VZJ sign up for etocs
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 27 April 2006
Published in Vadose Zone J 5:564-569 (2006)
DOI: 10.2136/vzj2005.0102
© 2006 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 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 (5)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Levin, J. M.
Right arrow Articles by Saiers, J. E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Levin, J. M.
Right arrow Articles by Saiers, J. E.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Levin, J. M.
Right arrow Articles by Saiers, J. E.
Related Collections
Right arrow Capillary Fringe Processes
Right arrow Water Retention/Capillary Pressure
Right arrow Laboratory Column Studies
Right arrow Colloids

NOTES

Colloid Mobilization from a Variably Saturated, Intact Soil Core

Janna M. Levina,*, Janet S. Hermanb, George M. Hornbergerb and James E. Saiersc

a Dep. of Physics, 7507 Reynolda Station, Wake Forest Univ., Winston-Salem, NC 27109-7507
b Dep. of Environmental Sciences, 291 McCormick Rd., Univ. of Virginia, Charlottesville, VA 22904-4123
c School of Environmental Studies, 205 Prospect St., Yale Univ., New Haven, CT 06511-2189
* Corresponding author (levinjm{at}wfu.edu)

Received 19 August 2005.



   ABSTRACT
TOP
 EXECUTIVE SUMMARY
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Colloids may transport contaminants through unsaturated soils to the groundwater, and colloid mobilization is associated with transient hydrological events and sustained, steady flow. We examined mechanisms of colloid mobilization from an intact, unsaturated soil core during relatively steady flow. We collected the core from a site where soils contain 26% clay and infiltration occurs only at capillary-pressure heads above –20 cm. We measured the colloid-mass flux during consecutive, 1.5-mo periods (P1, P2, and P3) distinguished by capillary-pressure heads ({Psi}B) of –18.5, –11.5, and –18.5 cm, respectively. Mean mass flux values were 0.0886, 0.197, and 0.171 mg h–1 for P1, P2, and P3, respectively. Intervention analysis showed a significant increase in the mass flux of 0.079 mg h–1 as {Psi}B became less negative. Results indicate that the number of soil pores through which water flows has a greater influence on colloid mobilization than do shear forces associated with elevated pore water velocities. Thus, colloid mobilization most likely is affected by a diffusion-limited, colloid-supply mechanism dependent on the number of pores contributing to flow.

Abbreviations: AR, autoregressive • MA, moving average


   INTRODUCTION
TOP
 EXECUTIVE SUMMARY
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
UNDERSTANDING MECHANISMS of colloid (suspended particles with at least one dimension in the size range of 10 nm to 10 µm) mobilization is important because colloids that are mobilized during infiltration events can carry adsorbed contaminants through the vadose zone and toward drinking-water aquifers (de Jonge et al., 1998; Sprague et al., 2000; McGechan and Lewis, 2002). In addition, rapid vertical translocation of colloids within macroporous soils may enhance development of new soil horizons (McCarthy and Zachara, 1989; Rees, 1990).

Observations from laboratory and field experiments show that colloids are mobilized rapidly into the porewater shortly after the onset of infiltration and with the passage of the wetting front (Kaplan et al., 1993; Jacobsen et al., 1997; Ryan et al., 1998; Laegdsmand et al., 1999; Weisbrod et al., 2002). At the field scale, colloid concentrations reach hundreds of milligrams per liter just after initiation of rainfall experiments (e.g., Villholth et al., 2000). Colloid-mobilization rates decline as the wetting front passes and as flow approaches steady state, but slow colloid mobilization can persist as long as flow continues (El-Farhan et al., 2000; Schelde et al., 2002). Relatively continuous colloid mobilization over several hours of steady flow (i.e., when the volumetric moisture content, specific discharge, and average linear porewater velocity do not change with time) through intact cores can contribute thousands of milligrams of colloids to the porewater (Schelde et al., 2002). Thus, sustained colloid mobilization can contribute substantially to the total mass of colloids mobilized during an infiltration event.

The purpose of this research is to elucidate the mechanisms of sustained colloid mobilization during steady flow because, to date, these mechanisms remain unclear. Some studies invoke a diffusion mechanism to explain colloid mobilization, whereby colloids are supplied at a sustained rate from the walls of macropores (Laegdsmand et al., 1999; Schelde et al., 2002). In other studies, fluid shear on colloids attached to the soil matrix has been proposed as the mechanism of mobilization (Kaplan et al., 1993; Weisbrod et al., 2002). Thus, in vadose zone flows through structured soils, there may be at least two important mechanisms of mobilization, both of which are influenced by the magnitude of the capillary-pressure head ({Psi}): a shear mechanism by which increased velocities will increase mobilization, and a supply mechanism whereby the number of flowing pores available for colloid diffusion will limit colloid-mass fluxes.

We collected time series data during a colloid mobilization experiment on an unsaturated, intact soil core to test the following alternate hypotheses: first, colloid mobilization increases with a decrease in {Psi}B (i.e., capillary-pressure head applied at the base of the core); second, colloid mobilization decreases with a decrease in {Psi}B. Our experiment was completed using a relatively constant volumetric flow rate but with different values of {Psi}B. We selected values of {Psi}B for the experiment based on tension infiltrometry at the field site indicating that flow through the soil occurs only when {Psi} is greater than –20 cm (Waldeck, 1998). Under the experimental conditions, if a velocity mechanism were of overpowering importance, a lower (more negative) {Psi}B should dictate higher effluent colloid concentrations because the total flow would have to occur through fewer pores, resulting in higher pore velocities at the outflow. On the other hand, if a supply mechanism were favored, colloid concentrations should be higher for higher (less negative) {Psi}B because more pores would be available to carry a diffusion-limited colloid supply. An intervention analysis of time series data from the experiment shows significant, albeit small, increases in the colloid-mass flux at {Psi}B = –11.5 cm relative to {Psi}B = –18.5 cm. We use these findings to elucidate the mechanism, either colloid supply or fluid shear, that most likely limits colloid mobilization from the macroporous soil core used in this study.


   MATERIALS AND METHODS
TOP
 EXECUTIVE SUMMARY
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
Soil Characteristics
We collected the intact soil core used in the colloid mobilization experiment from the Muddy Creek field site (38°32.93' N, 78°57.24' W), located in the Shenandoah Valley, 20 km northwest of Harrisonburg, VA, USA. Soils at the field site (Typic Hapludult) originate from weathering of limestone bedrock (Hockman et al., 1982) and are 26% clay (predominantly illite and kaolinite), 28% silt, and 46% sand (Kauffman, 1998). Organic C content ranges from 1 to 2% in the upper 20 cm of the soil profile to <1% between depths of 20 and 30 cm (Johnson, 1997).

The area where we collected the intact core remains untilled, and infiltration of water into the soil in this location mostly is attributed to flow through macropores. The density of macropores at the field site is approximately 17 pores m–2 (Waldeck, 1998). Pores with diameters >0.15 cm transmit 45% of the total flow at {Psi} between 0 and –2 cm, pores with diameters between 0.06 and 0.15 cm conduct 28% of the total flow at {Psi} between –2 and –5 cm, and pores with diameters between 0.02 and 0.06 cm conduct 23% of the total flow at {Psi} between –5 and –14 cm (Waldeck, 1998).

Experimental Design
The intact core was collected by hammering a hollow, stainless-steel cylinder (20 cm in diameter, 23 cm in length) into the soil at the field site and removing the core (soil length of 20 cm) from the ground. The bottom of the steel-encased core was secured to a Teflon housing fitted with a water-saturated, 10-µm porous plate. A 2-mm-thick layer of acid-washed sand (0.35 mm) along the bottom edge of the core ensured complete contact between the soil and the porous plate. The base of the Teflon housing was sealed to a Plexiglas vacuum chamber, which was used to regulate the capillary pressure at the base of the core.

Ten holes (0.3 cm in diameter) drilled into the steel cylinder before core collection permitted air entry and escape during imbibition and drainage. Tensiometers [assembled as described by Etching and Hopmans (1993)] and soil-moisture probes (Delta-T Devices, Inc., Cambridge, UK) were inserted into two sets of larger holes positioned 6 and 16 cm from the top of the core. Measurements of {Psi} and volumetric moisture content ({Theta}) were recorded every 30 min during the mobilization experiment with a data logger (Delta-T Devices, Inc.).

The core was saturated from below with a colloid-free, 0.005 M NaCl solution. Once the core was saturated, flow was reversed and soil-core drainage was initiated by applying suction at the base of the column while introducing the 0.005 M NaCl solution to the top of the core at rate of 0.30 mL min–1 (14 mm d–1). A drip chamber containing evenly spaced, hollow needles distributed the inflow solution across the bare soil surface. The capillary-pressure head applied at the base of column ({Psi}B) equaled –18.5 cm during drainage and was maintained at this level for 1.5 mo to establish steady flow within the core.

Data Collection
Collection of effluent samples commenced after the 1.5-mo equilibration period, marking the start of the experiment. The experiment was conducted for 4.5 mo and was divided into three consecutive 1.5-mo periods (P1, P2, and P3), distinguished on the basis of {Psi}B. In particular, {Psi}B was set at –18.5, –11.5, and –18.5 cm for P1, P2, and P3, respectively. The inflow rate was held constant at 0.30 mL min–1 for the entire experiment.

A fraction collector (Eldex Laboratories, Inc., Napa, CA) placed within the vacuum chamber collected solution eluted from the bottom of the core. Volumetric discharge was determined by dividing the volume of water contained within an effluent sample by the collection time (30 min). Every eighth sample was analyzed for colloid concentration with a spectrophotometer (Spectronic 1001 Plus, Milton Roy, Ivyland, PA) at a wavelength of 750 nm. Measurements of light absorbance were converted to colloid concentrations (mg L–1) using standards prepared from excess soil removed from the base of the core during the column-assembly process. Colloid-mass flux rates (mg h–1) were calculated by multiplying the volumetric discharge by the colloid concentration.

Intervention Analysis
An intervention analysis was used to estimate changes in the time series of the mass flux of colloids from the core due to the two changes in {Psi}B (the interventions). The time series of colloid mass outflow is considered to be the sum of noise in the observations and a response due to the intervention. The general model for the time series is as follows (Box and Tiao, 1975):

Formula 1[1]
where t is time, yt is the colloid mass flux (or a suitable transform used to achieve a fixed mean level and a constant variance), Nt is the preintervention noise component, and f({kappa},{xi},t) is the intervention component that includes time effects, unknown parameters ({kappa}), and the intervention series ({xi}).

Nt is modeled as an autoregressive (AR) moving average (MA) process:

Formula 2[2]
where B is the backshift operator such that Byt = yt – 1, at is the normally and independently distributed white noise residual with mean 0 and variance {sigma}a2, and {varphi}(B) = 1 – {varphi}1B – {varphi}2B2 – ... – {varphi}pBp and {theta}(B) = 1 – {theta}1B – {theta}2B2 – ... – {theta}qBq are AR and MA polynomials in B of degrees p and q, respectively. Identification of alternative models for Nt is based on regression of the preintervention series on the intervention series and subsequent checks of the white noise residuals (Hipel et al., 1975). The optimal model for Nt has the greatest explanatory power and provides the whitest noise. Once the optimal model is selected, it is assumed that Nt does not change after the time of intervention (T).

Because experimental changes in the column were instituted at a point in time and held fixed following the change, the intervention event in our experiments is represented by a step function, {xi}t = S(t)(T) where

Formula 3[3]

A dynamic model for an intervention is written:

Formula 4[4]
where {xi}t is the intervention time series, the unknown parameters ({kappa}) are now denoted by {omega} and {delta}, and {omega}(B) = {omega}0 – {omega}1B – ... – {omega}uBu and {delta}(B) = 1 – {delta}1B – ... – {delta}rBr are operators of the transfer function of degrees u and r, respectively.

Parameters of the intervention function and noise model are estimated using the method of maximum likelihood using the Statistical Analysis System software package (SAS Institute, 1999). The standard error associated with each maximum likelihood estimate indicates the statistical significance of each parameter. The magnitude of {delta} in Eq. [4] describes the rate of growth or decay in the level of the series after the intervention event. The lag operator B in Eq. [4] controls the time at which a growth or decay is experienced. The magnitude of the effect on the mass flux series of an intervention event is measured by the steady-state gain (g) of the time-series intervention model, where positive values of g indicate a net increase and negative values of g indicate a net decrease in the mean level of the mass flux time series:

Formula 5[5]


   RESULTS
TOP
 EXECUTIVE SUMMARY
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
Relatively steady flow was obtained for much of P1, P2, and P3 (Fig. 1 ). The average outflow rate for all periods was 0.018 L h–1, although outflow rates varied by about an order of magnitude. Unsteady flow was observed for several days after changes in {Psi}B. Outflow rates following the second intervention (INT2), for example, fluctuated between about 0.03 and 0.001 L h–1 for approximately 330 h. Relatively steady flow eventually was achieved after each intervention event.


Figure 1
View larger version (43K):
[in this window]
[in a new window]
  		Fig. 1. Outflow rates (upper panel) and colloid concentrations (lower panel) as a function of time, where black circles, white squares, and black triangles represent colloid concentrations measured during P1, P2, and P3, respectively. Flow rates are represented by black lines. The gray lines connecting colloid concentrations are provided only to indicate the sequence of colloid concentrations with time. Times of intervention are indicated with vertical, black lines labeled as INT1 and INT2.

 
The mean effluent concentration of colloids measured during P2, when {Psi}B equaled –11.5 cm, was greater than the mean concentrations measured during P1 and P3, when {Psi}B equaled –18.5 cm (Table 1). The concentrations of mobilized colloids exhibited substantial temporal variation (Fig. 1), even within periods. During P3, for example, observed effluent concentrations ranged from 6 to 30 mg L–1. The most notable changes in the magnitude of the effluent colloid concentration occurred immediately after the times of intervention. The timing of small increases in the colloid concentration often coincided with the timing of changes in outflow rates.


View this table:
[in this window]
[in a new window]
  		Table 1. Colloid concentrations, measured mass fluxes, and hydrological conditions in soil core. Locations of hydrological measurements are reported as depth from the top of the core.

 
Colloid-mass flux rates generally were <0.3 mg h–1, but approached 1 mg h–1 for brief intervals (Fig. 2 ). Flux rates reached maximum levels immediately after the first intervention (INT1) as {Psi}B increased from –18.5 to –11.5 cm. The second intervention (INT2), delineated by the decrease in {Psi}B from –11.5 to –18.5 cm, also induced an ephemeral increase in colloid-mass flux, although not as great as that observed for INT1 (Fig. 2). These spikes in the mass flux are associated with spikes in colloid concentration and with changes in flow rates, suggesting that flow transients associated with the capillary-pressure change enhanced colloid mobilization, but that a quasi-steady mobilization rate was attained shortly following each intervention event.


Figure 2
View larger version (26K):
[in this window]
[in a new window]
  		Fig. 2. Colloid-mass flux as a function of time, where black circles, white squares, and black triangles represent mass-flux values measured during P1, P2, and P3, respectively. The gray lines connecting data points are provided only to indicate the sequence of colloid-mass flux values with time. Times of intervention are indicated with vertical, black lines labeled as INT1 and INT2. The decrease in the mass flux at about 2300 h may be linked to the eventual failure at about 2400 h of the pump that delivered inflow solution to the top of the core.

 
The following intervention model provides the best fit to the time series data on colloid-mass flux:

Formula 6[6]
where Nt is described by

Formula 7[7]

The relative errors in the estimates of the parameters of the intervention function and noise model are generally <10% (Table 2). Calculations of g made by substituting the maximum likelihood estimates of the parameters into Eq. [5] indicate that INT1 led to a 0.079 mg h–1 increase in the mean colloid-mass flux, while INT2 promoted a 0.05 mg h–1 decrease in the mean colloid-mass flux. Given that {Psi}B was the same before INT1 and after INT2, the differences in the magnitude of g for the two interventions suggest that the relationship between colloid-mobilization rates and {Psi}B was hysteretic.


View this table:
[in this window]
[in a new window]
  		Table 2. Maximum likelihood estimates of model parameters (M.L.E.), standard errors (SE) associated with M.L.E, t values, and p values associated with t tests. The intervention associated with each parameter is listed in the left-hand column.

 

   DISCUSSION
TOP
 EXECUTIVE SUMMARY
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
Both chemical (i.e., solution ionic strength and pH) and physical (i.e., flow rate) perturbations mobilize large concentrations of colloids (e.g., Ryan and Gschwend, 1994; Saiers et al., 2003; Saiers and Lenhart, 2002; Gao et al., 2004); however, sudden changes in chemical and physical conditions are not required for colloid mobilization. Steady flow of water through structured, macroporous soils for extended periods of time may induce a more-or-less steady mobilization of colloids (Kauffman, 1998; Laegdsmand et al., 1999; Schelde et al., 2002).

We investigated whether a fluid shear or a colloid supply mechanism of colloid mobilization dominated for the experiment on the intact core. Because the volumetric flow rate was relatively steady and the chemistry of the inflow solution did not change, we expected colloid mobilization to be influenced by the magnitude of {Psi}B. The intervention analysis of time series data from our soil core showed significant increases in the colloid-mass flux at {Psi}B = –11.5 cm relative to {Psi}B = –18.5 cm. This finding clearly shows that the supply mechanism of colloid mobilization dominated in our experiment. Under our conceptual model, colloid mobilization should increase when {Psi}B is less negative because nearly all the pore spaces are saturated and, therefore, are capable of transmitting a diffusion-limited colloid supply. Colloid mobilization should decrease when {Psi}B is more negative because flow is constrained to fewer soil pores, thereby decreasing the total number of pores capable of transporting colloids.

The soil underwent imbibition of water on initiation of INT1 and drainage on initiation of INT2. The hysteresis observed in the change in the colloid-mass flux is consistent with expectations for hysteresis in the capillary-pressure head–moisture content relationship under the supply-mechanism hypothesis. Under this hypothesis, we expect more water-filled pores on initiation of drainage than on initiation of imbibition of the core.

Although we found significant differences in the colloid-mass flux among experimental treatments, these differences were relatively small. Work done by Kjaergaard et al. (2004) suggests that these small differences in colloid mobilization among experimental treatments could be explained by the high clay content of the soil. These authors showed that, for soil with a clay content >25%, as is the case with our soil, irrigation to initially very wet or only moderately wet soil results in equivalent leaching of colloids. Our volumetric moisture content data, furthermore, indicate that only modest desaturation occurred in the core as a result of the imposed changes in {Psi}B (Table 1). Our observation that soil moisture varied only slightly when {Psi}B was changed from –11.5 to –18.5 cm is consistent with measurements on other structured soils (Langner et al., 1999). Thus, the small differences in the colloid-mass flux among our experimental treatments could be due to the high clay content of the soil we used combined with the small changes in {Psi}B.

The supply of mobile colloids in our experiment was not depleted, even though flow through the core was essentially continuous for several months. During this time, only about 0.007% of the total mass of soil was lost to colloid elution, assuming a uniform bulk density of 1.4 x 103 kg m–3 and a total soil volume of 6.3 x 10–3 m3. The colloid concentrations that we observed are comparable to those ranging from several to tens of milligrams per liter for experiments on intact, sandy loam soil cores (Laegdsmand et al., 1999). Our colloid concentrations are relatively low, but fall within the range of several to hundreds of milligrams per liter reported by others who have worked with soil from our field site (Kauffman et al., 1998; El-Farhan et al., 2000). This comparison highlights the variability in the colloid mobilization response for soils from the same field site. Our results, furthermore, suggest that a steady supply of colloids may migrate downward through macroporous soils during long-duration rainfall and infiltration events. Movement of colloids under such conditions may be of considerable importance for soil illuviation and for facilitated transport of contaminants through the vadose zone to the water table.


   ACKNOWLEDGMENTS
 
This research was supported by grants from the National Science Foundation (EAR-9909491), the Geological Society of America, and the University of Virginia's Department of Environmental Sciences. We are grateful for the technical advice provided by Aaron Mills and John Lenhart and the laboratory assistance provided by Elizabeth Dubovsky, Rosie Patterson, and Drew Gower. We thank Robert Yaffee and Terry Woodfield for help with the intervention analysis.


   REFERENCES
TOP
 EXECUTIVE SUMMARY
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 




This article has been cited by other articles:


Home page
 Vadose Zone JHome page
P. Sharma, H. M. Abdou, and M. Flury
Effect of the Lower Boundary Condition and Flotation on Colloid Mobilization in Unsaturated Sandy Sediments
Vadose Zone J., August 1, 2008; 7(3): 930 - 940.
[Abstract] [Full Text] [PDF]

Home page
 Vadose Zone JHome page
M. Flury and H. Qiu
Modeling Colloid-Facilitated Contaminant Transport in the Vadose Zone
Vadose Zone J., May 27, 2008; 7(2): 682 - 697.
[Abstract] [Full Text] [PDF]

Home page
 Vadose Zone JHome page
J. C. Seaman, P. M. Bertsch, and D. I. Kaplan
Spatial and Temporal Variability in Colloid Dispersion as a Function of Groundwater Injection Rate within Atlantic Coastal Plain Sediments
Vadose Zone J., May 17, 2007; 6(2): 363 - 372.
[Abstract] [Full Text] [PDF]

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 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 (5)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Levin, J. M.
Right arrow Articles by Saiers, J. E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Levin, J. M.
Right arrow Articles by Saiers, J. E.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Levin, J. M.
Right arrow Articles by Saiers, J. E.
Related Collections
Right arrow Capillary Fringe Processes
Right arrow Water Retention/Capillary Pressure
Right arrow Laboratory Column Studies
Right arrow Colloids

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