 2006)pport or size (m) of the disk of the permeameter on saturated hydraulic conductivity (K s § , m s -1 ) of soil (data modified fromGupta et al., 2006) § SD: Standard Deviation; CV: Coefficient of Variation; Min: Minimum value; and Max: Maximum value

Table 4 .
Bagarello and Provenzano, 1996)y (K s § , m s -1 ) determined by using the packed soil columns and undisturbed soil samples of different sizes in the laboratory and the Guelph permeameter method under dry and wet antecedent soil water content (Adapted fromBagarello and Provenzano, 1996)CHP: Constant head permeameter; and large and small cores were collected in 0.085 m-diameter × 0.11 m-high and 0.05 m-diameter × 0.05 m-high stainless cylinders, respectively; † GP: Guelph permeameter; and SWC: Soil water content § GM: Geometric mean; SD: Standard deviation; and CV: Coefficient of variation; *