| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Dep. of Soil and Water Sciences, The Hebrew Univ. of Jerusalem, P.O. Box 12, Rehovot 76100, Israel
* Corresponding author(yonachen{at}agri.huji.ac.il).
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONTHEORYMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
eq1) was calculated, assuming that for the reference soil the contact angle eq0 = 0. Increasing the leaching fraction of an initially WRS resulted in a decrease in eq1 values. For an initially wettable soil, increasing wetting–drying cycles with DOM solutions resulted in an increase in eq1. The variations in coseq1 were reflected in a similar ratio for the capillary-pressure–saturation relations (CSR). The prediction from scaling the "reference soil" CSR curve by coseq1 was found to be satisfactory, but more accurate where the effective saturation S <0.5. At S > 0.5, organic matter is more likely to detach and dissolve, changing the properties of the soil solution by altering its surface tension and the soil particles' surface, i.e., eq1.
Abbreviations: CSR, capillary pressure–saturation relations • DOC, dissolved organic carbon • DOM, dissolved organic matter • MED, molarity of ethanol droplet • OM, organic matter • WDPT, water drop penetration time • WRS, water-repellent soil
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONTHEORYMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
The hydrological consequences of a WRS layer are: retardation of infiltration rates (Letey, 1969; Wallis and Horne, 1992; Feng et al., 2001) and associated enhancement of surface runoff (Burch et al., 1989) and soil erosion (Shakesby et al., 2000); an unstable wetting front, which results in fingered, preferential flow paths (Hendrickx et al., 1993; Ritsema and Dekker, 1996; Wang et al., 1998; Carrillo et al., 2000a, 2000b); and hysteresis in water retention (Bauters et al., 1998; Ritsema et al., 1998).
Several methods have been suggested to quantify the degree of soil hydrophobicity (Wallis and Horne, 1992; Letey et al., 2000; Bachmann et al., 2003). The most common ones, however, are the water drop penetration time (WDPT) test (Letey et al., 1962) and the molarity of ethanol droplet (MED) test (Watson and Letey, 1970; King, 1981). Whereas the WDPT test serves as a measure for the stability of soil water repellency (Letey et al., 2000), the MED test is a measure of its instantaneous breakdown and therefore can be used as an indication of its initial degree.
Both of these tests are based on the formation of an apparent initial advancing contact angle (in) >90° at the soil surface. The in can be calculated from measurements of the initial stage of capillary rise or via the Wilhelmy plate method (Bachmann et al., 2003). Reported values of in for natural and hydrophobized WRSs (Woche et al., 2005; Arye et al., 2006; Bachmann et al., 2006) range from 0 to 140°. Therefore, in measurements are also suitable for distinguishing soils that exhibit initial subcritical water repellency (i.e., 0° < in < 90°).
Water repellency depends primarily on the type and quantity of OM in the soil. For a given WRS, however, the expression and extent of its water repellency depends mainly on soil water content (King, 1981; Dekker and Ritsema, 1994; de Jonge et al., 1999; Regalado and Ritter, 2005) and the soil's wetting and drying history (Doerr and Thomas, 2000). The governing mechanism for this dependence is associated with the reconfiguration or reorientation of the amphipathic OM compounds when they interact with water (Ma'shum and Farmer, 1985; Swift, 1989; Horne and McIntosh, 2000; Ellerbrock et al., 2005). When soils are wet, polar groups interact with water molecules, but as they dry and water is lost, polar groups interact with each other and OM presents largely nonpolar groups on its surface.
This mechanism serves to partially explain a scenario in which all of the OM compounds that are present initially interact when the soil dries. Nevertheless, field and laboratory data indicate that hydrophobicity is not necessarily reestablished to its initial level and it may even disappear during, or just after, the soil dries (Doerr and Thomas, 2000; Regalado and Ritter 2005). Moreover, if the WRS layer is subjected to irrigation or rain, some OM compounds may dissolve and leach out of this layer. Therefore, in such common instances, the previously suggested mechanism may not be valid and reestablishment of the soil's water repellency to its initial degree is doubtful.
The effect of hydrophobicity continues to act after water penetrates the soil during infiltration and when water is retained in the soil. In fact, one of the primary effects of WRS is a reduction in the rate of water infiltration. Even in soils that appear to wet "normally" (i.e., subcritical water repellency), the infiltration rate into some WRS is greatly reduced, at times even by an order of magnitude (Wallis et al., 1991). The infiltration-rate pattern observed with WRS is different from that in wettable soils (Letey et al., 1962; Tillman et al., 1989; Clothier et al., 2000; Feng et al., 2001), and shows a slow rather than high initial phase of infiltration and an increase rather than decrease with time. Letey et al. (1962) attributed this behavior to the OM coating dissolving into the infiltrating water and reducing the surface tension (
L) of the soil solution.
Soil solution contains varying amounts of DOM, mostly composed of complex macromolecules of high molecular weight, namely humic substances (HS); (Kalbitz et al., 1999). The simultaneous presence of both hydrophilic and hydrophobic sites in HS suggests that, like synthetic surfactants, these compounds will exhibit significant surface activity (Anderson et al., 1995). Dissolved organic matter originating from humic acids, fulvic acids, and soil extracts (Chen and Schnitzer, 1978; Tschapek et al., 1978; Anderson et al., 1995; Yates and von Wandruszka, 1999) exhibit significantly lower L values than pure water (L = 72 mN m–1). This reduction has been found to depend on the origin and concentration of the DOM, pH, ionic strength, and the valence of metal ions present in the soil solution. Hence, during infiltration through a WRS layer, OM compounds that are dissolved in the soil solution leach out, reducing the DOM concentration in the soil solution and thus increasing L.
The relationship between water content and soil hydrophobicity has been studied mainly using measurements from WDPT and MED tests. These measurements have been performed on moist soils under field conditions before and after air drying or oven drying or after setting a water content by adding different amounts of it to dry soils (King, 1981; Dekker and Ritsema, 1994; Dekker et al., 2001; Goebel et al., 2004; Regalado and Ritter, 2005).
Doerr and Thomas (2000) challenged the established view of a two-way soil moisture–hydrophobicity. In their study, they examined the temporal aspect of hydrophobicity relative to soil moisture during wetting and drying cycle. The results of their study clearly demonstrated that hydrophobicity cannot be assumed to become reestablished. They noted, however, the limitation of the WDPT test in discriminating between different levels of highly hydrophobic soils due to impractical measurement times. The limitations of the WDPT and MED tests must also be considered for soils with initial low levels of hydrophobicity or moist soils when 0 < in < 90°.
The primary purpose of this study was to explore the reestablishment of hydrophobicity via the measurements of the equilibrium (static) contact angle and the capillary pressure–saturation relations. The specific objectives of this study were to: (i) examine the reestablishment of hydrophobicity of two naturally hydrophobic soils after wetting with different leaching fractions; (ii) examine the establishment of hydrophobicity of wettable soils after different wetting–drying cycles with two different DOM solutions; (iii) measure the capillary pressure–saturation relationship (CSR) for these soils; and (iv) apply the concept of "similar soils" (see below; Miller and Miller, 1956) to the measured CSR curves.
|
THEORY |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONTHEORYMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
in can be calculated from the early stage of capillary rise according to Washburn's equation (Washburn, 1921):
 | [1] |
[M T–1 L–1] is the viscosity of the liquid.
When in is calculated for soils, it is commonly assumed that during the measurements there is no interaction between water and the soil-particle surface. Since the measurements of in are rapid (in the order of 10–20 s) and changes in in are limited, this assumption may be justified (Roy and McGill, 2002; Arye et al., 2006). For a larger time scale, however, this assumption no longer holds.
When an equilibrium state is achieved (i.e., dh/dt = 0), the maximum liquid height (Heq) is
 | [2] |
eq (°) is the equilibrium (static) contact angle, 
[M L–3] is the density of the liquid, and g [L T–2] is the acceleration due to gravity.
From Eq. [2], when water is retained in the soil and any interaction between water and soil has reached an apparent equilibrium, the measure for soil hydrophobicity is deduced to be eq rather than in.
The difficulty in calculating eq from a measured Heq and known properties of the wetting liquid (L and ) is the unknown value of the equivalent radius (r) of the soil's capillaries. The common solution (Letey et al., 1962; Watson and Letey, 1970; Bachmann et al., 2003) is to use a reference liquid that is assumed to form eq = 0. Hence, for a given soil, the following relation can be obtained:
 | [3] |
0 = 22.4 mN m–1, 0 = 0.789 g cm–3), which is assumed to wet all soils with eq = 0 (Letey et al., 1962). Hence, from measured values of Heq1 for water and Heq0 for ethanol, one can calculate eq1.
An alternative solution for calculating eq1 from Eq. [3] is to use a "reference soil" that is assumed to form eq = 0 when water is the wetting liquid (Morrow, 1976; Nakaya et al., 1977; Bauters et al., 1998; Bachmann et al., 2002). Namely, the reference soil is assumed to have the same pore-size distribution as the WRS (i.e., r0 = r1) but different surface properties because of the absence of a hydrophobic organic coating on its surface.
In studies in which the WRS is artificially created by adding hydrophobic substances such as humic acid (Nakaya et al., 1977) or octadecyltrichlorasilane (Bauters et al., 1998) to a quartz sand, the reference soil is the untreated quartz sand. For naturally occurring WRS, however, a reference soil needs to be established by removing all of the OM compounds, which impart the hydrophobic properties.
For the concepts of both "reference liquid" and "reference soil," the assumption eq0 = 0 is difficult to demonstrate directly from a static or dynamic capillary-rise experiment. Moreover, in the case of a reference soil, one has to explicitly assume that 0 = 1. Namely, in both soils the retaining liquid is assumed to be water (L = 72 mN m–1). From our various measurements (data not shown) of L for the first pore volume displaced by distilled water from different WRS, however, we found that L can be as low as 50 mN m–1. If we consider, for simulation purposes, that this is the value for 1, then the ratio 0/1 = 1.44. Accordingly, an increase of about 44% in the calculated value of coseq1 will be obtained, which will result in a lower value for eq1 than that using the assumption of 0 = 1.
If our goal then, is to calculate eq1 for a given soil, then deviation from the above assumptions should be considered, and the calculated value of eq1 be regarded as only a relative value. Nevertheless, this relative value can provide a useful measure for soil hydrophobicity since it accounts for both the solid and soil-solution properties when water is retained in the soil.
Under equilibrium conditions of capillary rise (where the flow is zero or negligible), the hydraulic potential is equal everywhere in the system and hence, for any point along a soil column, the capillary height (H) is equal to the absolute value of the capillary pressure (
). Therefore, water content (
) distribution along a soil column will depend on the value of , where saturated water content (s) is located at the water table ( = 0) and the residual water content (r) is located at the maximum liquid height of capillary rise ( = –Heq). Therefore, from the measurements of along a soil column in equilibrium, one can derive the imbibition CSR curve.
Given the CSR curve for a WRS and its reference soil, Eq. [3] can be further expanded to examine the "similarity" (Miller and Miller, 1956) of their hydraulic properties. Namely, the capillary pressure of the reference soil can be scaled with the calculated coseq1 to predict the CSR for the WRS. It should be noted, however, that a unique value is assumed for eq1 regardless of the water content across the soil column.
Two soils can be assumed hydraulically similar if the relationship between the scaled capillary pressure and the effective degree of saturation (S) are not statistically different (Lenhard and Brooks, 1985), S being defined as
 | [4] |
Theoretically, one can select any point along the 1(S) and 0(S) imbibition curves and calculate eq1, which can then be used to scale the rest of the measured points of 1(S). For artificially hydrophobized quartz sand, Bauters et al. (2000) suggested that the best identifiable point is the "water-entry value," which, for wettable soils (i.e., in1 < 90°), can be identified from the "knee" in the CSR curve near saturation, and for WRS (i.e., in1 > 90°), from the "knee" in the curve approaching air-dried values. If two soils are hydraulically similar, however, any point along the CSR curve should do. In this study, for all soils, we selected the measured Heq to calculate coseq1, independent of the resultant measured CSR curve.
The scaled 0(S) curve, which is used to predict the 1(S) curve, can be given by
 | [5] |
From Eq. [5], it can be seen that even if the calculated value of eq1 is treated as a relative value, i.e., disregarding differences between 0 and 1, it is physically significant as a scaling factor. Hence, this scaling factor accounts for both liquid and solid properties, and is eventually calculated from the measured Heq1/Heq0 ratio.
To describe the experimental CSR curves, we adopted the widely used van Genuchten (1980) parametric equation (further referred to as the VG equation):
 | [6] |
:
 | [7] |
, n, and m are empirical fitting parameters affecting the shape of the CSR curve and derived from it.
To account for differences in eq or L, the VG equation can be modified by incorporating Eq. [5] into Eq. [7] (the VG equation):
 | [8] |
0, n0, and m0 are empirical fitting parameters calculated for the reference soil. |
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONTHEORYMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
All soils were air dried and passed through a series of sieves (2, 1, and 0.5 mm) before use. The 0.5-mm fraction was further analyzed for particle-size distribution (Table 1), hygroscopic water content (H) by heating at 105°C for 24 h, and OM content by loss-on-ignition at 400°C for 8 h (Ben-Dor and Banin, 1989).
|
Under natural conditions, the wetting of a WRS layer by rain or irrigation water normally results in leaching of the DOM solution from this layer. This, in turn, may alter the initial hydrophobic nature of the soil to some degree, since some hydrophobic organic compounds can leach out. To simulate such a scenario, the wetting phase of the naturally occurring WRS was examined by leaching these soils at different rates.
We used a 10-cm-diameter, 10-cm-high Buchner funnel, with a glass filter placed at the bottom to retain the soil and allow the soil solution to leach out. Since in the preliminary experiments we noticed that attempts to leach the air-dried WRS result in preferential flow, leaving a significant portion of the soil fraction unleached, we removed the initial effect of water repellency (the cause for the preferential flow) by first wetting the soils. An air-dried soil sample (500 g) was weighed into a plastic bag and 50 g of distilled water added. The soil and water were gently mixed until homogeneous wetting was observed. The wetted soil was packed into the Buchner funnel to a depth of 5 cm and covered with filter paper to achieve uniform wetting. A Marriott bottle was used through a port located above the soil surface to add distilled water and maintain a constant 3-cm depth of ponding water.
In the first treatment, before the soil solution started to leach from the bottom of the funnel, the ponded water was rapidly removed and the leaching stopped. Hence, the soil was assumed to be saturated, without any DOM loss. This treatment is referred to as L0. The same procedure was conducted for the subsequent treatments, except that the soil solution was allowed to leach to an equivalent volume of about 1, 2, and 4 pore volumes, referred to as L1, L2, and L4, respectively. During leaching, soil solution was collected in increments of 50 mL (equivalent to about 0.5 pore volumes) and measured for L using a Fisher semiautomatic Model 21 Tensiomat, and for dissolved organic carbon (DOC) using a total organic C analyzer (Skalar-Formacs Combustion TOC Analyzer, Skalar, Breda, the Netherlands).
To limit the degradation of OM by microorganisms, we first dried the soils at 65°C for 24 h and only then left them to reach an air-dried state at room temperature. These samples were also measured for OM and H.
The reestablishment of wettability of the completely wettable dune sand was examined after the addition of the two different DOM solutions. The extraction procedure was performed as decribed for the WRS, except that the leaching was stopped after displacing about two pore volumes. This procedure was repeated until the desired volume was obtained. The solution was then centrifuged and stored at 4°C until use.
The DOM solutions were added to the dune sand at three different wetting–drying cycles. In each cycle, 200 mL of DOM solution were added to a plastic container containing 500 g of air-dried dune sand spread in about a 2-cm layer. The dune sand and the DOM solution were first mixed manually with a glass rod until the sand was uniformly wetted and then the container was placed on a rotary shaker for 20 min. The sand–DOM mixture was then left to dry at 65°C for 24 h, then at room temperature for about 1 wk, until completely air dried. The samples were resieved through a 0.5-mm sieve. The different wetting–drying cycles were achieved by repeating this procedure one, two, and four times and are referred to as A1, A2, and A4. The addition of DOM extracted from compost and leonardite is referred to as COM and LEO treatment, respectively.
Establishment of a Reference Soil
The basic assumption of this study is the existence of a reference soil assumed to have eq = 0. It is also assumed that the OM present as a coating on soil particles or mixed between them imparts the soil's hydrophobic properties.
A common procedure for determining the OM content in soil is measuring its loss on ignition. Various regimes of heating time and temperature have been proposed for the removal of OM from soils with minimal effect on mineral composition (Davies, 1974; Goldin, 1987; Ben-Dor and Banin, 1989). The loss of weight due to ignition can be divided into the following stages: hygroscopic water removal (50–105°C); OM removal (100–400°C); dehydroxylation of phyllosilicates (200–700°C); and decarboxylation of carbonates (700–1000°C). Note that quartz and feldspars are not known to lose weight during heating up to 400°C (Ben-Dor and Banin, 1989).
For arid and semiarid soils, Ben-Dor and Banin (1989) showed that further changes in weight loss after 8 h of ignition at 400°C are marginal. Therefore, to establish a reference soil for the natural WRS, we adopted this procedure to remove the OM. Since the dune sand had no OM, it was used as a reference soil for the artificially made WRS, with no further treatment.
Capillary Pressure–Saturation Measurements
The main imbibition relationship between water saturation and capillary pressure was obtained by the capillary-rise experiment. To be able to sample the soil at various heights at the end of the imbibition time, a 50-cm transparent polyethylene (PE) tube (i.d. = 15 mm, o.d. = 19 mm) was used. The outside of the PE tube was protected by a rigid polyvinyl chloride housing pipe (i.d. = 19 mm, o.d. = 20.7 mm), which was cut to the desired lengths, such that it would stand upright. The bottoms of the tubes were sealed with cheesecloth to retain the soil and allow water to infiltrate from the bottom. Air-dried soil samples were packed into the PE tube by continually filling the tube through a funnel from the top. The soil column was gently tapped until no further change in soil height was observed and the bulk density was determined.
The soil columns were immersed in a container filled with distilled water. The water table was maintained constant by a Mariott bottle connected to it with a Tygon tube. The immersion depth for the EQL and ORC soils was 10 cm and for the artificial WRS, 5 cm.
At the end of the apparent equilibrium time, which was set at 48 h, the soil column was rapidly placed horizontally to minimize drainage from the bottom. The exposed surface area of the PE tube was removed using a scalpel and the soil along the column was then collected in 1-cm segments. Each segment was measured gravimetrically for water content. All experiments were conducted in duplicate.
|
RESULTS AND DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONTHEORYMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
The effect of OM content on b and H is shown in Fig. 1
. Both properties were linearly correlated to OM content (R2 = 0.91 for b and 0.97 for H). Even though the different treatments imposed some changes on the OM content, for a given soil these changes resulted in a limited range of both H and b values. The increasing trend of H as a function of OM content can be explained by the significantly higher vapor adsorption capacity of OM substances (Chen and Schnitzer, 1976).
|
b values, whereas for the COM and LEO soils, variations in b were insignificant; for the ORC soil, b rose from 1.12 to 1.42 g cm–3 and for the EQL soil, from 1.30 to 1.40 g cm–3. This appears to indicate some structural differences between the reference soil and the WRS. Given the rates of variation in b as a function of OM content and the fact that after each treatment all soils were resieved to 0.5 mm, however, the increase in b can be explained by a dilution effect of OM on b values rather than aggregate formation, which can affect pore-size distribution but was not observed.
Surface Tension and Dissolved Organic Carbon Concentration of Soil-Solution Leachates
The hydrophobic nature of the DOM solution leachates can be assessed by measuring their surface tension (L). Generally, the lower the L, the higher the surface-active nature of a solution, and therefore it can be defined as more hydrophobic. The result of L as a function of the cumulative leachate volume (V) or DOC concentration for both ORC and EQL soils is presented in Fig. 2
.
|
L vs. V curves, one can evaluate the hydrophobic nature of the leached solutions. For both soils, the lowest values of L (60–65 mN m–1) were obtained for the L1 treatment (0–200 mL). Therefore, these L values can be assumed to prevail in the soil solution if no leaching from the soil layer has occurred, and it can be further assumed that for these soils, the magnitudes of the L values for the L0 treatment are similar.
With increasing volume of water percolating through the soil layer, L values increase and approach the value of L for pure water (72 mN m–1). For the L2 treatment (0–400 mL), the obtained L values (63–67 mN m–1) were still distinctly lower than water, possibly reflecting the L values in the soil solution under the L1 treatment. It is only for the L4 treatment (after cumulative leaching of 900 mL) that we observed L values similar to water. This observation is in agreement with the prediction of eq1 based on Eq. [3] while assuming 1 = 0.
The main factor controlling L in the soil solution is the DOC concentration, as indicated by the L vs. DOC curves (Fig. 2), which exhibit a quasisurfactant behavior. Namely, the higher the DOC concentration, the lower the L, with a decreasing effect of DOC at increasing concentrations. Nevertheless, L is not approaching a constant value, as one would expect from a surfactant.
The decreasing DOC concentration, due to its transport in and out of a given soil-layer depth, provides a quantitive explanation for the increasing values of L in the percolating water. The hydrophobic nature of the surface-active compounds may be an additional controlling factor, however. Namely, the same DOC concentration in different DOM solutions can result in different L values, as can be seen by comparing the COM and EQL soils. Whereas the lowest value of L (
60 mN m–1) for the EQL soil is obtained at a DOC of 192.3 mg L–1, a similar value is obtained for the ORC soil at a DOC of 90 mg L–1. It should be noted, however, that other properties, such as pH, ionic strength, and the valence of metal ions, could alter the conformation and orientation of DOM compounds and, consequently, L.
Reestablishment of Hydrophobicity: Effect on Contact Angle
The values of Heq and eq1 as calculated from Eq. [3] are presented in Fig. 3
for the two soils subjected to leaching (ORC and EQL) and the two soils subjected to the addition of water-extracted DOM solution (COM and LEO). For the natural soils, the rightmost column refers to the initial WRS, which has not been subjected to any treatment, and for all soils, the leftmost columns show the respective reference soil.
|
The results clearly demonstrate that increasing the water volume applied to the soil surface resulted in a reduction in eq1 for both ORC and EQL soils when rewetted. The leftmost points, attributed to the reference soil, are constrained by the assumption eq1 = 0.
Since we did not find any significant differences in the water content at saturation (as measured from the weight differences of the Buchner funnel before and after leaching) for either soil at the end of each leaching treatment, this effect on the reestablishment of soil hydrophobicity can be considered similar among treatments. Hence, the reduction in eq1 can be mainly attributed to hydrophobic DOM compounds that leached out of the soil. In the case of L0, however, since we assume that all initially present OM compounds are involved in reestablishing hydrophobicity, the reduction in eq1 may be attributed to a mechanism related to the reorientation or reconfiguration of these compounds. Nevertheless, hydrophobicity did not resume its initial level in either treatment.
Moreover, for the L0 treatment in the ORC soil, the reduction in eq1 followed a general trend that could be attributed to the absence of hydrophobic compounds due to leaching. In the case of L0 treatment for the EQL soil, a sharp decrease in eq1 was obtained. After the different leaching treatments, however, hydrophobic levels were restored, albeit to lower levels than in the untreated soil.
As already mentioned, in calculating eq1 we assumed that 1 = 0 = water. From Fig. 2, however, we see that in the soil solution, L values are significantly lower than those of water. Moreover, a well-documented phenomenon based on field and laboratory studies is that DOC concentration increases following rewetting after a dry period (Kalbitz et al., 1999). As shown (Fig. 2), an increase in DOC concentration resulted in a decrease in L. Therefore, the values obtained for L in the wetting phase are likely to further decrease when the soils are rewetted after a drying phase.
If the organic compounds detached from the soil-particle surface by wetting or those dissolved from particulate OM are present in the soil solution during the rewetting phase, then 1 < 0 and, according to Eq. [3], a reduction in the calculated eq1 will be obtained. After leaching and removal of these compounds, 1 values increased (Fig. 2) and consequently, eq1 values increased as well.
This suggested mechanism may imply the involvement of different types of solid phases in the two soils. Namely, if hydrophobicity of the EQL soil is restored after leaching, then the fraction of solid OM, which gives it its hydrophobic nature, is not readily dissolved in water. Therefore, after leaching of the DOM fraction, the predominant factor is the solid OM. Accordingly, for the ORC soil, the predominant OM fraction that initially imparts hydrophobicity is the DOM.
Regarding the DOM-addition treatments (COM and LEO), the results in Fig. 3 should be observed from left to right: here, we are simulating a scenario in which an initially hydrophilic sand exhibits an increasing amount of DOM with an increasing number of wetting–drying cycles (A1, A2, and A4). For both soils, increasing amounts of added DOM resulted in increasing eq1 values. The eq1 values obtained for the COM treatments, however, were higher than for the LEO treatments. This can be explained quantitatively by the higher amount of DOC extracted from the compost (952 mg L–1) vs. the leonardite (275 mg L–1). Moreover, given the severe hydrophobic nature of leonardite, its water extract is likely to contain DOM with a small hydrophobic/hydrophilic ratio relative to the water extract of the compost. It can be seen that even after the A4 treatment, eq1 obtained for the LEO addition is significantly lower (eq1 = 24°) than that from the COM A1 treatment (eq1 = 46°). Therefore, in addition to the total amount of DOM added, the hydrophobic/hydrophilic ratio of its compounds can play a role in imparting hydrophobic properties to an initially wettable soil; and seems to be more relevant than the total amount of DOM.
In summary, the DOM solution seems to play a major role in the short-term reestablishment of hydrophobicity in soils owing to its surface-active nature, which can result in a decrease in L compared with water, or due to the kinetics and specific kind of adsorption–desorption to the soil particle's surface. When it is leached out of the soil layer, however, the absence of DOM seems to be the main factor controlling the reestablishment process.
Capillary Pressure–Saturation Relations
We have shown that the level of reestablishment of soil hydrophobicity can be assessed from a calculation of eq1. Ultimately, however, eq1 is calculated from the measured ratio Heq1/Heq0. According to Eq. [3], one cannot explicitly distinguish which effect is predominant in controlling Heq1—the intrinsic reduction in 1 or the reduction in eq1. The net effect, however, will eventually be reflected in Heq1 and consequently, in the calculated value of eq1, which should be considered as the outcome of the net effect, rather than the intrinsic eq.
Given the physical significance of eq1, it needs to be further examined whether this value can be related to any water content in the imbibed soil. The questions whether chemical or physical changes of both solid and liquid phases, due to leaching or addition of DOM solution, will result in equivalent variations in eq1 for any given water content need to be specifically addressed. To determine this, further information on the reestablishment process can be derived from the CSR of the soils. Namely, if from one scaling factor (i.e., coseq1) one can accurately predict the hydrophobic CSR curve from the reference soil CSR curve, then eq1 can be considered as a single value.
In Fig. 4 and 5 , the CSR curves are presented for the leaching and addition treatments, respectively. Nonlinear least-squares optimization (SigmaPlot Version 9, Systat Software, San Jose, CA) was used to fit the VG parameters to the CSR curves of all soils. For scaling purposes, however, we used only the VG-fitted line of the reference soil. The purpose of fitting the VG parameters to the rest of the soils will be discussed below.
|
|
, n, and m) and r and s (with and without fixing) we could not establish a good fit in the domain where (or S) approaches the air-dry water content (H) at the maximum capillary height (Heq). Therefore, to fit the VG equation to the reference soil, we used r as an additional fitting parameter with the constraint r < 0, m = 1 – 1/n, and a fixed value of s. This procedure provides an appropriate description of the measured CSR curves, even though the optimization of the negative r was found to be insensitive.
The optimization with unfixed values of r > 0, m, n, , and a fixed measured value of s (Table 2) compared with unfixed values of r < 0, n (m = 1 – 1/n), , and a fixed measured value of s (Table 2) is demonstrated for each reference soil (upper graphs in Fig. 4 and 5). For the first optimization (r > 0), the r values obtained for all references soils were zero. Consequently, S was calculated by substituting r = 0 and the measured value of s. For the second optimization, (r < 0), S was calculated from the r value obtained from the VG model at the point where 0 = Heq0 and the measured value of s.
|
The measured CSR curve and the VG equation fitted to the reference soils were scaled by coseq1 (Table 2) according to Eq. [5] and [8], respectively, and both are presented in Fig. 4 and 5. On each scaled curve, two points have to be fitted to the hydrophobic CSR curve: (S = 0) and (S = 1). For the rest of the scaled points, an agreement between the two curves is not dependable.
The results presented in these figures show that regardless of the treatment applied to the natural soils or the dune sand, adequate agreement can apparently be achieved from only one scaling factor (i.e., coseq1). In some instances, however, when approaching the saturated domain, the agreement between the two curves decreases somewhat. With respect to the natural WRS (Fig. 4), the disagreement between the two curves can be clearly observed with the L0 treatments for both ORC and EQL soils. For the dune sand, in most cases, the differences between the two curves occurred in a narrower range of effective saturation, approaching S = 1. In the domains where 1(S) > 0(S), a higher scaling factor should be considered, and a lower one where 1(S) < 0(S).
In an attempt to generalize these observations for all CSR curves, the scaling factor (coseq1) was calculated from the ratio 1(S)/0(S) at a given S. Since the measured points of 0(S) are not always superimposed on the measured points of 1(S) at a given S, the VG-fitted line was used. For each soil, the calculated values of coseq1 were normalized to the value initially used for scaling (i.e., Heq1/Heq0). The results of the normalized coseq1 as a function of S [excluding (S = 0) and (S = 1)] are given in Fig. 6
.
|
eq1 provides a measure for the validity of the scaling for every point of all CSR curves. Based on the resultant pattern, we suggest that the whole range of saturation degrees can be roughly divided into two domains: Domain 1, in which S ranges from 0 to 0.5; and Domain 2, from 0.5 to 1. In Domain 1, the normalized coseq1 is slightly scattered around 1 with relatively small deviations (excluding the EQL soil). Namely, in Domain 1 there is good agreement between the scaled reference soil CSR curve and the hydrophobic CSR curve. When S > 0.5, however, even though some points can still be considered close to 1, a gradual increase in scattering around 1 is shown, reaching its maximum deviation near saturation.
The points below and above the normalized coseq1 = 1 may imply higher and lower than expected contact angles. This suggests that in Domain 1, the capillary pressure depends mainly on the contact angle. In Domain 2, OM is more likely to detach from the soil particles and be dissolved by the water into the soil solution, This, in turn, may cause a decrease in L and eq1. It is the net effect of these two values, however, that we are interpreting as a lower or higher contact angle. The above-suggested mechanism can be clearly observed from the results of the COM and LEO soils. For these soils, all OM was initially extracted by water and, therefore, during rewetting was water soluble.
Correlations between van Genuchten Fitting Parameters and the Cosine of the Contact Angle
The VG model (Eq. [5]) is commonly used to model water retention and transport in variably saturated porous media. For a measured () obtained from the capillary-rise experiment, the measured () was shown to deviate from the model when approaching air-dried soil. Below this point, however, the VG model fits well with the measured (). Nevertheless, attempts we made to fit the VG model to the measured () resulted in sensitive and stable values of the fitted parameters with independent n, m = 1 – 1/n, , and fixed values of the measured s and r (Table 2). From the low standard errors of both n and (P << 0.01), we deduce that the fitted model is sensitive to the fitted parameters. These conditions provide statistically significant and reliable parameters that can now be correlated with coseq1.
From the results, we deduce that the VG parameter n is insensitive to variations in eq; its normalized range, however, is relatively low (0.8–2). On the other hand, the VG parameter –1 exhibits a significant linear relationship with coseq1.
The relationship of –1 vs. coseq1 and the ratio 0/1 vs. coseq1 are presented in Fig. 7
. The parameter –1 and the ratio 0/1 reflect an individual soil and a generalization for "similar" soils, respectively.
|
0/1 vs. coseq1, we suggest that 0/1 = coseq1. A similar relationship was assumed by Bachmann et al. (unpublished data, 2006) and Deurer and Bachmann (unpublished data, 2007), who implemented it for modeling water movement in WRS. The validity of their assumption is clearly demonstrated from the results of this study. |
CONCLUSIONS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONTHEORYMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
in). Theoretically, the values of in may range from 0 to 180°. When water is retained in the soil pores, the measure for soil hydrophobicity is the equilibrium contact angle (eq < 90°), which can be calculated relative to a "reference soil," under the assumption eq = 0 for this soil.
By using eq as a measure of the degree of soil hydrophobicity, we showed that for the soils examined in this study, reestablishment to its initial level could not be achieved when rewetting in the short term.
The decrease in soil hydrophobicity was found to depend on the amount of DOM leached out of the relevant soil layer. This can be explained in two ways: (i) leaching out of the soil profile of hydrophobic compounds, which will result in lower levels of hydrophobic surfaces when the soil is dried; and (ii) an increase in L values following the reduction in DOM concentrations, which can be interpreted as a reduction in the calculated eq.
When DOM is added to soils, it is more likely that the degree of hydrophobicity will increase, depending on the amount and the type of DOM added.
The variations in coseq values due to leaching or addition of DOM will be reflected in a similar manner in the capillary pressure when rewetting. Hence, from the measurements of Heq for a target WRS and "similar" hydrophilic soil, one can predict the variation in the CSR for the target soil. Moreover, for modeling purposes, one can assume that the ratio of the VG parameters 0/1 = coseq1. This relation seems to be more accurate where the effective saturation ranges from 0 to 0.5. Above 0.5, OM is more likely to detach and dissolve into the soil solution, and therefore, variations in the properties of both soil particles and soil solution will be more pronounced.
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONTHEORYMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Received for publication June 22, 2006.
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONTHEORYMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  R. Bashir, J. E. Smith, and D. F. Stolle Surfactant-Induced Unsaturated Flow: Instrumented Horizontal Flow Experiment and Hysteretic Modeling Soil Sci. Soc. Am. J., September 30, 2008; 72(6): 1510 - 1519. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Bachmann, M. Deurer, and G. Arye Modeling Water Movement in Heterogeneous Water-Repellent Soil: 1. Development of a Contact Angle Dependent Water-Retention Model Vadose Zone J., August 1, 2007; 6(3): 436 - 445. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
