| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Environmental Technology Division, Pacific Northwest National Lab., P.O. Box 999, Richland, WA 99352
b Geologic Consulting, Stevensville, MT 59870
* Corresponding author (george.last{at}pnl.gov)
All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher.
Received 7 December 2006.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMaterials and MethodsDiscussionREFERENCES |
|---|
Abbreviations: HBGIS, Hanford Borehole Geologic Information System
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsDiscussionREFERENCES |
|---|
|
Efforts are being made at the Hanford Site to standardize, manage, and analyze borehole geologic data to produce both regional sequence-stratigraphic hydrogeologic models and site-specific geostatistical representations of spatial distributions of lithofacies and their hydrogeologic and geochemical properties. Although the regional stratigraphic units for the Hanford Site are well documented, the differentiation of smaller-scale depositional elements and inherent flow-and-transport properties has been limited and difficult to quantify. Collecting new subsurface data is time consuming and expensive. Thus, it is important to extract as much information as possible from existing borehole data. This has led to the development of data standardization and translation techniques and methods for data integration, as well as the design and construction of a prototype, Web-based data extraction and management system.
This paper describes these data-standardization and translation techniques and the data-management system and presents an example of how these newly developed data sets are being used to develop a site-specific lithofacies and hydrofacies model for the vadose zone beneath one of Hanford's more complex waste sites.
|
Materials and Methods |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsDiscussionREFERENCES |
|---|
Three general approaches are commonly used to produce realistic simulations of the subsurface for application in vadose and groundwater models: (i) process-imitating, (ii) structure-imitating, and (iii) descriptive approaches (Koltermann and Gorelick, 1996). Hybrid approaches (Koltermann and Gorelick, 1996) minimize the influence of individual approach limitations. An example of a hybrid approach is given by Weissmann and Fogg (1999), wherein they combine descriptive sequence stratigraphic characterization, which provides stratigraphic framework and facies architecture, with transition probability geostatistics (a structure-imitating approach) to generate spatial variability for high-resolution three-dimensional realizations of the heterogeneity in an alluvial fan.
One of the most important elements of developing a lithofacies-based representation of the subsurface is the verification of facies geometries, lengths, and juxtapositions present within each depositional element or facies assemblage. Computer visualization technology, in combination with analysis of data from multiple boreholes, outcrop studies, and basinwide geologic studies, provides the basis for estimating the geometries and spatial distribution of lithofacies within the major stratigraphic units (Last et al., 2005b). Previous work at Hanford has focused on understanding vertical stratigraphic relationships because there has seldom been sufficient data to address the lateral facies changes and facies dimensions required to more accurately define geostatistical models of the subsurface characteristics.
The recognition and mapping of genetically linked lithofacies packages provide a framework for predicting textural and diagenetic changes and quantitative variations in contaminant distribution. A three-dimensional stratigraphic framework with unconformity-bounded sequences, combined with lithofacies modeling and even finer scale analyses of heterogeneity and supported with borehole geologic data, provides an improved quantitative method to represent the spatial distribution of flow-and-transport parameters and associated uncertainty.
The first critical element in developing a subsurface model for a given site is to gather and integrate all available borehole, outcrop, and geophysical data available for that site (Fig. 2 ). At the Hanford Site, by far the largest data set is borehole geologic data—there are over 7500 boreholes on site. In contrast, geologic outcrops are sparse and generally limited to shallow excavations. Surface geophysical data are also scarce and of limited value due to the great thickness (up to >100 m) of the vadose zone, the very dry and unconsolidated nature of the subsurface sediments, and the abundance of cultural features (e.g., fence lines, power lines, steel-cased boreholes). Thus, efforts are being made to increase the usefulness and value of information that can be extracted from the large amount of available borehole data.
|
Standardization of these descriptive observations results in an internally consistent set of data that can be integrated with other borehole geologic data sets (e.g., geophysical logs) and used to generate more quantitative, reproducible, and traceable interpretations of the subsurface framework. The majority of recent laboratory and borehole geophysical data already exists in a quantitative form and is located in various electronic and hard-copy sources. In contrast, driller's and geologist's logs are traditionally recorded on handwritten log sheets and vary widely in the quality and detail of their lithologic descriptions. To standardize these observations and to convert them into an electronic semiquantitative form, a series of logic rules and peer-reviewed procedures were developed. Based on these rules, the descriptive text information contained in log records is parsed and translated into a unified set of categorical data classes (e.g., particle size classes, color, clast lithology) on a standardized electronic template. Throughout the translation and standardization process, all data sets are carefully reviewed for completeness and traceability to original records and for compliance to translation and standardization procedures and are then entered into a relational database.
A prototype borehole geologic data-management system was developed to manage and integrate the newly interpreted categorical data sets with existing electronic raw databases (Fig. 2). The Hanford Borehole Geologic Information System (HBGIS) is a relational database management system consisting of a Microsoft Structured Query Language (SQL) Server back-end database that is accessed by end users via a Web-based graphical user interface (Last et al., 2005a). It is designed to be a completely extensible, multi-user system with multi-access that directly interfaces with existing databases at the Hanford Site to import "snapshots" of relevant data sets. This system is capable of direct data entry, automated data import, data query, and data export in the form of spreadsheet reports and graphical log plots (Fig. 2). Additional data analysis and visualization capabilities are being developed in parallel with other fluid flow and contaminant transport parameter databases, such as the Vadose Zone Hydraulic Property database (Freeman et al., 2002; Freeman and Last 2003) and the Hanford Contaminant Coefficient Database (Cantrell et al., 2003).
The HBGIS also provides traceability and documentation of changes (change control) for interpreted stratigraphic contacts. Thus, in addition to the raw borehole geologic data and the standardized borehole geologic data sets, the HBGIS also manages a database of interpreted stratigraphic contact depths for each borehole.
Ideally, the development of a regional stratigraphic contacts database should be tied explicitly and unambiguously to the raw geologic data. The raw data are being translated into electronic data sets, and interpretation and translation tools are being developed that can use these electronic data to identify and define the stratigraphic contacts in a reproducible and defensible way. To facilitate the development of these tools and to provide a benchmark for the development and refinement of a regional contacts database, an initial contacts database was assembled for the most extensively studied, central portion of the Hanford Site (Bjornstad, 2004). This initial contacts database was compiled from published and unpublished contact data sets that were then transposed to standardized stratigraphic nomenclature based on that defined by USDOE (2002) and Lindsey (1995). The resulting database identified many discrepancies between contacts subjectively picked by different geologists. The stratigraphic contacts are maintained within and are accessible through the HBGIS. Multiple contacts are stored for any given stratigraphic unit and borehole, providing traceability back to the individual interpretations.
Development of a Sequence-Specific Lithofacies Model for the 216-Z-1A Tile Field
First-order spatial distributions of flow-and-transport parameters generally adhere to the major stratigraphic units (represented by geologic formations and/or members). The internal structure of these major stratigraphic units can be further subdivided into discrete sedimentary packages or sequences, which allows a second-order distribution of flow-and-transport parameters to be defined.
Although the geology of the Hanford Site has been studied for more than 60 years, developing a regional stratigraphic contacts database and sequence stratigraphic model has been challenging. Besides the very heterogeneous nature of the sediments, many other ancillary sources of uncertainty are associated with the borehole data and interpretation of the geologic units, their lateral continuity, and their thicknesses (Oostrom et al., 2006a). These uncertainties include the spacing and accuracy of depth-discrete observations and analyses, ground-surface elevation at the time of measurement, and the geometric shape of the geologic units, particularly in the chaotic catastrophic Ice Age flood deposits of the Hanford formation. These inherent uncertainties, together with project-specific subjective interpretations, have led to multiple (often conflicting) stratigraphic interpretations and data sets within and between different projects.
Spatial delineation of the lithofacies within discrete stratigraphic units begins with a synergistic analysis of all borehole data, including field-derived drill logs, geological descriptions, and geophysical logs, along with laboratory-derived analyses of physical and geochemical properties (Fig. 2). Standardized geologic data residing in the HBGIS provide an easily retrievable and internally consistent data set for quantitative analysis. We have used these data to generate individual one-dimensional graphical well logs of all relevant geologic and petrophysical data (Last et al., 2006b) and have used logic rules to identify individual lithofacies in a traceable and reproducible way (Bush et al., 2005). We have also developed integrated interpretations of the sequence stratigraphy and facies architecture by constructing two-dimensional multi-well stratigraphic cross-sections and three-dimensional solid model interpretations (Oostrom et al., 2006a, b).
To develop a subsurface model for one of Hanford's more complex waste sites (the 216-Z-1A tile field), we combined descriptive stratigraphic characterization and facies architecture with indicator or transition probability geostatistics (a structure-imitating approach). Using these techniques, populated with standardized borehole geologic data, we generated multiple high-resolution three-dimensional realizations of the lithofacies-scale heterogeneity beneath the 216-Z-1A tile field (Bush et al., 2005). The geostatistical realizations were constrained to a physical domain located within the Hanford formation so that no major stratigraphic boundaries were crossed.
Minor stratigraphic units have been identified locally within the Hanford formation (Lindsey et al., 1994a, b) with the designation, from top to bottom, of the H1a, H1, H2, H3, and H4 units. However, these units are primarily based on variations in the amount of gravel present, and no well-defined markers separate those units. The units also tend to be laterally variable, so we elected to simulate the distribution of lithofacies within the Hanford formation as a whole (Bush et al., 2005), using ordinary kriging to account for changes in the proportions of the lithofacies (Journel and Rossi, 1989).
We grouped the flood deposits of the Hanford formation into five general facies types at the 216-Z-1A site with standardized geologic data that were entered and then integrated using the HBGIS database. The five general facies types (silty sand, fine sand, coarse sand, gravelly sand, and sandy gravel) were selected to represent sedimentary bodies that could be easily distinguished with the standardized data and that were expected to have similar hydrologic properties (based on grain-size similarity). The standardized borehole data included sediment classifications of driller's log data from 46 wells. These classifications, based on the Folk (1968) classification of clastic sediments, were assigned on the basis of the order in which the driller listed the sediments. For example, if the driller documented the sediment as sand, silt (mud), and gravel, it was assumed that the first type of sediment was the greatest percentage of the sample and decreased in the order it was listed. Laboratory-measured particle-size distributions were also used in this study for 16 wells where data were available in the HBGIS database. The laboratory-measured particle-size data were also used to classify the sediments using the Folk (1968) system. It was assumed that the laboratory-measured particle-size data were a better representation of the true classification of the sediment at each location than the driller's log data, so if both were available, the particle-size data were used. However, there were a number of wells with both types of data. A cross-calibration was performed, and the frequency with which driller's log data were associated with the different sediment classes identified by particle-size data was used to estimate the probability for a given sediment class to be present, given only the drill-log data. For example, for the fine sand facies, the drill-log data classified samples the same as the particle-size data 35% of the time (Table 1). Although the classification of fine sand from the drill-log data was often incorrect, the calibration still provides valuable information about the correct classification that was incorporated in the soft-data file generated for the geostatistical simulations. For example, if the drill log at a particular location suggested that fine sand was present, then the soft probabilities associated with that data were 56% that the location was actually muddy sand, 35% that it was actually fine sand, 6.7% that it was gravelly sand, and 2.5% that the sediment at the location was actually sandy gravel. Similar soft probabilities were derived from the cross-calibration for the classification of each facies by the drill log. The calibration results allow the use of the drill-log data as "soft data" with a measurable degree of confidence.
|
Excavation sites near the study area provided a qualitative assessment of the lateral extent and heterogeneity of the facies (Fig. 3 ). At Pit 30, located near the center of the Hanford Site, three primary lithofacies types occur within three distinct sedimentary packages. These include (i) massive silt to fine sand (Fm) at the top of the lower most gravel-dominated sediment package, (ii) horizontally laminated medium to coarse sand to pebbly sand [Sh(c)] making up the middle-sand–dominated sediment package, and (iii) large-scale, planar-tabular (foreset) bedded silty-sandy pebble to boulder gravel (Gp) making up the upper and lower gravel-dominated sediment packages. The lithofacies designations used here are from USDOE (2002). Note the lateral extent, heterogeneity, and apparent anisotropy within each of these facies, as well as the erosional unconformity at the top of the sand-dominated sediment package.
|
We used the conditional indicator simulation program sisim (Deutsch and Journel, 1998) to produce 100 realizations of the distribution of our five lithofacies. Uncertainty in the spatial distribution of the lithofacies was quantitatively assessed by analyzing the suite of realizations. The realizations that included the maximum amounts of silty sand and sandy gravel were chosen to represent the extreme distributions of facies likely to affect flow and transport, and the modal value at each node was used to identify the most likely distribution of lithofacies (Fig. 4 ). This set of realizations can be used as input for flow-and-transport modeling and should capture the range of behavior in flow-and-transport predictions.
|
|
Discussion |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsDiscussionREFERENCES |
|---|
Vadose zone flow-and-transport processes are profoundly influenced by textural and mineralogical changes in the sediments of the unsaturated zone. These changes are often linked in predictable ways to small-scale lithofacies heterogeneity and larger-scale stratigraphic packaging. In addressing the standardization and integration of borehole data, we recognize the need to provide data for analyses and process simulations that go beyond homogeneous, layered geohydrologic models populated with static parameters.
Contaminant transport and fate in the vadose zone are highly controlled by inter- and intralithofacies-scale heterogeneity, and thus, the spatial distribution of lithofacies is essential to mapping the distribution and heterogeneity of key subsurface properties that control vadose zone contaminant-transport processes (Last et al., 2005b). The effects of lithofacies heterogeneity are generally thought to enhance lateral spreading and impede downward migration of contaminants (Last et al., 2006a) but may also promote funneled or fingered flow.
Difficulties in developing lithofacies models include adequately imaging and predicting the shapes and sizes of geometrical facies (particularly thin stringers), assessing the juxtaposition of highly permeable beds, assigning values and boundary conditions to features such as cross-cutting clastic dikes in glacio-fluvial sediments, and assessing biogeochemical properties of intra-unit unconformities. Challenges also remain in properly scaling physical properties (e.g., effective permeability, porosity, moisture-retention characteristics, anisotropy, dispersivity) to larger modeled units and assessing the effects of small-scale structures on larger-scale effective flow-and-transport parameters (Ward et al., 2006; Oostrom et al., 2006a, b).
We anticipate improving our ability to represent the spatial distribution of effective hydrologic and geochemical properties by systematically evaluating available borehole geologic data to develop stratigraphic and lithofacies models and provide soft data for pedotransfer functions and upscaling approaches. This will enhance the resolution and defensibility of numerical flow-and-transport predictions in the vadose zone.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and MethodsDiscussionREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  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 | 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 | |||
