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


     

Published online 18 July 2005
Published in Vadose Zone J 4:505-521 (2005)
DOI: 10.2136/vzj2004.0067
© 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 Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (6)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Springer, E. P.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Springer, E. P.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Springer, E. P.
Related Collections
Right arrow Statistics
Right arrow Hydraulic Conductivity

SPECIAL SECTION: LOS ALAMOS NATIONAL LABORATORY

Statistical Exploration of Matrix Hydrologic Properties for the Bandelier Tuff, Los Alamos, New Mexico

Everett P. Springer*

Mail Stop J 495, Atmospheric, Climate and Environmental Dynamics Group Los Alamos National Laboratory, Los Alamos, NM 87545
* Corresponding author (everetts{at}lanl.gov)

Received 26 April 2004.



   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION AND CONCLUSIONS
 REFERENCES
 
The Bandelier Tuff forms the top of the thick vadose zone at Los Alamos National Laboratory (LANL). Subsurface pathway performance or risk assessments at Los Alamos require characterizing the unsaturated hydrologic properties of the Bandelier Tuff. The objective of this study was to analyze available unsaturated zone matrix hydrologic properties data for the Bandelier Tuff from the Los Alamos area for the purpose of developing estimates of properties and parameters for modeling vadose zone flow and transport. The hydrologic properties bulk density, saturated hydraulic conductivity, and the van Genuchten equation parameters, residual and saturated water contents, {alpha}, and n, were measured or estimated from measured data. A nonparametric statistical approach was used because of the small sample size leading to nonnormal distribution of the samples. The following differences were investigated: (i) between boreholes for a given unit within a site and (ii) between sites for a given unit. Results for borehole analyses within a location found that properties were different at one location and could be combined at another. The testing between locations found that samples could be readily combined. Linear correlation analyses of the one geologic unit that was the most consistent across the Los Alamos site found that there was essentially no relationship between matrix hydrologic properties or parameters to allow surrogate properties to be used for prediction. This preliminary analysis indicated that investigations at LANL will require more detailed sampling on a site-specific basis because properties cannot be transferred or estimated with any confidence.

Abbreviations: LANL, Los Alamos National Laboratory


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION AND CONCLUSIONS
 REFERENCES
 
LOS ALAMOS NATIONAL LABORATORY (LANL) is located on the Pajarito Plateau in north-central New Mexico. The plateau is bounded on the west by the Jemez Mountains and on the east by the Rio Grande (Fig. 1) . Erosion of the plateau has created a number of finger-shaped mesas that trend west to east, and these mesas are separated by steep canyons that contain intermittent and ephemeral streams. The cap rock of the Pajarito Plateau is the Bandelier Tuff that was deposited in a series of eruptions from the Jemez volcanic center, with the last episode occurring approximately 1 Ma (Griggs, 1964). The water table for the regional aquifer beneath mesas at the Laboratory lies at depths ranging from 365 m along the western edge to 183 m along the eastern edge of LANL. This creates a thick vadose zone at LANL that has been used by the Laboratory for disposal of municipal, hazardous, and radioactive wastes, with radioactive waste disposal continuing today. The vadose zone is an important element of the LANL environmental restoration activities because many of the legacy sites are located in the Bandelier Tuff. Knowledge of vadose zone hydrologic properties is critical to assess risks from past, present, and future operations at LANL. In addition, the work at LANL adds to the general knowledge of vadose zone hydrologic response.



View larger version (80K):
[in this window]
[in a new window]
  		Fig. 1. Map of Los Alamos National Laboratory with the locations of the boreholes used in the unsaturated hydrologic properties analysis.

 
The process of describing risks associated with contamination or waste disposal in the vadose zone requires a model to predict travel times of contaminants of concern. The unsaturated zone hydrologic properties required by mathematical models include the water retention curve (water content versus pressure head) and hydraulic conductivity vs. pressure head curve, the saturated hydraulic conductivity (Ks), and porosity. These properties have been shown to be spatially variable (Nielsen et al., 1973)—a property such as Ks can vary by orders of magnitude in a few meters (Koltermann and Gorelick, 1996). Both the water retention and hydraulic conductivity curves are nonlinear, which means that the hydraulic conductivity can vary by several orders of magnitude for a small change in pressure head at low water contents. Unsaturated properties of the soil or rock matrix are derived mostly by laboratory sample analyses; these procedures can be tedious and expensive, so characterizing the variability is important to ensure efficient sampling.

The stratigraphy at LANL by Broxton and Vaniman (2005) shows that the upper units of the vadose zone are in the Bandelier Tuff. Groundwater pathway assessments at Los Alamos focus on the Bandelier Tuff because of its location in the stratigraphic column and because contaminant sources are located on or in the Bandelier Tuff. The Bandelier Tuff is composed of two members: the upper member is the Tshirege and the Otowi is the lower member. The Cerro Toledo interval is a deposition unit formed during the period between the eruptions of the Otowi and Tshirege Members, and it has not been extensively characterized. The Tshirege Member has been divided into cooling units based on lithological analyses, and most sampling for hydrologic properties has used cooling units as an initial classification scheme for sample locations. This is similar to the approach taken by Flint (1998) for Yucca Mountain, Nevada, where initial classification of hydrogeologic units was based on degree of welding.

Risk assessments at LANL for groundwater pathways require statistical analyses of unsaturated zone hydrologic properties because of the limited sampling and large spatial variability in these properties. A number of samples of unsaturated zone hydrologic properties have been collected at Los Alamos National Laboratory over the years as site investigations have progressed. These samples have been used on a site-specific basis for assessments (Hollis et al., 1997; Birdsell et al., 2000), but no coherent analyses of these samples have been undertaken to examine statistical similarity across LANL. The goal of the unsaturated zone hydrologic parameter estimation project at Los Alamos is the same as that given by Flint (1998): to provide statistical properties of the unsaturated zone properties to support vadose zone modeling. The objective of this study was to analyze the existing data collected from different locations at LANL to determine if more accurate estimates of unsaturated zone matrix properties can be obtained by combining or pooling values from a sample location and across locations.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION AND CONCLUSIONS
 REFERENCES
 
Sample Locations
The boreholes from which the samples were collected are listed in Fig. 1. The borehole designator (e.g., 49-2-700) usually indicates in which LANL technical area the borehole is located. Figure 1 shows that most boreholes are located in the eastern and northern parts of the Laboratory where many environmental restoration and waste management activities have occurred. A number of boreholes are located at Technical Area 54, and these boreholes are further identified in Fig. 2 . Canyons were assumed to represent a different setting for this analysis because of the wetter conditions and reduced stratigraphy that affect weathering of the tuff. Samples from three canyons, Cañada del Buey, Mortandad, and Potrillo, are included in the data set, and these canyons are identified in Fig. 1. Sampling of canyons is not limited to the canyon locations as the LADP-3 borehole from Technical Area 21 is a canyon setting borehole.



View larger version (81K):
[in this window]
[in a new window]
  		Fig. 2. Map of Los Alamos National Laboratory Technical Area 54 with the locations of the boreholes used in the unsaturated hydrologic properties analysis.

 
Geology
The Bandelier Tuff is composed of the lower Otowi Member, which is a massive pumiceous ash-flow tuff, and the Tshirege Member, which is composed of four ash-flow tuff cooling units (Fig. 3) . Because the Cerro Toldeo interval is a volcanoclastic sediment it is not considered part of the Bandelier Tuff. Analyses of hydrologic properties use the basic Tshirege cooling units in Fig. 3 because further delineation for many of the boreholes was not made. Although vertical fractures are present in the Tshirege Member and these fractures can be hydrologically active with clay filling found to a depth of 3 m below the surface (Soll and Birdsell, 1998), this study deals only with matrix hydrologic properties of the Bandelier Tuff. Further description and details on geology and stratigraphy of the Bandelier Tuff can be found in Broxton and Vaniman (2005). The nomenclature in Fig. 3 is used in this paper, so the Tshirege Member is identified with a Qbt and then the cooling unit is designated. For instance, Tshirege Cooling Unit 2 is Qbt 2. The Otowi Member is not differentiated, and it is designated as Qbo.



View larger version (41K):
[in this window]
[in a new window]
  		Fig. 3. Stratigraphic nomenclature for the Bandelier Tuff. In this study, the major cooling units such as Qbt 3 in the Tshirege Member were used, and no identification of the subunits was made.

 
Sample Characterization
Hydrologic measurements used in this study are basically from three data sources, and a variety of techniques have been used to determine Ks and the moisture characteristic. Nyhan (1979) determined Ks on 12-cm-long cores from Technical Area 21 using constant head methods, and these are designated MDAT in the Appendix. Kearl et al. (1986) measured porosity using helium injection, Ks using nitrogen injection, and moisture retention with a centrifuge on Boreholes LGM-85-06, LLC-85-11, LLC-85-14, LLC-85-15, LLC-86-22, LLM-85-01, LLM-85-02, and LLM-85-05 from Technical Area 54. Samples from the other boreholes listed in the Appendix were characterized by Daniel B. Stephens and Associates in Albuquerque, NM. The methods used by Daniel B. Stephens and Associates varied between samples sets submitted by LANL; the details are too lengthy for this paper. The Ks was measured using both constant head and air injection techniques. Bulk density was obtained from oven-dry analysis of samples and porosity ({phi}) was calculated from {rho}b. The moisture release curve was obtained by pressure plate, submerged pressure outflow cells, or thermocouple psychrometer, depending on the magnitude of the pressure.

Dimensions for the properties are {rho}b (g cm–3), Ks (cm s–1), {alpha} (cm–1), and the water content values are (cm3 cm–3).

Data Analyses
The water content and pressure head data were fitted to the van Genuchten (1980) moisture characteristic equation using the computer code RETC (van Genuchten et al., 1991). The water retention relation is described by

[1]
where Se is the effective saturation, {theta} is the volumetric water content (cm3 cm–3), {theta}r is the residual volumetric water content (cm3 cm–3), {theta}s is the saturated volumetric water content (cm3 cm–3), h is pressure head (cm), {alpha} is a fitting parameter (cm–1), and n and m are fitting parameters, with m = 1 – 1/n.

The change in hydraulic conductivity with water content is described by the following equation:

[2]
where Ks is the saturated hydraulic conductivity (cm s–1), and {ell} is a pore-connectivity parameter (set to 0.5 for these analyses).

van Genuchten and Nielsen (1985) demonstrated the importance of the slope of the water retention curve (Eq. [1]) near saturation on predicted hydraulic conductivity (Eq. [2]) for the entire range of water content values. {theta}s can be measured using imbibition, but this approach was not performed with these data. van Genuchten and Nielsen (1985) presented a rationale that any model such as Eq. [1] obscures the description of the water content at saturation, leaving the definition of {theta}s model dependent. Springer et al. (2000) found no statistical difference for {alpha} and n from Eq. [1] when using a fitted vs. a fixed value such as porosity for {theta}s. There were significant differences for {theta}r, but these were not viewed as critical because for most cases {theta}r is zero.

Rogers and Gallaher (1995) analyzed the Bandelier Tuff data available at that time. They went through each moisture retention data set and censored points that were obvious outliers. The parameters derived by Rogers and Gallaher (1995) were used in this study and are reported in the Appendix.

Statistical Analysis
The statistical analyses used nonparametric methods to test for differences in matrix hydrologic properties between boreholes within a technical area, between technical areas, and between mesa top and canyon settings. For conditions where only two samples were compared, the Mann–Whitney U test was used, and when more than two groups are available, the Kruskal–Wallis analysis of ranks test was used (Sokal and Rohlf, 1969). All statistical analyses were performed using the STATISTICA software (StatSoft, Tulsa, OK; http://www.statsoftinc.com/). A post-hoc analysis of the means for the Kruskal–Wallis test is available to further define differences between the samples (Sokal and Rohlf, 1969).

Although nonparametric tests were used, the values for both Ks and {alpha} were transformed using the base 10 logarithm (log10) to reduce the skew of their distribution and to correct for hetroscedasiticity. Ks has been shown to be lognormally distributed (Nielsen et al., 1973).


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION AND CONCLUSIONS
 REFERENCES
 
Sample Size
The number of samples available is critical for determining statistical properties. The number of samples for each cooling unit and technical area for Ks are listed in Table 1. Sample sizes for Ks are the maximum available for all parameters used in the analyses that follow, so Ks represents the best case because there are fewer samples for {rho}b, {theta}s, {theta}r, n, and {alpha}.


View this table:
[in this window]
[in a new window]
  		Table 1. The number of samples for each technical area (TA) and unit for log10 saturated hydraulic conductivity (Ks).

 
Information in Table 1 and Fig. 1 indicate a skewed distribution of samples from the eastern part of LANL. The sums in the last row of Table 1 reveal that more than 60% of the samples for Ks and a similar percentage for the other matrix parameters (not shown) come from two locations, Technical Areas 21 and 54. The sampling of these eastern technical areas may bias the estimates of the matrix hydrologic properties if applied to other locations around LANL because distance from the source area, differences in stratigraphy, and weathering affect the hydrologic properties. The data in Table 2 reveal the unequal sampling of Bandelier Tuff cooling units that has occurred. Unit 4 (Qbt4) is sampled only at Technical Area 49 because it is not found to the east. Although better distributed, Qbt3 is sampled at three locations and is not found at Technical Area 54. The last column in Table 1 provides the sample numbers across cooling units, and it appears to be relatively evenly distributed except for Qbt4. There are not a large number of samples in Table 1, nor are these samples well distributed across LANL. Statistical analyses represents a logical approach to determine whether the matrix hydrologic properties from the various locations, or even within a location if there are multiple boreholes, can be combined to estimate hydrologic parameters for modeling assessments.


View this table:
[in this window]
[in a new window]
  		Table 2. Differences between boreholes for matrix hydrologic properties from Technical Area 21 for each cooling unit of the Tshirege and the Otowi Members of the Bandelier Tuff using the Mann–Whitney test (significant if p ≤ 0.05).

 
Residual Water Content
All values for {theta}r were estimated with RETC. An alternative approach proposed by Flint (1998) measured {theta}r as the differences between over-dry weight at 105°C and weight from drying at 60°C and 65% relative humidity. A total of 200 estimates for {theta}r were performed, and 126 (63%) of these were estimated to be zero. The resulting distribution is skewed with zero having a significant finite probability of occurrence. This mixed distribution (Yevjevich, 1972) is difficult to transform, and a more appropriate approach for estimating statistical properties of {theta}r is through an empirical cumulative density function. Although results will be reported for {theta}r in the following sections, they have to be weighed against its distribution. One option for unsaturated flow modeling of the Bandelier Tuff is to set {theta}r to zero given the large number of zero values that were estimated with this data set.

Technical Area Analysis
Within a Technical Area
The initial analysis examined whether the data from boreholes in a given technical area can be pooled. From Table 1, it is readily observed that Technical Areas 21 and 54 have the most samples, and within–technical area analyses were performed for these sites. The samples were sorted by Tshirege cooling units and the Otowi geologic unit, and the applied nonparametric test was dependent on the number of sample locations within the technical area.

The analyses for Technical Area 21 are presented in Table 2. There were a total of four sampling locations at this technical area, three boreholes (LADP-3, LADP-4, and 21-2523) and Nyhan (1979) samples from MDAT. The number of sample locations for any given unit was only two so the Mann–Whitney test was used for analyses.

The log10Ks values for Tshirege Units 2 and 3 and the Otowi samples were shown to be significantly different (Table 2). The inability to combine the samples of this key flux property, Ks, for a site-wide estimate of this property's distribution indicates that subsurface pathway assessments at Technical Area 21 for Qbt 2, Qbt 3, and Qbo will require either (i) collection of additional samples of Ks or (ii) making additional assumptions about the distribution of this property using these data. The analyses in Table 2 for Qbt 1v and Qbt 1g indicated that matrix hydrologic properties and parameters for these units can be combined for Technical Area 21, but these results must be viewed in relation to the limited sample size, which was 9 for Qbt 1v and 6 for Qbt 1g. In addition to the log10Ks, both {rho}b and {theta}s for the Otowi Member tested as different for the two locations. The samples for Obo were obtained from one borehole that was drilled in a mesa top (LADP-4), while LADP-3 is a canyon setting borehole, making this test an initial examination of the differences between these two settings where the canyon settings have higher moisture content because of recharge by streamflow. Summarizing the analysis of Technical Area 21, the matrix hydrologic properties and parameters cannot be readily merged for Tshirege Cooling Units 2 and 3 and the Otowi Member of the Bandelier Tuff, constraining unsaturated zone modeling to a site-specific basis in Technical Area 21 because model parameters are nonuniform.

Analyses of samples from Technical Area 54 are presented by Tshirege Member cooling unit in Tables 3, 4, and 5. From these tables, the larger number of boreholes at Technical Area 54 (17 total) leads to a more limited sample per borehole with only one or two samples per cooling unit as compared with Technical Area 21 (Table 2). The reduced effective sample size per borehole at Technical Area 54 decreases the ability to detect significant differences in matrix properties, especially when variability is large. The only significant difference that was detected for Technical Area 54 was for the van Genuchten n parameter for Unit 1v (Table 4). All remaining parameters for the three units, Qbt 2, Qbt 1V, and Qbt 1g, can be combined into a single sample for the data provided. The Otowi Member has not been sampled from a mesa top location at Technical Area 54.


View this table:
[in this window]
[in a new window]
  		Table 3. Statistics for Bandelier Tuff Tshirege Member Cooling Unit 2 (Qbt 2) for Technical Area 54 by borehole.

 

View this table:
[in this window]
[in a new window]
  		Table 4. Statistics for Bandelier Tuff Tshirege Member Cooling Unit 1v (Qbt 1v) for Technical Area 54 by borehole.

 

View this table:
[in this window]
[in a new window]
  		Table 5. Statistics for Bandelier Tuff Tshirege Member Cooling Unit 1g (Qbt 1g) for Technical Area 54 by borehole.

 
Using Technical Areas 21 and 54 as surrogates for the LANL site because these locations had the largest sample sizes available, it was found that there was a limited ability to combine or pool matrix hydrologic properties by Bandelier Tuff unit for a given technical area. The analyses indicated that the flux parameter, log10Ks cannot be combined at Technical Area 21, and this presents a major limitation for parameter estimation because vadose zone transport calculations are very sensitive to Ks values. Further analyses by geologic unit across the LANL site are limited by these results.

Comparisons between Technical Areas
The next step of the analysis was to examine the relationships between properties for different technical areas for selected units of the Bandelier Tuff on the basis of the previous analyses. Given the borehole analyses, the ability to pool samples for LANL estimated matrix hydrologic properties from the technical areas is limited. This analysis used the same nonparametric statistical approaches as the analysis within technical areas, except that technical area was the main effect.

Tshirege Unit 3 (Qbt 3).
This unit is found at Technical Areas 16, 21, and 49, and Kruskal–Wallis analyses results are presented in Table 6. For Technical Area 21, the properties and parameters other than log10Ks are from Borehole 21-2523. All properties tested as significantly different for this unit. As noted above, log10Ks has already tested significantly different for Technical Area 21. Further insight into differences between locations can be obtained from Fig. 4 , which contains the median and range for {rho}b (Fig. 4A) and log10{alpha} (Fig. 4B).


View this table:
[in this window]
[in a new window]
  		Table 6. Statistics and Kruskal–Wallis results for Bandelier Tuff Tshirege Member Cooling Unit 3 (Qbt 3) by technical area.

 


View larger version (28K):
[in this window]
[in a new window]
  		Fig. 4. Box and whiskers plots of the median, 25 to 75% quartile, and range for (A) bulk density ({rho}b) and (B) log10{alpha} from Bandelier Tuff Tshirege Unit 3 (Qbt 3) at designated technical areas of Los Alamos National Laboratory.

 
The range of {rho}b at Technical Area 16 is large compared with either Technical Area 21 or 49 (Fig. 4A), and this pattern is repeated for the other parameters in this group. Technical Area 16 is the western-most sampled location in Fig. 1, and it is the closest to the volcanic source area of the Bandelier Tuff. Also, at Technical Area 16 additional Tshirege Member cooling units are found above Qbt 3 that will affect its welding, cooling history, and weathering. No overlap exists for {rho}b between Technical Area 21 and 49, and these two locations can be viewed as different. The same description for log10{alpha} (Fig. 4B) reveals a large range for Technical Area 16, but Technical Area 21 and 49 do overlap more that for {rho}b. As observed with the borehole analysis, the unsaturated zone matrix hydrologic properties for Qbt 3 cannot be pooled to create a larger data set.

Tshirege Unit 2 (Qbt 2).
Samples for Qbt 2 were from Technical Areas 21, 49, 53, and 54 (Table 1), and, with two samples available from Technical Area 53, this location was eliminated from the analyses. It has already been determined that log10Ks cannot be pooled (Table 2), so this property was eliminated from the analysis. The Kruskal–Wallis results are presented in Table 7. {rho}b, {theta}s, and {theta}r are significantly different between technical areas, while the van Genuchten parameters n and log10{alpha} can be pooled for these areas. The median, quartile, and range of {rho}b and log10{alpha} are presented in Fig. 5A and 5B , respectively. For {rho}b that was significant, the range for Technical Area 54 is narrower, and overall values are lower than either Technical Area 21 or 49 (Fig. 5A). For log10{alpha}, the ranges at all three technical areas substantially overlap each other (Fig. 5B), providing a basis for combining the values from the technical areas to create a LANL estimated {alpha} for unsaturated zone modeling.


View this table:
[in this window]
[in a new window]
  		Table 7. Statistics and Kruskal–Wallis results for Bandelier Tuff Tshirege Member Cooling Unit 2 (Qbt 2) by technical area.

 


View larger version (32K):
[in this window]
[in a new window]
  		Fig. 5. Box and whiskers plots of the median, 25 to 5% quartile, and range for (A) bulk density ({rho}b) and (B) log10{alpha} from Bandelier Tuff Tshirege Unit 2 (Qbt 2) at designated technical areas of Los Alamos National Laboratory.

 
Unlike Qbt 3, the van Genuchten parameters for Qbt 2 can be pooled to provide a LANL estimate for these technical areas. It is not possible to extend these LANL estimated parameters to the west of Technical Area 49 without further sampling, but the large region covered by these technical areas provides parameter estimates to a substantial portion of LANL.

Tshirege Unit 1v (Qbt 1v).
Technical Areas 21, 49, 53, and 54 (Table 1) were sampled, and only two samples were available from Technical Areas 49 and 53 so these two units were excluded from the analyses even though the statistics for these technical areas are presented in Table 8. A Mann–Whitney test was used rather than the Kruskal–Wallis because only two technical areas were analyzed. The analysis of this unit for Technical Area 54 indicated that the van Genuchten n parameter was significantly different, so this parameter was excluded from the analysis. The statistics and the resulting significant differences are presented in Table 8, with {rho}b and log10{alpha} being statistically different between technical areas. The number of samples from Technical Area 54 was approximately four times larger than from Technical Area 21. This is the first analysis where log10Ks could be evaluated across technical areas. The plot of log10Ks in Fig. 6 reveals a two order of magnitude range for the Technical Area 54 log10Ks values, which is consistent with the extent of this property reported by others (Koltermann and Gorelick, 1996). Also the log10Ks values for Technical Area 21 are inside the range of the Technical Area 54 values supporting the merging of the two samples. A normality analysis using the STATISTICA Shapiro–Wilk W (not shown) of the log10Ks values indicates that a Gaussian distribution does not describe the log10Ks values from Technical Area 54, but when the log10Ks values from Technical Area 21 are included, the log10Ks values are Gaussian.


View this table:
[in this window]
[in a new window]
  		Table 8. Statistics and Kruskal–Wallis results for Bandelier Tuff Tshirege Member Cooling Unit 1v (Qbt 1v) by technical area.

 


View larger version (27K):
[in this window]
[in a new window]
  		Fig. 6. Box and whiskers plots of the median, 25 to 75% quartile, and range for log10Ks from Bandelier Tuff Tshirege Unit 1v (Qbt 1v) at designated technical areas of Los Alamos National Laboratory.

 
Tshirege Cooling Unit 1g (Qbt 1g).
There were samples for Unit 1g from Technical Areas 21, 49, 53, 54, Mortandad Canyon, Cañada del Buey, and Portrillo Canyon (Table 1), and with one sample available from Technical Area 53, this area was excluded from the analysis. This was the first unit where the canyon setting became a factor. Canyons at Los Alamos are wetter than the mesa tops, and this may lead to different weathering of the Bandelier Tuff and differences in matrix hydrologic properties. An analysis of the boreholes in Mortandad Canyon indicated that log10Ks cannot be considered homogeneous, but the other properties and parameters can be. Each canyon was treated as a single technical area for these analyses, even though the boreholes may be spread over considerable distances (Fig. 1).

The statistics by location and Kruskal–Wallis results for Qbt 1g are presented in Table 9. All matrix hydrologic properties except {rho}b are significantly different among locations. Box and whisker plots of log10Ks and van Genuchten n parameter are presented in Fig. 7 to determine if there are any readily apparent differences between canyon and mesa top settings. For log10Ks in Fig. 7A, there is overlap, especially for the 25 to 75% quartiles with high median values occurring at a mesa top, Technical Area 21, and a canyon, Cañada del Buey. In Fig. 7B, the 25 to 75% quartiles for the n parameter overlap, except for Technical Area 54, which does have overall larger values for n.


View this table:
[in this window]
[in a new window]
  		Table 9. Statistics and Kruskal–Wallis results for Bandelier Tuff Tshirege Member cooling unit 1g (Qbt 1g) by technical area.

 


View larger version (41K):
[in this window]
[in a new window]
  		Fig. 7. Box and whiskers plots of the median, 25 to 75% quartile, and range for (A) log10Ks and (B) van Genuchten n parameter from Bandelier Tuff Tshirege Unit 1g (Qbt 1g) at designated technical areas of Los Alamos National Laboratory.

 
Otowi Member (Qbo).
The Otowi Member was sampled at Technical Areas 21, 49, 53, Mortandad Canyon, Cañada del Buey, and Portrillo Canyon (Table 1), and this unit contains both canyon and mesa samples. Table 2 shows that significant differences were found at Technical Area 21 for log10Ks among boreholes, so this property will not be pooled. Again, canyons were treated as a single technical area for these analyses, consistent with the approach for Qbt 1g.

The Kruskal–Wallis analysis indicated that {rho}b was the only matrix hydrologic property that was not significantly different for the Qbo (Table 10). From a geologic perspective, the Otowi appears to be relatively homogeneous throughout the Los Alamos area, so this result was not expected. The statistics are also provided in Table 10. Technical Area 21 presents a special case in this analysis because samples were obtained from both mesa and canyon settings. Box and whisker plots of log10{alpha} in Fig. 8 show that Technical Area 21 is very different for the Obo than the other locations. Further breakdown in Fig. 9 by separating the Technical Area 21 canyon and mesa samples with the canyon now designated as Los Alamos Canyon indicate that the low values are consistent among the settings and the results presented in Fig. 8 are not necessarily a function of the sampling of the different settings.


View this table:
[in this window]
[in a new window]
  		Table 10. Statistics and Kruskal–Wallis results for Bandelier Tuff Otowi Member (Qbo) by technical area.

 


View larger version (30K):
[in this window]
[in a new window]
  		Fig. 8. Box and whiskers plots of the median, 25 to 75% quartile, and range for log10{alpha} from Otowi Member (Qbo) of the Bandelier Tuff at designated technical areas of Los Alamos National Laboratory.

 


View larger version (34K):
[in this window]
[in a new window]
  		Fig. 9. Box and whiskers plots of the median, 25 to 75% quartile, and range for log10{alpha} from Otowi Member (Qbo) of the Bandelier Tuff at designated technical areas of Los Alamos National Laboratory. The canyon samples from Technical Area 21 are labeled as Los Alamos Canyon.

 
Correlation Analysis
Relationships between matrix hydrologic properties and/or parameters make it possible to use a surrogate property or parameter to estimate another more difficult to measure parameter. Table 11 is the correlation matrix for Qbt 1v, which was the only unit where the properties and parameters other than {rho}b were consistent across the site. Significant correlations were found between Ks and both {alpha} and log10{alpha}, n and log10{alpha} (interestingly not n and {alpha}), and the transformations of Ks and {alpha}, which are functional in nature. Correlations are introduced by the fitting of Eq. [1] and RETC reports correlations between parameters with high correlations often occurring between {alpha} and n. The correlations in Table 10 do not provide any additional capability to predict matrix hydrologic properties for Qbt 1v across the Los Alamos area. These results combined with the previous analyses offer no relief in sampling requirements of the Bandelier Tuff at Los Alamos.


View this table:
[in this window]
[in a new window]
  		Table 11. Correlation matrix for unsaturated zone hydrologic properties from Bandelier Tuff Tshirege cooling unit 1v (Qbt 1v).{dagger}

 

   DISCUSSION AND CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION AND CONCLUSIONS
REFERENCES
 
The importance of the Bandelier Tuff at Los Alamos National Laboratory to waste management and environmental remediation means that the unsaturated zone hydrologic properties are critical in defining hydrologic response for either short or long time assessments. Unsaturated zone flow and transport modeling is the most readily available technique to perform required assessments at LANL, but these models require parameters and properties that describe the hydrologic and chemical properties of the geologic material. Bandelier Tuff unsaturated flow properties have been collected at a number of locations within the LANL site, and this study examined the potential of these data being combined using nonparametric statistical analyses to provide the required information at other locations across the LANL site.

Samples were collected and categorized using the geologic nomenclature from Broxton and Vaniman (2005), and this characterization is the starting point for the statistical analysis that was performed. Borehole locations were concentrated in the eastern portion of the LANL site because this is where the majority of the remediation and waste management activities have occurred. Samples were collected in an ad hoc mode from the boreholes, and there was no attempt at systematic sampling in any location. The uneven distribution in the number of samples per Bandelier Tuff Tshirege Member cooling unit and Otowi Member and the variability limit the conclusions that can be drawn from this study.

The first analyses examined the consistency of matrix hydrologic properties between boreholes in a given technical area. Statistically significant differences were found between boreholes at Technical Area 21, and the key property log10Ks was found to be significantly different for most geologic units. At a second site, the differences were not as prevalent, but the larger number of boreholes, limited sampling, and variability confounded the analysis. These statistical analyses indicated that the matrix hydrologic properties within a technical area or location are not consistent even for a given geologic unit, and further characterization is required to define model parameters for site vadose zone assessments.

Statistical analyses across technical areas for each Tshirege Member cooling unit and the Otowi Member were performed to examine the potential for pooling data to extend sample size. This analysis revealed that data from technical areas cannot be combined, with only Qbt 1v demonstrating any potential for pooling.

Differences in setting, canyon bottom vs. mesa top, were tested for Tshirege Unit 1g and the Otowi Member. Canyons are wetter than mesas, and matrix hydrologic properties can differ because of the weathering processes that occur with increased saturation. The results from the canyon–mesa comparison did not reveal any significant differences, and if there are, these differences may be masked by those between locations.

This study summarized the statistics for unsaturated matrix hydrologic properties for the Bandelier Tuff at Los Alamos National Laboratory. As indicated above, the sample size in this study was limited, and in many cases <10 samples were available for a geologic unit at a location. If the probability density or cumulative density function must be defined, then at least 30 samples are needed. This is one of the rationales for pooling of the data from different technical areas. What is clear from this study is that one cannot as a general rule combine samples for matrix hydrologic properties from different technical areas at Los Alamos National Laboratory. This means that either more sampling is needed from a technical area to further refine the local distribution function or it is necessary to use the current data with associated uncertainties.

An attempt was made to use a statistical approach to define the hydrogeologic units. The approach proposed by Flint (1998) is more logical when preparing a site investigation. The sampling of basic matrix hydrologic properties at Los Alamos has not focused on obtaining data on a regular or depth or spatial interval to describe the lithostratigraphy. This sampling approach would improve parameterization for flow and transport assessments by (i) better defining hydrogeologic units using an indicator such as porosity, proposed by Flint (1998), and (ii) providing data for geostatistical analyses that will lead to understanding spatial dependencies (in both the lateral and vertical directions).


Table A1. Listing of Bandelier Tuff samples and properties used in this study.

Borehole
 Technical Area
 Unit
 Setting
 Depth
 {rho}b
 {theta}s
 Ks
 Log10 Ks
 {theta}r
 n
 {alpha}
 Log10 {alpha}
m g/cm3 cm/s cm–1
LADP-3 21 Otowi 2  56.5 1.69 0.2570 0.00000220 –5.6576 0.0419 3.1668 0.0007 –3.1549
LADP-3 21 Otowi 2  73.2 1.25 0.3980 0.00003100 –4.5086 0 1.9818 0.0012 –2.9208
LADP-3 21 Otowi 2  83.0 1.30 0.3490 0.00002000 –4.6990 0 1.7494 0.0022 –2.6576
LADP-3 21 Otowi 2  84.0 1.29 0.3920 0.00002300 –4.6383 0 2.8065 0.0007 –3.1549
LADP-3 21 Otowi 2  89.4 1.25 0.3680 0.00001500 –4.8239 0 1.688 0.0033 –2.4815
LADP-3 21 Otowi 2  91.9 1.28 0.3460 0.00002200 –4.6576 0 1.6437 0.0047 –2.3279
LADP-4 21 2 2   4.9 1.49 0.3360 0.00002200 –4.6576 0.0243 1.9272 0.0035 –2.4559
LADP-4 21 2 1   8.3 1.78 0.2640 0.00002500 –4.6021 0.012 2.0162 0.0044 –2.3565
LADP-4 21 2 1   9.3 1.78 0.2820 0.00001800 –4.7447 0.0172 2.5176 0.0026 –2.5850
LADP-4 21 2 1  17.3 1.67 0.2860 0.00000600 –5.2218 0 2.3058 0.0008 –3.0969
LADP-4 21 2 1  21.3 1.51 0.3840 0.00001400 –4.8539 0 1.6076 0.0033 –2.4815
LADP-4 21 2 1  24.5 1.52 0.4060 0.00012000 –3.9208 0 2.2886 0.0022 –2.6576
LADP-4 21 1g 1  49.1 1.19 0.5720 0.00013000 –3.8861 0 1.5712 0.0056 –2.2518
LADP-4 21 1g 1  62.5 1.22 0.4680 0.00022000 –3.6576 0 1.6556 0.0056 –2.2518
LADP-4 21 1g 1  74.9 1.20 0.5220 0.00033000 –3.4815 0 1.4612 0.0156 –1.8069
LADP-4 21 1v 1  31.4 1.28 0.4090 0.00016000 –3.7959 0 1.8463 0.0035 –2.4559
LADP-4 21 1v 1  46.1 1.23 0.5470 0.00005100 –4.2924 0 1.3938 0.0063 –2.2007
LADP-4 21 Otowi 1  94.4 1.16 0.4640 0.00003300 –4.4815 0 2.0705 0.0027 –2.5686
LADP-4 21 Otowi 1  96.7 1.18 0.4000 0.00003200 –4.4949 0 3.1988 0.0008 –3.0969
LADP-4 21 Otowi 1  97.6 1.14 0.4460 0.00004600 –4.3372 0 1.8835 0.0035 –2.4559
LADP-4 21 Otowi 1 100.7 1.14 0.4180 0.00003800 –4.4202 0.01 2.1415 0.0024 –2.6198
LADP-4 21 Otowi 1 103.1 1.11 0.4080 0.00006000 –4.2218 0 3.3755 0.0009 –3.0458
LADP-4 21 Otowi 1 104.5 1.16 0.4160 0.00001800 –4.7447 0 2.3811 0.0019 –2.7212
LADP-4 21 Otowi 1 105.9 1.16 0.4080 0.00002900 –4.5376 0 2.2625 0.0019 –2.7212
LADP-4 21 Otowi 1 107.4 1.15 0.4060 0.00003200 –4.4949 0 2.6222 0.0014 –2.8539
LADP-4 21 Otowi 1 125.4 1.44 0.3980 0.00007100 –4.1487 0.0145 1.7677 0.0041 –2.3872
LADP-4 21 Otowi 1 147.2 1.27 0.4010 0.00009100 –4.0410 0 1.6368 0.0045 –2.3468
LADP-4 21 Otowi 1 151.7 1.29 0.3500 0.00004000 –4.3979 0 1.6421 0.0041 –2.3872
LADP-4 21 Otowi 1 152.4 1.23 0.3810 0.00002900 –4.5376 0.0151 1.7429 0.0041 –2.3872
LADP-4 21 Otowi 1 160.8 1.34 0.3530 0.00000014 –6.8539 0 2.3134 0.0009 –3.0458
MDAT 21 3 1 0.00008100 –4.0915
MDAT 21 3 1 0.00009400 –4.0269
MDAT 21 3 1 0.00000830 –5.0809
MDAT 21 3 1 0.00005000 –4.3010
MDAT 21 3 1 0.00003900 –4.4089
MDAT 21 3 1 0.00003100 –4.5086
MDAT 21 3 1 0.00004200 –4.3768
MDAT 21 3 1 0.00002200 –4.6576
MDAT 21 3 1 0.00003600 –4.4437
MDAT 21 3 1 0.00002800 –4.5528
MDAT 21 3 1 0.00005000 –4.3010
MDAT 21 3 1 0.00008600 –4.0655
MDAT 21 3 1 0.00003300 –4.4815
MDAT 21 3 1 0.00008100 –4.0915
MDAT 21 3 1 0.00002500 –4.6021
MDAT 21 3 1 0.00000830 –5.0809
MDAT 21 3 1 0.00002500 –4.6021
MDAT 21 3 1 0.00003600 –4.4437
MDAT 21 3 1 0.00005300 –4.2757
MDAT 21 3 1 0.00004200 –4.3768
MDAT 21 3 1 0.00000560 –5.2518
MDAT 21 3 1 0.00006700 –4.1739
MDAT 21 3 1 0.00010000 –4.0000
MDAT 21 3 1 0.00004400 –4.3565
21-2523 21 2 1  27.8 1.39 0.3574 0.00024000 –3.6198 0 2.1917 0.0039 –2.4089
21-2523 21 2 1  28.4 1.37 0.3934 0.00046000 –3.3372 0.0084 1.9582 0.0089 –2.0506
21-2523 21 2 1  29.5 1.49 0.3380 0.00009500 –4.0223 0.0128 2.967 0.0051 –2.2924
21-2523 21 2 1  29.9 1.46 0.3170 0.00009000 –4.0458 0.00000 2.0666 0.0021 –2.6778
21-2523 21 2 1  31.9 1.64 0.2690 0.00002100 –4.6778 0.0138 2.5215 0.0029 –2.5376
21-2523 21 2 1  33.4 1.61 0.2839 0.00003400 –4.4685 0.0076 2.1169 0.0036 –2.4437
21-2523 21 2 1  43.9 1.67 0.2843 0.00003000 –4.5229 0.02086 1.89268 0.00393 –2.4056
21-2523 21 2 1  48.6 1.51 0.3640 0.00013000 –3.8861 0 1.8791 0.0018 –2.7447
21-2523 21 3 1   2.0 1.35 0.3121 0.00002400 –4.6198 0.0229 1.9264 0.0042 –2.3768
21-2523 21 3 1   3.8 1.36 0.3249 0.00013000 –3.8861 0.0097 1.8657 0.0069 –2.1612
21-2523