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


     

Published online 27 October 2005
Published in Soil Sci Soc Am J 69:1955-1966 (2005)
DOI: 10.2136/sssaj2005.0060
© 2005 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow An erratum has been published
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 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 Web of Science (7)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Regalado, C. M.
Right arrow Articles by Ritter, A.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Regalado, C. M.
Right arrow Articles by Ritter, A.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Regalado, C. M.
Right arrow Articles by Ritter, A.
Related Collections
Right arrow Hydrophobicity
Right arrow Volcanic Soils
Right arrow Water repellency

Soil Physics

Characterizing Water Dependent Soil Repellency with Minimal Parameter Requirement

C. M. Regalado* and A. Ritter

Instituto Canario de Investigaciones Agrarias (ICIA), Dep. Suelos y Riegos, Apdo. 60 La Laguna, 38200 Tenerife, Spain

* Corresponding author (cregalad{at}icia.es)


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Soil water repellency varies nonlinearly with soil moisture, describing a repellency versus water content curve. Despite this water dependence, soil repellency is usually characterized by repellency indexes measured at fixed soil water content. We provide a statistically robust yet simple method to determine water repellency at a range of soil water contents, by proposing alternative water-dependent soil repellency parameters derived from the molarity of an ethanol droplet (MED) test. Useful correlations are found between some of these proposed repellency parameters, such is the case of the integrated area below the repellency curve, which is found to be strongly correlated with the soil water content at minimum repellency. Consequently, an efficient strategy for describing the repellency versus water content curve is designed based on a combination of parameter interdependence and a minimum number of determinations necessary. Following this strategy, only 27 samples are sufficient (p < 0.05) to characterize the water-dependent repellency of the studied soil. The integrated area below the repellency curve is proposed as the key index for repellency characterization. Furthermore, a new combined parameter IRDI (Integrative Repellency Dynamic Index) is defined as a measure of the average water dependent repellency. This allows simple comparisons of the whole water repellency behavior among different soils by coalescing the repellency curve into a single characteristic value.

Abbreviations: CDF, cumulative distribution function • CV, coefficient of variation • IRDI, Integrative Repellency Dynamic Index • MED, molarity of an ethanol droplet • OM, organic matter • PP, probability plots • SD, standard deviation • WDPT, water drop penetration time


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
MOST SANDY, LOAMY, and clayey textured mineral soils and those of volcanic origin exhibit water repellency to some extent (Wallis and Horne, 1992; Doerr et al., 2000; Jaramillo et al., 2000). Water repellency may be caused by lowered surface free energy of the soil particles (Roy and McGill, 2002), due to organic coating by amphiphilic substances. These substances may originate from vegetation cover, microorganism activity, humic acids or lipids (Wallis and Horne, 1992). Investigating soil water repellency is important because it affects many other soil processes, such as infiltration (Wallis et al., 1990; van Dam et al., 1996), solute transport (Bauters et al., 2000), preferential flow (Jaminson, 1945; Ma'shum and Farmer, 1985; Wallis and Horne, 1992), overland flow, and soil erosion (Shakesby et al., 1994).

Wallis and Horne (1992) and Letey et al. (2000) reviewed many different measuring techniques and approaches used to characterize water repellency. Among these techniques, the water drop penetration time (WDPT) test (Letey, 1969) and the MED test (Roy and McGill, 2002) are suitable for moderately repellent soils. The former is a measure of hydrophobicity persistency, while the latter gives an indication of the degree of repellency. For slightly water repellent soils, the capillary-rise method (Letey et al., 1962), the tension infiltrometer (Tillman et al., 1989), and more recently the Wilhelmy plate method (Bachmann et al., 2003) can measure subcritical repellency (Tillman et al., 1989). Based on these methods, different repellency indexes have been derived (Letey et al., 2000).

Water repellency varies nonlinearly with the soil water content (King, 1981; Wallis et al., 1990; de Jonge et al., 1999; Goebel et al., 2004). Hence, repellency indexes are to be considered water dependent. Despite the interdependence between soil water content and hydrophobicity, repellency parameters are usually measured at one extreme water content only. Furthermore, there is no agreement about the parameters or indexes necessary for a complete characterization of the water dependency of soil repellency. In general soils are wettable close to saturation, becoming water repellent up to a maximum as moisture content decreases. After this maximum water repellency diminishes monotonically with decreasing water content, or rises again to a second local maximum (de Jonge et al., 1999; Goebel et al., 2004). The origin of this nonlinear behavior is not well understood, although some proposed hypotheses speculate about an enhanced microbial activity with increasing relative humidity, that is, soil water content (Roberts and Carbon, 1971; Jex et al., 1985); molecular conformational changes in the organic matter responsible for hydrophobicity (Wallis et al., 1990); the attachment/detachment of hydrophobic molecules from the soil mineral particles as water content varies (Doerr and Thomas, 2000); disruption of mineral and organic hydrophobic bonds by the energy release that comes from vapor condensation, and hence the displacement of hydrophobic moieties into soil pores (Doerr et al., 2002); reduction of soil particles surface free energy due to the formation of thin water films or the interaction between water and hydrophobic organic molecules (Derjaguin and Churaev, 1986; Goebel et al., 2004).

Repellency parameters are selected by the physical meaning of a particular soil water repellency aspect (surface tension, contact angle, or surface free energy of the solid) that is to be characterized. A thorough investigation of the statistical properties of these repellency parameters, specifically normality, variability, and minimal number of determinations, has not been performed up to date. Such properties are desirable if for example, soil water repellency data are to be included for extending solute transport, overland flow and erosion simulation models. In addition, repellency is usually measured at fixed water content, without considering its soil moisture dependency. Thereby, it would be useful to derive parameters from the repellency versus water content curve that integrate the water dependent behavior of soil repellency, which must be considered as a dynamic property. In this paper we discuss parameters derived from the MED test, because it provides physically meaningful data such as the contact angle or the surface free energy of the solid particles of the soil.

The objectives of this paper are: (i) to propose both static (water independent) and dynamic parameters to quantify the soil water repellency curve; (ii) to investigate their statistical properties and interdependency, determining the repellency parameters, which give the lowest variability and, based on these results, (iii) to provide an efficient strategy to fully characterize the repellency vs. water content curve with the smallest number of replicates.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Study Area
The area of study is a 43.7-ha watershed in the Garajonay National Park, La Gomera Island, Spain (Fig. 1a, b) . It is covered by a subtropical mature mountain cloud forest mainly composed of species of the Laureaceae family partly covered by epiphytic bryophytes (Golubic, 2001). The watershed altitude ranges from 1090 to 1300 m above sea level with steep topography (slopes range between 10 and 40%) (Ritter and Regalado, 1999). The National Park is situated under the influence of a stratocumulus cloud layer caused by the trade wind temperature inversion that produces fog as it reaches the central high plateau of the island. These fog conditions provide high relative humidity and small temperature fluctuations for almost all of the year. Mean annual temperature and relative humidity range from 13 to 15°C, and 75 to 90%, respectively, while annual rainfall is between 600 and 800 mm. Soils are of volcanic origin and may be classified as Fulvudand. Their andic character confers particular physicochemical and hydraulic properties, such as high permeability, low bulk density, large microporosity, and water retention capacity (Warkentin and Maeda, 1980).



View larger version (81K):
[in this window]
[in a new window]
  		Fig. 1. (a, b) Location of the experimental site and sampling design: (c) random sampling for soil physicochemical properties, and (d) uniform for molarity of an ethanol droplet (MED) determinations.

 
Sampling Design
A preliminary soil physicochemical characterization was performed with samples (n = 32) taken randomly at four depths: 0 to 0.03, 0.03 to 0.23, 0.23 to 0.43, 0.43 to 0.63 m (Fig. 1c). Molarity of an ethanol droplet tests used samples from a second sampling (n = 140) consisting of a regular rectangular grid within the watershed (Fig. 1d) and collected at a soil depth of 0 to 0.03 m, after removing the top layer of decaying tree leaves. Soil samples were placed in plastic bags and kept at field moisture for transportation to the laboratory. Sampling campaign lasted 5 d. After the soils were brought to the laboratory they were hand sieved to <2 mm at field moisture for all subsequent analysis.

Soil Physicochemical Properties
Standard methods (Dane and Topp, 2002) were used for determining the physicochemical properties. Soil texture was determined by the method of the Bouyoucos densimeter, with hexametaphosphate as the dispersing agent. The bulk density and porosity were determined by gravimetry in undisturbed samples. pH was determined both in water and NaF (Blakemore et al., 1981). The organic matter content was measured with the Walkey-Black method (Schnitzer, 1986). Lipid content was determined by extraction with Dicloromethane/methanol (9:1 v/v). Soil water retention at –33 kPa (field capacity) and –1500 kPa (wilting point) were determined with Richards porous plates. Soils were saturated with a 0.005 M CaSO4 and thymol solution to minimize clay disaggregation and avoid degradation of organic matter (Dane and Hopmans, 2002). Since we observed significant soil shrinking as samples dried, individual weights of 0.7 kg were placed on top of each Richard's sample to assure close contact between the soil and the ceramic plates following Klute (1986). Soil surface area (Se) was determined by placing 1 g of soil in weighing bottles and dried in vacuum to constant weight over diphosphorus pentaoxyde. These were then saturated to constant weight by adsorption in a sulfuric acid atmosphere (Newman, 1983).

Water Repellency Determinations
About 230 mL of each soil sample were placed on Petri dishes (0.14 m diam. and 0.02 m deep) and wetted from field moisture to saturation by manually spraying distilled water, homogenizing with a teaspoon, and leaving at room temperature for moisture equilibration for at least 24 h. The wetting process was performed in several steps for 2 to 3 d, during which Petri dishes remained closed to minimize evaporation. After saturation, repellency measurements were performed in desorption steps of about 3 g in weight and at least at 10 steps for each sample. In the early stages repellency was measured in air-dried samples leaving the Petri dishes open at room temperature. In the final stages (water content, {theta}g < 0.20 kg kg–1) samples were oven dried at 55, 60, and 105°C, leaving the samples to cool down at room temperature before starting with the repellency measurements.

Water repellency was determined with the MED test (Roy and McGill, 2002) using solutions of the following ethanol concentrations: 1, 2, 3, 4, 5, 6, 8, 10%, and from 12.5 to 100% in 2.5% steps. Before each MED measurement, soil samples were weighed, homogenized, and leveled with a teaspoon and shaking. Ethanol molarity, M (mol L–1), data was then converted to contact angle, {alpha}, by means of (Roy and McGill, 2002):

[1]
where {gamma}w is the water-air surface tension (71.27 mN m–1), and the 90° surface tension, {gamma}90°, (mN m–1) is given by (King, 1981):

[2]
In Eq. [1] {alpha} ranges from 90 to 109° (Roy and Mc. Gill, 2002). A wettable soil would have a contact angle {alpha} {approx} 90°, while for a water repellent soil {alpha} > 90°. The {gamma}90° is related to the solid-air surface tension or surface free energy of the solid, {gamma}s (mN m–1), by the following simple proportionality relation (Carrillo et al., 1999):

[3]
where {Theta} is a constant that varies with the molecular properties of the solid and liquid and has values ranging from 0.5 to 1.15. The molecular interaction parameter {Theta} was recently investigated by Arye (G.A. Arye, personal communication, 2005), who propose average {Theta} values for soils around 0.6.

Additionally, the WDPT test (Letey, 1969) was also used to characterize water repellency persistency with soil depth in a small subset of air-dried samples (n = 32).

Repellency Parameters
The description of the water dependent repellency curve in terms of quantitative parameters is scarce (de Jonge et al., 1999). We have analyzed the following 10 parameters to characterize the repellency versus water content curve (Fig. 2) : (1) both the increasing (s+) and, (2) modulus of the decreasing (|s–|) repellency slopes (° kg–1 kg); (3) the trapezoidal integrated area below the {alpha}({theta}g) curve, S (° kg kg–1); (4) the maximum contact angle measured, {alpha}max [°]; (5) the water content at {alpha}max, {theta}g-max (kg kg–1); (6) the contact angle in the dried soil, {alpha}105°C [°], or potential repellency (Dekker and Ritsema, 1994); (7) the lowest {theta}g at which repellency is negligible ({alpha} {approx} 90°), {theta}g-min (kg kg–1); (8) the difference between maximum and potential repellency, {alpha}err [°] (i.e., {alpha}err = {alpha}max–{alpha}105°C); (9) the surface free energy of the solid at minimum water content, {gamma}s-105°C (mN m–1); and (10) when repellency is maximum, {gamma}s-max (mN m–1) (i.e., minimum surface free energy of the solid).



View larger version (21K):
[in this window]
[in a new window]
  		Fig. 2. Mean repellency curves, indicating the parameters selected for their description, (a) in terms of contact angle, {alpha}; or (b) surface free energy of the solid, {gamma}s. Bars indicate 95% confidence interval. For repellency parameter definitions see caption in Table 2.

 

View this table:
[in this window]
[in a new window]
  		Table 2. Main statistics: mean, standard deviation (SD), range, and normality transformation ({tau}), of the repellency parameters (n = 140). Contact angle in the dried soil ({alpha}105°C); surface free energy of the solid at minimum water content, {gamma}s-105°C; and when repellency is maximum, {gamma}s-max; maximum contact angle ({alpha}max); {alpha}err = {alpha}max–{alpha}105°C; water content at {alpha}max ({theta}g-max); lowest {theta}g at which repellency is negligible ({theta}g-min); increasing (s+) and modulus of the decreasing (|s–|) repellency slopes; integrated area below the repellency {alpha}({theta}g) curve (S); organic matter content (OM).

 
Statistical Analysis
Probability Plots
Probability plots (PP) represent a useful tool for discerning between different statistical distributions that may fit a given data set. The original data set, X, was thus ordered such that Xi > Xi-1, and plotted against the expected value, Zpi, of the hypothesized cumulative distribution function CDF to be fitted. If the distribution was to be normally distributed then Zpi was approximated by the inverse of the standard normal CDF given by Stedinger et al. (1992):

[4]
where the plotting position was defined as pi = (i – b)/(n + 1 – 2b), with n being the number of samples and b a plotting position parameter. A traditional choice is Blom's plotting position, b = 3/8, which ensures unbiased quantiles (Stedinger et al., 1992). Visual inspection of the plot of Xi versus Zpi will thus reveal that if the data followed the hypothesized distribution, it should yield a straight line through the origin (Stedinger et al., 1992). Probability plots were computed with SYSTAT 10 (SPSS Inc., 2000).

Box-Cox Normality Plots
Visual inspection of probability plots is however rather subjective, and hence a more objective procedure was needed. A particularly useful family of transformations to achieve normality is the Box-Cox transformation:

[5]
where {tau} is the transformation parameter and X*i the transformed value, however, for {tau}->0, the transformation reverts to the natural logarithm of the data. Given the above Box-Cox transformation, it is helpful to define a measure of the normality of the resulting transformation by means of the correlation coefficient of a normal probability plot. The correlation is computed between the vertical and horizontal axis variables of the PP and is an objective measure of the linearity of the PP. The Box-Cox normality plot is a plot of these correlation coefficients for various values of the {tau} parameter (NIST/SEMATECH, 2005). The value of {tau} corresponding to the maximum correlation on the plot is then the optimal choice for {tau}. Box-Cox normality plots were computed with Dataplot (Filliben et al., 1978).

Minimum Number of Samples
Another important criterion when deciding between different measuring methods is the number of samples necessary to estimate the population mean of the parameter. This question may be solved in terms of the accepted range ± d about the mean. The number of samples required to estimate the mean of a population with a standard deviation {sigma} is given by (Warrick and Nielsen, 1980):

[6]
n is the number of samples and Z0.05 is the normalized difference from the mean for a 95% error probability (i.e., Z0.05 = 1.96).


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Soil Characterization
USDA soil texture is loamy-clay (18.1% clay, 22.6% silt, 59.3% sand) and soil pH is acid (5.5 ± 0.4). Water retention is high at both field capacity (1.25 ± 0.24 kg kg–1) and wilting point (0.43 ± 0.05 kg kg–1), as it is typical of volcanic soils. Bulk density is low (600 ± 100 kg m–3) because of the large microporosity and high organic matter content (42.3 ± 12.1 g 100 g–1). Mean surface area is 197 ± 21 m2g–1. The Alp/Alo and Alo+ 1/2Feo relations and the organic matter content, constitute an organomineral soil in the top horizon. Both organic matter and lipid content decreased with soil depth, as did WDPT and MED contact angle. Below 0.23 m, soils were wettable at air-dry moisture. pHNaF was lower in the first centimeters of the soil, which is consistent with the organomineral character of the topsoil (Table 1).


View this table:
[in this window]
[in a new window]
  		Table 1. Water repellency and chemical properties of the soil profile (mean ± standard deviation).

 
Qualitative Description of the Water Repellency Curve
Repellency varied nonmonotonically with water content (Fig. 2). In general, soils were wettable at saturation. At about field capacity, contact angle increased up to a maximum before the wilting point, and then decreased monotonically, but remained water repellent when dried at 105°C. Water repellency was thus present within the water range relevant to plants. The repellency behavior is the inverse if expressed in terms of surface free energy of the solid, {gamma}s, instead of contact angle, {alpha} (Fig. 2). Additionally when repellency is expressed in terms of {gamma}s instead of {alpha}, uncertainty arises about the value of the molecular interaction parameter {Theta} in Eq. [3]. However in the following this only affects the absolute value of {gamma}s but not the statistical analysis and conclusions derived, and for the former a value of {Theta} = 0.6 was assumed (G.A. Arye, personal communication, 2005). There is a fundamental problem that may arise when using the MED on relatively moist soils: the dilution of the ethanol solution by soil pore water as soon as the drop comes in contact with the soil. Thus, surface free energy and contact angle values calculated on the basis of MED test results may become less reliable as soil water content increases. This implies higher data dispersion as water content increases (Fig. 2).

King (1981) also found increasingly sharp water repellency when water content varied from air-dry to near the wilting point, where it reached a maximum, and then decreased sharply to zero close to field capacity. Although hysteresis cannot be discarded in our case, rewetting oven-dry soil samples may bias the results, because of the heat pretreatment (de Jonge et al., 1999). This is particularly relevant in volcanic soils, where drying can have irreversible effects (Warkentin and Maeda, 1980). From air- to oven-dry we found that water repellency was essentially unchanged in most soil samples (70%), while 30% showed a decrease in water repellency within this moisture regime. King (1981) also observed that water repellency remained unchanged as soil moisture increased from oven to air-dry. De Jonge et al. (1999) and Goebel et al. (2004) reported a nonlinear behavior of water repellency. However, in their case most soils were wettable at very low water contents. In our case, only two soil samples recovered wettability after being oven dried (105°C). The drying process did not have an intensifying effect on the repellency, as found by de Jonge et al. (1999), where a second repellency maximum was detected for very low water contents as a consequence of the high temperatures. Heating may well affect the stereochemistry of the hydrophobic compounds or volatilize organic matter, and this may have some effect on the shape of the repellency curve at very low water contents ({theta}g < 0.20 kg kg–1). However, these effects are not relevant in field conditions and even then most of the conclusions derive here refer to water contents {theta}g > 0.20 kg kg–1. For some soil samples we observed only a slight increase in water repellency in the 105°C dried measurements, compared with that at the previous water content (60°C), but this cannot be attributed to temperature effects since similar behavior was observed by de Jonge et al. (1999) in freeze-dried samples.

Quantitative Description of the Water Repellency Curve
Statistics of the Repellency Parameters and Minimum Number of Measurements
Repellency parameters proposed to characterize the repellency versus water content curve have been introduced in the Materials and Methods section and Fig. 2. Standard statistics of these parameters and for the soil organic matter content (OM) are summarized in Table 2 and 3. A first look at these statistics shows that variability was different for the various parameters. In general, repellency parameters showed low variability, as compared with hydraulic or electrical conductivity with CV > 100% (Table 9.1 in Mulla and McBratney, 2002). While repellency slopes, s+ and |s–|, are relatively highly variable (CV > 50%), {gamma}s-105°C, {gamma}s-max, {theta}g-max, {theta}g-min, {alpha}err, S, and OM show moderate variability (10 < CV ≤ 31%), and {alpha}105°C, {alpha}max are little variable (CV < 3%). Notice that although {alpha}105°C and {gamma}s-105°C, {alpha}max, and {gamma}s-max are equivalent parameters, their CV is quite different (Table 3). This is because the relationship between {alpha} and {gamma}s is highly nonlinear (see Eq. [1]–[3]). From a methodological point of view, those parameters that show lower variability are preferable, since the minimum number of measurements necessary to characterize confidently the repellency curve is smaller. Notice that the estimation of some parameters with d = 5%, such as the repellency slopes, would require an extremely large number of samples if the mean is to be computed with a 95% probability (Table 3).


View this table:
[in this window]
[in a new window]
  		Table 3. Minimum number of measurements required to obtain an estimate of the repellency parameters (mean ± d) for a 95% probability. For repellency parameter definitions see caption in Table 2.

 
Normality
Many statistical tools rely on a normal distribution of the data. Normality was first investigated by means of PP after the parameter set was either log or power transformed. Figure 3 shows the best transformation found for each parameter after visual inspection of the PP. Some parameters are Gaussian distributed ({theta}g-max, {theta}g-min, OM, and S); for s+ and |s–| normality improves after a log-transformation is applied, while {alpha}err is Gaussian distributed after taking square root of this parameter. Hallett et al. (2004) found that in the determination of the repellency index given as the ratio of 95% ethanol (SE) to water (Sw) sorptivities, Sw data were log-transformed, while SE was Gaussian distributed after transformation to the square root. Probability plots of both {alpha}105°C and {alpha}max indicate binning into classes, more evident in the former (results not shown). This is also the case for {gamma}s-105°C and {gamma}s-max (Fig. 3). This may be a desirable property if repellency is to be characterized in terms of hydrophobicity classes (Letey, 1969; King, 1981). Alternatively {alpha}err, which is a linear combination of {alpha}105°C and {alpha}max, does not show binning (Fig. 3). The parameters {alpha}105°C and {alpha}max did not improve normality under any of the applied transformations. A way to overcome this is to use {gamma}s-105°C and {gamma}s-max instead, via Eq. [2] and [3]. However, {gamma}s-105°C and {gamma}s-max showed a tailing effect, especially at the upper end of the distribution (Fig. 3).



View larger version (37K):
[in this window]
[in a new window]
  		Fig. 3. Probability plots of repellency parameters. Vertical axes show the expected value, Zpi.

 
The distribution followed by each parameter may give us some clues about their origin and genesis. A principle sometimes stated is that dynamic or flow related properties may be approximately log-normal and that soil static properties are normally distributed (Hopmans and Overmars, 1986; Horowitz and Hillel, 1987). This is in agreement with the distribution of both repellency slopes, s+ and |s–|, which in some sense refer to a dynamical property as the rate of repellency changes. Additionally, a log-normal data distribution would imply that these arise from the product of many independent and identically distributed random variables, and thus one can hypothesize about the processes involved in their genesis. By contrast, a Gaussian distribution results from an additive contribution of processes, and this would be the case of {theta}g-max, {theta}g-min, OM, and S.

One of the limitations of PP is that suitable transformations to achieve normality are not searched systematically. Additionally, interpretation of probability plots is rather subjective, since the best Gaussian transformation is determined by their visual inspection. A more systematic procedure is that of Box-Cox normality plots. Figure 4 depicts Box-Cox normality plots for the investigated parameters. Inspection of these shows that the log or power transformation is generally only an approximation of the best Gaussian distribution. The surface free energy of the solid at maximum repellency, {gamma}s-max, approaches normality after raising this parameter to a power of –1, although improvement is small; {gamma}s-105°C follows a normal distribution after a power transformation with exponent {tau} = –2. In most cases the Box-Cox curve rises sharply from {tau} = –2, until reaching its maximum. Thus finding a suitable normalizing transformation, based on the visual inspection ("eyeballing") of a probability plot would have not been difficult in these cases. By contrast, the S and {theta}g-min Box-Cox curves appear flatter, rising slightly from {tau} = –2 (r2 {approx} 0.92) up to a maximum between {tau} = 0 and {tau} = 1. Certainly, in these two examples, a decision based on the visual judgment of the probability plots would have not concluded that the underlying distributions were different for {tau} = 0 (log-normal) or {tau} = 1 (normal). {theta}g-max and OM are considered Gaussian, since the maximum correlation is achieved near {tau} = 1.



View larger version (23K):
[in this window]
[in a new window]
  		Fig. 4. Box-Cox normality plots of repellency parameters. Vertical axes show the correlation coefficient. For repellency parameter definitions see caption in Table 2.

 
Localization of the Repellency Maximum and Minimum
The dependence of wettability on soil water content may be explained by a combination of two processes (Goebel et al., 2004). At very low water contents (oven-dry regime) water adsorption is controlled by the low free energy caused by organic coating of soil particles. In fact at 105°C the surface free energy of the coated-solid, {gamma}s-105°C, is given by Eq. [3]. The mean value of {gamma}s-105°C is 10.4/{Theta}2 mN m–1 (i.e., {gamma}s-105°C ranges from 41.6 to 7.9 mN m–1 for {Theta} from 0.5 to 1.15). Assuming {Theta} = 0.6 (G.A. Arye, personal communication, 2005) this renders a mean {gamma}s-105°C = 28.98 ± 3.98 mN m–1. Such low {gamma}s-105°C of the coated-soil particles results in a weak attraction between the solid and the liquid phase, that is, water repellency. As soil moisture increases so does the number of water monolayers adsorbed on the soil particles. Previous results suggest that three layers of water molecules (bound water) are affected by the polar groups and negative charges of the soil minerals (Sposito and Prost, 1982; Mulla et al., 1984). The fraction of bound water, {theta}bw (kg kg–1), can be computed from (Dirksen and Dasberg, 1993):

[7]
where l is the number of water monolayers of thickness {delta} (3.5 Å) bounded to the soil particles; {rho}w is the water density, and Se (m2g–1) is the soil surface area. In a subset of n = 18 soil samples, we determined both the wilting moisture and the water content at –33 kPa or field capacity, {theta}fc, and the surface area. Assuming three sheets of bound water (l = 3) in Eq. [7], the resulting mean {theta}bw is 0.21 kg kg–1. From this data subset, the following relation was found between {theta}bw and {theta}g-max (R2 = 0.658):

[8]
Thus, since the slope in Eq. [8] is not unity, we can conclude that coating of the soil solid particles exerts a water repellency action further away from three sheets of water layers. Long-range forces on water layers adjacent to interfaces are well illustrated in the literature. For example, it has been demonstrated the existence of a sharp water birefringence interface at about 40 to 60 nm from solid substrates (Derjaguin and Churaev, 1986). A reduced water dielectric permittivity of {epsilon}0 = 23 to 25 (instead of 70 for bulk water) has been measured in montmorillonite interlayers of thickness 50 to 80 Å, that is, 7 to 11 water diameters apart from the montmorillonite surface (Derjaguin and Churaev, 1986).

Additionally we found the following relation between {theta}g-max and the wilting moisture, {theta}wp in gravimetric units (R2 = 0.715) (Fig. 5a) :

[9]



View larger version (10K):
[in this window]
[in a new window]
  		Fig. 5. Correlation of the repellency parameters {theta}g-max and {theta}g-min with the soil water retention parameters (a) {theta}wp and (b) {theta}fc.

 
This last result is not surprising if we combine Eq. [8] and [9], and we take into account that a similar empirical relation has been previously proposed for {theta}bw in volumetric units (Newton, 1977; Wang and Schmugge, 1980):

[10]
in terms of the soil wilting volumetric moisture.

Wettability is fully recovered at about field capacity. A plot of {theta}g-min versus {theta}fc shows this, as the correlation equation between the two is not far from the 1:1 line (dash line in Fig. 5b).

Parameter Correlations
Based on the need for a multi-parameter characterization of the water repellency curve, and given the distinct variability of such parameters, we have explored the possibility that some of these may be correlated. This may provide an efficient approach for characterizing the {alpha}–{theta}g curve. Table 4 summarizes the Spearman's rank correlation coefficient between the different parameters, and also the soil organic matter content. Relevant correlations are underlined. One may observe that S is correlated with most parameters, in particular, with {theta}g-max, {theta}g-min, s+, |s–| and OM, where modulus of Spearman's rank correlation coefficient is above 0.725 (Fig. 6 and Table 4). De Jonge et al. (1999) also found a good correlation between S and OM. However, the results from different studies show that generally repellency is not correlated with the total amount of OM but with its "type" and degree of decomposition (Doerr et al., 2000; Ellerbrock et al., 2005). It is reliable the almost exact match of S and {theta}g-min (R2 = 0.997). This linear relationship S = S({theta}g-min) is expected, given the following approximate analytical expression of S obtained by triangularization:

[11]


View this table:
[in this window]
[in a new window]
  		Table 4. Spearman's rank coefficient correlation matrix (p < 0.01). Highly relevant correlations are marked by underlining. For repellency parameter definitions see caption in Table 2.

 


View larger version (22K):
[in this window]
[in a new window]
  		Fig. 6. Correlation of the integrated area below the repellency curve, S, with other repellency parameters such as the water content at maximum, {theta}g-max, and minimum repellency, {theta}g-min, and with soil organic matter content (OM).

 
From a conceptual point of view, S is a rather convenient parameter since it integrates the water-dependent repellency characteristics of a soil. However, for its estimation one must determine the whole {alpha}-{theta}g curve, and this is a rather cumbersome and time-consuming experiment. The above correlations may be helpful in this sense, since these may provide a way to extrapolate S from measurements of {theta}g-max, {theta}g-min, or OM. We may however stress that correlation does not negate the need to conduct certain measurements, but it must be view as a guiding strategy, and at its best as an estimate.

The above results motivate the following combined repellency parameter:

[12]

Notice that the parameter IRDI (Integrative Repellency Dynamic Index) is a measure of the average repellency function {alpha} = {alpha}({theta}g) within the water interval [0, {theta}g-min], in contact angle units. It allows simple comparisons of the whole water repellency behavior among different soils by coalescing the repellency curve into a single characteristic value. We may stress the low variability of IRDI in our case, CV = 1.64%, which is certainly a consequence of the linear correlation between S and {theta}g-min shown in Eq. [11] and Fig. 6b.

Strategy for Repellency Characterization
We thus propose the following efficient strategy to characterize the repellency properties of the current soil (Fig. 7) . This strategy needs to be proved in other soils before becoming a general procedure. S can be determined either from the OM of the soil sample (Fig. 6a) or {theta}g-min (Fig. 6b). The former is not part of the repellency measurement protocol, but it is not unusual that this may have been measured, so it can provide us a rapid estimation of the S value, although with higher uncertainty (R2 = 0.597). To be able to determine the water content at which {theta}g-min is localized, at least three repellency measures at decreasing steps of water content from saturation are needed. However, this process is simplified by taking into account that {theta}g-min is placed near the soil field capacity (Fig. 5b), and this may be already known or can be estimated from for example, soil texture. Combining S and {theta}g-min the IRDI value can be computed. Once S is estimated, one can predict the moisture content at which repellency is maximum, that is, {theta}g-max (R2 = 0.544). Additional information as the soil surface area of the sample or the wilting point may provide us a way to check the validity of this result, by using Eq. [7], [8], and [9]. The inverse result is also valid and one may be able to provide an estimate of Se and {theta}wp from {theta}g-max making use of the inverse relations described by Eq. [7], [8], and [9]. We computed the root mean square error (RMSE) of the two possible strategies for estimating {theta}g-max that is, either from OM (Path A in Fig. 7) or from {theta}g-min (Path B in Fig. 7), as a mean to quantify in one single measure the error involved in two consecutive determinations. As expected, RMSE of Path A is higher than B (0.147 versus 0.124). Finally once {theta}g-max is identified, one can measure {gamma}s-max or {alpha}max by performing the MED test at such fixed water content. Also from S one can estimate two dynamic repellency properties as s+ and |s–| (Fig. 7). Table 3 can help us in providing the minimum number of samples necessary in each case. The direct determination of s+ and |s–| would require a rather intensive measuring campaign, not only because of their large variability (CV = 54.83 and 55.41%, respectively; Table 3), but also because for each slope computation at least three repellency measurements would be necessary for each soil sample. Thus the above extrapolation strategy may be a way to get round this problem. Additionally, a value of repellency when the soil sample is dried at 105°C can provide both the potential repellency and also an estimation of {alpha}err (Table 4).



View larger version (23K):
[in this window]
[in a new window]
  		Fig. 7. Flow diagram of the strategy proposed for characterizing the repellency curve with a minimal number of determinations. IRDI: Integrative Repellency Dynamic Index; OM: organic matter; Se: surface area; {theta}fc and {theta}wp: water content at field capacity and wilting point, respectively. For repellency parameter definitions see caption in Table 2.

 

   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
Water repellency was quantified in the top horizon of a forest watershed with the MED test in decreasing steps of water content (from saturation to oven-dry). Repellency varied non-monotonically with water content, and was relevant within the plant available water range. Water-dependent soil repellency parameters were used to describe the repellency vs. water content curve. The location of the water content for minimum and maximum repellency ({theta}g-min and {theta}g-max) was related through the soil water status at field capacity and wilting point, respectively. Useful correlations were found between the area below the curve (S) and {theta}g-min or {theta}g-max. It is appealing the strong correlation between {theta}g-min and S, which we have shown can be derived with an explicit equation. Soil OM content was correlated with {theta}g-min, s+, and S. Some parameters show higher variability than others. While the curve slopes, s+ and |s–|, are relatively highly variable (CV > 50%), {gamma}s-105°C, {gamma}s-max, {theta}g-max, {theta}g-min, {alpha}err, S, and OM show moderate variability (10 < CV ≤ 31%), providing different minimum number of measurements necessary in each case. The parameters {theta}g-max, {theta}g-min, OM, and S were found to be closely Gaussian; s+ and |s–| were highly skewed, log-normally distributed; {alpha}err, {gamma}s-105°C, and {gamma}s-max improved normality after a power transformation. These results have been obtained from desorption experiments. Since hysteresis is not discardable it cannot a priori be expected that the important sorption process will exhibit the same statistical behavior as described here.

By contrast to other soil properties, such as the water retention curve, which is described by generally accepted van Genuchten's parameters, there is no agreement about the parameters necessary for a complete characterization of the repellency curve. The integrated area below the repellency curve is an optimal candidate, since it encompasses both the water dependency of soil repellency and the correlation with most dynamic and static parameters investigated. In addition, S exhibits favorable statistical properties, such as normality and moderate variability. Taking into account all this information, we designed an efficient strategy to fully characterize the repellency curve, where S is the key index for repellency characterization. Following this strategy, S can be estimated in this soil through {theta}g-min with only 27 samples. Furthermore, S and {theta}g-min may be combined into a single parameter, IRDI as a measure of the average water dependent repellency.

The above conclusions have been obtained for one sampling site and it is not known how applicable the results are in a wider context. However the underlying assumption behind the current study is that some physicochemical processes support the shape of the repellency curve (and therefore the parameters that characterize it), since this has been also confirmed in previous studies from other World regions and soil types. In this sense some clues have been provided here in terms of the free energy of the organic coating of soil particles and its modification by the number of bound water layers. What may not be applicable in other studies is the "exact" numbers derived here, but the strategy, parameter sets to be investigated and mathematical techniques used.

Temporal variability is another issue not specifically addressed in this paper. In the field, soil water repellency is affected by physicochemical and biological effects throughout a wide variety of timescales. It is however worth stressing that some underlying temporal variability is implicit in the current study throughout the inclusion of the soil water status, since soil moisture content in the field is time dependent. As part of a hydrological study currently being performed in the studied watershed, field soil water content data is available. The inclusion of field soil moisture data into the above derived parameters may provide us with a more dynamic view of the repellency status of the soil in field conditions. Also field spatial variability of repellency parameters has not been addressed here. This will be presented elsewhere.

This study shows the importance of an appropriate experimental design and statistical analysis to obtain representative repellency indexes. In addition, such analysis may be useful for mean comparisons, spatial variability analysis, modeling, etc. A similar analysis as the one performed here, with parameters derived from the MED test, can be applied also to others, such as the soil repellency persistency measured with the WDPT test.


   ACKNOWLEDGMENTS
 
The authors thank A.R. Socorro and A. Pérez (ICIA) for their help in the field survey and laboratory analysis; Dr. G. Aschan (University of Duisburg-Essen, Germany) for his contribution in the vegetation description; S. Armas and Dr. J.M. Hernández (University of La Laguna, Spain) for the determination of andic properties; Dr. D.F.W. Naafs (Utrecht University, The Netherlands) for the soil lipid content determinations. We also thank A. Fernández and L.A. Gómez (Garajonay National Park) for their support. This work was financed with funds of the INIA Project Nr. RTA 01-097.

Received for publication February 25, 2005.


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




This article has been cited by other articles:


Home page
 Vadose Zone JHome page
C. M. Regalado and A. Ritter
A Soil Water Repellency Empirical Model
Vadose Zone J., February 10, 2009; 8(1): 136 - 141.
[Abstract] [Full Text] [PDF]

Home page
 Soil Sci.Home page
C. M. Regalado and A. Ritter
Cokriging Spatial Interpolation of Soil Water Dependent Repellency Parameters Determined with Two Different Tests
Soil Sci. Soc. Am. J., October 30, 2008; 72(6): 1683 - 1693.
[Abstract] [Full Text] [PDF]

Home page
 Soil Sci.Home page
K. Kawamoto, P. Moldrup, T. Komatsu, L. W. de Jonge, and M. Oda
Water Repellency of Aggregate Size Fractions of a Volcanic Ash Soil
Soil Sci. Soc. Am. J., September 28, 2007; 71(6): 1658 - 1666.
[Abstract] [Full Text] [PDF]

Home page
 Vadose Zone JHome page
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 page
 Soil Sci.Home page
G. Arye, I. Nadav, and Y. Chen
Short-term Reestablishment of Soil Water Repellency after Wetting: Effect on Capillary Pressure-Saturation Relationship
Soil Sci. Soc. Am. J., April 5, 2007; 71(3): 692 - 702.
[Abstract] [Full Text] [PDF]

Home page
 Soil Sci.Home page
C. M. Regalado and A. Ritter
Geostatistical Tools for Characterizing the Spatial Variability of Soil Water Repellency Parameters in a Laurel Forest Watershed
Soil Sci. Soc. Am. J., May 23, 2006; 70(4): 1071 - 1081.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow An erratum has been published
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 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 Web of Science (7)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Regalado, C. M.
Right arrow Articles by Ritter, A.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Regalado, C. M.
Right arrow Articles by Ritter, A.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Regalado, C. M.
Right arrow Articles by Ritter, A.
Related Collections
Right arrow Hydrophobicity
Right arrow Volcanic Soils
Right arrow Water repellency

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Journal of Natural Resources
and Life Sciences Education		 Vadose Zone Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome