VZJ Download to Citation Manager
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in 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 (3)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Gee, G. W.
Right arrow Articles by Ward, A. L.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Gee, G. W.
Right arrow Articles by Ward, A. L.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Gee, G. W.
Right arrow Articles by Ward, A. L.
Related Collections
Right arrow Water Flow Models
Right arrow Recharge
Right arrow Disposal Facilities
Published in Vadose Zone Journal 4:32-40 (2005)
© 2005 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA

ORIGINAL RESEARCH

Measurement and Prediction of Deep Drainage from Bare Sediments at a Semiarid Site

G. W. Gee, J. M. Keller* and A. L. Ward

Pacific Northwest National Laboratory, Box 999, K9-36, Richland, WA 99352
* Corresponding author (jason.keller{at}pnl.gov)

Received 24 May 2004.



   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 Hanford Site Background
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 SUMMARY
 REFERENCES
 
In desert environments, nonvegetated (bare) soils and sediments can act as recharge basins, allowing water infiltration but restricting evaporation. When such sediments are located over buried wastes, drainage can transport vadose zone contamination downward to groundwater. Lysimeters were used to quantify drainage from bare sediments at the U.S. Department of Energy's Hanford Site in Washington state, USA. Drainage varied widely from zero to more than half of the annual precipitation for sediments ranging from fine silts to coarse rock fragments. Decade-long drainage records were used to develop two empirical models relating annual drainage and textural properties of bare sediments. A 22-yr drainage record for bare, coarse sand was tested, and the calibration developed for the past 10 years (1995–2004) was found to reliably predict drainage from the previous 12 years. The texture models were also compared against Darcy's Law drainage estimates (i.e., unsaturated hydraulic conductivity) for coarse sand and found to outperform Darcy's Law estimates of the long-term drainage average. The texture models reasonably predicted annual drainage rates for bare sediments containing significant fines (materials less than 50 µm), but significantly overpredicted drainage rates for clean rock and gravels with little or no fines. The failure of the textural models with coarse materials containing minimal fines was attributed to advective drying. Drainage predictions using the textural models indicate that to minimize drainage only modest quantities of fines need to be added to the coarse sediments to substantially reduce the potential for groundwater contamination.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
Hanford Site Background
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 SUMMARY
 REFERENCES
 
THE RISK OF LEAVING WASTES in the ground often depends on how much water they are exposed to and for how long. The amount of water that drains through the soil and recharges the underlying aquifer in a desert environment is often difficult to predict. Studies of drainage, which leads to recharge in desert soils such as those found at the U.S. Department of Energy's Hanford Site in Washington State, have been ongoing for more than two decades (Gee, 1987; Gee and Hillel, 1988; Gee et al., 1992, 1994; Fayer et al., 1999). These studies have dispelled the myth that hot, dry desert conditions prevent deep drainage. In reality, significant drainage can occur under desert conditions, even when potential evaporation rates greatly exceed precipitation, particularly at locations where soils have been disturbed and vegetation removed. Studies have shown that accelerated movement of water into the deep subsurface, around and beneath buried waste, can be attributed to gravel-covered barren surfaces (Gee, 1987; Smoot et al., 1989; Gee et al., 1992; Ward et al., 1997). It should be noted that waste sites at Hanford are often kept barren to minimize any potential for radionuclide uptake by animals and plants. While biotic (plant and animal) uptake of radionuclides and other contaminants is minimized with a surface treatment of bare gravel, the disadvantage is that drainage conditions are optimized (Gee et al., 1992).

What is not known with any certainty is the depth and extent of the contaminant plumes underneath desert waste sites. For example, just how rapidly contaminants are moving beneath leaking tanks is the subject of ongoing research (Ward et al., 1997; McKinley et al., 2001; Johnson and Chou, 2001). Vadose-zone plumes are largely driven by advective water flow with gravity drainage being the dominant transport mechanism. The location of mobile contaminants such as tritium, nitrate, and technetium-99 found at depth beneath buried waste tanks is intimately linked to the drainage (percolation) rates of meteoric water that infiltrates through the bare soil surfaces (Ward et al., 1997; Gee and Ward 2002; DePaolo et al., 2004).

This paper provides an analysis of drainage under conditions where surfaces are primarily coarse-textured and barren. Estimates of unsaturated hydraulic conductivity are used to predict drainage rates, and two simple models are developed and tested against measured field data to estimate gravity drainage for waste sites at Hanford.


   Hanford Site Background
TOP
 ABSTRACT
 INTRODUCTION
 Hanford Site Background
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 SUMMARY
 REFERENCES
 
Drainage or deep percolation of water at nonirrigated desert sites is controlled by three major factors: climate, soils, and vegetation. These controlling factors are described for the Hanford Site, specifically highlighting the features unique to waste sites where drainage information is needed to assess contaminant migration rates.

Climate
The Hanford Site can be characterized as semiarid (Hoitink et al., 2003), the dominant climatic features being cool, wet winters coupled with hot, dry summers (Stone et al., 1983; Rickard et al., 1988). While there are no site-specific, long-term weather records for individual waste burial sites, a reliable source of meteorological data (e.g., precipitation, wind speed, air temperature, humidity, solar radiation) is the Hanford Meteorological Station (HMS), located within 5 km of the majority of the waste sites. Weather records have been kept since the late 1940s, covering the time period of interest for all 200 area waste-site operations at Hanford. For the past 22 yr, precipitation has averaged 182 mm yr–1, about 5% above the historical average (Hoitink et al., 2003). During this time, annual precipitation amounts have varied by more than a factor of three, ranging from a record high of 313 mm in 1995 to a low of 95 mm in 1999. While variations in both seasonal and annual precipitation strongly affect drainage rates at the waste sites, the most dominant factor is winter (November–March) precipitation (Gee and Simmons, 1979; Gee et al., 1992). Winter precipitation during the past 22 yr at Hanford has varied from a record high of 224 mm in 1997 to a low of 58 mm in 1990. Winter months typically have the greatest amount of precipitation (more than 62% of the annual average) and the lowest amount of evaporation.

Soils
Almost all waste sites at Hanford have surfaces consisting of coarse-textured backfill materials. At some Hanford waste sites, commercial "road-base" materials (gravelly sands) have been added to cover the surface contamination and provide operational stability for worker access. These well-drained surfaces transmit water readily to the underlying sediments of the Hanford formation, which are generally coarse materials, often having less than 10% fines, where fines are materials less than 50 µm (Tallman et al., 1979). While there have been a number of studies of hydraulic properties for deep sediments at Hanford (Khaleel et al., 1995; Khaleel and Heller, 2003), only limited information on the hydraulic properties of surface materials is available, particularly the backfill soils that currently cover many of the waste sites. However, textural data are currently available for surface sediments at some of the waste sites (Smoot et al., 1989).

Vegetation
Most waste sites (e.g., tank-farms, cribs, and trenches) where radionuclides are stored are kept vegetation free by applications of herbicides and soil sterilants (Gee et al., 1992). The control of vegetation is important in the waste burial operations because of concerns about radionuclide uptake by plants (e.g., radioactive tumbleweeds). Without vegetation, evaporation is the sole mechanism for upward removal of water.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 Hanford Site Background
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 SUMMARY
 REFERENCES
 
Several methods have been used to estimate drainage rates in Hanford sediments (Fayer et al., 1996). These include direct measurement of drainage with lysimeters, tracer tests, and computer modeling of the surface water balance. In addition, estimates of the drainage rate can be made by applying Darcy's Law to flow in the vadose zone. In this paper, we describe and compare two methods, namely the use of Darcy's Law (or Richards' equation) and an empirical model approach, for estimating drainage rates from bare soil surfaces.

Study Sites and Soils
Three sites which have extensive drainage records were selected for the study.

The 300 North Lysimeter Site
The 300 North (300 N) Lysimeter Site is located about 10 km north of Richland, WA, near the southern boundary of the U.S. Department of Energy's Hanford Site. At this location, several large drainage lysimeters (2.7 m in diameter, 7.6 m deep) were constructed in 1978 (Gee, 1987). Two of these lysimeters, called the North Caisson and South Caisson, are virtually identical in design and content and have been monitored periodically for the past 26 yr. Soil taken from the 300 N lysimeters were analyzed by Rockhold et al. (1988) for unsaturated hydraulic conductivity. In addition, field tests were run in the lysimeters to measure the dependence of unsaturated hydraulic conductivity on water content and soil-water pressure. Details of testing for unsaturated hydraulic conductivity at this and other sites at Hanford are provided by Rockhold et al. (1988). Rockhold et al. (1988) deployed a number of methods to determine the K({theta}) for the 300 N lysimeter soil. These included (i) the particle-size method of Arya and Paris (1981), (ii) the steady state column method of Klute and Dirksen (1986), (iii) the Guelph permeameter method of Reynolds and Elrick (1985), and (iv) the instantaneous profile method of Watson (1966), where water contents were measured with neutron logging and pressure profiles measured with a nest of tensiometers. More recently, Gee and Ward (2002) also reported values of the unsaturated conductivity of the 300 N lysimeter soil using the ultracentrifuge method of Nimmo et al. (1994). In addition, we have continued to monitor profiles of soil-water pressure in both lysimeters using tensiometers (Sisson et al., 2002) and soil-water content by capacitance methods and gravimetric sampling.

The Field Lysimeter Test Facility
The Field Lysimeter Test Facility (FLTF) is located adjacent to the HMS near the middle of the Hanford Site, about 35 km northwest of Richland, WA. The facility contains a number of lysimeters designed to measure water balance of a wide range of surface cover materials ranging from coarse gravels to silt loam soils. Details of the lysimeter construction are provided by Gee et al. (1989) and Fayer et al. (1992)(1999). The FLTF site contains three sets of lysimeters: 14 cylindrical drainage lysimeters, 2 m in diameter and 3 m deep; 6 cylindrical drainage lysimeters, 0.3 m in diameter and 3 m deep; and 4 square weighing lysimeters, 1.5 m wide by 1.5 m deep. While most of these lysimeters were used to test the effectiveness of a silt loam cover in storing both natural and elevated precipitation, four of them (two of the large-diameter and two of the small-diameter lysimeters) were used to measure drainage of coarse sediments under natural precipitation conditions. Of these, only one had vegetation, which over the course of the past 10 yr was a very sparse cover of cheatgrass (Bromus tectorum L.) with a very shallow (less than 30 cm) root zone, and for most of the year the vegetation was inactive. For these reasons, the lysimeter was treated as a bare soil in the analysis described below.

The Hanford Solid Waste Landfill
The Hanford Solid Waste Landfill (SWL) is located midway between the 300 N and the FLTF facilities. The SWL is instrumented with a 6.5 m–deep, basin (pan-type) lysimeter with a capture area of 85 m2 (DynCorp, 2000). The lysimeter was placed at the bottom of the landfill trench in 1992 and subsequently filled over the next 3 yr. Drainage has been collected from this site since July 1996 (DynCorp, 2000). The waste material at the SWL contains little organic matter and is heavily mixed with Hanford formation sands and gravels such that hydrologically the landfill acts much like the surrounding soils and sediments. The waste at the lysimeter location is {approx}6 m deep, covered with about 0.5 m of sandy-gravel backfill. The site has not been revegetated but over a period of time vegetation in the form of weedy species, including Indian wild rice [Achnatherum hymenoides (Roemer and J.A. Schultes) Barkworth], have invaded the site. However, the present plant cover on the surface of the lysimeter is estimated to be less than 5% of the total surface area.

At the 300 N lysimeter site, FLTF, and SWL the texture of the surface soils was determined using both dry and wet sieving followed by hydrometer analysis (Gee and Bauder, 1986). Table 1 lists the lysimeter facilities used for the analysis and the soils and key textural characteristics for each of the lysimeters reported. Figure 1 displays the particle size distribution curves for the lysimeter soils.


View this table:
[in this window]
[in a new window]
  		Table 1. Calculated grain-size statistics for the lysimeter soils and sediments. dg and {sigma}g are geometric mean particle diameter and geometric standard deviation, respectively.

 


View larger version (20K):
[in this window]
[in a new window]
  		Fig. 1. Particle size distribution curves for the lysimeter soils and sediments.

 
Precipitation was measured at the HMS, located directly adjacent to the FLTF. The HMS is approximately 27 km from the 300 N site and 13 km from the SWL. Noncontiguous precipitation records from the 300 N site have shown that the precipitation measured by HMS provides a good approximation of precipitation at the 300 N site (Gee, 1987). Because of the SWL's proximity to the HMS, precipitation at the SWL is believed to be well represented by HMS records.

Drainage Estimation Methods
Darcy's Law Approach
A simple method to approximate vadose zone transport in draining soils is to assume that flow is gravity driven and that unit gradient conditions persist. In such cases, the drainage flux can be estimated by the unsaturated hydraulic conductivity. Formally, a steady state drainage rate can be approximated in one dimension using the equivalent of Darcy's Law for the unsaturated zone (Richards, 1931, 1950) as:

[1]
where K({theta}) is the unsaturated hydraulic conductivity at the depth of interest, {theta} is water content, {psi} is the soil water pressure head, and z is the vertical direction, taken positively downward. When the head gradient is near or equal to zero, the total head-gradient approaches unity (i.e., the so-called unit gradient), and the drainage flux approaches the value of the unsaturated hydraulic conductivity, K({theta}). While the Darcy's Law approach is straightforward, the disadvantage of this method relates to the great uncertainties in estimating K({theta}).

In spite of its limitations, this method is often employed as a first-cut estimate of drainage, thus requiring an estimate of the unsaturated hydraulic conductivity. Methodologies for determining K({theta}) functions include estimates from soft data (grain size, texture, water retention), laboratory core tests, and field-scale tests using instantaneous-profile methods. A comprehensive treatise on methods to measure unsaturated hydraulic properties of soils is provided by van Genuchten et al. (1999).

Texture Models
Deep drainage at waste burial sites at Hanford is best analyzed by assessing the complete water balance of the surface soils. Drainage is an integral component of the water balance, which in its simplest form can be written as:

[2]
where D is the drainage, P is the precipitation input, S is the storage change, ET is the evapotranspiration, Ro is run-on, and Roff is runoff.

Topography also can affect drainage because of runoff or run-on. The Hanford Site waste sites, in general, have highly permeable surfaces and are on relatively level ground or are bermed so that runoff or run-on are controlled or minimized. Based on the above description of climate, soils, and vegetation, we developed a simple water balance model for surface soils at waste sites at Hanford. The model assumes the following conditions:

  1. Winter precipitation dominates the net infiltration process.
  2. Water runoff and run-on volumes are negligible.
  3. Annual water storage changes are negligible.
  4. Soil texture (e.g., particle-size distribution) controls the amount of water retained in the surface and influences the overall evaporation rate.
  5. The soil surface remains unvegetated (upward water movement is by evaporation only—i.e., no transpiration or water uptake by plants). The barren surfaces persist because of herbicide and/or soil sterilants treatments, as explained previously.
  6. Water storage in the bare soils is largely confined to the top meter of soil.

Based on the above assumptions, the surface water balance can be written as:

[3]
where P1 is the winter (November–March) precipitation, P2 is non-winter (April–October) precipitation, E1 is the winter evaporation, and E2 is the non-winter evaporation, for the same time period.

Combining terms leads to the following expression,

[4]
where Ef = (E1 + E2) – P2 is an evaporation factor.

Ef is dependent on soil texture and precipitation. Based on an analysis of lysimeter records (Gee et al., 1992; Fayer et al., 1999; Fayer and Szecsody, 2004) and subsequent measurements made at the lysimeter sites through April 2004, we determined the impact of precipitation and soil texture on drainage for an assortment of surface conditions, ranging from clean, washed gravels to fine-textured silt loam soils.

Relationships between Ef and grain-size statistics were evaluated using drainage data from the lysimeters listed in Table 1 with the exception of the SWL, which was used to test the texture model. The evaporation factor (Ef) was calculated for each of the four lysimeters using the average lysimeter drainage data from 1 Jan. 1995 to 1 Apr. 2004, and the average winter precipitation (November–March) for the same period of time. Note that at the FLTF, monitoring of the loamy sand lysimeter drainage began in 1990 while monitoring of drainage from the silt loam and sandy gravel lysimeters began in 1994. Three different analyses of the relationship between Ef and grain-size statistics were performed, including: (i) Ef versus percentage of fines, as was done by Gee and Ward (2002); (ii) Ef vs. percentage of fines multiplied by the soil's D10 value; and (iii) a multivariate analysis using percentage of fines, geometric mean particle diameter dg, and geometric standard deviation {sigma}g, as the independent variables.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 Hanford Site Background
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
SUMMARY
 REFERENCES
 
Darcy's Law Estimates of Drainage
Table 2 shows Darcy's Law drainage estimates from K({theta}) derived from the five different methods and the measured drainage for the 300 N lysimeter sandy soil. Field-measured soil-water pressures confirm that unit gradient conditions exist in the 300 N lysimeter (Gee, 1987; Sisson et al., 2002) so use of the unsaturated conductivity value at field water content and soil-water pressures is justified. The results indicate that the five hydraulic conductivity functions varied widely in their estimates of drainage at a soil-water pressure head of –0.4 m and a field water content of 0.09 m3 m–3. The pressure head of –0.4 m is typical of what is found below the top 2 m in this bare sandy soil at the 300 N lysimeters (Gee, 1987).


View this table:
[in this window]
[in a new window]
  		Table 2. Darcy's Law drainage estimates from unsaturated hydraulic conductivity functions for 300 North Lysimeter Site (300 N) sand. Assumes a unit-gradient condition at a soil-water content of 0.09 m3 m–3 and a soil-water pressure head of –0.40 m.

 
The five methods used to generate the unsaturated conductivity functions resulted in differences of more than three orders of magnitude. Compared to lysimeter (direct) measurements, the instantaneous profile method of Watson (1966) provided the best estimate of drainage, while the Guelph permeameter provided the worst estimate, with particle size (Arya–Paris), laboratory column, and the centrifuge estimates of hydraulic conductivity exhibiting intermediate values between the two extremes. The Guelph Permeameter assumes a Gardner-type exponential function which, when fitted to data near saturation, tends to overestimate K({psi}) at lower water contents and pressures. The steady state laboratory column also overestimated drainage with the K({psi}) function, but slightly underestimated the drainage with the K({theta}) function. This may be accounted for, in part, by the hysteretic nature of the water-retention characteristic. The soil-water pressure measured in the field is consistently higher (less negative) than that measured in the laboratory, suggesting that the field soil is on a wetting characteristic curve while the laboratory is on a drying characteristic. It is clear that the hysteretic nature of soil water must be considered when using the laboratory-derived hydraulic functions for estimating field drainage. Similar observations of hysteresis effects on drainage have been observed for other Hanford Site soils (Tyler et al., 1999).

Texture Models
The soils used to develop the Ef functions range from a fine silt loam to a coarse sandy gravel. Three of the soils contain similar percentage of fines (fraction less than 50 µm), yet are markedly different when comparing the total grain-size distribution. This is apparent when reviewing the calculated soil statistics (Table 1) and particle size distribution curves (Fig. 1). For example, dg for the sand is 0.506 mm, which is appreciably smaller than the dg for the sandy gravel and SWL soils, each being 7.529 mm and 4.687 mm, respectively.

The cumulative drainage for the 300 N and FLTF lysimeters for the time period used for the model calibration and the cumulative drainage for the SWL are shown in Fig. 2 . Note that measurements of drainage at the SWL site did not begin until July 1996. Breaks in the 300 N drainage data points represent interruptions in project funding. During these times, drainage water remained in the base of the lysimeter until funding allowed for measurements to resume. Table 3 shows the average winter precipitation, as measured at the HMS, the average drainage, and the calculated evaporation factor Ef, for the five lysimeters during the time period used for the model calibration. In addition, average winter precipitation, average drainage, and the calculated Ef for the sand lysimeter is shown from January 1982 to March 1993.



View larger version (19K):
[in this window]
[in a new window]
  		Fig. 2. Cumulative drainage for the 300 North (300 N) Lysimeter Site and Field Lysimeter Test Facility (FLTF) lysimeters from 1 Jan. 1995 to 1 Apr. 2004, and cumulative drainage for the Hanford Solid Waste Landfill (SWL) lysimeter from 1 July 1996 to 1 Apr. 2004.

 

View this table:
[in this window]
[in a new window]
  		Table 3. Average winter (November–March) precipitation, average drainage, and calculated evaporation factor, Ef, from lysimeter records. Measurements for the sand soils are from the same lysimeter but for different time periods.

 
For the soils used in developing the Ef relationship, drainage decreases as the soil becomes finer, with the silt loam soil experiencing no drainage. This trend is consistent with an increase in field capacity as soil texture becomes finer. The similarity of the calculated Ef for both the early and later time sand data sets illustrates the stability of Ef with differing precipitation conditions.

Figure 3 shows the relationships between Ef and two separate soil textural descriptors, with the developed Ef functions presented in Table 4. The relationship between Ef and percentage of fines used by Gee and Ward (2002) does not illustrate the difference in observed drainage from the sand and sandy gravel soils. While both soils possess the same fraction of fines, the presence of coarser soil particles in the sandy gravel decreases its overall storage capacity. To correct for the presence of a coarser soil fraction, two approaches were explored: (i) multiply the percentage of fines by the soil's D10 value and (ii) perform a multivariate analysis using percentage of fines, dg, and {sigma}g as the independent variables. From Fig. 3, use of D10 in combination with percentage of fines allows the sandy gravel soil to be better represented. This improves the Ef relationship, yet retains the influence of the fine soil fraction in the overall evaporation rate. From Table 4, the use of multivariate analysis results in a r2 value of unity due to the degrees of freedom of the regression being equal to zero. While this essentially renders statistical evaluation of the regression model meaningless, the model is included in the overall analysis to demonstrate the potential of using different statistical models to estimate Ef values. It is expected that as data from additional sites become available the regression models will be strengthened by their addition to the calibration data set. Future Ef regression models should include analysis such as the Akaike Information Criterion (Sakamoto et al., 1986; Yu et al., 1997) to identify when the addition of independent variables is offset by a significant reduction in regression error, while still retaining enough degrees of freedom to evaluate the regression model.



View larger version (13K):
[in this window]
[in a new window]
  		Fig. 3. Relationship between the evaporation factor, Ef, and a) percentage of fines; b) percentage of fines x D10. The solid line represents the developed Ef function.

 

View this table:
[in this window]
[in a new window]
  		Table 4. Relationship between Ef and soil textural descriptors, goodness of fit, r2, predicted 300 North Lysimeter Site (1981–1993) sand and Hanford Solid Waste Landfill drainage from developed Ef functions, and the corresponding measured drainage.{dagger} The 300 N sand data set is from a 12-year record not used in the model calibration.

 
Application of the texture model using the two separate Ef functions for the 300 N sand drainage during a 12-yr period not included in the calibration data set and the independent drainage data from the SWL are presented in Table 4. For the 300 N sand, drainage was accurately predicted using the multivariate Ef function and, to a lesser degree, from the percentage of fines and D10 Ef function. Comparison of the developed Ef functions versus the independent drainage data from the SWL shows that for both relationships, the model underpredicts the observed Ef of 70 mm yr–1, resulting in an overpredicted drainage. The Ef function developed using multivariate analysis provided the best estimate of SWL drainage, calculating a drainage of 66 mm yr–1 compared to a measured drainage rate of 51 mm yr–1. The Ef function using the percentage of fines x D10 descriptor performed similar to the multivariate analysis, estimating a drainage of 70 mm yr–1.

No attempt was made to keep the surface of the SWL vegetation free, resulting in the potential for water uptake by plants. Indian wild rice has invaded the site and while sparse, it is a perennial growth form that has a relatively deep rooting depth and most likely has contributed to the lower drainage rates observed at this site. Because the Ef functions were developed using drainage data from nonvegetated soils, the presence of vegetation would result in the underprediction of Ef as was seen in the results. Figure 4 shows the SWL Ef from 1996 through 2004 calculated from measured drainage and precipitation. A general increase in Ef with time is observed, correlating with a decrease in drainage. The trend is likely because drainage will continue to decrease with further establishment of vegetation at the SWL. These data illustrate the significance of vegetation in influencing water loss rates from surface soils. Where surfaces remain barren, such as at radioactive waste sites at Hanford, texture modeling is expected to provide reasonable predictions of drainage. Texture modeling was designed to be used at waste sites where the percentage of fines range from a few to more than 60% (i.e., coarse gravels to silt–loam surfaces).



View larger version (9K):
[in this window]
[in a new window]
  		Fig. 4. Hanford Solid Waste Landfill (SWL) Ef over time as calculated from SWL drainage and precipitation records.

 
Not presented here are data from lysimeters containing coarse rock fragments or gravels with little or no fines. Open, coarse sediments with no fines are conditions at Hanford (see climate discussion above) that allow for significant advective heating during late spring through fall because of high thermal loading of the soil surface (Gee et al., 1997; Ward and Gee, 2000; Gee et al., 2002). Evaporation rates for clean gravels have been measured by Kemper et al. (1994), and diffusion theory has been applied to estimate vapor losses as a function of the gravel size, wind speed, and humidity. Such an analysis does not apply directly to the bare sediments that are described in our tests because the materials are mixed with fines, and thus the experimental conditions are different. Because the surfaces of waste sites at Hanford generally contain at least a few percentage of fines, lysimeter data on open-work rock (e.g., basalt rip rap) and gravel materials were not included in the analysis. Where such open-work materials might exist, such as on sideslopes on above-grade surface covers at some waste sites, the texture model developed here would tend to overpredict the drainage and thus provide a conservative estimate for the overall drainage at a given waste site.

There has been some effort over the past 10 yr to stabilize the surface of Hanford waste sites for worker protection against radioactivity and for ease of access. The surfaces of a number of the tank farms have been recovered with commercial road-base material. While no exact specifications are available for the size distribution of this gravelly material, the major specification is that it contains what is called 3/4 minus material, meaning that the majority of the material passes through a 3/4-inch square sieve. Since retrieving material from within the tank farms is difficult, we collected some road-base material from adjacent to the FLTF and used the texture analysis from this material to predict the average annual drainage rates for such materials that might exist over current waste sites at Hanford. Table 5 shows the size-distribution statistics and the expected drainage rates for a bare soil with 3/4-inch road-base and two synthetic soils consisting of road-base material with the addition of 10% fines and 20% fines. Predictions indicate that drainage would be substantial at all sites where road-base surfaces now exist over Hanford waste sites. However, modest increases in the percentage of fines significantly reduce the drainage. An increase in percentage of fines from 4 to 14% reduces the predicted drainage by as much as a factor of four. An additional increase in percentage of fines to 24% further reduces the predicted drainage to less than 20 mm yr–1. Calculations of drainage from the synthetic soils assume that the thickness of the synthetic materials is at least 1 m. As a first approximation, if the synthetic layer is less than 1 m the storage would be altered in proportion to the layer thickness; storage and hence drainage are functions of layer thickness, which is implicit in the empirical models presented here.


View this table:
[in this window]
[in a new window]
  		Table 5. Grain-size statistics and predicted drainage for commercial road-base material, representing tank farm surface cover, and two synthetic soils, each consisting of road-base material with an addition of 10 and 20% fines.

 
The two empirical texture models have obvious limitations. These include restriction to bare surfaces and assumptions of local climate conditions and soil layer thicknesses. In spite of these limitations the models do surprisingly well in predicting drainage from bare waste sites. Where more site data are available, more sophisticated models may be justified. However, it should be noted that employing deterministic codes such as UNSAT-H (Fayer and Gee, 1997), HELP (Schroeder et al., 1994), or a recently developed transient water budget model (Simmons and Meyer, 2000) all require hydrologic data such as water-retention characteristics and unsaturated hydraulic conductivities of the surface soil, presently unavailable for any of the tank farms and a number of the solid-waste burial grounds, waste cribs, or trenches. Even if hydrologic data were available, calibration is still required for such models to correctly predict drainage. For example, the HELP model (Schroeder et al., 1994) uses an empirically derived factor, called the evaporative depth, to scale the potential evaporation for a given soil and plant type. There is no theoretical basis for this factor. For the Hanford Site it was found that the evaporative depth in the HELP code changed with time, overpredicted drainage, and to calibrate the model, HELP required an unrealistic evaporative depth deeper than the 1.5 m–deep lysimeters used in the study (Martian, 1994; DOE, 1996). The physically based UNSAT-H code also overpredicted drainage from bare soils (Fayer and Gee, 1997; Tyler et al., 1999). These results suggest that the ability of water-balance codes to estimate evaporation from the surfaces of desert soils using deterministically measured parameters is still inadequate, and the use of empirical models that are less stringent may be satisfactory at the present time until better methods can be developed to obtain more robust and reliable estimators for evaporative losses from desert soils.

Several options are available for improving drainage predictions for waste sites at Hanford. First, it is important to continue to maintain the lysimeter records on which simple texture models are based. Continued monitoring of drainage under a wider range of climate variables (i.e., more extremes in precipitation and temperatures) over longer periods of time would build confidence in the range over which the model will perform. Second, as a minimum, it will be important to obtain textural data and layer thickness data directly from waste burial grounds. Installation of lysimeters within waste sites is an option that has been considered, but at present, no full-scale lysimeter installation exists in a Hanford waste site. Instrumenting and monitoring waste sites will help for verifying predictions. Then, as more information is available, the model can be modified to incorporate the known layer thicknesses for the bare surfaces explicitly. In the interim, the texture model should provide at least first-order approximations of expected drainage for bare waste sites at Hanford.

Estimates of drainage rates from Darcy's Law using hydraulic property data may have large uncertainties, unless those estimates are tempered by calibration with actual field drainage data. Direct measures of drainage either at waste sites of concern or adjacent to them but with similar surface soil conditions will be useful in calibrating both simple models, such as those described here, or more complex models that require well-defined hydraulic properties. While the texture models developed here are specific to the Hanford Site, the general approach may be applicable to other sites where similar conditions exist (i.e., bare soil surfaces where drainage is dominated by winter precipitation events).


   SUMMARY
TOP
 ABSTRACT
 INTRODUCTION
 Hanford Site Background
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 SUMMARY
REFERENCES
 
Drainage estimates using Darcy's Law, with five separate methods to estimate the hydraulic conductivity of a sand at the Hanford Site, varied over three orders of magnitude. The best estimate of actual drainage was from field-measured instantaneous-profile data, followed by the steady-state column method, the particle-size method, and the ultracentrifuge method. The worst estimate was obtained from the Guelph Permeameter. The drainage of meteoric water was also estimated at several waste burial grounds at the Hanford Site using two texture models. The models were based on climatic variables and surface soil textures specific to the Hanford Site. Winter (November–March) precipitation records for the past 22 yr were combined with soil-texture data obtained from waste-site soil samples.

The models, calibrated using lysimeter drainage data from bare surface sediments ranging from silt loams to coarse gravel and rock fragments, were tested against a 12-yr record for bare sand and found to predict the drainage with precision of better than 15%. In the absence of hydraulic property information for bare surface sediments, textural information appears to be useful for estimating drainage at other waste sites that are kept barren. Multiyear predictions for an uncalibrated site, the SWL, indicated agreement with the multivariate analysis texture model but with an overprediction of drainage by about 30%. The overprediction was attributed to the sparse vegetation at the SWL site, which decreased the actual drainage over the years of testing.

The results illustrate that both percentage of fines (<50 µm) and presence of vegetation contributed to decreased drainage, and the addition of relatively small quantities of fines to the surface soil can promote higher evaporation, which leads to reduced drainage, and potentially limiting contaminant migration. The textural model may be useful in cases where direct measurements are absent, and it can help determine how much fine soil could be added to significantly reduce the drainage rate in the future.


   ACKNOWLEDGMENTS
 
The Pacific Northwest National Laboratory is operated for the U.S. Department of Energy (DOE) by Battelle under Contract DE-AC05-76RL01830. Funding for this work was provided by the DOE Remediation and Closure project for the Richland Operations Office of the Department of Energy and the Remediation Decision Support Task of the Groundwater Remediation Program managed by Fluor Hanford, Inc. for the U.S. Department of Energy. We acknowledge the help of Dr. John Nimmo of the U.S. Geological Survey, Menlo Park, California, and his staff in supplying the ultracentrifuge analysis.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 Hanford Site Background
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 SUMMARY
 REFERENCES
 




This article has been cited by other articles:


Home page
 Vadose Zone JHome page
G. W. Gee, M. Oostrom, M. D. Freshley, M. L. Rockhold, and J. M. Zachara
Hanford Site Vadose Zone Studies: An Overview
Vadose Zone J., November 20, 2007; 6(4): 899 - 905.
[Abstract] [Full Text] [PDF]

Home page
 Vadose Zone JHome page
D. G. Meadows, M. H. Young, and E. V. McDonald
Estimating the Fine Soil Fraction of Desert Pavements Using Ground Penetrating Radar
Vadose Zone J., May 26, 2006; 5(2): 720 - 730.
[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 (3)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Gee, G. W.
Right arrow Articles by Ward, A. L.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Gee, G. W.
Right arrow Articles by Ward, A. L.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Gee, G. W.
Right arrow Articles by Ward, A. L.
Related Collections
Right arrow Water Flow Models
Right arrow Recharge
Right arrow Disposal Facilities

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