SSSAJ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (22)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Gerke, H. H.
Right arrow Articles by Köhne, J. M.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Gerke, H. H.
Right arrow Articles by Köhne, J. M.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Gerke, H. H.
Right arrow Articles by Köhne, J. M.
Related Collections
Right arrow Structure and Properties
Right arrow Hydraulic Conductivity
Soil Science Society of America Journal 66:26-36 (2002)
© 2002 Soil Science Society of America

DIVISION S-1 - SOIL PHYSICS

Estimating Hydraulic Properties of Soil Aggregate Skins from Sorptivity and Water Retention

Horst H. Gerke*,a and J. Maximilian Köhneb

a Institut für Bodenlandschaftsforschung, Zentrum für Agrarlandschafts- und Landnutzungsforschung (ZALF) e.V., Eberswalder Straße 84, D-15374 Müncheberg, Germany
b Institut für Geologische Wissenschaften und Geiseltalmuseum, Martin-Luther-Universität Halle-Wittenberg, Fachgebiet Umweltgeologie, Domstraße 5, D-06108 Halle, Germany

* Corresponding author (hgerke{at}zalf.de)


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Skins of soil aggregates often consist of clayey or clay-organic coatings which may affect preferential flow in aggregated soils. The objective was to determine hydraulic properties of samples with intact and removed (cut) skins and interior/skin hydraulic conductivity ratios for estimating mass transfer parameters in dual-permeability models. Soil aggregates from the Csd-horizon of a clay-loam glacial till soil (Stagnic Calcaric Regosol) were analyzed. A tension-imbibition apparatus was used for measuring water uptake of multiple aggregates at boundary matric potential heads of -1 and -5 cm. Sorptivities were used to calculate mean weighted water diffusivities and final water contents to fit wetting retention functions. Water retention and hydraulic conductivity functions for the skin layer were derived from differences in water contents and hydraulic resistances between intact and cut samples. Water absorption rates were generally smaller for intact than for cut aggregates. The water retention function of cut was shifted towards smaller water contents compared with intact samples. Mean water diffusivity of intact was 4.5 times smaller than that of cut samples. The interior/skin ratio in unsaturated hydraulic conductivity was about 12 in the measured matric potential head range. The ratio was up to 70 near water saturation and dropped below unity for soil water potentials smaller -1000 cm of water. Aggregate skins may be regarded as a separate porous domain whose hydraulic properties may control water transfer between inter- and intraaggregate pore domains in structured soils.

Abbreviations: Corg, organic C


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
SOIL AGGREGATES are porous blocks (peds) of various shapes and sizes that are separated from the surrounding soil by interaggregate pores. Primary peds, formed by solid constituents, may be hierarchically arranged into secondary and tertiary peds (Brewer, 1964) and are ranging in size between 0.2 and 5 cm for polyhedral and 1 and 30 cm for prismatic shapes (Scheffer, 1979). Intra- and interaggregate domains typically exhibit a sharp contrast in porosity, pore size, and pore continuity. While the intraaggregate domain can store most of the soil water, the interaggregate pore network may serve as pathway for rapid downward movement of water and dissolved chemicals (e.g., Beven and Germann, 1982; White, 1985). Thus, the observed preferential flow in structured soils (e.g., Booltink and Bouma, 1991; Flury et al., 1994; Hart and Lowery, 1996) has mostly been attributed to the differences in hydraulic conductivity and water retention properties between the intra- and interaggregate pore domains. Attempts to describe preferential flow are based on the dual-porosity model concept (Barenblatt et al., 1960), which assumes separate hydraulic properties for the two domains. A number of two-region mobile-immobile type transport models (e.g., van Genuchten and Wierenga, 1976; van Genuchten and Dalton, 1986) and dual- or multiple-permeability flow and transport models (e.g., Jarvis et al., 1991; Gerke and van Genuchten, 1993a; Gwo et al., 1995) have been proposed which differ mainly with respect to the descriptions of water movement in the macropore or interaggregate pore domain and of water exchange between the domains. The latter models all use the Richards equation for describing flow in a soil matrix domain, which in structured soils is mainly represented by the properties of the soil aggregates.

However at the local scale, even the primary soil aggregates cannot be regarded as a homogeneous porous medium. In particular, aggregates formed in illuvial or argillic horizons of Alfisols and Ultisols are covered by fine-textured skins or cutans. The translocated clays in cutans are typically more homogeneous and finer than matrix clays; complex cutans may include absorbed organic matter and Fe oxides while compound cutans consist of alternate layers of mineralogically and chemically different substances of different fabrics (Cady et al., 1986). Slickensides and pressure faces on ped surfaces are characteristic features in Vertisols or fine-textured glacial till soils (McCormack and Wilding, 1974). While the pedogenesis of clay void coatings (e.g., Bouma and van Schuylenborgh, 1969; Brinkman et al., 1973; Ransom et al., 1987) and the mineralogical (e.g., Hiller et al., 1993; Feijtel et al., 1989; Jongmans et al., 1998) and physicochemical (e.g., Kaiser and Wilcke, 1996; Celis et al., 1997; Kaiser and Zech, 1998) properties of coatings have been intensively studied, only little is known about the hydraulic properties of aggregate skins.

Wetting and drying induced stresses on aggregates may be corresponding with rearrangement of pores and particles (Horn et al., 1995), and lead to increasing clay contents at the outer skin (Horn, 1987). The proportion of fine pores was found to be higher near the surface of soil aggregates than in the center (Horn, 1986). Matric potential differences between inner- and outer-pore regions of soil aggregates could be observed at short distances using microtensiometers (Türk et al., 1991). These differences were larger for the denser aggregates of polyhedrical and subangular blocky than for weaker ones of prismatic shapes. Such structure effect was attributed to the unsaturated hydraulic conductivity functions of aggregates near saturation which for the denser aggregates had steeper slopes and were dropping faster with decreasing matric potentials than that of the bulk soil (Horn, 1994). In several studies (e.g., Gunzelmann et al., 1987; McKenzie and Dexter, 1996), the hydraulic functions of soil aggregates as a whole were analyzed, however, not separately for the skin layer. Youngs et al. (1994) and Leeds-Harrison et al. (1994) used artificially remoulded stabilized clay aggregates, which showed no difference between skin layer and interior.

For most natural aggregates, however, soil aggregate skins may act as a resistance to water flow from the interaggregate pore system into the center and vice-versa. A numerical sensitivity study, imitating a two-dimensional cross-sectional matrix-fracture system, showed that an interfacial hydraulic resistance is strongly increasing the duration of nonequilibrium in pressure head between the domains and thereby controlling the rate of preferential water movement during transient flow (Soll and Birdsell, 1998). Using a dual-permeability model, Gerke and van Genuchten (1993b) found that for typical aggregate sizes of ~1 cm, significant pressure head nonequilibrium and preferential flow could only be obtained by reducing the transfer term conductivity by orders of magnitudes below that of the soil matrix. Hydraulic conductivities of mineral fracture coatings were determined by comparing results from porous tuff samples with and without coatings (Thoma et al., 1992). However, the applicability of these methods is limited, mainly because of the different mineralogical composition and spatial dimensions of the rock compared with the soil system.

The objectives of this paper were to determine the hydraulic properties of samples with intact and removed skins and to derive interior/skin hydraulic conductivity ratios for estimating mass transfer parameters in dual-permeability models.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Soil samples were from the tile-drained glacial till agricultural field-site ‘Bokhorst’ located about 30 km south of the city of Kiel, Germany, where significant preferential flow was observed using Br- and dye-tracer experiments (Wichtmann et al., 1998; Lennartz et al., 1999). Soils exhibited a relatively large spatial variability, i.e., subsoil textures varied at places within a single meter distances between clay loam and silty sand. Soil types ranged from Stagnic Calcaric Regosols to Stagnic Luvisols according to the FAO classification scheme (FAO, 1990) (i.e., Calcaric or Eutric Epiaquent and Argillic Epiaqualf of the U.S. system [Soil Survey Staff, 1975]).

Undisturbed soil blocks of 20-cm edge length were sampled from the Csd horizon at 35- to 50-cm depth. Most of the blocks were composed of polyhedral and prismatic aggregates, while unstructured-sandy soil regions could also be found (Köhne, 1999). For this study, 42 primary aggregates of ~1- to 3-cm edge length were carefully detached from one block (5B) with a more fine-textured soil (Calcaric Regosol). The field-moist aggregates were then cut into cubic or prismatic forms so that geometries allow the assumption of one-dimensional flow and volumes could be determined without submerging and treating the samples before the beginning of the infiltration experiment. At one side of half of the aggregate samples, skins were left intact while skin layers of the other 21 samples, of at most 1-mm thickness, were manually removed using a scalpel. Other methods previously used for separating skin layers from aggregates (e.g., Kayser et al., 1994; Santos et al., 1997) were not applicable here. For each of the prepared soil aggregate samples, the surface area for water absorption was determined by vertically projecting the shape of the absorption side of the aggregates onto a piece of paper. The projected area was cut out of the paper and weighted. The area was calculated according to a regression relation between area and weight established for the particular type of paper. The field-moist volume, Vfield, of the aggregate samples was obtained from sample heights multiplied by projected areas. For the quality paper (80 g m-2) we used, the relation between weight and area was highly linear (coefficient of determination was 0.999). The projected areas were relatively precise; sample height could only be determined with a precision of ±0.5 mm, which appears large compared with an average sample height of about 6 mm.

To prevent evaporation and to mechanically stabilize the samples, all sides of the shaped aggregates were then covered with wax. Wax coating was done manually using a brush in such a way that wax could not enter soil pores and that top and bottom remained open. Wax cover weight was 3.5 g for 21 cut and 4.04 g for 21 intact aggregates. The prepared samples were placed in a porous kaolin-silt bed of a suction plate apparatus and equilibrated to an initially uniform matric potential head value of hi = -210 cm. Water contents were derived from gravimetrically determined mass changes, {Delta}Mi,j, of the samples between two different pressure head steps. Water absorption was measured using a tension-imbibition plate (see below) by first applying a matric potential head of h0 = -5 cm. The samples were then equilibrated again to hi = -210 cm and a second absorption experiment applying h0 = -1 cm followed. After that, the samples were water saturated from the bottom at about h = -0.1 cm and desorption water retention was determined at 10 pressure steps by using standard ceramic plates and kaolin beds for hanging water columns of up to 210 cm and pressure chambers for 1000 and 15000 cm of water. The samples were then oven-dried at 105°C for 2 d to obtain the dry mass. Finally, the volumes of the oven-dry samples, Vdry, were determined using the clod method (Blake and Hartge, 1986a) by immersing completely waxed samples in water.

Capillary water absorption at the bottom of 21 soil aggregates was measured using a tension-imbibition plate (Fig. 1 , numbers that follow the description correspond to Fig. 1). It consisted of a porous-membrane apparatus with an elastic foam bed (1), a bed of fine sand (2), a nylon membrane (HYDRO-BIOS Inc., Kiel, Germany) with a uniform pore size of 0.03 mm (3), a plastic foil evaporation-protection cover (4) with 21 holes in which the aggregate samples (5) were placed, an additional plastic foil evaporation-protection cover on top of the aggregate samples (6), a seal to prevent lateral evaporation water losses from the sand bed (7), and a 0.2-kg foam-coated cover (8) which slightly pressed the soil aggregates onto the sand bed to establish the hydraulic contact. Total water uptake by all 21 aggregates was determined gravimetrically by weighing the change of water mass in a storage bottle of 11-cm i.d. using an automatically recording balance. Considering the relatively small sums of total water uptake by 21 samples for the two pressure steps (i.e., 2.5 g and 3.4 g for intact and 3.2 g and 4.8 g for cut samples, respectively), the decrease of water level in the storage bottle was <0.05 cm during the imbibition experiments and considered to be negligible.



View larger version (31K):
[in this window]
[in a new window]
  		Fig. 1. Schematic cross-section of the capillary imbibition pressure-plate apparatus for multiple soil aggregates: elastic foam bed (1), sand bed (2), 30-µm pore-size membrane (3), plastic foil with holes for samples (4), aggregate samples (5), plastic foil (6), lateral sealing (7), foam-coated cover (8). A water-storage bottle of 11-cm i.d. placed on a balance is indicated on the left side.

 
Considering macroscopic transfer-term parameters, water absorption is determined here for the sum of an ensemble of 21 aggregates and not for individual samples. Water uptake rates of individual aggregate samples, of at most 0.01 to 0.1 g min-1, would have been too small for detailed transient measurements considering the sensibility of imbibition apparatus and balance (0.01 g). However, the contributions of individual samples to the overall sums of water uptake were determined at the end of each imbibition experiment gravimetrically. For estimating possible swelling at saturation, wet aggregate volumes, Vwet, were calculated using dry mass, particle density, and water volume at h = -0.1 cm assuming zero air-filled porosity. Soil texture was determined by wet sieving and sedimentation using pipette method (Gee and Bauder, 1986) with Köhn-Pipette method after organic C destruction with H2O2 and chemical dispersion using Na2PO7 (Hartge and Horn, 1992). Organic C content was calculated as the difference between total C, determined by elemental analysis of C after dry combustion at 1250°C (Deutsche Industrie Norm International Standardization Organization, 1994), and carbonate C, determined after application of phosphoric acid by gas-chromatographic analysis of CO2 evolution. Particle density was determined using the pycnometer method (Blake and Hartge, 1986b) and Xylol as submersion fluid.

Calculation of Hydraulic Properties of Intact and Cut Aggregates
The determination of unsaturated hydraulic conductivity is based on measurements of sorptivity, S [LT-0.5], of soil aggregates. Sorptivity as a function of the boundary and initial water contents {theta}0 and {theta}i is defined (Philip, 1957) as:

[1]
where I is cumulative horizontal infiltration [L3 L-2] and t is time [T]. Values of S may also be approximated by Eq. [1] for early stages of vertical water absorption by relatively small aggregates. The weighted mean soil water diffusivity, D [L2 T-1], (Dirksen, 1975) for sorptivity values obtained from experiments with different boundary conditions, is calculated (Klute and Dirksen, 1986) as:

where {gamma} = 0.67, {theta}i (which is required to be constant) is determined gravimetrically after each sample equilibration at hi = -210 cm, and {theta}0 (at the infiltration surface) is assumed to be identical with the aggregate water content, {theta}(h0), for imposed heads, h0, determined gravimetrically at the end of each absorption experiment (i.e., after water-uptake rates approached zero). In other studies (e.g., Bohne and Tille, 1985), values of {theta}0 have been determined immediately after linearity in the relation between I and t0.5 was observed using a thin soil layer from the contact surface of the sample, thus requiring a different sample for each absorption experiment with a different imposed head to derive an empirical function of S versus {theta}0. In this study, the function S ({theta}0, for constant {theta}i) is evaluated from two sorptivity measurements using the same sample for each imposed value of h0 and using the slope of the (here linear) regression of log S2 in terms of {theta}0. Here, the two D({theta}0)-values obtained with Eq. [2] are used to calculate a single mean soil-aggregate water diffusivity:

[3]
as a function of a mean boundary-water content:

[4]
where superscripts s indicate the wetting curve of the water-retention function {theta}s(hs) and the subscripts 0,1 and 0,2 denote the imposed matric potential heads at the boundary for the first and second imbibition experiments, respectively. Since the function D({theta}) is hysteretic, the wetting curve of the water retention function was used for calculating a hydraulic conductivity value at the mean water content as:

[5]
where the soil water capacity, C = , was evaluated at hs(0) using the retention function model of van Genuchten (1980) (VG-type)

[6]
where {theta}s [-] and {theta}r [-] are saturated and residual water contents, respectively; {alpha} [L-1], n [-], and m = 1 - n-1 are empirical coefficients. Both, the boundary drying and a single scanning wetting retention functions were fitted to desorption and wetting data, respectively, using the RETC-nonlinear optimization routine (van Genuchten et al., 1991). Since the number of wetting retention data was limited to four, the fitted wetting function was used here only for interpolating between measured matric potential head and water content values.

The relative hydraulic conductivity function of the VGM-type (Mualem, 1976; van Genuchten, 1980) was obtained using parameters of the drying retention curve, {theta}d (hd), as:

[7]
where Se = is the reduced water content. The K(0) value obtained with Eq. [5] from the sorption experiments was used to derive a desorption K({theta}) function assuming that, in contrast to the K(h) function, hysteresis in K({theta}) is small (e.g., Si and Kachanoski, 2000; Gilham et al., 1976; Rogers and Klute, 1971). The evaluated K value was matched with Kr obtained with Eq. [7] at the mean water content as:

[8]
where Ks is the hydraulic conductivity at water saturation [LT-1]. Here, the Ks value was used only for fitting the Kr function to the matching point K; and K(h) was obtained using the inverse of the VG-type retention model Eq. [6].

As an alternative method, compared with Dirksen (1975), White and Perroux (1989) proposed approximate relations for evaluating (h) from measured values of S as:

[9]
and as:

[10]
where = 0.5, = 0.5, {Delta}h = h1 - h2, {Delta}S = S1 - S2, {Delta}0 = 0 - {theta}i, and with 0 as the mean boundary water content (Eq. [4]). The value of M in Eq. [9] may vary between M = 1 (Green-Ampt-model) for structureless soils and M = {cong} 0.78 (weakly nonlinear Knight-model) for structured field soils. Values for , , {Delta}, {Delta}, and {Delta}0 were obtained from two absorption experiments.

Estimation of Skin Layer Hydraulic Properties
The approaches of both, Dirksen (1975) and White and Perroux (1989), result in effective values of the hydraulic conductivity, Keff, which are related to the total aggregates. A direct measurement of hydraulic properties of only the skin layer, separated from the aggregate interior, did not seem to be practicable. However, skin hydraulic properties could, for instance, be estimated indirectly by inverse modeling of the two-layer infiltration process (e.g., Smunek et al., 1998) or, more directly, by comparing hydraulic properties of cut with those of intact samples. Here, the concept of resistivity, R [T], previously used for describing hydraulic effects of soil crusts (e.g., Hillel and Gardner, 1969), was utilized.

Considering that the original soil aggregates were shaped to one-dimensional columns of relatively simple geometries, the volumes could be represented by lengths (Fig. 2) as:

[11]
where L [L] is the length of intact samples that includes the aggregate interior, Linterior, and the thickness of the skin, Lskin, (here 0.1 cm). Defined by the experimental conditions, here Lskin was identical with the thickness of the removed skin layer of the cut aggregates, while the mean length (Table 2) of the cut aggregates (Lcut = 0.6 cm) was different from the mean length of the interior part of the intact aggregates (Linterior = 0.49 cm). Since the soil properties of the cut samples are essentially identical with those of the interior of intact samples, the parameters {alpha} and n of the VG-type water retention function of the interior were assumed as {alpha}interior = {alpha}cut and ninterior = ncut. The saturated water content parameters (subscript s), however, were assumed proportional to interior and skin thickness of intact samples (Fig. 2) as:

[12]
where {theta}interiors = {theta}cuts. Similarly, the residual water content (subscript r) parameters, {theta}skinr and {theta}interiorr, could be obtained from {theta}cutr and {theta}r. The VG-type water retention function (Eq. [6]) of the interior of intact aggregates, {theta}interior(h) was obtained using desorption parameters, {alpha}d and nd, of the cut aggregates (Table 5). The water retention function of the intact aggregate skin, {theta}skin(h), was calculated as a length-weighted difference between intact and interior water contents.

[13]
for a number of N matric potential steps, hi. Finally, the skin retention function parameters were obtained by fitting the tabulated values, obtained with Eq. [13], to the VG-type function using the RETC-program (van Genuchten et al., 1991).



View larger version (18K):
[in this window]
[in a new window]
  		Fig. 2. Schematic picture of the physical system indicating soil-aggregate sample preparation and used symbols. Cubic or prismatic samples with intact (intact) and cut skins (cut) were prepared out of irregularly shaped (original) soil aggregates leaving the material of the removed skin layer of the cut samples (skin layer). Superscript symbols are used to distinguish between the lengths of the interior and skin parts of intact aggregate samples and the length of the cut samples.

 

View this table:
[in this window]
[in a new window]
  		Table 2. Sums, arithmetic means (Mean), standard deviations (SD) of volumes, V, and mass changes, {Delta}M, of 21 cut and 21 intact soil-aggregate samples used in the imbibition experiments. Vfield is the paper-measured area multiplied by the height of field-moist samples, Vwet is the sum of solid and water volumes of saturated samples (at h = -0.1 cm), Vdry and Mdry are volume and mass of oven-dry (105°C) samples, respectively; Mfield is field-moist mass, and {Delta}Mi,j are water mass changes between initial, i, and applied, j, matric potential heads.

 

View this table:
[in this window]
[in a new window]
  		Table 5. Parameters of the van Genuchten-type retention function (m = 1 - n-1) for desorption and sorption data of cut and intact samples obtained with the RETC-fitting routine. The fit to the regression function is indicated by the coefficient of determination, r2, and deviations between estimated and measured water contents by the sum of squares (SSQ).

 
By defining a hydraulic resistance, R [T], of the intact aggregate skin as:

[14]
with

[15]
the hydraulic conductivity of the skin was determined as:

[16]

where K and Kcut were both evaluated at the mean of the intact and cut desorption pressure head values at average boundary water contents, 0, of the two imbibition experiments. Note, that Eq. [16] assumes local equilibrium in matric potential between the interior and the skin of intact aggregates as well as the relation:

[17]
where Rinterior [T] is the hydraulic resistance of the interior of the intact aggregates. As for intact and cut aggregates, the function Kskin (h) was derived by fitting the Kr function to the matching point, Kskin = Kskin, here based on the reduced water content (van Genuchten et al., 1991) as:

[18]
which yields similar results without explicitly using a projected Ks value as in Eq. [8]. From the function Kskin(h), assuming a constant skin thickness, Lskin, and local pressure head equilibrium within the aggregates, values of the skin hydraulic resistances, Rskin(h), as a function of the matric potential head were obtained. In case different measured values of Lskin were available, interior/skin ratios in hydraulic conductivity, Kinterior/Kskin, could directly be recalculated by using Eq. [16] and [17] and keeping R and Kinterior = Kcut constant.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Texture data in Table 1 indicate relatively small differences between aggregate interiors and skins. The average clay (<2 µm) content of 25.8% and the fine- to medium-silt (2–20 µm) content of 33.2% were slightly larger than those of the interior of 24.4 and 29.6%, respectively. Average coarse-silt (20–63 µm) and fine-sand (63–200 µm) contents of the skins were 1.7 and 4.3% lower compared with the interior. Mean organic C (Corg) content (Table 1) of the skin material (0.87%) was about 1.6 times higher than that of the aggregate interior (0.54%). An enrichment of Corg in the skin layer may be characteristic for illuvial processes (Kaiser and Wilcke, 1996). Note, that data in Table 1 are based on 294 aggregate samples out of 7 soil blocks, including the 42 samples of the imbibition experiments, since skin material of the 21 cut samples was insufficient for carrying out texture and Corg analyses.


View this table:
[in this window]
[in a new window]
  		Table 1. Particle–size distribution, and organic C (Corg), and carbonate C (CCO3), contents (in % of weight) of the inner (core) and outer (skin) parts of soil aggregates; two replicates for each analysis using mixed soil material collected from 294 individual aggregate samples.

 
Areas, masses, and volumes as the sum of 21 intact and cut aggregate samples (Table 2) are used for evaluating the hydraulic properties. Mean values and standard deviations of the areas and the masses of field moist, Mfield, and dry, Mdry, aggregates are relatively similar for intact and cut samples (Table 2). The different volumes for dry and wet samples indicate that aggregate samples are subject to some swelling and shrinking. Differences between field moist, Vfield, and estimated water saturated, Vwet, volumes, however, may not be significant because of experimental uncertainties in sample height determination in case of Vfield and zero-air-volume assumption in case of Vwet. Mass changes in Table 2 show that for both imposed matric potential heads, {Delta}M-210,-5 and {Delta}M-210,-1, the water uptake sum of the 21 cut samples was larger than that of the 21 intact samples. Considering values of individual samples, the relatively large coefficients of variation (between 0.24 and 0.29) indicate that these differences in water uptake between cut and intact may not be significant.

At all moisture levels during the sorption experiment, mean gravimetric moisture contents, w, of the intact are larger than those of the cut samples (Table 3). Differences in w between intact and cut are more significant for the field moist and the initial conditions than for the -5- and -1-cm sorption steps considering standard deviations of 6 to 8% of the mean values. Bulk densities of dry samples, {rho}b_dry, appear to be lower for intact than for cut aggregates (Table 3). Standard deviations of bulk density of field-moist samples, {rho}b_field, are 13 to 14% of the mean value, which is about twice that of {rho}b_dry. Standard deviations in calculated porosities, {epsilon}field, are largest for field-moist samples. Measured solid particle densities, {rho}s, are slightly lower for intact than for cut aggregates (Table 3). Note that limited sample material allowed two replicate measurements of {rho}s only.


View this table:
[in this window]
[in a new window]
  		Table 3. Arithmetic means (Mean) and standard deviations (SD) of gravimetric moisture contents, w, bulk densities, {rho}b, calculated porosities, {epsilon}, and measured solid particle densities, {rho}s, of 21 cut and 21 intact aggregate samples used in the imbibition experiments. Field-moist conditions are indicated by subscript field, water saturation (at h = -0.1 cm) by subscript wet, and oven-dry by subscript dry, subscripts -210, -5, -1, and -0.1 refer to applied matric potential head steps between -210 cm and -0.1 cm of water.

 
Relative cumulative water uptake, I, each for group of 21 samples (Figure 3) , shows that water uptake is fastest for the cut aggregates. The reduced water absorption rates of intact samples point to a skin hydraulic resistance. Mass changes, {Delta}M, at the end of the imbibition experiments (Table 2) confirmed that each individual sample participated in water absorption, i.e., no outliers resulting, for instance, from poor hydraulic contact, were found. The differences in water uptake rates between intact and cut samples are somewhat larger for h0 = -5 cm compared with h0 = -1 cm of water (Fig. 3). In a visual inspection, a few cylindrical pores and planar cracks were observed at intact surfaces which may have reduced the skin resistance effect at h0 = -1 cm more than at h0 = -5 cm. Microscopic observations of cut aggregates showed that sliding and pressing of the scalpel blade created local patterns of rupture cracks alternating with more smeared areas. The cut surfaces appeared more level than the intact ones. These observations mainly suggest that water absorption through cut surfaces may start faster compared with intact ones, which seems to correspond with results in Fig. 3. Since cutting increased the permeability of only a thin layer at the surface, capillary-water uptake should be governed by the much larger unaffected pore system of the remaining interior. Water-drop penetration tests indicated that the delayed start of the intact samples (Fig. 3) to absorb water could not be explained by water repellency. Swelling could not visually be observed during the imbibition experiments. After water saturation (h = -0.1 cm), the wax mantles of some samples were cracked, possibly as a result of swelling. Values of I > 1.0 in Fig. 3, as for the cut samples at h0 = -1 cm, indicate losses because of evaporation.



View larger version (17K):
[in this window]
[in a new window]
  		Fig. 3. Relative cumulative water absorption by intact and cut soil aggregate samples of Soil Block 5B as a function of the square root of time (in min) for applied pressure head values of h0 = -1 cm and h0 = -5 cm of water. Each data series represents absorption rates obtained from a group of 21 individual soil aggregates relative to the total area-related water uptake, qin, at the end of the imbibition experiment (Table 4). Values of the sorptivity, S, indicated by solid lines, were obtained from the initial (t < 2 min) linear parts of the absorption curves.

 

View this table:
[in this window]
[in a new window]
  		Table 4. Data of the imbibition experiment: Applied pressure head, h0, initial water content, {theta}i, for initial pressure of hi = -210 cm of water, final water content, {theta}0, total water uptake, qin, sorptivity, S, and Dirksen-diffusivity, D, for cut and intact aggregate samples; each value represents a group of 21 individual samples using the sums in Table 2.

 
Sorptivities, S, in Table 4 are calculated using the initial periods of the water absorption measurements, i.e., of up to 1 to 2 min of the experiments, indicated by solid lines in Fig. 3. For the cut samples, the linear regressions of cumulative water absorption versus square root of time could be fitted through the origin. For the intact samples, only the linear slopes beginning from the first absorption measurement were used for calculating sorptivities. The analyses of the sorptivity data according to Dirksen (1975) result in larger D({theta}0) values for cut than for intact aggregates at both h0 steps (Table 4).

For calculating water retention of intact and cut aggregates, field-moist volumes, Vfield, were used at all pressure steps (Fig. 4) . The fitted drying and wetting retention curves are based on the sums of water masses and sample volumes of each group of aggregates to remain consistent with the imbibition data (Fig. 3). Mean values and 95% confidence intervals of the water-retention data show generally larger water contents at the same matric potential for intact compared with cut samples. These differences reflect that the skin layer may have a somewhat higher porosity than the aggregate interior, which corresponds with a lower {rho}b_dry of intact compared with cut samples (Table 3). The differences between {theta}(h = -0.1 cm) = 0.4058 for intact and {theta}(h = -0.1 cm) = 0.3688 for cut samples (Fig. 4) and the porosities, {epsilon}field, calculated using measured particle densities, {rho}s (Table 3) are larger for intact than for cut samples. The retention functions may to a certain extent be influenced by a lack of data on sample volumes for each matric potential step. Differences between intact and cut water retention would be smaller; e.g., at h = -15000 cm, {theta}cut would increase more than {theta}intact when using Vdry instead of Vfield and at h = -0.1 cm, {theta}intact would decrease more than {theta}cut when using Vwet instead of Vfield (Table 2). However, a combined step-wise determination of aggregate volumes and hydraulic properties was beyond the scope of this study.



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 4. Drying (solid symbols) and wetting (open symbols) water retention functions of intact and cut aggregate samples. Large symbols represent retention data obtained from the summed field-moist volumes and masses (Table 2) for each group of samples. Small symbols and errors bars indicate the arithmetic means and the 95% confidence limits obtained from individual samples. Solid lines are the fitted VG-type (van Genuchten, 1980) retention functions.

 
The VG-retention model (van Genuchten, 1980) assuming m = 1 - n-1 fits the desorption data of cut and intact samples when assuming a value of the residual water content parameter of {theta}r = 0.0 (Fig. 4). Differences between {theta}(h = -0.1 cm) (see above) and fitted {theta}s values (Table 5) are small. Each of the sorption curves is based on only three data points plus {theta}(h = -0.1 cm). The fitted wetting curves, showing relatively large {theta}r and {alpha} values (Table 5), are therefore limited to the interpolation of water contents between h0 = -5 cm and h0 = -1 cm to evaluate K in Eq. [5]. The sum of squares (SSQ) are given for completeness. The sorption curves (Fig. 4) indicate hysteresis of the aggregate water retention functions in the matric potential head range -210 cm <= h < 0 cm, however, data are insufficient for evaluating hydraulic properties in the wetting direction. Junkersfeld and Horn (1997) demonstrated that water retention and pore system of single soil aggregates depend on the initial intensity and frequency of desiccation subject to the allowed amount of swelling and shrinking. They postulated that irreversible changes of the pore geometry during desiccation limit hydraulic conductivity measurements with the same sample for repeated drying and wetting cycles. Here, the infiltration experiment started with relatively moist samples at hi = -210 cm of water covered by a wax mantle to keep such effects relatively small. Since the initial water contents (Table 3) at hi = -210 cm for the second imbibition step were nearly identical to those for the first step, we believe that the pore system remained similar during the two sorption experiments.

The arithmetic mean soil water diffusivity, , evaluated at the mean water content, 0, of the intact aggregates was ~4.5 times smaller and the mean hydraulic conductivity, K, was about 6 times smaller than that one of the cut samples (Table 6). The hydraulic conductivities at saturation, Ks, are resulting from matching the Kr functions through the evaluated K values. Values of Ks, which are only about 1.5 times smaller for intact compared with cut samples (Table 6), are given for completeness, however, may not be valid since projections of K using the VGM-type model from a matching point towards saturation are strongly affected by pore structure (e.g., Kosugi, 1999; Schaap and Leij, 2000).


View this table:
[in this window]
[in a new window]
  		Table 6. Evaluated soil water diffusivities, , and hydraulic conductivities, K, for cut and intact soil aggregate samples obtained from the imbibition experiments according to Dirksen (1975) at a mean water content, 0. The hydraulic conductivity, K), was evaluated using the sorption, hs, water retention; the hydraulic conductivity at saturation, Ks, was estimated using desorption, hd, retention function.

 
The hydraulic conductivities, , evaluated according to White and Perroux (1989) were slightly smaller than those evaluated according to Dirksen (Table 7). However, the mean matric potential head values of both approaches were slightly different, i.e., s = -3 cm instead of hs = -2.4 cm (cut) and -2.25 cm (intact) in Table 6. The sensitivity of the coefficient M in Eq. [9] is relatively small (Table 7). Although aggregate water retention is not required, Eq. [9] of White and Perroux, using a value of M = 0.9 leads to nearly the same results than the more comprehensive approach in Eq. [10] (Table 6) and those of Dirksen (1975).


View this table:
[in this window]
[in a new window]
  		Table 7. Hydraulic conductivity values, , evaluated from the imbibition experiments according to Eq. [9] and [10] (White and Perroux, 1989) at an average pressure head of = -3 cm water using sorption water contents and different values of M in Eq. [9].

 
The water-retention functions of the interior and the skin layer of the intact aggregate samples (Fig. 5) are evaluated from the fitted cut and intact curves (Fig. 4). The hydraulic conductivity of the skin, Kskin, evaluated at d = -28.17 cm is about 12 times smaller than the value of the aggregate interior (Table 8); compare the 6 times smaller value of K of intact versus cut samples (Table 6). The hydraulic resistance (Table 8) of the 0.1-cm skin is 2.4 times larger than that of the 0.5-cm interior of the intact samples. The parameters of the fitted VG-type function for the skin (Table 8) show a shift towards higher values of the water contents, which may partly result from differences in porosities between intact and cut samples (Fig. 4) and partly reflect an effect of higher Corg–content of the skin (Table 1).



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 5. Water-retention functions of interior and skin parts of the intact aggregate samples derived from the arithmetic mean values of 21 aggregate samples of Soil Block 5B. Solid circles represent the difference between the water contents of intact samples and interior (Eq. [13]). The dotted line indicates the fitted VG-type water-retention function of the skin.

 

View this table:
[in this window]
[in a new window]
  		Table 8. Evaluated hydraulic parameters of skin and interior parts of the intact aggregate samples. Values of R and K are evaluated at d = -28.17 cm.

 
Hydraulic conductivity functions for intact and cut aggregates are compared in Fig. 6 with that of the skin layer. The values (White and Perroux, 1989) are fitting closely to the K(hd) functions that were matched using Dirksen-evaluated K values for intact and for cut samples. The K(hd) functions of intact and cut samples are of similar shape, however, K(hd) values of cut samples are generally larger than those of the intact samples. The K(hd) values of the skin layer (Fig. 6) are smaller in the near saturation range above h = -200 cm of water compared with the intact samples and above h = -1000 cm compared with the cut samples. For h < -1000 cm, the K(hd) of the skin layer becomes increasingly larger than K(h) of intact and cut samples (not shown). The relatively small value of the relation of about 17 at water saturation between the interior (cut in Table 6) and the skin layer will not be further interpreted here because of the K(hd) projection from the matching point, as discussed above. For the unsaturated hydraulic conductivity, however, the interior/skin ratio could be used as a first estimate for the dual-permeability mass transfer term parameter. The relation Kinterior(hd)/Kskin(hd) is ranging between 50 and 70 near water saturation, ~10 at h = -30 cm, and drops below unity for h < -1000 cm. In case, the skin layer is smaller than the experimentally defined value of 0.1 cm, the Kinterior(h)/Kskin(h) relation would, of course, increase since the measured hydraulic resistance of the total intact sample (R in Table 8) remains constant. Sullivan and Koppi (1991), for instance, report values of skin thickness for a desert loam soil of 0.02 cm for clay and 0.03 cm for silt coatings, while composite coatings had a thickness of up to 0.1 cm. Assuming, for example, a value of Lskin = 0.05 cm with Linterior = 0.55 cm yields values of Rskin = 1.361 d, Kskin = 0.037 cm d-1, and a Kinterior/Kskin relation of 22.6 at the matching value of d = -28.17 cm.



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 6. Hydraulic conductivity as a function of the matric potential head of intact and cut aggregate samples of Soil Block 5B as well as of the skin of intact samples. The symbols represent hydraulic conductivity values evaluated from data of the imbibition experiments according to Dirksen (1975) using Eq. [2] through [5] and White and Perroux (1989) using Eq. [10]. The VGM-type relative hydraulic conductivity functions were evaluated using desorption water retention and ‘matched’ at Dirksen-evaluated values.

 
For better understanding aggregate skin hydraulic properties, experimental techniques and analyses could be improved, for instance, with respect to volume determination during hydraulic experiments and interactions between mechanical and hydraulic properties of aggregates (Horn et al., 1995), and with respect to inverse modeling of water absorption (e.g., Smunek et al., 1998) and prediction of hydraulic conductivity from water retention (e.g., Kosugi, 1999; Schaap and Leij, 2000). For studying water transfer between the intra- and interaggregate pore domains, it should be considered that aggregate skins in structured soils in situ may possibly exhibit different hydraulic effects than skins of detached individual samples.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
Hydraulic properties of soil aggregate skins can be evaluated by comparing water sorptivity and retention properties of intact with skin-removed (cut) aggregate samples. Soil aggregates skins can significantly reduce water absorption rates of soil aggregates during tension-controlled imbibition. For aggregate samples with intact skin, the effective hydraulic conductivity was 6 times smaller than for cut samples. Similar values were obtained when using the more approximate methods of White and Perroux (1989) instead of the approach of Dirksen (1975). Hysteresis of soil aggregate water retention needs to be accounted for when evaluating the hydraulic properties of the skin layer. The water retention function of the skin layer indicates a somewhat higher porosity and finer texture compared with that of the bulk aggregate which may be correlated with higher organic C content of the skin. The skin hydraulic conductivity was estimated to be about 12 times smaller than that of the interior at d = -28 cm, and was predicted up to 70 times smaller close to water saturation and larger for matric potentials heads below -1000 cm.

As for mineralogical and chemical properties, the results of this study indicate that aggregate skins differ also with respect to hydraulic properties from the bulk soil matrix. Skins may even be regarded as a separate porous domain whose water retention and hydraulic conductivity properties control water transfer between inter- and intraaggregate pore domains in structured soils. The findings may help improving the description of water transfer in dual-permeability models where often a constant transfer term conductivity parameter is used instead of a matric potential dependent function. Future work will be required to further verify the results for different soils and to transfer the skin hydraulic properties obtained for groups of separate aggregates to aggregated soil volumes in situ.


   ACKNOWLEDGMENTS
 
This study was financially supported by the German Research Council (DFG), Bonn (contract Wi 671/9-1). We thank Sigrid Meyer-Windel, University of Halle, Germany, for valuable comments, Harald Grehl, University of Kiel, Germany, for technical assistance, Prof. Dr. P. Widmoser and Dr. B. Lennartz, Kiel, for helpful support during the project, Sarah Ellgen for polishing the English, and Dr. Ed McCoy and anonymous reviewers for constructive comments and suggestions. Parts of the manuscript were prepared during an OECD-fellowship of the first author at the University of Hawaii at Manoa.

Received for publication July 27, 2000.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES