United States
Environmental Protection
Agency
Office of Ftesearch and
Development
Washington DC 20460
EPA/540/R-95/519
August 1995
&EPA
Rapid Optical Screen
Tool (ROST™)
Innovative Technology
Evaluation Report
SUPERFUND INNOVATIVE
TECHNOLOGY EVALUATION
-------
CONTACT
Lary Jack and Steve Rock are the EPA contacts for this report. Lary Jack is presently with the new
Characterization Research Division (formerly the Environmental Monitoring Systems Laboratory) in Las
Vegas, NV, which is under the direction of the National Exposure Research Laboratory with headquarters in
Research Triangle Park, NC.
Steve Rock is presently with the new Land Remediation and Pollution Control Division (formerly the Risk
Reduction Engineering Laboratory) in the newly organized National Risk Management Research Laboratory
in Cincinnati, OH.
-------
EPA/540/R-95/519
August 1995
Rapid Optical ScreenTool (ROST™)
Innovative Technology Evaluation Report
NATIONAL RISK MANAGEMENT RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
NATIONAL EXPOSURE RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
LAS VEGAS, NEVADA 89193
Printed on Recycled Paper
-------
Notice
The information in this document has been funded wholly or in part by the U.S. Environmental Protection Agency in partial
fulfillment of Contract No. 68-CO-0047, Work Assignment No. 0-40, to PRC Environmental Management, Inc. It has been
subject to the Agency's peer and administrative review, and it has been approved for publication as an EPA document. The
opinions, findings, and conclusions expressed herein are those of the contractor and not necessarily those of the EPA or other
cooperating agencies. Mention of company or product names is not to be construed as an endorsement by the agency.
-------
Foreword
The U.S. Environmental Protection Agency is charged by Congress with protecting the Nation's land,
air, and water resources. Under a mandate of national environmental laws, the Agency strives to formulate and
implement actions leading to a compatible balance between human activities and the ability of natural systems
to support and nurture life. To meet this mandate, EPA's research program is providing data and technical
support for solving environmental problems today and building a science knowledge base necessary to manage
our ecological resources wisely, understand how pollutants affect our health, and prevent or reduce environmental
risks in the future.
The National Risk Management Research Laboratory is the Agency's center for investigation of
technological and management approaches for reducing risks from threats to human health and the environment.
The focus of the Laboratory's research program is on methods for the prevention and control of pollution to air,
land, water and subsurface resources; protection of water quality in public water systems ; remediation of
contaminated sites and ground water; and prevention and control of indoor air pollution. The goal of this research
effort is to catalyze development and implementation of innovative, cost-effective environmental technologies;
develop scientific and engineering information needed by EPA to support regulatory and policy decisions; and
provide technical support and information transfer to ensure effective implementation of environmental
regulations and strategies.
This publication has been produced as part of the Laboratory's strategic long-term research plan. It is
published and made available by EPA's Office of Research and Development to assist the user community and
to link researchers with their clients.
E. Timothy Oppelt, Director
National Risk Management Research Laboratory
-------
Abstract
In August 1994, a demonstration of cone penetrometer-mounted sensor technologies took place to evaluate then-
effectiveness in sampling and analyzing the physical and chemical characteristics of subsurface soil at hazardous waste
sites. The effectiveness of each technology was evaluated by comparing each technology's results to the results
obtained using conventional reference method technologies. The demonstration was developed under the
Environmental Protections Agency's Superfund Innovative Technology Evaluation Program.
Three technologies were evaluated: the rapid optical screening tool (ROST™) developed by Loral Corporation and
Dakota Technologies, Inc., the site characterization and analysis penetrometer system (SCAPS) laser induced
fluorescence sensor developed by the Tri-Services (Army, Navy, arid Air Force), and the conductivity sensor
developed by Geoprobe® Systems. These technologies were designed to provide rapid sampling and real-tune,
relatively low cost analysis of the physical and chemical characteristics of subsurface soil to quickly distinguish
contaminated areas from noncontaminated areas.
Three sites were selected for the demonstration, each contained varying concentrations of coal tar waste and petroleum
fuels, and wide ranges in soil texture.
This demonstration found that the ROST™ technology produced screening level data. Specifically, the qualitative
assessment showed that the stratigraphic and chemical cross sections were comparable to the reference methods. The
quantitative assessment showed that during the 1994 demonstration, the ROST™'s data could not be used as a reliable
predictor of actual contaminant concentration. Based on this study, the ROST™ appears to be capable of rapidly and
reliably mapping the relative magnitude of the vertical and horizontal extent of subsurface contamination when that
contamination is fluorescent. This type of contamination includes petroleum fuels and polynuclear aromatic
hydrocarbons. The design of the ROST™'s fluorescence detection system also allows this technology to identify
specific waste types, such as jet petroleum (JP-4), diesel fuel, or coal tar. This chemical mapping capability, when
combined with the stratigraphic data produced by the cone penetrometer creates a powerful tool for site
characterization.
IV
-------
Table of Contents
Section
Notice _ jj
Foreword '.'.','.'.'.'.'.'.'.'.'.'. jjj
Abstract :.*...'.'.'.'.'.'.'.'.'.'.'. iv
List of Figures viii
List of Tables , vjjj
List of Abbreviations and Acronyms '.'.'.'.'.. ix
Acknowledgments '.'.'.'.'.'.'.'.'.'.'. xi
1 Executive Summary 1
2 Introduction 3
Demonstration Background, Purpose, and Objectives 3
Demonstration Design 4
Qualitative Evaluation 4
Quantitative Evaluation ..... 6
Deviations from the Approved Demonstration Plan '.'.'.'.,'.'. 7
Site Descriptions '.'.'.'.".'. 7
3 Reference Method Results g
Reference Laboratory Procedures g
Sample Holding Times \ g
Sample Preparation 9
Initial and Continuing Calibrations 10
Sample Analysis 10
Detection Limits 11
Quality Control Procedures 11
Confirmation of Analytical Results 12
Data Reporting 12
Quality Assessment of Reference Laboratory Data 12
Accuracy 12
Precision 12
Completeness 12
Use of Qualified Data for Statistical Analysis 13
Chemical Cross Sections 13
Atlantic Site \' |' 13
York Site '.'.'.'.'.'.'. 13
Fort Riley Site '.'.'.'.'.'.'.'.'. 17
Quality Assessment of Geotechnical Laboratory Data 17
Geotechnical Laboratory 17
Borehole Logging 17
Sampling Depth Control 17
-------
Table of Contents (Continued)
Section
Page
Stratigraphic Cross Sections 17
Atlantic Site 19
York Site 19
Fort Riley Site 21
4 Rapid Optical Screening Tool 22
Background Information 22
Components 22
Cone Penetrometer Truck System 22
ROST™ Technology 23
Nd:YAG Laser 23
Tunable Dye Laser 23
Fiber Optic Cable 24
Detection System 24
Control Computer 25
General Operating Procedures 25
Training and Maintenance Requirements 26
Cost 26
Observations ; 26
Data Presentation 28
Chemical Data 28
Atlantic Site 28
York Site 31
Fort Riiey Site 33
Cone Penetrometer Data x 36
Atlantic Site 36
York Site 36
Fort Riley Site 37
5 Data Comparison 39
Qualitative Assessment 39
Stratigraphic Cross Sections 39
Atlantic Site 39
York Site 40
Fort Riley Site 40
Summary 40
Chemical Cross Sections 41
Atlantic Site 41
York Site 42
Fort Riley Site 43
Total Organic Carbon 44
Quantitative Assessment 44
6 Applications Assessment 51
7 Developer Comments and Technology Update 53
Loral Comments (April 1995) 53
DTI Comments (May 1995) 56
Technology Update 57
Converting Rapid Optical Screening Tool (ROST™) Fluorescence Intensities to
Concentration Equivalents 57
Calibration Derived from Site Materials with In Situ Fluorescence Measurements 58
VI
-------
Section
Calibration Derived from Synthetic Standards with "Above Ground"
Fluorescence Measurements 58
Approach 1: Designation of POL and Soil Type 59
Approach 2: Specific POL Material, Designated Soil Type 59
Approach 3: Specific POL Material and Soil from the Site 59
60
8 References
Appendix
A Qualitative, Quantitative, Geotechnical, and TOC Data 61
vii
-------
List of Figures
Figure
Page
2-1 Typical Transect Sampling Line and Stratified Random Sampling Grid 6
3-1 TPH Reference Method Chemical Cross Section - Atlantic Site 14
3-2 PAH Reference Method Chemical Cross Section - Atlantic Site 14
3-3 TPH Reference Method Chemical Cross Section - York Site 15
3-4 PAH Reference Method Chemical Cross Section - York Site 15
3-5 TPH Reference Method Chemical Cross Section - Fort Riley Site 16
3-6 PAH Reference Method Chemical Cross Section - Fort Riley Site 16
3-7 Reference Method Stratigraphic Cross Section - Atlantic Site 18
3-8 Reference Method Stratigraphic Cross Section - York Site 18
3-9 Reference Method Stratigraphic Cross Section - Fort Riley Site 19
4-1 System Components 24
4-2 ROST™ Chemical Cross Section - Atlantic Site 30
4-3 Typical WTM - Atlantic Site 32
4-4 ROST™ Chemical Cross Section - York Site 32
4-5 Typical WTM - York Site 33
4-6 ROST™ Chemical Cross Section - Fort Riley Site 34
4-7 FVD - Fort Riley Site 35
4-8 Typical WTM - Fort Riley Site 35
4-9 Cone Penetrometer Stratigraphic Cross Section - Atlantic Site 37
4-10 Cone Penetrometer Stratigraphic Cross Section - York Site 37
4-11 Cone Penetrometer Stratigraphic Cross Section - Fort Riley Site 38
5-1 Normalized LIF and Qualitative Reference Data - Atlantic Site 42
5-2 Normalized LIF and Qualitative Reference Data - York Site 43
5-3 Normalized LIF and Qualitative Reference Data - Fort Riley Site 44
List of Tables
Table Page
2-1 Criteria for Data Quality Characterization 5
3-1 Comparison of Geologist's Data and Geotechnical Laboratory Data - All Sites 20
4-1 Quantitative ROST™ Data - Atlantic Site 29
4-2 Quantitative ROST™ Data -York Site 29
4-3 Quantitative ROST™ Data - Fort Riley Site 30
5-1 Regression Analysis Results for Initial ROST™ Push and Reference Methods - All Sites 46
5-2 Regression Analysis Results for Averaged ROST™ Push and Reference Methods -All Sites . 47
5-3 Data for Mean ROST™ -All Sites 49
7-1 Summary of TPH Results for Quantitative Evaluation 54
VIII
-------
List of Abbreviations and Acronyms
ASTM American Society for Testing and Materials
bgs below ground surface
BTEX benzene, toluene, ethylbenzene, and xylene
CCAL continuing calibration
cm centimeter
cm/s centimeter per second
DQO data quality objective
DSO digital storage oscilloscope
DTI Dakota Technologies, Inc.
EPA Environmental Protection Agency
ERA Environmental Resource Associates
ETS Environmental Technical Services
FID flame ionization detector
FMGP Former Manufactured Gas Plant
FVD fluorescence versus depth
GC gas chromatograph
HPLC high performance liquid chromatography
Hz pulses per second
ICAL initial calibration
ITER innovative technology evaluation report
LCS laboratory control samples
LIF laser induced fluorescence
MDL method detection limit
Method OA-1 University of Iowa Hygienics Laboratory Method
/4J/kg micrograms per kilogram
VQ/L microgram per liter
mg/L milligram per liter
mg/kg milligram per kilogram
mg/mL milligram per milliliter
Mj millijoules
mL milliliter
mm millimeter
MMTP Measurement and Monitoring Technologies Program
MS matrix spike
MSD matrix spike duplicate
NDSU North Dakota State University
Nd:YAG neodymium-doped yttrium aluminum garnet
NRMRL National Risk Management Research Laboratory
NERL-CRD National Exposure Research Laboratory-Characterization Research Division
nm nanometer
ns nanosecond
%D percent difference
%RSD percent relative standard deviation
PAH polynuclear aromatic hydrocarbon
PE performance evaluation
ix
-------
List of Abbreviations and Acronyms (Continued)
PID photoionization detector
PMT photomultiplier tube
POL petroleum, oils, and lubricants
ppb parts per billion
ppm parts per million
PRC PRC Environmental Management, Inc.
PRL PACE reporting limit
PTI Photon Technology, Inc.
QA quality assurance
QAPj'P quality assurance project plan
QC quality control
ROST™ Rapid Optical Screening Tool
RPD relative percent difference
SCAPS Site Characterization and Analysis Penetrometer System
SITE Superfund Innovative Technology Evaluation
TER technology evaluation record
TOC total organic carbon
TPH total petroleum hydrocarbon
TPM technical project manager
USCS Unified Soil Classification System
USDA United States Department of Agriculture
VOC volatile organic compound
VPH volatile petroleum hydrocarbon
WTM wavelength-time matrix
-------
Acknowledgments
We wish to acknowledge the support of all those who helped plan and conduct this demonstration, interpret data, and
prepare this report. In particular, for demonstration site access and relevant background information, Dean Harger
(Iowa Electric Company), Ron Buhrman (Burlington Northern Railroad), and Abdul Al-Assi (U.S. Army Directorate
of Engineering and Housing); for turn-key implementation of this demonstration, Eric Hess, Darrell Hamilton, and
Harry Ellis (PRC Environmental Management, Inc.); for editorial and publication support, Suzanne Ladish and Frank
Douglas; for peer and technical reviews, Dr. T. Vo-Dinh (Oak Ridge National Laboratory), Grace Bujewski (Sandia
National Laboratories), and Jeff Kelley (Nebraska Department of Environmental Quality); and for EPA project
management, Lary Jack (National Exposure Research Laboratory-Characterization Research Division) (702) 798-
2373). In addition, we gratefully acknowledge the participation of the technology developers Loral Corporation and
Dakota Technologies (Rapid Optical Screening Tool) (612) 456-2339 and (701) 237-4908, respectively).
XI
-------
-------
Section 1
Executive Summary
Recent changes hi environmental site characteriza-
tion have resulted in the application of cone penetro-
meter technologies to site characterization. With a
variety of in situ physical and chemical sensors, this
technology is seeing an increased frequency of use hi
environmental site characterization. Cone penetrometer
technologies employ a wide array of sampling tools and
produce limited investigation-derived waste.
The Environmental Protection Agency's (EPA)
Monitoring and Measurement Technologies Program
(MMTP) at the National Exposure Research Laboratory,
Las Vegas, Nevada, selected cone penetrometer sensors
as a technology class to be evaluated under the
Superfund Innovative Technology Evaluation (SITE)
Program. In August 1994, a demonstration of cone
penetrometer-mounted sensor technologies took place to
evaluate how effective they were hi analyzing the
physical and chemical characteristics of subsurface soil
at hazardous waste sites. Prior to this demonstration,
two separate predemonstration sampling efforts were
conducted to provide the developers with site-specific
samples. These samples were intended to provide data
for site-specific calibration of the technologies and
matrix interferences.
The main objective of this demonstration was to
examine technology performance by comparing each
technology's results relative to physical and chemical
characterization techniques obtained using conventional
reference methods. The primary focus of the
demonstration was to evaluate the ability of the
technologies to detect the relative magnitude of
fluorescing subsurface contaminants. This evaluation is
described in this report as the qualitative evaluation. A
subordinate focus was to evaluate the possible
correlations or comparability of the technologies
chemical data with reference method data. This
evaluation is described hi this report as the quantitative
evaluation. All of the technologies were designed and
marketed to produce only qualitative screening data.
The reference methods for evaluating the physical
characterization capabilities were stratigraphic logs
created by a geologist from soil samples collected by a
drill rig equipped with hollow stem augers, and soil
samples analyzed by a geotechnical laboratory. The
reference methods for evaluating the chemical
characterization capabilities were EPA Method 418.1
and SW-846 Methods 8310 and 8020, and University of
Iowa Hygienics Laboratory Method OA-1. In addition,
the effect of total organic carbon (TOC) on technology
performance was evaluated.
Three technologies were evaluated: the rapid
optical screening tool (RQST™) developed by Loral
Corporation and Dakota Technologies, Inc. (DTI), the
site characterization and analysis penetrometer system
(SCAPS) developed by the Tri-Services (Army, Navy,
and Air Force), and the conductivity sensor developed
by Geoprobe® Systems. Results of the demonstration
are summarized by technology and by data type
(chemical or physical) hi individual innovative
technology evaluation reports (ITER). In addition to the
three technology-specific ITERs, a general ITER that
examines cone penetrometry, hydraulic probe samplers,
and hollow stem auger drilling hi greater detail has been
prepared.
The purpose of this ITER is to chronicle the
development of the ROST™, its capabilities, associated
equipment, and accessories. The report concludes with
an evaluation of how closely the results obtained using
the technology compare to the results obtained using the
reference methods.
The ROST™ evolved from U.S. Government
Department of Defense funded research performed at
North Dakota State University (NDSU). The funding
was sponsored by the U.S. Department of Defense Tri-
Services SCAPS committee. The technology is being
commercialized and marketed by a consortium of
government and industry led by the Loral Corporation.
Loral Corporation owns the marketing rights to ROST™
with development assistance provided by DTI, Tri-
-------
Services, and the U.S. Advanced Research Projects
Agency. The technology was generally designed to
provide rapid sampling and real-time, relatively low cost
screening level analysis of the physical and chemical
characteristics (primarily petroleum fuels and coal tars)
of subsurface soil to quickly distinguish contaminated
areas from noncontaminated areas. The ROST™ mea-
sures fluorescence and is attached to a standard cone
penetrometer tool, which provides a continuous reading
of subsurface physical characteristics. This is translated
by software into various soil classifications. This
capability will allow investigation and remediation
decisions to be made more efficiently on site and will
reduce the number of samples that need to be submitted
for costly confirmatory analyses.
One hazardous waste site each was selected in Iowa,
Nebraska, and Kansas to demonstrate the technologies.
The sites were selected because of their varying concen-
trations of coal tar waste and petroleum fuels, and
because of their ranges hi soil textures.
This demonstration found that the ROST™ produces
screening level data. Specifically, the qualitative
assessment showed that the stratigraphic and the chemi-
cal cross sections were comparable to the reference
methods. The ROST™ showed advantages relative to the
reference methods in that the technology does not require
the collection of samples for analysis because analysis
occurs hi situ. This capability helps the technology
avoid the problems with sample recovery encountered
with the reference methods during this demonstration.
The relatively continuous data output from the ROST™
eliminated the data interpolation required for the refer-
ence methods, and it provided greater resolution. The
ROST™ can also be used to identify changes in waste
type during a site characterization. Through the use of
a wavelength-time-matrix (WTM), the ROST™ can
identify classes of contaminants, such as gasoline, diesel,
jet petroleum (JP-4), and coal tar. The qualitative
assessment showed that relative to the degree of contam-
ination; for example, low, medium, and high, the
technology's data and the reference data were well
correlated. Changes in TOC concentration did not
appear to affect the technology's performance.
The hi situ nature of the ROST™ minimized the
altering of soil samples, a possibility inherent with
conventional sampling, transport, and analysis. Further-
more, the cone penetrometer rods are steam cleaned
directly upon removal from the ground, reducing
potential contamination hazards to field personnel. In
addition, the continuous data output for both the chemi-
cal and physical properties of soil produced by the
ROST™ appears to be a valuable tool for qualitative site
characterization.
The quantitative assessment found that the ROST™
data exhibited little correlation to any of the reference
data concentrations of the target analytes. The lack of
correlation for the quantitative evaluation cannot be
solely attributed to the technology. Rather, it is likely
due to the combined effect of matrix heterogeneity, lack
of technology calibration, uncertainties regarding the
exact contaminants being measured, and the age and
constituents hi the waste. Based on the data from this
demonstration, it is not possible to conclude that the
technology can or cannot be quantitative hi its current
configuration. Based on the effects listed above, a high
degree of correlation should not be expected hi compari-
sons with conventional technologies.
Verification of this technology's performance should
be done only on a qualitative level. Even though it
cannot quantify levels of contamination or identify
individual compounds, it can produce qualitative contam-
inant distribution data very similar to corresponding data
produced by conventional reference methods, such as
drilling and laboratory sample analysis. The general
magnitude of the technology's data is directly correlated
to the general magnitude of contamination detected by
the reference methods. The performance of the ROST™
during this demonstration showed that it could generate
site characterization data faster than the reference
methods and with little to no waste generation relative to
the reference methods. The cost associated with using
this technology to produce the qualitative data used hi
this demonstration was approximately $41,000 which
included the cone penetrometer track and cone pen-
etrometer sensor, and the ROST™. Due to the increased
quality control and visitor distractions, it is likely that
the actual "production mode" cost of the ROST™
operation would be less than that exhibited during this
demonstration. This can be compared to the approxi-
mate $55,000 used to produce the reference method
cross sections, which were not available until 30 days
after the demonstration. The ROST™ cost less than the
reference methods, it produced almost 1,200 more data
points (continuously), and provided data hi a real-tune
fashion.
The question that this demonstration can not answer
is whether or not it is better to have fewer data points at
the highest data quality level or more data points at a
lower data quality level. Issues such as matrix heteroge-
neity may greatly reduce the need for definitive level
data hi an initial site characterization. Sampling and
analysis must always be done to effectively use the
ROST™ and critical samples will always require defini-
tive analysis.
-------
Section 2
Introduction
The purpose of this ITER is to present information
on the demonstration of the ROST™, a technology
designed to analyze the chemical characteristics of
subsurface soil. Since the ROST™ must currently be
used in conjunction with a cone penetrometer truck, the
geological data collection abilities of the cone penetro-
meter truck also were evaluated during this demonstra-
tion.
This technology was demonstrated hi conjunction
with two other sensor technologies: (1) the SCAPS
sensor designed by the Tri-Services (the U.S. Army,
the U.S. Air Force, and the U.S. Navy), and (2) the
conductivity sensor developed by Geoprobe® Systems.
The results of the demonstration of the other two tech-
nologies are presented in individual ITERs similar to this
document. An additional general ITER was prepared
which discusses the history, sampling, and other capabil-
ities of cone penetrometry, hydraulic probe samplers,
and hollow stem auger drilling. Complete details of the
demonstration, descriptions of the sites, and the experi-
mental design are provided in the final demonstration
plan for geoprobe- and cone penetrometer-mounted
sensors (PRC 1994). This information is briefly sum-
marized for this document.
This section summarizes general information about
the demonstration, such as the purpose, objectives, and
design. Section 3 presents and discusses the validity of
the data produced by the reference methods in the
evaluation of the ROST™ technology. Section 4 dis-
cusses the ROST™ technology, its capabilities, equip-
ment and accessories, and costs. Section 5 evaluates
how closely the results obtained using the ROST™
compare to the results obtained using the reference
methods. Section 6 discusses the potential applications
of the technology. Section 7 presents the developer's
comments on this ITER as well as an update on the
current application of the technology. Section 8 provides
complete references for the documents cited in this
report.
Demonstration Background, Purpose,
and Objectives
The demonstration was developed under the MMTP.
The MMTP is a component of the EPA's SITE Pro-
gram. The goal of the MMTP is to identify and demon-
strate new, viable technologies that can identify, quan-
tify, or monitor changes hi contaminants at hazardous
waste sites or that can be used to characterize a site less
expensively, better, faster, and/or safer than reference
methods.
The ROST™ uses laser induced fluorescence (LIF)
to detect the presence and absence of fluorescing com-
pounds, such as petroleum fuels and coal tar wastes.
The technology is incorporated into a standard CP sensor
and advanced into the soil with a standard cone pen-
etrometer truck.
The ROST™ was designed to provide rapid sampling
and real-time, relatively low cost screening level analysis
of the physical and chemical characteristics of subsurface
soil. The ROST™ was designed to analyze the chemical
characteristics of the subsurface soil by quickly identify-
ing the presence or absence of contamination, and
possibly, approximate concentrations. Since the ROST™
can be deployed with a CP sensor, it also is possible to
obtain physical properties of subsurface soils as the
ROST™ sensor is advanced. These capabilities allow
investigation and remediation decisions to be made more
efficiently and quickly, reducing overall project costs
such as the number of samples that need to be submitted
for confirmatory analyses and the need for multiple
mobilizations.
The primary focus of the demonstration was to
evaluate the ability of die technologies to detect the
relative magnitude of fluorescing subsurface contami-
nants, and hi some cases their ability to measure subsur-
face stratigraphy. This evaluation is described hi this
-------
report as the qualitative evaluation. A secondary focus
was to evaluate the possible correlations or comparability
of the technologies chemical data with reference method
data. This evaluation is described in this report as the
quantitative evaluation. All of the technologies were
designed and marketed to produce qualitative screening
data.
There were three objectives for the qualitative
evaluations, and one objective for the quantitative
evaluations conducted during this demonstration. The
first qualitative objective evaluated for the ROST™ was
its ability to vertically delineate subsurface soil contami-
nation and physical properties of the soil. Cross sections
of subsurface contaminant plumes and soil stratigraphy
produced by ROST™ were visually compared to corre-
sponding cross sections produced by the reference
methods. The second qualitative objective evaluated the
ability of the ROST™ to characterize physical properties
of subsurface soils. The third qualitative objective was
to evaluate reliability, ruggedness, cost, and range of
application of the ROST™. The ROST™ was quantita-
tively evaluated on how its data compared to the refer-
ence methods, and an attempt was made to identify its
threshold detection limits.
Demonstration Design
The experimental design of this demonstration was
created to meet the specific quantitative and qualitative
objectives described above. The experimental design
was approved by all demonstration participants prior to
the start of the demonstration. This experimental design
is detailed in the final demonstration plan (PRC 1994).
Sample results from the ROST™ were compared to
results from the reference methods. The reference
methods are commonly used means of obtaining the
same data as that produced by an innovative technology.
For this demonstration, the reference methods included
standard SW-846 methods for measuring petroleum
hydrocarbons and polynuclear aromatic hydrocarbons
(PAH), and borehole logging and sampling by a geolo-
gist using continuous samples from hollow stem auger
drilling. These comparisons were used to determine the
quality of data produced by the technology. Two data
quality levels were considered during this evaluation:
definitive and screening data. These data quality levels
are described in the EPA's "Data Quality Objectives
Process for Superfund - Interim Final Guidance" (1993).
Definitive data are generated using rigorous analyti-
cal methods, such as approved EPA reference methods.
Data are analyte-specific, with confirmation of analyte
identity and concentration. Methods produce tangible
raw data (e.g., chromatograms, spectra, digital values)
in the form of paper printouts or computer-generated
electronic files. Data may be generated at the site or at
an off-site location, as long as the quality
assurance/quality control (QA/QC) requirements are
satisfied. For the data to be definitive, either analytical
or total measurement error must be determined.
Screening data are generated by rapid, less precise
methods of analysis with less rigorous sample prepara-
tion. Sample preparation steps may be restricted to
simple procedures, such as dilution with a solvent,
instead of elaborate extraction/digestion and cleanup.
Screening data provide analyte identification and quanti-
fication, although the quantification may be relatively
imprecise. At least 10 percent of the screening data are
confirmed using analytical methods and QA/QC proce-
dures and criteria associated with definitive data.
Screening data without associated confirmation data are
not considered to be of known quality.
Since this technology is new and innovative, ap-
proved EPA methods for in situ LIF analysis do not
exist. For the purpose of this demonstration, the lack of
approved EPA methods did not preclude ROST™ from
being considered a definitive level technology. The
evaluation of this technology as to its quantitative
capabilities was included to provide potential users a
complete picture of the technology's capabilities hi its
present configuration during the demonstration. In the
configuration demonstrated, the developer never claimed
the technology was quantitative. Recent developer
advances hi data interpretation may increase the likeli-
hood that the technology can be quantitative. The main
criteria for data quality level assignment was based on
the comparability of the technology's data to data
produced by the reference methods. Table 2-1 defines
the statistical parameters used to define the data quality
levels produced by ROST™.
The sampling and analysis methods used to collect
the baseline data for this demonstration are currently
accepted by EPA as providing legally defensible data.
This data is defined as definitive level data by Superfund
guidance. Therefore, for the purpose of this demonstra-
tion, these technologies and analytical methods were
considered reference methods.
Qualitative Evaluation
Qualitative evaluations were made through observa-
tions and by comparing stratigraphic and chemical cross
sections from the technology to cross sections produced
from the reference methods. The reference methods for
the stratigraphic cross sections were continuous sampling
with a hollow stem auger advanced by a drill rig and
-------
TABLE 2-1. CRITERIA FOR DATA QUALITY CHARACTERIZATION
Data Quality
Level
Statistical Parameter
Definitive
Screening
r2 - 0.80 to 1.0, and the slope3 and y-intercept are statistically similar to 1.0 and 0.0, respectively the
precision is less than or equal to 20 percent and inferential statistics indicate the two data sets' are
statistically similar.
r2 = 0.80 to 1.0, the precision RSD is greater than 20 percent, and the technology meets its developer's
performance specifications, normal deviate test statistics on the two regression parameters indicate the
two data sets are statistically not similar; or in the case where the regression analysis indicates the data
is of definitive quality, but the inferential statistics indicate the data sets are statistically different
Notes:
r2
RSD
a
Coefficient of determination.
Relative standard deviation.
Since the ROST™ did not produce data in equivalent unils to the reference methods, the slope cannot be
used to assess accuracy, however, comparability can still be evaluated.
borehole logs created by a geologist. In addition, the
technology's ability to determine subsurface soil texture
at discrete intervals was further compared to data
produced by an off-site geotechnical laboratory. The
reference methods for the chemical cross sections were
subsurface sampling using a drill rig and off-site sample
analysis by EPA Method 418.1 and SW-846 Method
8310. EPA Method 418.1 produces data on total
petroleum hydrocarbon (TPH) concentrations. EPA
Method 8310 produces data on PAH concentrations.
These reference methods were selected for the qualita-
tive evaluations based on recommendations made by the
developer, consideration of the types of fluorescing
target analytes, and the project objectives. In addition,
soil samples were analyzed for (TOC) using the
90-3 Walkley-Black Method; and soil texture analysis
was performed by American Society of Testing Materi-
als (ASTM) Method D-422.
To qualitatively assess the ability of the ROST™ to
identify the presence or absence of contamination and
produce contaminant distribution cross sections, the
technology was required to continuously sample at five
points located along a transect line at each of the demon-
stration sites (Figure 2-1). These points were called
sample nodes. The transect line was placed across an
area of known subsurface contamination identified
during predemonstration sampling activities and previous
investigative sampling at these sites. A 6-foot by 6-foot
area was marked around each sample node. This area
was subdivided into nine sections of equal size, identified
as Sections A through I. At least one sample node per
site was placed outside the area of contamination.
Once each 6-foot by 6-foot area was marked,
sampling points for each technology and the reference
methods were assigned randomly at each node. This
produced a stratified random sampling design. This
design does not result in a predictable or fixed sampling
pattern for the technologies or the reference methods.
Sections for sampling at each node were only used once.
The stratified random sampling design was used
since the distribution of target analyte information was
believed to be heterogeneous throughout a given sam-
pling interval, and since the information's distribution
was not controlled by the demonstration.
The potential effect of organic matter was evaluated
qualitatively by TOC analysis of soil samples. This
geotechnical evaluation was intended to examine poten-
tial interferences from TOC on fluorescence response.
The chemical and geotechnical data generated by the
ROST™ in conjuction with its CP advancement platform
was used to produce qualitative data regarding contami-
nant and stratigraphic cross sections along each transect
line. These cross sections were compared to cross
sections generated by the reference methods results from
soil samples collected with a drill rig. The comparison
of contaminant cross sections involved visual compari-
sons with overlays, and minimum and maximum depths
of contamination at each location along a transect line.
Other factors that underwent qualitative evaluation
were technology costs, ease of operation, ruggedness,
instrument reliability, environmental sampling capability,
and production rates. PRC assigned an observer to work
with the developer to become knowledgeable hi the use
and application of the ROST™. With this training, PRC
was able to assess these operational factors.
During the demonstration, a total of 78 soil samples
were collected and analyzed by the reference methods,
-------
FIGURE 2-1. TYPICAL TRANSECT SAMPLING LINE AND STRATIFIED RANDOM SAMPLING GRID
«jTR*Tinro R»Mnnu SAMPLING CRIP
7RANSFCT S^UPI INC I INE
NUMBERED SAMPLING NODE
CONTINUOUS VERTICAL MEASURING POINT FOR
QUALITATIVE ASSESSMENT
TARGETED 6 INCH DEPTH INTERVAL FOR
QUANTITATIVE ASSESSMENT
and used in the qualitative data evaluations. These
samples were distributed as follows: 28 from the Atlantic
site, 26 from the York site, and 24 from the Fort Riley
site. Only sample data reported as positive values were
used in the evaluation. Sample data reported as "not
detected" was not used.
Quantitative Evaluation
The ROST" was evaluated quantitatively on its
ability to chemically characterize subsurface soil con-
tamination relative to classes of contaminants and
specific contaminants. This evaluation consisted of com-
paring data generated using the technology to data
obtained using reference analytical methods over a wide
range of concentrations. The reference method for the
chemical cross sections sampling was hollow stem
drilling. The University of Iowa Hygienics Laboratory
Method OA-1 volatile petroleum hydrocarbon (VPH),
SW-846 Method 8310 (PAH), SW-846 Method 8020
benzene, toluene, ethylbenzene, and xylene (BTEX), and
EPA Method 418.1 (TPH) were used as the reference
analytical methods. This allowed technology evaluation
relative to VPH, TPH, PAH, and BTEX concentrations.
This demonstration attempted to determine if the results
from the ROST™ could be correlated to results from the
reference methods, and if the technology was able to
differentiate between different types of contamination,
such as PAHs and BTEX. In addition, PRC attempted
to determine the detection thresholds of the technology
for these classes of contaminants.
To quantitatively assess the comparability of the data
produced by the ROST™ technology to the reference
methods' data, the final demonstration plan (PRC 1994)
required each technology to conduct its most accurate
and precise measurements at discrete depths at each
sampling node. These depths represented zones of initial
contaminant detection, medium, and high fluorescence.
However, at the start of the demonstration, the develop-
ers of both ROST™ and SCAPS technologies informed
PRC that the data produced during their standard dy-
namic push modes was the most accurate data, they could
produce. Therefore, the technologies' data for qualita-
tive evaluation was the same as that used in the quantita-
tive evaluations.
-------
The locations for the reference method sampling for
the quantitative evaluation were selected after reviewing
both the ROST™ and SCAPS data for a site. Sample
intervals that showed similar data from both technologies
were selected as reference method sampling intervals.
Reference method sampling intervals represented zones
of initial contaminant detection, medium, and high
fluorescence. The data produced at these intervals was
used to quantify contamination, identify contaminants,
establish a technology's precision and resolution, and
establish a technology's contamination detection thresh-
olds.
For the quantitative evaluation, data produced by the
ROST™ was averaged over a 12-inch push interval
corresponding to intervals sampled for reference method
analysis. This data was used to determine a mean
fluorescence over that interval. This data was com-
pared to corresponding mean reference method concen-
trations for any given interval. To create these mean
reference method concentrations, PRC collected five
replicate samples from the 12-inch intervals identified as
reference method sampling intervals based on the
ROST™ and SCAPS data. Each replicate sample was
collected from a randomly assigned section at each
sample node. The mean fluorescence for the ROST™
was compared to the mean constituent concentration for
the same interval, as generated by the reference method
analysis and the replicate sampling.
The data developed by the ROST™ was compared to
reference method data for the following compounds or
classes of compounds: TPH, total BTEX, VPH, total
PAH, total naphthalene (naphrnalene, 1-methylnaphtha-
lene, and 2-methylnaphthalene) and individual com-
pounds (BTEX, naphthalene, 1-methylnaphthalene,
2-methylnaphthalene, acenaphthene, fluoranthene,
pyrene, benzoapyrene, and anthracene). These compari-
sons were described in the August 1994 final demonstra-
tion plan.
Method precision also was examined during the
demonstration. The ROST™ was required to produce
10 separate readings or measurements at given depths
without moving the sensor between readings. From
these 10 measurements at each discrete depth, precision
control limits were established. This data also allowed
an examination of technology resolution and precision.
For the quantitative evaluation, a total of 103 soil
samples were collected and analyzed by the reference
methods. The distribution of these samples was as
follows: 8 replicate sampling intervals producing
38 samples at the Atlantic site, 7 replicate sampling
intervals producing 35 samples at the York site, and
30 isamples ;from 6 replicate sampling intervals at the
Fort Riley site. Only sample data reported as positive
values were used in the evaluation. Sample data re-
ported as "not detected" was not used.
Deviations from the Approved
Demonstration Plan
The primary deviation from the demonstration plan
(PRC 1994) dealt with the statistical analysis for the
quantitative evaluation.
Since the technology did not produce data directly
representing the concentration of contaminants, or data
in the same units as the reference method analysis, the
Wilcoxon Rank Sum Test could not be used, and the
comparison of the technology's data to 99 percent
confidence intervals was not made. In addition, the
effect of soil moisture was not examined due to the fact
that the bulk of the contaminated zones at each site were
at or near saturation. Finally, the demonstration plan
identified a hydraulic probe sampler as the reference
method for collecting the soil samples used in the
quantitative evaluations. However, due to sample matrix
affects (running sands), the hydraulic probe samples
could not meet the soil sampling objectives regarding
sample volume. The inability of this method to produce
full sample recovery was caused by the saturated fine
sands encountered at many of the target sampling depths.
To allow for adequate sample volume, PRC changed die
reference method for this soil sampling to hollow stem
augering and split spoon sampling.
Site Descriptions
The demonstration took place at three sites within
EPA Region 7. The three sites are the
(1) Atlantic-Poplar Street Former Manufactured Gas
Plant (FMGP) site (Atlantic site), (2) York FMGP site
(York site), and (3) the Fort Riley Building 1245 site
(Fort: Riley site). Brief summaries for each site are
given below. Complete details are located in the August
1994 final demonstration plan.
The Atlantic site is located in Atlantic, Iowa. The
site is surrounded by gas stations, grain elevators, a seed
supply company, and a railroad right-of-way. All
strucitures associated with the FMGP have been demol-
ished. A gas station now operates on the location of the
FMGP.
The Atlantic Coal Gas Company operated the
FMGP from 1905 to 1925. During that time, an un-
known quantity of coal tar was disposed of on site. In
-------
addition to the coal tar waste, more recent releases of
petroleum from two nearby gas stations also have
occurred. An investigation conducted at the site from
1990 to 1992 identified the following primary contami-
nants: BTEX and PAHs. The local groundwater con-
tains free petroleum product and pure coal tar.
The York site is located hi York, Nebraska. The
site encompasses nearly a half acre in an industrial
section of the city. The site is bordered by a former
railroad right-of-way, a concrete company, a seed
company, and a farm supply store. The site is nearly
level, and one building occupied by the FMGP is still
present. The York Gas and Electric company operated
the FMGP from 1899 to 1930. Coal tar waste was
disposed of at the site. Current information on the site
suggests that coal tar waste and its constituents should be
the only waste encountered.
The Fort Riley site is located at Building 1245 on
the east side of the Camp Funston area at Fort Riley,
Kansas. Between 1942 and 1990, five 12,000-gallon
steel underground storage tanks were located at this site.
The tanks were used to store leaded and unleaded
gasoline, diesel fuel, and military operations gasoline.
Soil at the site is contaminated with gasoline and diesel
believed to be the result of past petroleum fuel releases
from the underground storage tanks.
;8
-------
Section 3
Reference Method Results
All soil samples collected during this demonstration
were submitted to PACE, Inc. (PACE), for chemical
and geotechnical analysis. The PACE laboratory hi
Lenexa, Kansas, performed the methods 418.1, 8020,
and OA-1 analyses, while the PACE laboratory hi St.
Paul, Minnesota, performed the Method 8310 analyses.
PACE subsequently subcontracted the geotechnical
analyses to Environmental Technical Services (ETS),
Petaluma, California. The chemical data supplied by the
reference laboratory, the geotechnical data supplied by
the geotechnical laboratory, and the data produced by the
on-site professional geologist are discussed in this
section.
Reference Laboratory Procedures
Samples collected during this demonstration were
homogenized and split for the following analyses:
• TPH by EPA Method 418.1 (EPA 1986)
• PAH by EPA SW-846 Method 8310 (EPA
1986)
• BTEX by EPA SW-846 Method 8020 (EPA
1986)
• Total VPH as gasoline by University of Iowa
Hygienic Laboratory Method OA-1 (University
Hygienic Laboratory 1991)
• Soil texture and TOC by the 90-3 Walk-
ley-Black Method (Page 1982)
The results of these analyses are summarized hi
Appendix A, Tables A-l and A-9. The results are
reported as wet-weight values as required hi the
demonstration plan (PRC 1994). The data is grouped by
analytical method, site, and whether the data is intended
for qualitative or quantitative data evaluation. The
chemical cross sections produced by the qualitative data
are presented and discussed later hi this section.
The data from the PACE laboratory was internally
reviewed by PACE personnel before the data was
delivered to PRC. PRC personnel conducted a data
review on the results provided by PACE following EPA
guidelines (1991). PRC reviewed the raw data and
checked the calculated sample values.
The following sections discuss specific procedures
used to identify and quantitate TPHs, VPHs, PAHs,
BTEX, and TOC. Most of these procedures involved
requirements that were mandatory to guarantee the
quality of the data generated.
In addition to being generally discussed in this
section, all of the reference method results used to assess
the ROST™ are presented in Appendix A, Tables
A-l through A-9.
Sample Holding Times
The required holding tunes from the date of sample
receipt for each analytical method used to analyze the
soil samples were as follows: University of Iowa
Hygienics Laboratory Method OA-1 (Method
OA-1), 14 days for extraction and analysis; EPA
SW-846 Method 8020 (BTEX), 14 days for extraction
and analysis; EPA Method 418.1 (TPH), 14 days for
extraction and 40 days for analysis; EPA
SW-846 Method 8310 (PAH), 14 days for extraction and
40 days for analysis; and 90-3 Walkley-Black Method
(TOC), 28 days for extraction and analysis.
All holding times for the samples were met during
this demonstration.
Sample Preparation
Preparation of soils for TPH analysis was performed
following EPA Method 418.1. This method uses a
Soxhlet extraction as stated in SW-846 Method
9071. The soil sample extracts were analyzed for TPH
using SW-846 Method 418.1.
-------
Extracts for VPH analysis were prepared following
Method OA-1. The BTEX sample preparation require-
ments were carried out as specified hi that method.
The preparation of soil samples for TOC analysis
were carried out as specified hi the 90-3 Walkley-Black
Method.
Sonication extraction, Method SW-846 3550, was
used for the preparation of soil samples for
SW-846 Method 8310 analysis. The preparation of
samples for PAH analysis by SW-846 Method 8310 were
carried out according to the method requirements.
Initial and Continuing Calibrations
Initial calibrations (ICAL) were performed before
sample analysis began. ICALs for SW-846 Methods
8020, 8310, and 418.1 consisted of the analysis of five
concentrations of standards. Method OA-1 required the
analysis of three concentrations of standards for the
ICAL. Linearity for these ICALs was evaluated by
calculating the percent relative standard deviation
(%RSD) of the calibration factors. The %RSD QC limit
for SW-846 Methods 8020 and 8310 and Method
OA-1 was 20 percent. The calibration factors were
calculated by dividing the response (measured as the area
under the peak or peak height) by the amount of
compound injected on the gas chromatograph (GC)
column. The 90-3 Walkley-Black Method for TOC
required a daily calibration to a reference sulfate
solution. This ICAL was performed hi duplicate. All
initial calibrations met the respective method
requirements.
Continuing calibrations (CCAL) were performed on
a daily basis to check the response of the detector by
analyzing a mid-concentration standard and comparing
the calibration factor to that of the mean calibration
factor from the ICAL.
Calibration factors were monitored hi accordance
with the SW-846 and OA-1 Methods. No CCAL was
performed for the 90-3 Walkley-Black Method. Six
CCALs exceeded the 15 percent difference (%D) criteria
for various BTEX compounds. This resulted hi sample
results being qualified as estimated (J) and usable for
limited purposes. Various PAH compounds hi six
SW-846 Method 8310 CCALs exceeded 15 %D for one
of the two detectors. Sample results for the compounds
exceeding 15 %D were qualified as estimated (J) and
usable for limited purposes. SW-846 Method 8310 uses
two detectors, an ultraviolet detector and a fluorescence
detector. Since one detector's CCAL response was
within QC guidelines, this data is considered useable.
Retention tunes of the single analytes were
monitored through the amount of retention tune shift
from the CCAL standard as compared to the ICAL
standard. The retention tune windows for SW-846
Method 8310 were set by taking three tunes the standard
deviation of the retention times that were calculated from
the ICAL and CCAL standards. The retention time
windows for SW-846 Method 8020 were set by PACE at
plus or minus 0.07 minutes for benzene, ethylbenzene,
m-xylene, and plus or minus 0.10 minutes for toluene.
No CCAL retention tunes for the individual PAH
analytes were outside the retention tune windows.
CCAL retention times for the individual BTEX analytes
were observed outside the retention tune windows as set
by the ICAL. No samples were qualified based on this
QC criteria because the retention tune shifts were
adjusted appropriately by PACE for sample identification
and quantitation.
Following the ICAL, a method blank was analyzed
to verify that the instrument met the method
requirements. Following this, sample analysis may
continue for 24 hours. As stated hi SW-846 Method
8000, a CCAL must be analyzed and the calibration
factor verified on each working day. Sample analysis
may continue as long as CCAL standards meet the
method requirements.
Sample Analysis
Specific PAH and BTEX compounds were identified
hi a sample by matching retention tunes of peaks found
after analyzing the sample with those compounds found
hi PAH and BTEX standards. VPH was identified hi a
sample by matching peak patterns found after analyzing
the sample with those compounds found hi VPH
standards. Peak patterns may not always match exactly
because of the way the VPHs were manufactured or
because of the effects of weathering. When peak
patterns do not match, the analyst must decide the
validity of the identification of VPHs. For this reason,
peak pattern identification is highly dependent on the
experience and interpretation of the analyst.
Quantitation of PAHs, BTEX compounds, TPHs,
and VPHs was performed by measuring the response of
the peaks hi the sample to those same peaks identified hi
the ICAL standard. The reported results of this
calculation were based on wet weights (except for
PAHs), as required by PRC. PAH data was reported on
a dry-weight basis. PRC converted this data to
wet-weight based results dividing the dry-weight result
by the percent moisture of the original wet sample.
Quantitation of TOC was performed by measuring the
volume of potassium dichloride (K2Cr2O7) titrated and
calculating the milliequivalents of K2Cr2O7 titrated.
10
-------
This value was then multiplied by conversion factors
defined in the method and subsequently divided by the
grams of sample. TOC results were reported on a
wet-weight basis.
Sample extracts can frequently exceed the
calibration range determined during the ICAL. When
this occurred, the extracts were diluted to obtain peaks
that fall within the linear range of the detector. For
BTEX compounds and VPHs, this linear range was
defined as the highest standard concentration response of
the analytes of interest analyzed during the ICAL. The
linear range for TPHs was defined as an absorbance
maximum of 0.8. For PAHs, as defined in SW-846
Method 8310, the linear range was from 8 tunes the
method detection limit (MDL) to 800 tunes the MDL
with the following exception: benzo(ghi)perylene
recovery at 80 tunes and 800 times the MDL are low.
Once a sample was diluted to within the linear range, it
was analyzed again. Dilutions were performed when
appropriate on the samples for this demonstration.
Detection Limits
The PACE reporting limit (PRL) for PAHs was
calculated by multiplying the calibration correction
factor based on dry weight, times the MDL for each
specific PAH. PRLs for BTEX compounds were
determined by the lowest concentration standard of the
ICAL. The BTEX ICAL concentration range was from
10 micrograms per liter Og/L) to 100 //g/L. The PRL
for benzene, toluene, and ethyl benzene was 50
micrograms per kilogram (Mg/kg) and 100 #g/kg for
total xylene. The three levels of standard concentrations
for the VPH ICAL ranged from 2 milligrams per
milliliter (mg/mL) to 8 mg/mL. The PRL for VPH was
5 milligrams per kilograms (mg/kg). For TPH, the
calibration range was calculated by calibrating the
infrared detector using a series of working standards. A
plot was then prepared of absorbance versus milligram
petroleum hydrocarbons per 100 milliliter (mL) solution.
The PRL for TPH was 10 mg/kg. The MDL for TOC
analysis was 10 mg/kg wet weight.
Quality Control Procedures
A number of QC measures were used by PACE as
required by SW-846 Methods 8310 and 8020, EPA
Method 418.1, Method OA-1, and the
90-3 Walkley-Black Method. These QC measures
included the analyses of method blanks, instrument
blanks, laboratory control samples (LCS), matrix spike
(MS) and matrix spike duplicates (MSD), and the use of
sample surrogate recoveries.
All method and instrument blanks met the
appropriate QC criteria, except for two method blanks
analyzed by Method 418.1. TPH was reported as
slightly above the PRL of 10 mg/kg hi two method
blanks. Due to the low values reported in the two
method blanks, the sample results were not qualified.
Surrogate standards were added to all samples,
method blanks, MSs, and LCSs for the SW-846 Methods
8310 and 8020, and Method OA-1. All surrogate
recoveries for SW-846 Method 8020 were within the QC
acceptance criteria of 42 to 140 percent for soil. Seven
samples were qualified as estimated (J) and usable for
limited purposes based on surrogate recoveries for the
Method OA-1. The QC acceptance criteria for surrogate
recovery for Method OA-1 was 67 to 127 percent.
Thirty soil samples for SW-846 Method 8310 analysis
were qualified as estimated (J) and usable for limited
purposes based on surrogate recoveries observed outside
the QC limits of 58 to 140 percent. Two surrogates
were used for Method 8310. Samples were qualified
only when both surrogates were outside the QC limits
and no dilution analysis was performed. Numerous soil
samples required dilution for the Method 8310 analysis
because of petroleum interference. Dilution of these
samples resulted hi corresponding reductions hi
surrogate concentrations. When this occurred, the
resultant concentration of surrogate was below its MDL.
In cases where dilution resulted hi failure to detect the
surrogate, no coding of the data was implemented.
MS samples are samples to which a known amount
of the target analytes are added. There were 10 MSs
performed during the analysis by Method 418.1. Eight
of the MS samples were affected by high concentrations
of target analytes hi the spiked samples. No samples
were qualified. Eleven MSs were performed during the
analysis of samples by Method 8310. All but three MSs
and MSDs were outside the QC limits for percent recov-
ery and relative percent difference (RPD). These QC
exceedences were due to petroleum matrix interference.
The data associated with the QC samples was not qual-
ified because EPA guidelines state that samples cannot
be qualified based on MS and MSD results alone
(1991). There were seven MS and MSD samples
analyzed by Method 8020 and five by Method
OA-1. The MSs and MSDs analyzed by Method
8020 did not meet QC acceptance criteria for percent
recoveries or RPDs. No samples were qualified based
on these MS and MSD results due to the reasons stated
above. All MS and MSDs analyzed by Method
OA-1 and 90-3 Walkley-Black Method met all QC
acceptance criteria and were considered acceptable.
11
-------
All LCSs met QC acceptance criteria and were con-
sidered acceptable for all soil samples analyzed by
SW-846 Method 8310, Method OA-1,
90-3 Walkley-Black Method, and Method 418.1. One
soil LCS analyzed by Method 8020 was outside the QC
control limits. The soil LCS percent recovery for
toluene was below the QC limit. Twenty soil samples
were qualified as estimated (J) and usable for limited
purposes.
Also, three equipment rinsate blanks and one trip
blank were analyzed to assess the efficiency of field
decontamination and shipping methods, respectively.
There was no contamination found above PRLs in any of
these blanks, indicating decontamination procedures
were adequate.
Confirmation of Analytical Results
Confirmation of positive results was not required by
any of the analytical methods performed except
SW-846 Method 8310. Confirmation of positive PAH
results by Method 8310 was performed by the use of two
types of detectors. Both an ultraviolet detector and a
fluorescence detector were used in the analysis of PAHs.
The only requirement for using either detector for quan-
titation was that they meet the QC criteria for linearity
(ICAL) and %D (CCAL). If either detector failed either
of these criteria, it could not be used for quantitation,
but it could be used for confirmation of positive results.
Data Reporting
The results reported and qualified by the reference
method contained two types of qualifier codes. Some
data was coded with a " J," which is defined by PACE as
detected but below the PRL; therefore, the result is an
estimated concentration. The second code, "MI," was
defined as matrix interference. Generally, the effect of
a matrix interference is to reduce or enhance sample
extraction efficiency.
Quality Assessment of Reference
Laboratory Data
This section discusses the accuracy, precision, and
completeness of the reference method data.
Accuracy
Accuracy of the reference methods were
independently assessed through the use of performance
evaluation (PE) samples purchased from Environmental
Resource Associates (ERA) located in Arvada,
Colorado, containing a known quantity of TPH. In
addition, LCSs and past PE audits of the reference
laboratory were used to verify analytical accuracy.
Based on a review of this data, the accuracy of the
reference methods were considered acceptable.
Precision
Precision for the reference method results was
determined by evaluating field duplicates, laboratory
duplicate, and MS and MSD sample results. Precision
was evaluated by determining the RPDs for sample
results and their respective duplicate sample results.
The MS and MSD RPD results for the PAH
compounds averaged 25 percent for all of the 11 MS and
MSD sample pairs. However, there was one MS and
MSD sample pair that had a RPD of 99.9 for
1-methyl-naphthalene. If this point was removed, the
overall average would decrease to 20 percent. The
average RPD for the seven BTEX MS and MSD sample
pairs was less than 25 percent. Only four BTEX RPDs
were outside advisory QC guidelines defined by the
reference laboratory's control charts. All five VPH MS
and MSD sample RPDs met advisory QC guidelines set
by the reference laboratory's control charts. The
10 TPH MS and MSD sample pairs were considered
acceptable.
Laboratory duplicate samples are two separate
analyses performed on the sample. During the analysis
of demonstration samples, 10 TPH laboratory duplicate
samples were prepared and analyzed. All TPH
laboratory duplicate RPD result values were less than
25 percent. This was considered to be acceptable.
Completeness
Results were obtained for all of the soil samples.
PACE J-coded values that were detected below the PRL,
but above the MDL. As previously discussed, samples
were J-coded based on one or more of the advisory QC
guidelines not being met (i.e., surrogate and spike
recoveries). Also, some samples were J-coded based on
BTEX CCALs not meeting QC guidelines. PRC did not
consider this serious enough to preclude the use of this
data because the %Ds for the CCALS did not exceed the
QC guidelines by more than 10 percent. The analytes
with %Ds outside the QC guidelines were not detected
hi most of the samples associated with the CCALs. The
J-coded data is valid and usable for statistical analysis
because the QC guidelines were based on advisory
control limits set by either the reference analytical
methods or by PACE. This data set should be
considered representative of data produced by
conventional technologies. For this reason, the actual
completeness of data used was 100 percent.
12
-------
Use of Qualified Data for Statistical Analysis
One hundred percent of the reference laboratory
results were reported and validated by approved QC
procedures. The data review indicated that J-coded data
was acceptable for meeting the demonstration objective
of providing baseline data to compare against the
demonstrated technologies.
None of the QA/QC problems was considered
serious enough to preclude the use of J-coded data for
this demonstration. The surrogate and spike recovery
control limits were for advisory purposes only, and no
corrective action was required for the surrogate
recoveries that were outside of this range. RPD results
for MSs and MSDs that did not meet advisory QC
control limits were common when the matrix contained
a high concentration of petroleum. Again, these were
advisory limits and no corrective action was required.
These same general results would be seen by any
laboratory using the reference analytical methods on
such highly contaminated samples.
Also, rejection of a large percentage of data would
increase the apparent variation between the reference
laboratory data and the data from the technology. This
apparent variation would probably be of similar
magnitude to that introduced by using the data. For
these reasons, the J-coded data was used.
Chemical Cross Sections
Chemical cross sections were created from the
reference analytical data produced for the qualitative
data evaluation (see Appendix A, Tables A-l through A-
6). These samples were collected by a professional
geologist on site during the logging of boreholes. The
cross sections were hand contoured. The contour
intervals were selected to best represent a conventional
approach to the delineation of subsurface contamination.
The cross sections are presented on Figures 3-1 to
3-6. A written interpretation of these cross sections is
presented below.
Atlantic Site
The five sampling nodes formed a northwest to
southeast trending transect across the site (Figure
3-1). Node 1 on the far northwestern edge of the cross
section represented an area that was not impacted by the
contamination from the Atlantic site. Just southeast of
this location at Node 2, two distinct layers of
contamination were identified. The upper zone extended
from approximately 1 foot to 5 feet below ground
surface (bgs). This zone was characterized by TPH
contamination ranging from 100 to 10,000 parts per
million (ppm). The lower zone of contamination
extended from approximately 22 feet to 28 feet bgs. The
TPH concentrations in this zone ranged from 100 to
greater than 10,000 ppm. These two zones expanded
and blended together as Node 3 was approached.
Around Nodes 3 and 4, the thickness of the TPH plume
remained fairly constant, extending from approximately
1 foot to 31 feet bgs. The central portion of this zone
exhibited TPH contamination greater than 1,000 ppm.
The remainder of this zone exhibited TPH contamination
in the range of 100 to 1,000 ppm. As the far
southeastern edge of the transect was approached at
Node 5, the highest concentrations hi the center of the
plume pinched out, leaving a contamination zone that
extended from just below the ground surface to
approximately 27 feet bgs. This zone exhibited
contamination in the range of 100 to 1,000 ppm.
The total PAH cross section along this same transect
exhibited a slightly different distribution (Figure
3-2). As with the TPH cross section, the total PAH
cross section began at Node 1 in the area exhibiting no
signs of contamination. At Node 2, again two zones of
contamination were detected. The upper zone extended
from the ground surface to approximately 7 feet bgs.
This zone deepened toward the east. The concentrations
of total PAHs hi this zone ranged from 10 to greater
than 100 ppm. The lower zone extended from
approximately 14 to 30 feet bgs. The concentrations of
total PAHs in this zone ranged from 10 to greater than
100 ppm. Concentrations greater than 100 ppm were not
exhibited at this depth in the nodes occurring further
east The distribution of the 10 to 100 ppm dipped
below the ground surface at progressive depths farther
east of Node 2. At Node 5, this upper zone began at
approximately 7 feet bgs. This zone also reached its
majdmum depth around Nodes 3 and 4, approximately
30 feet bgs. Around Nodes 3 and 4 were two lenses of
total PAH contamination hi excess of 100 ppm. The
largest of these zones appeared to be thickest around
Node 3, extending from approximately 7 to 16 feet bgs.
This zone thinned to the east and pinched out between
Nodes 4 and 5. A smaller lens of greater than 100 ppm
total PAH contamination was exhibited at Node 4. This
zone extended between 7 to 9 feet bgs. This zone was
not detected hi Nodes 3 or 5.
York Site
The five sampling nodes formed a north to south
trending transect. The TPH and total PAH distributions
appeared to be similar, except at Node 5, at the York
site (Figures 3-3 and 3-4). At Node 5, the TPH
contamination was more extensive, extending from 1 to
25 feet bgs. At this same locations, the PAH
contamination extended from 13 to 21 feet bgs.
13
-------
FIGURE 3-1. TPH REFERENCE METHOD CHEMICAL CROSS SECTION - ATLANTIC SITE
SOUTHEAST
NODES
0-
-2-
-3-
-4 —
-3-
-8-
-7-
-8-
•B-
-10-
-11-
-12-
-13-
-14-
-ts-
-1B-
-20-
-21-
-22-
-23-
-24-
-25-
-28-
-27-
-28-
-29-
-30-
-31-
-32-
-33-
-34-
-33-
-38-
-37-
NODE4
NODE3
NODE2
DISTANCE (FEET)
NORTHWEST
NODE1
— 2
— 3
—4
--S
— 7
--8
--•
--10
— 11
— 14
— IS
— 18
— 17
— 18
— IB
"i?
— 22
— 23
— 24
— 25
— 28
— 27
— 28
— 29
— 30
— 31
--32
— 33
— 34
--35
— 38
— 37
LEGEND
;<100 PPU
- 100 - 1.000 PPU
123 ISO
1.000 - 10.000 PPM
> 10,000 PPU
200
225
NO - NOT DETECTED
PPU - PARTS PER UILUON
250
• - QUANTITATIVE
REFERENCE DATA
NOTE: QUANTITATIVE REFERENCE DATA USED BECAUSE OF POOR SAUPLE RECOVERY
FIGURE 3-2. PAH REFERENCE METHOD CHEMICAL CROSS SECTION - ATLANTIC SITE
SOUTHEAST
NODES
-8-
-7-
-8-
-10-
-11-
:!§:
-14-
-17-
-18-
-1B-
-20-
-21-
-22-
-23-
i
-28-
-29-
-30-
-31-
-32-
-33-
-34-
-35-
NODE4
NODE3
NODE2
NORTHWEST
NODE1
-0
—2
—3
—4
—5
—8
—7
—10
—11
—12
—13
—14
—15
—16
—17
—18
—IB
—20
—21
—22
—23
—24
—25
—28
—27
—28
—29
--30
—31
—32
—33
—34
—35
—38
—37
DISTANCE (FEET)
LEGEND
a< 10 PPM
150
110 - 100 PPM
H> 100 PPM
T - < 1 PPM
PPM - PARTS PER MILLION • - QUANITATIVE REFERENCE DATA
NOTE: QUANITATIVE REFERENCE DATA USED BECAUSE OF POOR SAMPLE RECOVERY
14
-------
FIGURE 3-3. TPH REFERENCE METHOD CHEMICAL CROSS SECTION - YORK SITE
NORTH
0-
-1-
-2-
-3-
-4-
-5-
-6-
-7-
-8-
-9-
-1O-
-11-
-12-
-13-
-14-
-15-
-16-
-17-
-18-
-19-
-20-
-21-
-22-
-23-
-24-
-25-
-26-
NODE1
NODE2
NODE3
NOOE4
NODES
377
NO
DISTANCE (FEET)
0 10
LEGEND
60
< 10 PPM
> 10,000 PPM
S/jyXl'O - 1OO PPM
NO - NOT DETECTED
L-..-_HIOO - 1.000 PPM F
r_ _ J £
PPM - PARTS PER MILLION
100 110
11.000 - 10,000 PPM
SOUTH
-0
—1
--2
—3
—4
—5
--6
—7
—8
—9
—10
—11
—12
—13
—14
—15
—16
—17
—18
—19
—20
--21
--22
—23
—24
—25
—26
FIGURE 3-4. PAH REFERENCE METHOD CHEMICAL CROSS SECTION - YORK SITE
NORTH
0-
-1-
-2-
-3-
-4-
-5-
-6-
-7-
-8-
-9-
-10-
-11-
-12-
-13-
-14-
-15-
-18-
-17-
-18-
-19-
-20-
-21-
-22-
-23-
-24-
-25-
-26-
NODE1
NODE2
NODE3
NODE4
NODES
DISTANCE (FEET)
70
90
SOUTH
-0
—1
—2
—3
—4
—5
—6
—7
--8
—9
--10
—11
—12
--13
--14
—15
—16
--17
--18
--19
—20
--21
--22
—23
--24
--25
--26
LEGEND
; < 10 PPM
[
•10-100 PPM
: 100 - 1.OOO PPM
• > 1.000 PPM
T - < 1 PPM
ND - NOT DETECTED
PPM - PARTS PER MILLION
15
-------
FIGURE 3-5. TPH REFERENCE METHOD CHEMICAL CROSS SECTION - FORT RILEY SITE
SOUTH
0-
-1-
-2-
-3-
-4-
-5-
-8-
-7-
-8-
-9-
-10-
-11-
-12-
-13-
-14-
-13-
-18-
-17-
-18-
-19-
-20-
-21-
-22-
-23-
-24-
-25-
-28-
-27-
-28-
-29-
-30-
-31-
NODE2
NODES
NODE3
NODE4
NORTH
-2
•-3
-4
•-5
•-«
•-7
•-8
•-9
—10
—11
—12
—13
—14
—15
—18
—17
—18
—19
—20
—21
--22
—23
—24
—25
—28
—27
—28
—29
—30
—31
DISTANCE (FEET)
ISO 160
LEGEND
2«/>910 - 100 PPM
NO - NOT DETECTED
p-X-|ioo - 1.000 PPM g|
PPM - PARTS PER MILLION
a 1.000 - 10.000 PPM
FIGURE 3-6. PAH REFERENCE METHOD CHEMICAL CROSS SECTION - FORT RILEY SITE
SOUTH
0-
-1-
-2-
-3-
-4-
-5-
-6-
-7-
-8-
-9-
-10-
-11-
-12-
-13-
-14-
-15-
-18-
-17-
-18-
-19-
-20-
-21-
-22-
-23-
-24-
-25-
-28-
-27-
-28-
-29-
-30-
-31-
NODE2
NODES
NOOE3
NODE4
NORTH
-0
—1
—2
—3
—4
—5
—8
—7
—8
—9
—10
—11
—12
—13
—14
—15
—16
—17
—18
—19
—20
—21
—22
--23
—24
—25
--28
—27
—28
—29
—30
—31
DISTANCE (FEET)
150
LEGEND K8$< 10 PPM t-X-lio - 100 PPM
ND - NOT DETECTED PPM - PARTS PER MILLION
1100 - 1.000 PPM
T - <1 PPM
16
-------
All of the nodes for this transect occurred in areas
that were impacted by the contamination associated with
this site. The contamination at this site appeared to
occur in a single band extending from approximately 10
to 22 feet bgs for total PAH and 2 to 25 feet bgs for
TPH contamination. This band of contamination thinned
from south to north across the transect. At the north end
of the transect, the TPH contamination thinned to a zone
extending from 12 to 19 feet bgs. Concentrations of
TPH in this zone ranged from 10 to 10,000 ppm, and the
concentrations of total PAH contamination range from
10 to 1,000 ppm. TPH contamination exhibited its
maximum concentrations hi a lens around Nodes 3 and
4. This lens extended from approximately 12 to 16 feet
bgs, and exhibited TPH concentrations greater than
1,000 ppm. This lens tended to thin and deepen from
south to north. Node 4 exhibited the greatest TPH and
total PAH contamination. Two narrow lenses of total
PAH concentrations greater than 1,000 ppm existed at
approximately 14 to 16 feet bgs and 18 to 20 feet bgs. A
narrow lens of TPH contamination greater than
10,000 ppm was detected at approximately 18 to 19 feet
bgs at Node 4.
Fort Riley Site
The five sampling nodes formed a south to north
trending transect. The TPH and total PAH distributions
appeared to be similar at the Fort Riley site (Figures
3-5 and 3-6). Node 4, situated at the far north end of
the transect, was not affected by contamination. All of
the remaining nodes for this transect occurred hi areas
that were impacted by the contamination associated with
this site. The contamination at this site appeared to
occur hi a single zone extending from approximately
1 to 25 feet bgs for total PAH and 0 to 30 feet bgs for
TPH contamination. This zone of contamination
exhibited relatively constant thickness across Nodes 2, 3,
and 5. Concentrations of TPH hi this zone ranged from
100 to greater than 10,000 ppm, and the concentrations
of total PAH contamination ranged from 10 to 300 ppm.
Total PAH contamination exhibited its maximum
concentrations in a lens around Nodes 2, 3, and 5. This
lens extended from approximately 5 to 8 feet bgs at
Node 3, becoming thicker and deeper at Node 2 where
it extended from 10 to 20 feet bgs. The TPH
concentrations exhibited two lenses of concentration at
greater than 10,000 ppm. These lenses did not appear to
be extensive and their occurrence was limited to the
areas around single nodes. Node 5 in the center of the
transect exhibited one of these lenses of highest TPH
contamination extending from 10 to 13 feet bgs. Node
2 has the other such lens which extended from 17 to
19 feet bgs.
Quality Assessment of Geotechnical
Laboratory Data
This section discusses the data quality of the geo-
technical laboratory results, the data quality of the
borehole logging conducted by the on-site professional
geologist, and the soil sampling depth control. The
stratigraphic cross sections resulting from this logging
are presented and discussed later hi this section.
Geotechnical Laboratory
Soil samples submitted for textural determination
were analyzed by ASTM Method D-422 (1990). ETS,
Petaluma, California, conducted these analyses. ASTM
Method D-422 does not define specific QAXQC criteria,
however, it specifies the use of certified sieves, and
calibrated thermometers and hydrometers. ETS
followed the approved method and complied with all the
equipment certification and calibration requirements of
the method. Based on this, the geqtechnical data was
determined acceptable.
Borehole Logging
The data quality of the on-site professional
geologist's borehole logs was checked through on-site
audits by a soil scientist, and by comparison of the
geologist's descriptions for intervals corresponding to
samples analyzed by ASTM Method D-422. This
comparison is discussed later hi this section.
Sampling Depth Control
At each site, random checks of the reference
sampling intervals were made. These checks consisted
of stopping drilling operations just before inserting the
split spoon sampler into the hollow stem auger to collect
samples. At this tune, a weighted tape measure was
used to measure the top of the sampling interval. The
measurement was checked against the intended sampling
depth. If the difference between the intended and actual
sampling depth had varied by more than 1 inch, the
borehole would have been redrilled. Depth checks were
made at a minimum of once per sampling day. None of
these depth checks resulted hi data exhibiting a greater
than 1 inch difference between intended and actual
sampling depth. Based on this, the reported sample
intervals were considered accurate.
Stratigraphic Cross Sections
Stratigraphic cross sections were based on the data
produced by a professional geologist during the logging
17
-------
FIGURE 3-7. REFERENCE METHOD STRATIGRAPHIC CROSS SECTION - ATLANTIC SITE
SOUTHEAST
NODES
NODE4
NODE3
NODE2
NORTHWEST
NODE1
-0
LEGEND
SILT
SILTY SAND
WELL GRADED SAND
POORLY GRADED SAND
( ) - LABORATORY CLASSIFICATION (USDA)
FIGURE 3-8. REFERENCE METHOD STRATIGRAPHIC CROSS SECTION - YORK SITE
NORTH
0-
-1-
-2-
-3-
-4-
-5-
-8-
-7-
-8-
-9-
-to-
-11-
-12-
-13-
-14-
-15-
-18-
-17-
-IB-
-19-
-20-
-21-
-22-
-23-
-24-
-25'
-28-
NODE1
NODE2
NODE3
NODE4
NODES
(SILTY CL
LOAII)
(SILT
LOAM)
(SANDY
LOAM)
(SANDY
LOAM)
DISTANCE (FEET)
40
50
60
LEGEND
CLAYEY SILT t SILT (ML)
WELL GRADED SAND
( ) - LABORATORY CLASSIFICATION (USDA)
80 90
I SILTY CLAY (CL)
POORLY GRADED SAND
SOUTH
-0
—1
—2
—3
—4
—S
—8
--7
—8
—9
--10
—11
—12
—13
—14
--15
--18
—17
—18
—19
—20
—21
—22
—23
—24
—25
—26
110 r.
11 SILT
\ SILTY CLAY (CH)
18
-------
FIGURE 3-9. REFERENCE METHOD STRATIGRAPHIC CROSS SECTION - FORT RILEY SITE
SOUTH
0-
-1-
-2-
-3-
-4-
-5-
-•-
-7-
-8-
-9-
-10-
-11-
-12-
-13-
-14-
-15-
-16-
-17-
-18-
-19-
-20-
-21-
-22-
-23-
-24-
-25-
-26-
-27-
-28-
-29-
-30-
-31-
NODE1
NODE2
NODES
NODE3
NODE 4
(LOAM)
(SANDY
LOAM)
NORTH
-0
--1
--2
--3
--4
--5
—6
--7
--8
--9
--10
--11
--12
—13
--14
—15
—16
—17
—18
—19
—20
—21
—22
—23
—24
—25
—26
—27
—28
—29
—30
DISTANCE (FEET)
—31
100 110 120 130 140 150 160 170 180 190 200 210 220
LEGEND
; SILTY CLAY (CL)
! SILTY CLAY (CH)
ff
( ) - LABORATORY CLASSIFICATION (USDA)
NW CLAYEY SAND
WELL GRADED SAND
POORLY CRAOED SAND
of boreholes during the demonstration. The cross
sections were intended to represent a conventional
approach to the delineation of subsurface stratigraphy.
The cross sections are presented on Figures 3-7 to
3-9. A verbal interpretation of these cross sections is
presented below by site. QA/QC consisted of the
collection of samples for textural analysis by a
geotechnical laboratory (see Appendix A, Tables
A-l through A-9). These samples are discussed at the
end of each site-specific discussion.
Atlantic Site
The Atlantic site is located on the flood plain of the
Nishnabotna River, which is located about 0.7 mile west
of the site. The flood plain is nearly level. The surface
soil at the site is a silty clay. These soils have most
likely formed from alluvium. Stratigraphic cross
sections produced during the demonstration from soil
borings indicated that the subsurface soil at the site
consisted of silts and clays and silty clay interfingered
with each other to a depth of approximately 21 feet bgs
on the northwest end (Node 1) and to 28 feet bgs on the
southeast (Node 5). See Figure 3-7 for a graphical
representation of the cross section. A layer of sand was
present from 18 to 36 feet bgs at Node 2. This zone
remained relatively uniform from Node 2 to Node 5.
Seven soil samples were collected to verify the
geologist's borehole logging at the Atlantic site. The
geologist's classification of soils matched the geotech-
nical laboratory's classifications six out of seven times
(Table 3-1). This one point of disagreement was sample
DR16 from the 2- to 3-foot interval at Node 1. In this
sample, the geologist identified silt as the predominant
size fraction of the sample, while the geotechnical
laboratory identified clay as the predominant size
fraction. This is a common point of variance between
field soil classification and laboratory classification.
These common differences are magnified hi grossly
contaminated soils, and when the geologist is forced to
wear plastic gloves during classification. This difference
was noted, and geologist's Stratigraphic borehole logs
meet the demonstration data quality objectives (DQO)
for screening level data.
York Site
The York site is located on the flood plain of Beaver
Creek, which is located 0.1 mile southwest of the site.
The site is situated on a nearly fiat lying terrace above
the river. The surface soils is a silt loam. These soils
mosit likely formed hi alluvium on stream terraces. A
Stratigraphic cross section based on soil borings during
the demonstration was prepared (Figure 3-8). The top
19
-------
TABLE 3-1. COMPARISON OF GEOLOGIST'S DATA AND GEOTECHNICAL LABORATORY DATA
ALL SITES
Site
Geologist Classification
Geotechnical Laboratory Classification
Match
Atlantic Silty Clay (ML)
Clayey Silt (CL)
Silt (ML)
Well Graded Sand (SW)
Clay (CL)
Silty Clay (CL)
Silty Clay (CL)
York Clayey Silt (ML)
Silty Clay (CL)
Well Graded Sand (SW)
Poorly Graded Sand (SP)
Clayey Silt (ML)
Sand (SW)
Fort Riley Poorly Graded Sand (SP)
Fill
Poorly Graded Sand (SP)
Clayey Silt (ML)
Poorly Graded Sand (SP)
Poorly Graded Sand (SP)
Poorly Graded Sand (SP)
Well Graded Sand (SW)
Sandy Lean Clay (CL) Noa
Clay or Silt (CL or ML) Yes
Silt or Clay (ML or CL) Yes
Well or Poorly Graded Sand (SW or SP) Yes
Sandy Lean Clay or Sandy Silt (CL or ML) Yes
Sandy Lean Clay or Sandy Silt (CL or ML) Yes
Silt or Clay (ML or CL) Yes
Silt or Clay (ML or CL) Yes
Silt or Clay (ML or CL) Yes
Silty to Clayey Sand (SM or SC) Noa
Poorly Graded Sand with Silt or Clay (SW-SC or SP-SC) Yes
Silt or Lean Clay (CL or ML) Yes
Silty or Clayey Sand (SM or SC) Moa
Silty or Clayey Sand (SM or SC) No
Silty or Clayey Sand (SM or SC) Yes
Silty or Clayey Sand (SM or SC) No
Silty or Clayey Sand with Gravel (SC or SM) Noa
Silty or Clayey Sand (SM or SC) Noa
Silty or Clayey Sand (SM or SC) Noa
Siity or Clayey Sand (SM or SC) Noa
Silty or Clayey Sand (SM or SC) Noa
Notes:
a
0
These failures to match were due to the geologist underestimating the percentage of fines in the sample.
Unified Soil Classification System two-letter code.
1 to 2 feet of the cross section was fill. From 2 to
14 feet bgs, the cross section consisted of clayey silt with
some lenses of silty clay and silt. At approximately
14 feet bgs, there were thick lenses of silt, sandy silt,
and sand. These lenses were approximately 7 feet thick
and were interfingered with each other. At
approximately 21 feet bgs, the material became
primarily well graded sand to the bottom of the section
at 25 feet bgs.
Six soil samples were collected to verify the
geologist's borehole logging at the York site. The
geologist's classification of soils matched the
geotechnical laboratory's classifications four out of six
tunes (Table 3-1). The two points of disagreement were
samples DR27 (Node 1, 15 to 15.5 feet bgs) and DR
29 (Node 3, 12 to 13 feet bgs). In both cases, the
geologist underestimated the percentage of silt and clay
size particles in the samples. This is a common point of
variance between field soil classification and laboratory
classification. These differences are magnified in
grossly contaminated soils, and when the geologist is
forced to wear plastic gloves during classification
activities. The variances described above are not
uncommon in environmental studies, and thus, the
geologist's stratigraphic borehole logs, while odiibiting
20
-------
some disagreement with the laboratory data, are
considered to meet the demonstration DQOs for
screening level data.
Fort Riley Site
The Fort Riley site is located on the flood plain of
the Kansas River, which is located 0.1 mile southeast of
the site. The site is situated on a nearly flat lying terrace
above the river. The surface soil is a silt loam. This
soil is most likely formed from deep alluvium. A
stratigraphic cross section based on soil borings
conducted during the demonstration is presented on
Figure 3-9. This cross section showed typical deposition
in an alluvial setting with interfingered beds of clay, silt,
silty clay, clayey silt and sand. In the center of the cross
section, the top 8 feet was fill. The northern and
southern edges of the cross section were silt or silty clay
at the surface. In the northern half of the cross section,
poorly graded sand was present from 5 to 17 feet bgs.
In the southern half of the cross section from 8 to
approximately 18 feet bgs, the cross section consisted of
interfingered lenses of clay, silty sand, and sand. Below
20 feet bgs, the cross section became primarily sand
with silt and clay lenses of various thickness intermixed
to the terminal depth of the cross section.
Eight soil samples were collected to verify the
geologist's borehole logging at the Fort Riley site. The
geologist's classification of soils matched the geo-
technical laboratory's classifications one out of eight
times (Table 3-1). Both classifications did correctly
identify the dominant particle size fraction. In all of the
cases of disagreement, the geologist underestimated the
percentage of silt and clay size particles. Small shifts in
the estimation of these particles can alter the descriptive
modifier used in classification. The variances described
above affect the accuracy of the reference stratigraphic
cross sections as far as the secondary classification
modifiers are concerned. The baseline classification as
to the dominant particle size is accurate. This data met
the demonstration's DQOs, however, decisions based
solely on differences in classification modifiers should be
qualified as semiqualitative Table 3-1.
21
-------
Section 4
Rapid Optical Screening Tool
This section describes the HOST™ technology
evaluated under this demonstration. The description
provided is based on information provided by the
developer, on information PRC obtained from reports
and journal articles written about the technology, and on
observations made during the demonstration. The
description includes background information on the
technology and its components. General operating
procedures, training and maintenance requirements, and
cost of the technology also are discussed.
Background Information
The Department of Defense, through the U.S. Air
Force's Armstrong Laboratory, has supported research
at NDSU to develop tunable dye laser systems for field
environmental analysis since 1989. A prototype tunable
system built at NDSU was integrated with a cone
penetrometer truck and demonstrated at Tinker Air
Force Base in 1992. Follow-up demonstrations of the
tunable dye laser systems for optical cone penetrometry
were carried out in late 1993 and 1994 at the following
air force bases: Plattsburgh, Patrick, and Dover. This
second round of demonstrations used an improved
prototype built by DTI. DTI is a small business formed
by two individuals who developed the original tunable
dye laser system at NDSU.
In 1993, a consortium comprised of Unisys
Corporation, DTI, NDSU, and the U.S. Air Force
Armstrong Laboratory received a Technology
Reinvestment Project award from the Advance Research
Projects Agency, which led to the development of the
ROST™. Loral Corporation acquired the portion of
Unisys responsible for the ROST™ development and
services on May 5, 1995. For this demonstration,
ROST™ was temporarily installed on a cone
penetrometer truck supplied by a subcontractor to PRC.
The subcontractor selected was Fugro Geosciences, Inc.
(Fugro); however, the developer stated that ROST™ is
compatible with almost any standard cone penetrometer
truck.
Components
This subsection describes the components of ROST™
and the cone penetrometer truck system.
Cone Penetrometer Truck System
A complete cone penetrometer truck system consists
of a truck, hydraulic rams and associated controllers,
and the CP itself. The weight of the truck provides a
static reaction force, typically 20 tons, to advance the
CP. The hydraulic system working against the static
reaction on force advances 1-meter-long segments of
3.57-centimeter-diameter threaded push rod into the
ground. The CP, which is mounted on the end of the
series of push rods, contains sensors that continuously
log tip stress and sleeve friction. The ROST™
technology's fiber optic cables, mirrors, supports, and
sapphire window are built into a "sub" which is fitted to
a standard CP. This CP is fitted with tip stress and
sleeve friction sensors. The data from CP sensors are
used to map subsurface stratigraphy. Conductivity or
pore pressure sensors can be driven into the ground
simultaneously with the modified CP. Soil, ground-
water, and soil gas sampling tools can also be used with
the cone penetrometer truck. These capabilities are
discussed hi greater detail in the general ITER.
Generally, sampling tools and sensors are not used
concurrently due to the necessity for sampler retrieval
after each sample is collected.
In favorable stratigraphies, push depths of 50 meters
or greater have been achieved. The CP can be pushed
through asphalt, but concrete must be cored prior to
advancing the CP. Advancing sensors and sampling
tools with the cone penetrometer truck may be difficult
or impossible in the following subsurface environments:
Gravel units
Cemented sands and clays
Buried debris
Boulders
Bedrock
22
-------
The cone penetrometer truck used with the ROST™
during this demonstration was fitted with a steam cleaner
to decontaminate the push rods as they were withdrawn
from the ground. The decontamination water is
contained in the decontamination apparatus and it can be
directly discharged into a storage container. The clean
water, pressure sprayer, and grouting pump are mounted
in a trailer that is towed by a support vehicle.
ROST™Technology
The main HOST™ system components are:
• Neodymium-doped Yttrium Aluminum Garnet
(Nd:YAG) primary laser
• Tunable dye laser pumped by the Nd:YAG
laser
• Fiber optic cable and CP "sub"
• Detection system comprised of a mono-
chromator, photomultiplier tube (PMT), and
digital storage oscilloscope (DSO)
• Control computer
In the prototypes that preceded ROST™, the
components of the tunable dye laser were arranged on a
122-centimeter (cm) by 61-cm optical breadboard, which
sat on top of a cart in which the other components were
placed. The ROST™ technology used in this
demonstration is more compact and integrated. The
ROST™ components are assembled into two half-height
instrumentation racks. Each rack is approximately
66-cm high by 51-cm wide by 61-cm deep. A diagram
that shows how the ROST™ system components are
arranged in the racks is shown on Figure 4-1. The main
ROST™ components are discussed in more detail below.
All of the electronic components are powered by a
gasoline generator, independent of the cone penetrometer
truck power system.
Nd:YAG Laser
The primary laser is a Nd:YAG laser with a second
harmonic generation option manufactured by Big Sky
Laser Technologies. It generates 532 nanometers (nm)
of light at a repetition rate of 50 pulses per second (Hz).
The near-Gaussian output beam is approximately
6.35 cm in diameter. The primary laser is rated at a
pulse energy of 50 millijoules per pulse (mJ). It requires
standard 110-volt line voltage; a dedicated 20^-amp
circuit is recommended. Because water to cool the
Nd:YAG rod is supplied by an internal circulator hi the
power supply, no external source of cooling water is
required. The light from the primary laser pumps the
dye laser.
Tunable Dye Laser
The dye laser converts the fixed wavelength light of
the Nd:YAG laser into wavelengths that can be
optimized for the contaminants) of interest. The dye
laser is pumped by the Nd:YAG laser output. The
output wavelength range of the dye laser depends on
which dye is used. The dye was Rhodamine 6G for this
demonstration. The wavelength range that can be
generated from Rhodamine 6G is approximately 560 to
600 nm.
The dye laser uses Bethune prism dye cells, which
are right angle prisms with a narrow bore through which
the dye flows. The collimated light from the Nd:YAG
laser enters the hypotenuse face of the prism. After
undergoing total internal reflection at the other two
faces, the light creates a highly uniform illumination of
the dye solution that is flowing through the Bethune
cells. The developer states that a major advantage of the
Bethune cells is that the Nd:YAG laser light does not
have to be focused into the dye cell. Collimated light (or
nearly so) is sufficient to maintain the total internal
reflection condition.
The wavelength of the monochromatic laser light,
which is produced in the oscillator of the dye laser, is
selected to be optimal for the detection of specific
compounds. The oscillator consists of the Bethune cell,
cavity mirrors, and a wavelength dispersing element (a
diffraction grating in this case). The broad Rhodamine
6G fluorescence (560 to 600 nm) that exits the bore of
the Bethune prism cell hits a feedback element that
reflects a portion of the light back through the Bethune
cell bore. The feedback element, which can be a
wedged piece of fused silica, also serves as the output
coupler of the dye laser. On this second passage, the
light is amplified by stimulated emission into a narrow
pencil of light. On the other side of the prism cell, the
pencil of light grazes the surface of a diffraction grating.
The diffraction grating disperses the light. That is, it
spatially separates light into its wavelength components
in the same fashion as a prism. A tuning mirror returns
the diffracted light back to the grating at which point the
return beam is re-diffracted. Only a narrow range of
wavelengths within the rediffracted beam follow the
proper path to pass back through the Bethune cell. This
monochromatic light is further intensified by stimulated
emission as it passes back through the dye cell. It exits
the laser cavity through the feedback element. By
varying the angle of the tuning mirror, one may select
the wavelength of light which finally exits the dye laser
cavity, which is why the dye laser is referred to as
tunable.
23
-------
FIGURE 4-1. SYSTEM COMPONENTS
A. Control rack
B. Spectrometer rack
C. Nd:YAG power supply
D. Optical breadboard
E. Frequency doubling crystal
F. Nd:YAG'laser head
G. Dye laser
H. Monochromator
I. Stepper motor controller/driver
J. Photomultiplier tube and housing
K. High voltage power supply
L. Stepper motor reset button
M. Water reservoir
N. Dye reservoir
O. Monitor
P. Keyboard
Q. Utility Drawer
R. Computer
S. Oscilloscope
T. Temperature Controller
The monochromatic laser light emerging from the
oscillator of the dye laser is amplified as it passes
through a second Bethune cell, which is also energized
from the Nd:YAG laser. The amplified beam then
passes through a frequency doubling crystal, which
converts a portion of the incoming light into photons of
half the wavelength, in this case in the range 280 to
300 nm. The unconverted visible light is removed by a
blocking filter and the remaining light is focused onto the
launch end of the fiber optic cable.
Fiber Optic Cable
The fiber optic cable delivers excitation light from
the dye laser through the sapphire window in the
modified CP and returns any fluorescence emission
arising from aromatics contamination in the soil back to
the PMT, which is a component of the detection system.
The sapphire window is typically 2 to
3 millimeters (mm) thick and 6.35 mm in diameter.
This window is mounted flush with the outside of the
standard CP sensor casing. The window is located about
20 to 25 centimeters above the terminal end of the CP
sensor casing. This window allows laser light to pass
from the ROST* and into the soil. The window also
transmits LJF from the soil back into a return fiber optic
cable. Sapphire is used due to its good optical
transmission characteristics and due to its abrasion
resistance. The fiber that transmits the laser light to the
sapphire window is referred to as the delivery fiber.
The return fiber (or fibers) is referred to as a collection
fiber (or fibers). The delivery and collection fibers run
from the dye laser through the center of the series of
push rods to which the CP is attached.
The fibers have three concentric layers. The light
travels through the innermost zone known as the core.
The middle layer is a cladding material that has a higher
reflective index than the core fiber. This difference in
reflective index causes internal reflection as light to
move through the core. In effect, the cladding traps
light within the core. The outer layer is a protective
abrasion-resistant buffer. For additional protection, the
fibers are inserted into furcation tubing.
For this demonstration, the developer used a single
delivery fiber and a single collection fiber. Sensitivity
could be increased with additional collection fibers.
Detection System
The detection system consists of a monochromator,
PMT, and a DSO. The monochromator incorporates
entrance and exit slits, a diffraction grating, and several
mirrors. The monochromator acts as a variable
wavelength narrow bandpass filter. By acquiring
fluorescence data at a series of wavelengths, the
fluorescence technician can determine the wavelength of
24
-------
maximum intensity in the fluorescence spectrum. The
light passing through the monbchromator at this
wavelength is converted to an electrical signal by the
PMT. The signal from the PMT is fed to the DSO,
which displays the waveform (fluorescence intensity as
a function of time following the excitation laser pulse).
Typically, the data are acquired over a window of 100 to
500 nanoseconds (ns). The DSO also averages the signal
from several consecutive laser shots. After the selected
number of laser shots have been averaged, the waveform
is downloaded to the control computer for permanent
storage and post-processing of the data.
Control Computer
The ROST™ technology is controlled by a personal
computer, which also is used for off-line data analysis.
Several ROST™ technology motors are controlled by the
computer: one to drive the tuning mirror of the dye
laser oscillator, one to position the frequency doubling
crystal, and one to drive the monochromator wavelength
selection. The control computer also sets the high
voltage on the PMT and provides communication with
the DSO via a GPIB bus.
General Operating Procedures
The sapphire window is located in a "sub" that is
fitted to a standard CP. These push rods are
approximately 1 meter long and must be continuously
added as the CP is advanced using a hydraulic ram. The
standard penetration rate is approximately 2 centimeters
per second (cm/s). The maximum depth of a push is
determined by soil matrix conditions and the static
reaction force of the cone penetrometer truck. A
standard 20-ton cone penetrometer truck was used for
this demonstration.
The CP also provides continuous monitoring of tip
stress and sleeve friction. These data are downloaded
onto a computer with software that classifies the soil
type.
When sensors are not attached to the push rods,
conventional soil, soil gas, and groundwater sampling
tools can be fitted to the push rods for environmental
sampling. In addition, cone penetrometer trucks can be
used to install small diameter piezometers.
Before collecting data at a site, the ROST™ is
calibrated. The ROST™ operator attaches a sealed vial
containing 10,000 ppm solution of gasoline to the
sapphire window. The emission wavelength of the laser
is set and multiple laser shots are fired at the solution
and the resultant fluorescence is measured. The data
system is then calibrated to read 100 percent fluo-
rescence based on the fluorescence of the standard at the
predetermined emission monitoring wavelength. All
subsequent data is reported as a percent fluorescence
relative to the standard.
The ROST™ technology can be operated hi both
dynamic (push) and static modes. In the dynamic mode,
the modified CP equipped with the sapphire window and
fiber optics is advanced at a rate .between 1.5 and
2.5 cm/s. In this mode, which the developer refers to as
FVD (fluorescence versus depth), the excitation laser
wavelength and fluorescence emission monitoring
wavelength are held constant. The fluorescence
emission intensity is plotted as a function of depth bgs.
The developer selected an excitation wavelength in the
• range 280 to 295 nm for this demonstration. This
wavelength range was selected because naphthalene, a
major PAH constituent of coal tar and diesel fuel,
fluoresces strongly under these conditions. The emission
monochromator was set at a wavelength determined
during the laboratory analysis of the predemonstration
samples. It was set at 400 nm for the Atlantic and York
sites, and at 360 nm at the Fort Riley site. With these
laser and detector settings, the ROST™ was configured
to detect the presence or absence of primary fluorescing
compounds associated with the petroleum fuel and coal
tar contamination expected during this demonstration.
For this demonstration, the laser repetition rate was
50 Hz. The DSO in the current ROST™ system
configuration averages the time-integrated signal from
these 50 pulses to produce a fluorescence datum for that
depth. The spatial resolution was, therefore, 2 cm. The
complete averaged fluorescence intensity versus depth
profile was archived for post-processing.
Once areas of significant contamination have been
identified in the dynamic mode, ROST™ can be operated
in the static mode to identify the general class of
contamination present. In this mode, the CP is held at
a fixed depth. The fluorescence technician, who is
observing the fluorescence signal visually, can simply
signal the hydraulic operator to halt the push. ROST™
also can operate in the static mode when additional push
rods are added to the string.
During the static mode, ROST™ can obtain
multidimensional data representations, called wave-
length-time-matrices (WTM). These differ from the two
dimensional FVDs that are plots of relative fluorescence
intensity versus depth. A WTM represents a
3-dimensional plot of relative fluorescence intensity
versus fluorescence lifetimes versus wavelength. WTMs
produce contaminant class specific three-dimensional
figures. Preliminary research indicates that this
25
-------
characteristic image can be used in a fashion similar to
fingerprinting to identify contamination. This data can
be used to identify contaminants present, for example,
the type of fuel that is present. For WTM acquisition,
either the excitation or the fluorescence emission
wavelength can be varied. Normally the excitation
wavelength is held constant and the emission monitoring
wavelength is varied. This was the procedure used
during the demonstration. For a WTM, the data from
100 to 200 laser pulses are averaged at each of a series
of return emission wavelengths separated by 10 nm,
between 50 and 300 nm. The acquisition of a WTM
takes about 5 minutes longer than the time required to
add an additional probe rod for continued probe
advancement.
Training and Maintenance Requirements
One person is needed to operate the ROST™
technology and two are needed to operate the cone
penetrometer truck. A crew chief operates the
hydraulics of the cone penetrometer truck and monitors
the push depth, cone tip resistance, sleeve friction, and
pore pressure or conductivity measurements if taken.
The rod person screws on additional push rods after
completing each 1-meter push. The fluorescence
technician operates the ROST™ technology.
The technology is presently offered only as a service
from Loral Corporation. Their operators are fully
trained in the operation and maintenance of the
equipment and the interpretation of the data. The typical
operator has a bachelor's degree hi a science or
engineering discipline, plus any prerequisite environ-
mental field training. Typically the ROST™ operators
have 30 days of training hi the use of the technology.
This training consists of classroom and on-the-job
training.
Cost
Because the ROST™ technology is still in the
developmental stage, a specific sales cost for the
technology has not been established. Currently, use of
the ROST™ technology is contracted out on a job-by-job
basis. The developer envisions ROST™ being sold as a
complete unit eventually. Training would likely be
incorporated into the sale price.
The developer estimates that the daily rate for use of
a cone penetrometer truck and ROST™ would be
between $5,000 and $5,500. Mobilization and operator
per diem costs are not included. The ROST™ technology
and one operator costs $2,800 per day if a cone
penetrometer truck is already being supplied. These
costs also do not include mobilization or per diem. The
ROST™ technology was used with a Fugro cone
penetrometer truck. The cone penetrometer truck costs
incurred during the demonstration were $2,350 per day,
which included mobilization for the Fugro cone
penetrometer truck and two operators. If the cost of
subcontracting the cone penetrometer truck is added to
the daily use charge of the ROST™ technology, a daily
use rate for a complete ROST™ rig, as used hi this
demonstration, would be approximately $5,150 per day.
The total cost of the three site characterizations
performed using ROST™ would have been approximately
$41,200. For comparison, the predemonstration
activities produced similar data at the three sites,
however, it required more personnel and on-site
analytical capabilities at an approximate cost of $43,000.
The predemonstration resulted hi fewer data points,
relative to the continuous data output of the ROST™ and
CP. In addition, the predemonstration activities only
produced one borehole log at each site. Another cost
comparison can be made relative to the costs accrued
producing the reference cross sections for this
demonstration. Data acquisition and production costs of
the reference cross sections cost approximately $55,000,
including approximately $30,000 for drilling services,
approximately $8,000 for the on-site geologist and
sampler, approximately $12,000 for off-site analytical
services, and approximately $5,000 for handling and
disposal of investigation derived waste.
Observations
An observer was assigned to the ROST™ technology
demonstration to assess the following operational factors:
• Cost
• Ease of operation, ruggedness, and reliability
• Sampling capabilities and production rates
A summary of the observations made during the
ROST™ demonstration is included hi the following
paragraphs. The developers' corrective actions and
responses to some of these observations are presented hi
Section 7.
The cone penetrometer truck equipment, as
previously stated, was supplied by Fugro Geosciences,
Inc., Houston, Texas. The cone penetrometer truck
consisted of a driver's cab and an enclosed 15- by 8-foot
rear compartment, which housed the cone penetrometer
truck pushing equipment and tools, and the electronic
equipment used by both the cone penetrometer truck and
ROST™. The technology's components were supplied by
the developer and included the equipment previously
identified hi this section of this report. In addition, the
ROST™ system used a printer to allow in-the-field
printouts of push data.
26
-------
Major consumables for the ROST™ demonstration
were water, grout, bentonite, cement, diesel fuel,
kerosene, and gasoline. Repair parts for the ROST™ or
the CP could be stored in the cone penetrometer truck or
sent by the manufacturer by overnight carrier.
Use of the ROST™ technology is limited mainly by
physical factors relating directly to the cone
penetrometer truck. These include terrain, both above
ground (clearance from overhead hazards), and below
ground (presence of gravel or rock fragments and buried
utilities). While attempting to push through the coarse
and fine sands, the overlying clay and silt layers
provided little lateral support to the cone penetrometer
truck rod, increasing the possibility of bending or
breaking the rod. This caused the operator to terminate
pushes in most cases at approximately 30 feet below
ground surface. The cone penetrometer truck operator
can overcome this problem by placing casing through the
clay and silt layer. The casing provides adequate lateral
support to the rod to allow the cone penetrometer truck
to push through the deeper sand layers.
The developer of ROST™ claims it is capable of
performing 300 feet of cone penetrometer truck pushes
per 10-hour work day. The demonstration showed that
this claim was accurate. Initial setup of the ROST™
technology required 4 hours for this demonstration, and
takedown and stowage time between site mobilizations
was approximately 2 hours for this demonstration.
Specific problems that occurred during the
demonstration included:
• Moisture fogging the sapphire window was
experienced during three pushes, twice at the
Atlantic site, and once at the York site. The
push rod was dismantled and the window was
cleaned with methanol. When the rod was
reassembled, great care was taken to avoid
introduction of moisture from ambient air into
the rod. The O-ring seals were also replaced to
ensure that the rod was leak-tight. Each time
this corrective action was taken, it required
approximately 3 hours of down time.
• The fiber optic cable broke during a push at the
Atlantic site. This may have been due to stress
applied to the cable during storage and push rod
handling. The fiber optic cable was replaced
and the system was recalibrated. This repair
and recalibration required approximately
4 hours of down time. A portion of the cable
lies on the floor of the cone penetrometer truck
during pushes and is always susceptible to
breakage from falling tools or from being
stepped on.
• The self-contained decontamination unit
frequently leaked during usage. This was due
to physical abrasion of the rubber gaskets used
to scrape soil and provide a water tight seal
around the push rods. This is a potential route
for contaminant migration down the probe hole.
The leaking water tended to migrate into the
push rod hole along the outside of the push
rods. Based on the performance, to prevent
leakage, the operator (Fugro) suggested that
these gaskets be replaced after every cone
penetrometer truck push, and checked for
leak-tightness before each cone penetrometer
truck push. Replacement of these gaskets
required approximately 0.5 hour of down tune.
• The depth-gauge cable for the push rods broke
during a push at the Atlantic site. Parts were
acquired from a local hardware store to fix the
device. This repair required approximately
2 hours of down tune.
• The independent depth-gauge and recording
device used by the ROST™ technology did not
correspond with the cone penetrometer truck
depth gauge after the initial two pushes at the
Atlantic site. Because accurate measurements
from the cone penetrometer truck push rod
depth gauge were available, depth corrections
were made. The failure of the depth recording
device was caused by slippage of the ROST™
device's depth measuring wheel during the
push. The depth recording wheel is made of a
smooth hard plastic. ROST™ technicians are
considering replacing the hard plastic with a
more slip resistant material.
• The depth of penetration of the CP was limited
to 30 to 35 feet bgs at the Atlantic and York
sites. This limitation was imposed by the cone
operator in an attempt to prevent damaging the
equipment. This decision was based on the
subsurface stratigraphy. During the pushes at
these sites, soft clay or silt layers were logged
by the CP as occurring above a harder sand or
stiff clay zone at the terminal end of the pushes.
Since the overlying soft clays and silts do not
provide lateral support for the push rods,
pushes were terminated as soon as any hard
zone, relative to overlying zones, were
encountered. Even with the relatively shallow
push depths, the ROST™ technology was able to
27
-------
apparently map the vertical extent of elevated
fluorescence at all three sites.
Data Presentation
To qualitatively assess the abilities of the ROST™
technology to identify the subsurface physical properties
of a site, it was necessary to collect soil physical data at
each of the five sample nodes at each site. The nodes
were arranged in a transect line across a known area of
subsurface soil contamination, which was identified
during predemonstration sampling and previous
investigations conducted at each site.
The soil texture data generated by the technology
was used to produce soil texture cross sections along
each transect line. A comparison of its data to that of
the reference methods is discussed in Section 5. The
following sections present chemical and physical data for
the ROST™ technology by site.
Chemical Data
The ROST™ sensor and CP sensors data is presented
and discussed as cross sections. The comparative
evaluation of this data against the reference methods is
discussed in Section 5. The ROST™ data used for the
quantitative evaluation hi Section 5 is presented in Tables
4-1, 4-2, and 4-3.
The ROST™ LIF logs are plotted as relative FVD
and are used to describe the relative distribution of
subsurface contaminants. The WTMs are used to
identify changes in contaminant type. The ROST™ did
not produce its own cross sections, therefore, PRC
prepared ROST™ LIF cross sections based on the FVD
data, and on scales that matched the ones used for the
reference methods. The contour intervals for the cross
sections were based on order of magnitude ranges. The
ROST™ technology's standard graphical outputs are
panel plots of relative fluorescent intensity versus depth
and WTMs. The written data evaluation of these cross
sections is presented below.
The ROST™ LIF data is reported as fluorescence
intensity relative to the fluorescence intensity of a
20,000 ppm gasoline standard. Theoretically, changes
in intensity relative to the standard can be used to assess
relative changes in concentrations of subsurface
fluorescing materials, such as the PAHs associated with
coal tars and petroleum fuels. In practice, as the LIF
intensity increases, the concentration of fluorescing
contaminants also may increase. One objective of this
demonstration was to directly evaluate the relationship
between changes in the ROST™ LIF intensity data and
changes hi contaminant concentrations.
The following data summary was provided by the
ROST™ personnel, and represents a typical narrative
data evaluation provided by ROST™. PRC edited this
evaluation and removed text that was not directly related
to cross section definition.
Atlantic Site
The ROST™ LIF and CP measurements at the
Atlantic site were carried out from approximately
mid-day on Sunday, August 14 to mid-day on August 16.
Three pushes were made on Sunday to test the ROST™
performance after it was integrated into the cone
penetrometer truck, which was provided and operated by
personnel from Fugro. A new device to log depth
information independent of the push rod depth gauge was
also tested for the first tune. It used a rotary encoder
which was attached to a roller wheel pressed up against
the push rods.
The final demonstration plan (PRC 1994) called for
two pushes at each node. Three pushes were made at
Node 2, but the data from the first push was discarded
because of a depth encoding error. All FVD readings
were measured at a fluorescence monitoring wavelength
of 400 nm, except for pushes at Node 1, which were
performed at 340 nm. The choice of 400 nm for the
monitoring wavelength was a compromise since a mix-
ture of coal tar and gasoline contamination was expected
at the site; the optimal (most sensitive) wavelength for
gasoline detection is around 350 nm, whereas coal tar
gives the strongest signal at about 500 nm.
The FVDs are discussed and grouped by node. All
of the FVDs were normalized by dividing the actual
measured fluorescence intensity at each depth by the
intensity of the 20,000 ppm gasoline standard. After the
FVD was interpolated to 0.1-foot depth intervals, it was
then background corrected. Once the data is background
corrected, negative values can be produced as artifacts
of background noise (variance), and should be con-
sidered zero fluorescence reading. This background
corrected data was used to produce the chemical cross
section presented on Figure 4-2.
Nodes 1 and 2
The official site demonstration activities began on
August 15 with a push at Node 1 on the west end of the
site. Based on the data from this push, Node 1 was in a
background region, meaning an area of low to no
contamination. This interpretation is based on the fact
mat the FVD was flat over the entire 33 feet of the push.
28
-------
TABLE 4-1. QUANTITATIVE ROST™ DATA - ATLANTIC SITE
Node
2
2
3
4
4
4
5
5
Note:
a
TABLE
Node
1
2
2
3
4
4
5
Note:
a
Fluorescence , Fluorescence
Depth No. of Reading Reading
(feet) Samoles Maximum3 Minimum3 Mean
21-22 11
24-25 11
16-17 11
6.5-7.5 11
10-11 11
27.5-28.5 11
16-17 11
23.5-24.5 11
Fluorescence readings are
128.40
7.04
69.55
108.40
223.30
559.20
20.89
174.20
reported as a percentage of the
2.22
-2.11
17.16
2.78
126.60
384.40
3.25
3.52
calibration
71.99
2.94
38.80
21.56
174.60
498.00
12.61
67.30
standard.
Standard
Deviation
42.82
2.83
20.34
31.56
26.87
50.31
5.63
69.26
4-2. QUANTITATIVE ROST™ DATA - YORK SITE
Depth No. of
(feet) Samples
15-16 11
13.5-14.5 11
17-18 11
17-18 11
14-15 11
18-19 11
1.5-2.5 11
Fluorescence Fluorescence
Reading Reading
Maximum3 Minimum3
5.51
172.00
40.93
152.90
107.76
1.16
74.95
0.08
102.30
4.95
29.09
90.38
-3.92
3.45
2.16
132.40
20.72
101.50
97.14
-1.37
23.59
Standard
Deviation
1.62
25.05
12.75
43.84
5.74
1.66
21.49
The cone penetrometer truck was then moved to
Node 2. During the first push at Node 2, a mismatch of
approximately half a meter between the cone
penetrometer truck and ROST™ system depth gauges was
noticed. After some experimentation, it was determined
that the ROST™ encoder was adversely affected by rod
vibration as the hydraulic ram's push-clamp was moved
during rod changes. A slight modification in Fugro's
rod handling procedures eliminated the problem. The
FVD of the second push at this location revealed
narrows band of contamination located at 0 to 2 feet bgs
and at 21.4 feet bgs. If the depth encoding error for the
first push is used to correct the first push data; then there
is a good match between the two FVDs. This indicates
that there is adequate back-up for the push-depth
monitoring system.
It was noted that the background levels for the first
several pushes on August 15 were higher than normal.
Based on this observation, the ROST™ operator decided
to examine the optical module. After the cone was
disassembled, droplets with the odor of fuel were
observed on surfaces in the optical module. The ROST™
operator believed that during a test push the previous day
near a monitoring well containing free phase gasoline,
29
-------
TABLE
Node
1
1
2
2
5
5
4.3. QUANTITATIVE ROST™ DATA - FORT RILEY SITE
Depth
fleet)
2-3
13-14
6-7
17-18
10.5-11.5
16-17
No. of
Samples
11
11
11
11
11
11
Fluorescence
Reading
Maximum3
44.13
0.35
57.26
143.90
216.90
208.20
Fluorescence
Reading
Minimum3
-0.32
-0.53
6.50
87.15
164.00
96.13
Mean
16.94
-0.20
34.88
120.20
188.40
171.07
Standard
Deviation
15.62
0.26
18.66
17.04
20.05
41.15
Note:
Fluorescence readings are reported as a percentage of the calibration standard.
FIGURE 4-2. ROST™ CHEMICAL CROSS SECTION - ATLANTIC SITE
SOUTHEAST
NODES
-2-
-3-
-10-
•^12-
-13-
-14-
-15-
-16-
-28-
-29-
-30-
-31-
-32-
-33-
-38-
-37-
NODE4
NODE3
NODE2
NORTHWEST
NODE1
-2
-3
-4
-5
-•
-7
-8
-9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
--28
--29
--32
--33
--34
DISTANCE (FEET)
25
LEGEND
0 - 10 (X OF STANDARD)
ISO
K»J 10-100
250
100 - 1.000
the cone passed through a thick zone of gross gasoline
contamination. Some of this fuel contamination
probably leaked into the cone and then worked its way
into the optical module overnight. The high, constant
background levels were believed to be caused by
gasoline contamination on the optical components inside
the sensor. The wavelength-time matrices discussed
below support this interpretation. Even after the original
background correction was applied, the effect of the
elevated background was still detected hi the form of a
noisier than usual baseline.
After the optical module was cleaned and
reassembled, a third push was made at Node 2. The
background had been reduced by more than a factor of
10 after the maintenance. In all aspects, the FVDs at
Node 2 were exceptionally consistent, probably as much
so as at any of the nodes at all three sites.
30
-------
The cone penetrometer truck was then returned to
Node 1 for a second push as a background check.
Nodes 3 and 4
Based on site-specific historical data, the heart of
the coal tar source was thought to lie slightly south of
Node 3. The FVDs for Node 3 reveal intense
fluorescence; maximum amplitudes reach nearly
1,000 percent of the standard. However, even greater
fluorescence intensity was found at Node 4. Not only
were the signal amplitudes much higher at Nodes 3 and
4 than at Node 2, but the contamination also extended
over much thicker depth intervals. Taken as a whole,
the four FVDs at Nodes 3 and 4 give a clear indication
of two separate, upper and lower, zones of
contamination. The fluorescence signal in the upper
zone extended from 10 to 17 feet bgs at Node 3 and it
was slightly broader (7 to 19 feet bgs) and stronger at
Node 4. The lower zone extended from 21 to 31 feet
bgs. The FVDs in the lower zone appeared to be more
variable than hi the upper zone.
Node 5
Node 5 is located across Poplar Street east of the
coal gasification site. The distance from Node 4 to
Node 5 is considerably greater than the separation
between Nodes 3 and 4. Based on historical data, the
subsurface groundwater flows to the north. This means
that the coal tar plume would have to move a
considerable distance cross gradient to reach Node 5. On
the other hand, Node 5 is closer to one of the sources of
gasoline contamination found at this site.
In general, the contamination at Node 5 was not
nearly as great as at Node 4 to the west. Between
24 and 29 feet bgs, there were several very narrow (less
than 0.5 feet thick) layers of high concentration
contamination. In the second push at Node 5, the
contamination reached over 1,500 percent of the
standard. It seems likely that it was associated with free
phase contaminants moving in a seam; however, the
corresponding cone penetrometer truck-generated
subsurface geological data offered no evidence of such
a seam. Both of the Node 5 pushes exhibited nearly
constant high or low fluorescence intensity from ground
surface to about 15 feet bgs. This phenomenon was not
related to the background elevation problems observed
at Nodes 1 and 2. As supported by the fact that the
signal returned to background levels from 18 to 24 feet
bgs.
Twenty-two WTM analyses were acquired during
the course of this demonstration. This data was used to
identify types or classes of contaminants. Most of the
WTMs were dominated by the fluorescence of coal tar
(Figure 4-3). The exceptions included the ones taken at
Node 5 near the source of the gasoline contamination,
the ones produced when samples of free-phase gasoline
from on-site monitoring wells were analyzed, and two
taken relatively near the surface at Nodes 4 and 5,
10.43 feet bgs and 10.79 feet bgs, respectively. The
WTM taken at Node 5 resembled free-phase gasoline
collected just east of Node 5 (Figure 4-3). The WTM
taken at the surface at Nodes 4 and 5 bore some
resemblance to gasoline and were distinctly different
from coal tar; they most likely represented a mixture of
coal tar and gasoline (Figure 4-3).
York Site
The LIF and CP measurements were carried out
sequentially from Node 1 to Node 5 with a couple of
exceptions. This data was used to produce the chemical
cross section presented on Figure 4-4. After the first
push at Node 1 showed that there was very little
contamination at this location, the cone penetrometer
truck was moved to Node 2. The two pushes at Nodes
2 and 3 were completed before returning for the second
push at Node 1. The two pushes at Node 4 went
smoothly. After the first push at Node 5, the fiber optic
probe suffered a cable break and stopped further data
gathering activities for the day.
WTM were obtained in an attempt to identify the
types or classes of. contaminants present at this site.
Based on WTM coal tar was the dominant contaminant
at the site (Figure 4-5). However, there is a thin seam
of contamination, apparently irregularly spaced around
the site, restricted to the upper foot or two of the ground
surface. This zone's WTM resembled some type of
waste oil.
Nodel
Minimal contamination was observed hi the two
pushes at Node 1. Each push revealed a slight increase
in relative fluorescence intensity between 0.0 and 1 foot
bgs. This zone exhibited a stronger relative intensity for
the fust push. The first push at Node 1 exhibited a
nearly constant background of approximately 6 percent
of the standard, dropping to baseline at about 12.5 feet.
Two narrow low intensity fluorescent areas, equivalent
to 5 to 7 percent of the standard, were observed at
15.0 and 16.8 feet bgs in both pushes.
Nodes 2 and 3
The two pushes at Node 2 revealed contamination
extending from 10 to 19 feet bgs. There was about a
31
-------
FIGURE 4-3. TYPICAL WTM - ATLANTIC SITE
too
1 50
100
100
150
300
350 400 450
Wavelength (nm)
500
300
350 400 450
Wavelength (nm)
500
300
350 400 450
Wavelength (nm)
500
Notes'. A. Node 2, 21'(approx.), Identified as Coal Tar
B. Node 4, 17.19', Identified as Coal Tar and Gasoline
C. Recovered Product (Gasoline)
FIGURE 44. ROST™ CHEMICAL CROSS SECTION - YORK SITE
NORTH
o-
-1-
-2-
-3-
-4-
-5-
-8-
-7-
-8-
-9-
-10-
-11-
-12-
-13-
-14-
-15-
-18-
-17-
-18-
-19-
-20-
-21-
-22-
-23-
-24-
-2S-
-28-
NODE1
NODE2
NODE3
NODE4
NODES
SOUTH
-0
"I
--a
--3
..9
•-10
•-11
-12
•-13
—14
-15
—18
—17
—18-
—19
—20
—21
--22
—23
--24
--IS
--28
DISTANCE (FEET)
100
LEGEND
30 - 10 (X OF STANDARD)
K.---J10 - 100
[1100 - 1.000
32
-------
FIGURE 4-5. TYPICAL WTM - YORK SITE
100
o
<»
M
O
E 50
300 350 400 450
Wavelength (nm)
Note: Node 3, 17.52'
Identified as Coai Tar
500
two-fold variation in the fluorescent intensity between
the two pushes. The highest fluorescent intensity for this
interval was 84 percent of the standard during the second
push.
The contamination at Node 3 resembled that at Node
2. The contamination occurred hi a slightly deeper
range, 12 to 20 feet bgs. The second push showed
signals only about half as great as those for the first
push.
Nodes 4 and 5
The relative intensities for the two pushes at Node
4 were hi good agreement, within 25 percent, but there
was more spatial variability as exhibited by FVDs. The
level of contamination was lower than at Nodes 2 and
3, being about 20 percent of the standard integrated over
the 10 to 20 foot bgs interval. The second push detected
a deeper zone of contamination extending from 19 to
22 feet bgs.
The first push at Node 5 was consistent with the
previous results. An increase in fluorescence was
detected between 11 and 16 feet bgs. However, nearly
as high contamination occurred in the 0.0 to 5 foot bgs
zone, and the signal remained elevated above
background between 5 and 11 feet bgs. Peak intensities
in both the 0- to 5-foot bgs and the 11- to 16-foot bgs
intervals exceeded 60 percent of the standard. The fiber
optical cable broke after the completion of the first push.
After an overnight repair, the investigation resumed in
the morning. Two more pushes were made at Node
5. These showed a very flat, broad zone of
contamination, after an initial contamination spike in the
first few feet. This was very different than the first push
at this location the previous day.
Fort Riley Site
Nodel
The two pushes at Node 1 showed some spatial
variability. The principal zones of contamination were
hi the intervals from 2 to 4 feet bgs, 15 to 22 feet bgs,
and below 22 feet bgs. This is shown hi the FVD
presented on Figure 4-7. This type of intra (FVD)
output is typical for the ROST™. The LIF CP
measurements were carried out sequentially from Node
1 to Node 5. This data was used to produce the
chemical cross sections presented on Figure 4-6. The
monitoring wavelength chosen for this site was changed
to 360 nm based on historical data regarding potential
contaminants present on site. This information indicated
that diesel fuel would be the primary contaminant found
at this site.
33
-------
FIGURE 4-6. ROST™ CHEMICAL CROSS SECTION - FORT RILEY SITE
SOUTH
0-
-1-
-2-
-3-
-8-
-10-
-1V
-12-
-13-
-14-
-15-
-1»-
-17-
-20-
-2V
-22-
-23-
-24-
-25-
-26-
-27-
-28-
-29-
-31-
NODE1
NODE2
NODES
NODE3
NODE4
DISTANCE (FEET)
150
160
LEGEND
- 10 (X OF STANDARD)
1------I 10-100
180 190 200 210
Hoc - 1.000
NORTH
-0
--1
--2
--3
--4
--5
--8
--7
--8
--9
--10
--11
--12
--13
--14
--KS
--10
--17
--IB
--19
--29
--21
--22
--23
--24
--25
--28
--27
--28
--29
--3D
--31
A fluorescence peak was detected at 2 to 4 feet bgs.
The overall fluorescence intensity, at this depth, for the
second push was more than five tunes greater than for
the first push, indicating spatial variability hi
contamination distribution.
A second fluorescence peak was detected at 15 to
22 feet bgs. The distribution of contamination hi this
zone agreed very well for the two pushes. There
appears to be a segmentation of this zone into an upper
part and a lower part. The upper part agreed very well
for the two pushes. However, the magnitude of the
contamination was approximately 10 tunes higher for the
second push in the lower half of the zone. The signal
returned to baseline at 22 feet. This depth corresponds
to a transition zone from clay to sand occurring between
22 and 25 feet bgs, as recorded by the cone penetro-
meter.
A third fluorescence peak was detected at depths
below 22 feet bgs. The lowest portion of this zone was
highly variable between the two pushes. The intensity
was much higher for the first push than for the second.
The contamination detected in the 22- to 23-foot interval
was 2 to 3 tunes more intense for push 1. Elevated
fluorescence intensity was only detected in the 23- to
24-foot interval hi the first push.
Significant contamination was found at Node 1. If
the contaminant is diesel, then concentration!! hi excess
of 10,000 mg/kg are present. However, this appeared to
be a highly variable region. It is possible that the close
proximity of the two pushes modified the contaminant
distribution relative to the sampling volume of the LIF
sensor.
Two WTM were recorded at Node 1 from 15 to
22 feet bgs. The wavelength tune matrices recorded hi
Node 1 hi the 2- to 4-foot bgs interval showed good
consistency and strongly resemble diesel fuel
(Figure 4-8). Spectrally, their distribution shifted to
somewhat shorter wavelengths than the WTM recorded
hi the 2- to 4-foot bgs interval. They still retained most
of the characteristics of diesel fuel; however, the slight
differences may indicate that the diesel fuel has been
slightly modified by transport. The shift to shorter
wavelength would be consistent with relative loss of the
larger, less mobile PAHs, whose emissions tended to
occur at longer wavelengths. One WTM was obtained
at 22.47 feet bgs hi the first push.
This WTM was completely different from all the
rest hi this node. The 340 nm and the entire intensity
distribution shifted to shorter wavelength. The WTM
strongly resembled that of JP-4 fuel. A third push was
made at Node 1 with emission monitoring shifted to
34
-------
FIGURE 4-7. FVD - FORT RILEY SITE
400
350
fsoo
2250
(A
0200
*W
1 100
50
0
0
K
,A|f
u
L
.
-
I 1
1
i
'
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40
Depth (ft)
FIGURE 4-8. TYPICAL WTWI-FORT RILEY SITE
100
£ 50
100
M 50
B.
300
350 400 450
Wavelongth (nm)
500
300
350 400 450
Wavelength (nm)
500
100
£ 50
100
50
300
350 400 450
Wavelength (nm)
500
300
350 400 450
Wavalangth (nm)
500
Notes:
A. Node 2, 21.95', Identified as Diesel Fuel
B. Node 3, 17.95', Identified as JP-4
C. 10,000 pptn Diesel Fuel Standard
D. 10,000 ppm JP-4 Standard
35
-------
340 nm. This change in monitoring wavelength was
made to optimize the system for the apparent JP-4
contamination. However, no changes in measured
fluorescence were observed at this altered monitoring
wavelength.
Node 2
The two pushes at Node 2 produced similar LIF
data, indicating relatively uniform contaminant
distribution. The maximum depth contamination is
comparable to that at Node 1. Six WTM were measured
at Node 2. There was little variation between them;
each one resembled diesel fuel.
NodeS
The pushes in this node were highly variable,
similar to Nodes 1 and 2. The pattern of fluorescence
for both pushes between 13 and 19 feet bgs showed good
similarity, but the amplitudes varied. The overall signal
was more than three times higher for the second push at
this node. However, in the second push there was a
shallow zone of contamination that reached above
40 percent of standard. This zone was between 6 and
12 feet bgs, and it was not present hi the FVD from the
first push.
Node 4
This node appeared to be located in a background
area. No fluorescence signal in excess of 2 percent of
standard was observed in either of the two pushes.
No WTM were acquired at Node 4 since it appeared
to be a background location. The Fort Riley site was a
difficult site to interpret. There was a great deal of
variability in pushes, even within the same node. It may
be possible that this was caused by the close proximity
of the paired pushes, however, this was not observed at
any of the other demonstration sites. Some of the paired
pushes had spatial separations of approximately
18 inches. This is well under the ASTM
recommendation that pushes be separated by 20 cone
diameters (32 niches). It is also possible that the
variability was representative of true spatial variability
of the contaminant distribution.
NodeS
Two pushes were made at Node 5. Node 5 lies
between Node 2 and Node 3. The reproducibility
between these two pushes was extremely good, the best
at any of the nodes at the Fort Riley site. The
contamination extended in a broad zone from
approximately 6.5 to 21 feet bgs. Some minor
contamination was also detected at 27 feet bgs.
Six WTM were obtained at Node 5. All of them
agreed extremely well. Their patterns were indicative of
diesel fuel with little or no alteration due to transport or
weathering.
Cone Penetrometer Data
The CP stratigraphic logs were used to construct
stratigraphic cross sections for each of the demonstration
sites. The CP produced individual stratigraphic logs for
each push. Fugro, PRC's subcontractor for cone
penetrometry services, does not produce cross sections
as a standard service. PRC transferred the individual
push stratigraphic data and plotted it as a cross section
on a scale that matched the one used for the reference
methods. The Fugro data package did not include a
narrative of the site-specific geology, therefore, a PRC
geologist provided the descriptions presented below by
site.
Atlantic Site
The transformed stratigraphic cross section for the
Atlantic site is presented on Figure 4-9. The CP
technology logged the upper portion of the cross section
as primarily clay with several silty clay lenses. The
bottom of the clay varied from 18 feet bgs hi the
northwest (Node 1) to 28 feet bgs hi the southeast (Node
5). In the south half of the cross section, the CP
technology identified a 3-foot thick silty clay layer at
about 13 to 16 feet bgs. The northern half contained a
3-foot thick silty sand layer with silty clay lenses
identified in Node 1. Below 28 feet bgs in the southeast
(Node 5) and 18 feet bgs hi the northwest (Node 1), the
cross section is predominantly sand and silty sand lenses.
York Site
The transformed stratigraphic cross section for the
York site is presented on Figure 4-10. The CP
technology logged the upper portion of the cross section
as clay with several silty sand and silty clay lenses hi the
upper 12 to 14 feet bgs. Below 12 to 20 feet bgs on the
south side (Node 5) and 14 and 20 feet bgs on the north
side the section became predominantly silty sand with
some silty clay lenses included. From 20 to 26 feet bgs
the cross section was identified as sand.
Fort Riley Site
The transformed stratigraphic cross section for the
Ft. Riley site is presented on Figure 4-11. The CP
36
-------
FIGURE 4-9. CONE PENETROMETER STRATIGRAPHIC CROSS SECTION - ATLANTIC SITE
SOUTHEAST
0-
-1-
-3-
-4-
-5-
-6-
-7-
-8-
-9-
-10-
-11-
-12-
-13-
-14-
-15-
-16-
-17-
-18-
-19-
-20-
-21-
-22-
-23-
-24-
-25-
-26-
-27-
-28-
-29-
-30-
-31-
-32-
-33-
-34-
-35-
-38-
-37-
NODE5
NODE4
NODE3
NORTHWEST
DISTANCE (FEET)
150
250
275
-0
— 1
— 2
--3
— 4
— 5
--6
--7
—8
— «
— 10
--11
--12
— 13
--14
— 15
—18
—17
--18
--19
--20
—21
--24
—25
—26
— 27
— 28
--29
--30
— 31
—32
--33
--34
300
LEGEND'
FIGURE 4-10. CONE PENETROMETER STRATIGRAPHIC CROSS SECTION - YORK SITE
NORTH
0-
-1-
-2-
-3-
-4-
-5-
-6-
-7-
-B-
-10-
-11-
-12-
-13-
-14-
-15-
-16-
-17-
-18-
-19-
-20-
-21-
-22-
-23-
-24-
-25-
-26-
NODE1
NODE2
NODE3
NODE4
NODES
DISTANCE (FEET)
SOUTH
-0
—1
—2
—3
—4
--5
—6
—7
—B
—9
—10
--11
—12
--13
—14
—15
--16
—17
—18
—19
--20
—21
--22
—23
—24
--25
—26
20 30
LEGEND
3SILTY CLAY
3SILTY SAND
37
-------
FIGURE 4-11. CONE PENETROMETER STRATIGRAPHIC CROSS SECTION - FORT RILEY SITE
SOUTH
0-
-1-
-2-
-3-
-4-
-5-
-8-
-7-
-8-
-8-
-10-
-11-
-12-
-13-
-14-
-15-
-15-
-17-
-18-
-1B-
-20-
-21-
-22-
-23-
-24-
-28-
-28-
-27-
-28-
-29-
-30-
-31-
NOOE1
NODE2
NODES
NODE3
NODE4
DISTANCE (FEET)
LEGEND
NO 120 130 140 150 160 170 180 190 200 210
JSILTY SANO
NORTH
0
1
-2
3
4
5
6
7
-8
9
10
11
n
13
-H
-IS
18
17
111
-10
20
21
22
23
24
23
-2I5
-27
213
29
30
31
technology logged most of the site as silty sand or sand
with some silty clay and clay lenses. The upper 12 feet
of the southern three quarters of the cross section was
logged as clay and silty clay with a 2-foot-thick bed of
silty sand running across it at 7 to 10 feet bgs. From
2 feet bgs in the north (Node 4) to 17 feet bgs in the
south (Node 1), the Fugro primarily logged silty sand or
sand. The sand became cleaner at depth.
38
-------
Section 5
Data Comparison
The data produced by the ROST™ technology and
the CP was evaluated using the criteria described hi
Section 1. The qualitative and quantitative data
evaluations are discussed separately. The qualitative
evaluation examined the chemical cross section produced
from the ROST™ data, and stratigraphic cross sections
produced from the CP data. The quantitative evaluation
statistically compares the ROST™ technology's data with
analytical data produced by the reference methods.
Qualitative Assessment
The qualitative assessment presents the evaluation of
both the stratigraphic and chemical mapping potential of
the ROST™ and CP relative to the reference methods.
In addition, the potential affects of TOC on the ROST™
technology's measurements are examined. Both the
reference and technology cross sections were produced
from collocated sampling areas as discussed in Section
1. Since these methods were sampling spatially different
locations, matrix heterogeneity will impact the
comparisons of both the physical and chemical cross
sections. Based on a review of the demonstration data,
this impact appears to have had minimal impact on the
qualitative data evaluation. The qualitative nature of the
comparisons, and the level of data quality may have
masked much of the effect of matrix heterogeneity for
this evaluation.
Stratigraphic Cross Sections
The following sections present descriptions of the
similarities and differences observed between the
stratigraphic cross sections produced from the CP data
and the reference method. For this comparison, PRC
used the CP sensor cross sections shown in Section 3.
These cross sections were produced directly from the
individual stratigraphic logs produced at each node.
These cross sections have been produced at the same
scale as the reference stratigraphic cross sections shown
hi Section 2. These comparisons are qualitative and, as
such, are subjective hi nature. However, these
comparisons were made by a certified professional
geolo^st (American Institute of Professional Geologists)
with over 17 years of experience hi this field.
Atlantic Site
The cone penetrometer truck subsurface geological
cross section compared favorably to the reference
methods' cross section as far as general mapping of
stratigraphy. The CP logged the upper portion of the
cross section primarily as clay with several silty clay
lenses. The corresponding portion of the reference
stratigiraphic cross section identified this portion as silty
clay and clayey silt with several silt lenses. The depth
of the clay and silty clay hi the CP cross section varied
from 19 feet bgs in the northwest (Node 1) to 28 feet bgs
in the southeast (Node 5), and the reference cross section
identified these same soil textures at 21 hi the northwest
(Node 1) and 28 feet bgs in the southeast (Node 5).
Below 28 feet bgs hi the southeast (Node 5) and 19 feet
bgs hi the northwest (Node 1), the CP cross section is
predominantly sand with silty sand lenses which closely
matches the reference cross section.
At. the Atlantic site, the technology showed good
correlation with the respective reference geological cross
sections. However, one difference was noted. The
reference methods had trouble collecting samples for
logging purposes in the running sands mat generally
occurred from 20 feet bgs to the termination of the
borehole. This lack of complete sample recovery is
common for this method of borehole logging. The
on-site geologist used indirect logging methods, such as
logging drill cuttings or monitoring drilling rates and
down pressure of the drill rig during drilling, to fill hi
the resultant gaps hi the borehole logs at depth. The CP
technology does not need to physically collect soil
samples to produce borehole logs, and thus, is not as
affected by running sands. This explains the greater
detail shown hi the CP cross section below
approximately 20 feet bgs.
39
-------
Seven samples were collected at the Atlantic site for
geotechnical analysis. Only six of these samples
corresponded to measured intervals by the CP. The
results of these analyses were compared to the
corresponding CP stratigraphic data. Three out of the
six samples showed intermethod agreement. The three
remaining samples were not matched due to the CP,
once overestimating the fraction of fines (DR19), and
twice underestimating the percentage of sand (DR17 and
DR18). This resulted hi the CP technology identifying
intervals as silty sand when it was identified by the
reference methods as a sand once, and identifying a
sandy lean clay as a clay or silty clay twice. This
indicates that the CP technology cannot resolve small
shifts in particle size distribution.
York Site
The CP stratigraphic cross sections showed good
correlation with the reference cross sections at the York
site. The CP did not identify the surface fill which was
identified by the reference methods as extending from
the ground surface to 2 feet bgs. Below this zone, the
CP logged primarily clay with several silty sand and silty
clay lenses extending from 12 to 14 feet bgs. The
reference methods logged this section similarly except
that the clayey silt extended from 15 to 19 feet bgs. The
CP logged the remainder of the section below 12 feet
bgs, on the south side (Node 5), and 14 feet bgs, on the
north side (Node 1), as predominantly silty sand and
sand with some silty clay lenses included. The reference
methods identified this same portion of the section as
well graded sand with silt and silty clay and sandy silt
lenses. The lack of correlation relative to the thin sand,
silt, and clay lenses may be more representative of the
reference methods' inability to resolve thin strata, in a
standard field logging mode. The detail of the reference
method can be increased by spending more time
examining sample cores, however, tune and cost factors
often prohibit fine detailed examination of sample cores.
The CP produces the same level of detail whenever it is
used. Running sands were not a problem at this site.
Six samples were collected at the York site for
geotechnical analysis. The results of these analyses were
compared to the corresponding CP stratigraphic data.
Four out of the six samples showed intermethod
agreement. The remaining two samples were not
matched due to the CP's lack of resolution relative to
detecting small increases in the distribution of coarse
particles (Node 1—18.5 to 19 feet bgs and Node 3—16.5
to 17 feet bgs). This indicates that the CP has trouble
resolving shifts in coarse particle size distribution in a
matrix dominated by silts and clays, relative to the
reference methods.
Fort Riley Site
The CP's cross sections showed good correlation
with the reference methods' cross sections at the Fort
Riley site. The CP technology logged most of the
section as silty sand or sand with some silty clay and
clay lenses as did the reference methods. The upper
12 feet of the southern portion of the CP cross section is
logged as clay and silty clay with a 2-foot-thick bed of
silty sand extending from 7 to 10 feet bgs. The
reference methods did not identify the sand layer, rather
it logged it as silty clay, silt and fill. From 2 feet bgs hi
the north (Node 4) to 17 feet bgs hi the south (Node 1),
the CP primarily logged silty sand or sand. The
reference methods logged this as poorly graded sand
starting at 5 feet bgs in the north (Node 4) and extending
to 19 feet bgs in the south (Node 1). The sand logged by
the CP became well graded at depth; this agrees with the
reference methods which identified all the sand below
17 feet bgs as well graded sand. The CP exhibited a
finer resolution of stratigraphy hi the saturated sands
identified at depth. This is similar to the findings at the
Atlantic site.
Eight samples were collected at the Fort Riley site
for geotechnical analysis. The results of these analyses
were compared to the corresponding CP stratigraphic
data. Six of the samples showed acceptable intermethod
matches. The two samples that did not match may have
ben caused by the CP's inability to resolve small change
in particle size distributions. The CP identified one
sample as a silty clay when the reference method
laboratory identified the sample as a silty sand or sandy
clay and the second sample was identified as a sand by
the CP and as a silty or clay sand with gravel by the
reference method (DR02 and DR05, respectively).
Summary
The CP and the reference methods produced similar
stratigraphic cross sections relative to the reference
method. Generally, the CP and the reference methods
showed good agreement in identifying the dominant
particle size hi soil. However, the CP did show
deviation from the reference methods when small shifts
in particle size distribution occurred. However, the CP
provided a finer resolution of strata, identifying more
thin stratigraphic units than the reference method. This
difference was magnified when running sands were
encountered at the Atlantic and Fort Riley sites. This
may be due to its ability to continuously acquire soil
textural data during a push, and the common limitations
of a geologist's logs where strata are less than several
inches thick. An additional difficulty with the reference
methods was their inability to retrieve samples from
running sands. This caused significant data gaps at
40
-------
depth. The CP does not require active soil sampling to
log a hole, and therefore, it is not as affected by running
sands, and may produce more representative subsurface
stratigraphic logs than the reference methods in running
sands.
Chemical Cross Sections
The following sections present descriptions of the
similarities and differences observed between the
chemical cross sections produced from the ROST™ data
and the analytical results from the reference methods.
Unless otherwise specified, the comparisons are made in
consideration of both reference cross sections, TPH and
total PAH. PRC used the ROST™ cross sections shown
hi Section 3 and the reference cross sections shown hi
Section 2 to conduct this evaluation. The ROST™ LIF
cross sections were made directly from the ROST™ LIF
data, and plotted to the same scale as the reference
method cross sections. These comparisons are
qualitative and, as such, are subjective in nature. The
effects of heterogeneity may influence this data
comparison, however, the qualitative nature of this
comparison should greatly reduce the potential impact of
heterogeneity hi contaminant distribution. These
comparisons were made by a soil scientist with over
9 years of experience hi site characterization activities.
Atlantic Site
Both the ROST™ and the reference methods showed
good correlation for background characterization. This
is exhibited by both the ROST™ and the reference
method's data showing Node 1 to be outside the area of
contamination. Both the ROST™ and the reference
methods identified shallow contamination intermittently
spaced across the cross section within 5 feet of the
ground surface. Both the ROST™ and the reference
methods detected the zone of contamination at Node
2, which extends from approximately 20 to 28 feet bgs
for TPH and from 16 to 31 feet bgs for total PAH. The
ROST™ identified this zone being from 2 to 9 feet
thinner than the reference method for TPH and total
PAH, respectively. This difference can be explained as
an artifact of data interpolation which was used for the
reference method to create the reference cross sections.
This is common when relatively few samples are used to
define zones of contamination. Both the ROST™ and the
reference methods identified a zone of elevated
contamination 1.5 feet bgs at Node 2.
Overall, the remaining zones of elevated ROST™
data corresponded with general zones of contamination
shown hi both reference cross sections. The differences,
such as upper boundaries of contamination and
delineation of distinct zones of contamination hi the cross
sections, can be attributed to a data interpolation artifact
of the reference methods.
The major differences between the ROST™ and the
reference two methods were exhibited at Node 5. At
Node 5, the ROST™ technology identified three distinct
plumes occurring from approximately 2 to 4 feet bgs,
14 to 17 feet bgs, and 24 to 30 feet bgs. The reference
method produced a single TPH plume at Node
5 extending from the ground surface to approximately
27 feet bgs. Historical data for this site indicated that
the zone of extended contamination shown hi the
reference TPH cross section reflected gasoline
contamination exclusively. The total PAH cross section
produced by the reference method identified two zones
of contamination: one extending from approximately
0.0 to 1 foot bgs, and a second zone extending from 6 to
17 feet bgs. These two zones overlap the upper two
zones identified on the ROST™ cross section, however,
their thicknesses and depth intervals vary. This variance
may be due to the difference hi data collection
techniques discussed above. The lack of correlation
between the deepest contamination detected by ROST™
at Node 5 and the reference method can be attributed to
the lack of sampling hi this interval by the reference
method. The lack of sampling was due to the failure to
collect samples hi the running sands below
approximately 19 feet bgs.
Another way to examine the relationship between
the ROST™ LIF data and the qualitative reference data
is to superimpose the two data types on a single plot of
fluorescence intensity and reference method
concentration against depth. To make the plot scales
meaningful, the ROST™ LIF and reference data had to
be normalized. The reference data was normalized to
the highest TPH and total PAH concentrations measured
during the qualitative sampling. The LIF data was
normalized to the average high reading measured over a
qualitative method reference sampling point at this site.
This normalization allows a general comparative
evaluation of the data.
Figure 5-1 shows the normalized data plots for
nodes where qualitative reference data was generated.
In all cases where the LIF data recorded increased
fluorescence relative to background, the reference data
showed TPH and PAH contamination. A detailed
review of this data shows that the qualitative reference
data and the ROST™ LIF data generally agree hi then-
identification of zones of high, medium, and low
contamination. Some exceptions to this can be seen hi
Node 3 (21 feet) and Node 4 (15.5 feet). In these cases,
the LIF data identification of high contamination did not
match the reference data. It is possible that these
41
-------
FIGURE 5-1. NORMALIZED LIF AND QUALITATIVE REFERENCE DATA - ATLANTIC SITE
c
0.0
13
3.0
43
9.0
73
00
105
— 120
t~"*15.0
10.5
100
103
21.0
223
24.0
253
27.0
295
30.0
JNODE - 2
INODE - 3|
NODE - 4|
|NODE - 5|
Pmcwilig.OfHIghRMding P.nat»ge Of High Reading Percentage Of High i Reading Percentaoe Of Hfch Reading
K 41) fa 0 SO 1DO 150 200 250 ° =" 1°° 15° 0 20 40 60 60 100 120
^~
0.0
1.5
3.0
43
9.0
7.5
D.O
10.5
—12.0
£.133
t15.0
183
18.0
1B.5
21.0
22.5
24.0
253
27.0
283
30.0
J
"p-
^
0.0
1.5
3.0
4.5
9.0
7.5
103
— 12.0
£133
§15.0
16.5
18.0
19.5
21.0
22.5
24.0
25.5
27.0
28.5
30.0
1^~
! Total PAH DTPH
. LIF Data |
0.0
1.5
3.0
4.5
6.0
7.5
«.o
103
-12.0
£133
§15.0
19.5
18.0
19.5
21.0
223
24.0
25.5
27.0
28.5
30.0
m
U-^"
^-
differences are due to matrix heterogeneity and/or slight
errors in relative sampling depth recording.
York Site
The ROST* cross section showed little correlation
to the reference method's cross sections at Node 1. The
ROST" technology only detected contamination from
0 to 1 foot bgs at this node, however, the reference
method detected both TPH and total PAH contamination
from 12 to 18 feet bgs at this node.
Nodes 2, 3, 4, and 5 showed better correlation
between the reference methods' cross sections and the
ROST™ cross section.
The differences between the reference cross section
and the ROST™ cross section could be the combination
of an artifact of data interpolation for the reference cross
section, and the finer definition provided by ROST™
which produces continuous profiles with a 2 cm
resolution. The variation between the ROST* and
reference methods cross sections could also be attributed
to the small sample volume used by the ROST™
technology relative to the reference method.
Figure 5-2 shows the normalized line graphs of the
five ROST" LIF pushes at the York site.
In all but one case, the LIF data exhibited elevated
fluorescence relative to background, at the point where
qualitative reference data reported TPH or total PAH
contamination above background. At Node 4 (18.5 feet
bgs), the reference data reported TPH and total PAH
contamination at 100 percent and 80 percent of the
highest reading, while the LIF data detected no
fluorescence above background. The reference TPH
concentration was 13,000 ppm, and the total PAH
concentration was 1,130 ppm. The reference sampling
point and the LIF sampling point was separated
horizontally by approximately 2 feet. An examination of
the corresponding CP log shows that at this LIF
sampling interval the LIF window was measuring
fluorescence inside a clay seam. It is possible that this
clay lens has resisted contaminant infiltration, supporting
the LIF data showing no contamination. This illustrates
the value of the combined LIF and CP data. The high
contamination detected by the reference sampling may
reflect its larger sample volume and sample
homogenization, or heterogeneity hi the geometry of the
clay lens.
A detailed review of the remaining data shows that
the relative magnitudes between the two types of data
were in agreement. Zones of high reference readings
corresponded to zones of high LIF readings. This
relationship appears to hold for medium and low zones
of contamination.
42
-------
FIGURE 5-2. NORMALIZED LJF AND QUALITATIVE REFERENCE DATA - YORK SITE
PtrartW Of High Racing PmxiUgXXHI^iRMding P«c«ig«0(HiohRMding »„_„ «,.«_*,, _^_ "
- « « • - « . » M . rT^r^ , „ '"^"HWR-*,
0.0
IS
3.0
4.5
0.0
7.5
10.S
120
f~15.0
1M
11.0
24.0
2S.S
27.0
2U
30.0
t
0.0
u
3.0
4.5
0.0
7.5
TO*
12.0
Il50
*s
21.0
24.0
35.8
27.0
20.S
30.0
"^
^1
0.0
1.S
3.0
4.8
0.0
7.5
10.S
12.0
f15.0
11.5
110
21.0
22.3
24.0
29.B
27.0
29.8
30.0
'
r
^^^^
[.Total PAH aTPH -UFIMi |
0.0
3.0
7.5
9.0
12.0
S15.0
185
11.0
21.0
27.0
2*5
' ' ' ' '
.f
^-^.
HH^^^K^ •-:
0.0
1.S
3.0
4.S
0.0
7.S
_10.S
—12.0
°15.0
22J
24.0
23.S
g^
..
£
>
?_
*"- '
~~~~_^_ _. —
>
T'i:
Fort Riley Site
The ROST™ cross section showed good correlation
to the reference methods' cross sections at all nodes with
the exception of the upper limit of contamination at Node
1 and the middle of Node 4.
At Node 1, the ROST™ technology found
contamination primarily from 15 to 25 feet bgs, while
the reference method's total PAH contamination was
detected from 1 to 24 feet bgs. The reference method
detected an isolated area of low TPH contamination at a
depth of approximately 15 bgs in Node 4. The ROST™
technology detected no zones of elevated contaminant
concentrations at Node 4. For Nodes 2, 3, and 5, the
relative depths, thicknesses, and intensities of the
contamination were similar between the reference
method and the ROST™ cross sections. The differences
noted at Node 1 and 4 involved low concentrations of
contaminants, and it is possible that the differences were
caused by matrix heterogeneity rather than due to
inaccuracies of the technology.
The greater definition of potential contaminant
lenses in the ROST* cross sections was most probably
due to the 2 cm sampling resolution provided by the
technology. The need to interpolate data for the
reference method reduced the potential for identifying
distinct smaller lenses of contamination.
Figure 5-3 shows these normalized line graphs of
the ROST™ LIF pushes at the Fort Riley site. Only
pushes where qualitative reference sampling was
conducted are shown on this figure.
In all cases where the qualitative reference sampling
detected contamination above background, the LIF data
also showed elevated fluorescence. A detailed review of
this data shows that for all pushes the general
contamination trends identified by the technology match
the trends detected by the qualitative reference data.
Similar zones of low, medium, and high contamination
were identified by the technology and the reference
methods.
Generally, the ROST™ technology showed a good
relative correlation with the reference methods' cross
sections. The ROST™ technology provided greater
resolution in identifying contaminated zones, relative to
the reference method due to the absence sample
collection the ROST™ data relative to the reference
methods. Sampling difficulties and cost restrictions
limited the number of samples collected and analyzed by
the reference methods. This forced the reference
method to use data interpolation to create the cross
section, In addition, the ROST™ data and quantitative
reference data were well correlated in then: identification
of zones of low, medium, and high contamination.
43
-------
FIGURE 5-3. NORMALIZED LIF AND QUALITATIVE REFERENCE DATA - FORT RILEY SITE
|NODE-1|
[NODE-2|
Percentage Of High Reading
50 100 150 200
Percentage Of High Reading
20 40 60 «0 100
[NODE - 3|
Percentage Of High Reading
10 20 30
00
U
12
4.7
82
77
«2
107
-112
£13.7
ftSJZ
19,7
155
22,7
242
25.7
37.2
tt.7
0.0
1.5
3.0
4.5
80
7.5
9.0
105
—^0
^.13.5
f1S.O
1B.S
1B.O
18.5
21.0
22.5
24.0
25.5
27.0
28.5
30.0
0.0
1.5
3.0
4.5
6.0
7.5
9.0
10.5
— 12.0
g.13.5
t15.0
18.5
18.0
18.5
21.0
22.5
24.0
25.5
27.0
as
30.0
[NODE-5|
Percentage Of High Reading
20 40 GO 80 100 120
.Total PAH oTPH
The observed differences between the ROST™ and
reference methods' cross sections could be caused by
several factors. The ROST™ sampling volume covers a
circle less than 0.5 cm in diameter, and approximately
only a few molecules thick (approximately 0.2 cubic
centimeter). This makes the technology hypersensitive
to the natural spatial variability of contaminant
distribution. The reference methods analyze 30 gram
aliquots from a homogenized sample of approximately
1,000 grams, which is several thousand times larger than
for the technology. This larger sample volume may
average out some of the smaller heterogeneities detected
by the ROST™. Some of this effect is canceled out by
the fact that the ROST™ collects much more data. In the
case of this demonstration, the reference method used a
total of 76 samples, compared to over 1,300 sample
points provided by the ROST™.
Total Organic Carbon
PRC compared the ROST™ intensity measurements
for areas free from contamination to the corresponding
TOG concentrations. This evaluation examined the
potential for gross mimics to affect ROST™ intensity
measurements. ROST™ data from the York, Atlantic,
and Fort Riley sites were reviewed. The samples
collected for this evaluation contained TOC
concentrations from not detected to over 3,000 ppm.
Based on the limited data base (11 samples), there
appears to be no effect of TOC on ROST™ intensity at
any of the three sites. This is based on tilie fact that
although the TOC concentrations varied over three
orders of magnitude, the ROST™ measurements
remained relatively constant. However, some literature
suggests that TOC becomes a potential interferant in the
presence of organic solvents or petroleum products.
This interference may be created by the contaminant's
activation of fluorescent properties in the TOC,
specifically humics. Isolation and examination of the
potential for this activation of fluorescence in humics
was beyond the scope of this demonstration.
Quantitative Assessment
This section presents the comparative evaluation of
the ROST™ data and the reference methods' analytical
data, and an evaluation of the precision of the
technology. The precision discussion is presented after
the regression analysis discussion.
The reference method sampling and analysis
identified considerable heterogeneity in the distribution
of contaminants in the soil matrix. The experimental
design of this demonstration expected heterogeneity and
intended to define it through co-located replicate
sampling. This sampling did define the heterogeneity.
However, in many cases the heterogeneity was greater
than expected. In almost 50 percent of the 21 quan-
44
-------
itative sample intervals the heterogeneity produced
ranges between maximum and minimum concentrations
in excess of one order of magnitude. This heterogeneity
coupled with the developer's inability to specifically
identify the compounds it is measuring, the lack of a
reference analytical method that monitors the exact suite
of the compounds measured by the technology, the
mixed distribution of constituents in the contamination,
and the varied age of the contaminants cause uncertainty
to be introduced into the point by point comparison of
data in the quantitative evaluation. Therefore, any
conclusions stated in this section should be considered as
trend indicators and not definitive statements on
technology performance. However, the conclusions are
likely to be duplicated if similar field in situ verification
is attempted.
The quantitative assessment evaluated ROST™ data
at distmct intervals relative to corresponding data from
the reference method. This evaluation was intended to
quantify relationships between the ROST™ data and
individual compounds or class-specific analytical data
produced by the reference methods. According to the
developer, due to spectral overlap of fiuorescing
compounds, it is virtually impossible to select excitation
and emission monitoring wavelenths to quantitate
individual compounds, however, quantitation for classes
of compounds is possible. Therefore, any correlations
noted for individual compounds are going to be site
specific and dependent on a consistent distribution of the
compound in the overall class of contaminant. The lack
of observed correlations for individual compounds does
not indicate a performance problem, rather, it is
probably due to spectral overlap interferences or the
random distribution of individual compounds over a
given site. This type of random distribution can be
caused by contaminant aging or changes in waste
generation processes.
The target analytes for the quantitative evaluation
were TPH, VPH, BTEX, total BTEX, naphthalene,
1-methylnaphthalene, 2-methylnaphthalene, acenaph-
thene, fluoranthene, phenanthrene, pyrene> benzo-a-
pyrene, total PAH, and total naphthalene. The TPH,
VPH, total naphthalene, total PAH, and total BTEX
groupings were made in an effort to most closely match
the ROST™ data. The developer felt that at the current
stage of this technology's development, classes of
compounds would show the closest match to the ROST™
data.
This data evaluation involved regression analysis of
the ROST™ data against the corresponding reference
data. As defined hi the final demonstration plan, a
coefficient of determination of 0.80 or better defines a
useable predictive model.
The ROST™ made two collocated pushes at each
node. The first push was intended to produce the
primary data for the qualitative and quantitative
evaluation. The second push was intended to examine
the ROST™'s precision. The second push also produced
continuous ROST™ data to depth. The primary data
evaluation focused on the data from the first push at each
node, however, PRC did examine the possible impact of
averaging the two pushes for the regression evaluation.
This averaging had very limited impact on the outcome
of the regression analysis and will only be discussed
where its findings differ from the single push data.
The data sets were initially examined as a whole and
then post-hoc techniques were used to eliminate data
outliers. The total data set for this evaluation consisted
of 21 ssimpling intervals, 8 at the Atlantic site, 7 at the
York site, and 6 at the Fort Riley site. Each one of
these intervals produced one data point for the regression
analysis, however, each of these data points represents
the mean concentration from five collocated samples.
Therefore, this evaluation was based on the analytical
results of 105 individual samples and analyses. The data
presented is based on nontransformed data. PRC
mirrored the analyses discussed below with
log-transformed data, however, in no case did the
correlations improve. This suggests that the high and
low concentration points did not disproportionally bias
the regression. PRC also examined the data in its raw
form prior to averaging the reference data. This
approach did not improve the correlation of the data.
The initial regression analysis examined the data set
of meam concentrations as a whole. In this analysis the
technology's data was considered to be a dependent
variable. From this evaluation, no coefficient of
determination of greater than 0.15 was observed (Table
5-1). An examination of the maximum and minimum
concentrations for each set of collocated samples
indicated that several locations at each site exhibited
considerable heterogeneity. This was expected and is
normal for environmental sampling. Using data points
from confirmatory sampling depths that exhibit wide
ranges in contaminant concentration introduces additional
uncertainty into the data evaluation. In these cases, it is
hard to define a representative mean concentration.
Concentrations are highly location dependant. In an
effort to reduce the impact of this heterogeneity on the
data evaluation, all data points exhibiting greater than
1 order of magnitude range between the maximum and
minimum were eliminated. This range was selected
after consultation with ROST™ operators. In most cases,
this resulted hi almost a 50 percent reduction hi useable
data. For this reason the subsequent data analyses
should be considered indicators of trends hi correlation,
and not well defined predictive models.
45
-------
TABLE 5-1. REGRESSION ANALYSIS RESULTS FOR INITIAL ROST™ PUSH AND THE
REFERENCE METHODS—ALL SITES
Compound
TPH
VPH
Benzene
Toluene
Ethylbenzene
Xylene
Naphthalene
1-Methylnaphthalene
2-Methylnaphthalene
Acenaphthene
Fluoranthene
Phenanthrene
Pyrene
Benzo-a-Pyrene
Total Naphthalene
Total PAH
Total BTEX
n
21
20
16
19
20
20
21
18
20
12
19
21
17
19
21
21
20
Initial
i2
0.04
0.09
0.06
0.10
0.06
0.13
0.01
0.00
0.00
0.05
0.00
0.01
0.00
0.02
0.00
0.00
0.12
Regression
slope
9.6
1.1
12.5
63.5
28.1
131.8
-0.05
0.01
-0.02
-0.08
0.00
-0.02
-0.01
-0.00
-0.12
-0.04
245
y-intercept
toom)
2,817
212
4,374
9,367
9,376
20,551
42.8
57.0
34.4
51.5
3.06
18.0
8.02
1.76
200
115
40,375
Final Post-Hoc Data
n
8
9
6
10
10
11
8
12
10
4
12
10
8
11
12
10
11
r2
0.06
0.45
0.68
0.55
0.63
0.38
0.66
0.09
0.21
0.12
0.02
0.15
0.69
0.00
0.32
0.37
0.33
slope
20.0
1.6
52.7
411
151
237
0.00
0.16
0.20
0.34
0.00
0.06
0.09
0.00
0.75
1.01
826
Reduction
y-intercept
(ppm)
2,311
-15.8
8.8
-3,930
-•3,212
-246
1.8
37.6
12.0
33.6
1.00
4.33
-0.37
0.80
28.0
24.2
14,108
After these data points were removed the regression
analysis was run again. No significant changes in the
regression parameters were observed (Table 5-1).
However, a post-hoc examination of the residuals
identified several outliers for each regression.
A final regression analysis was conducted on the
data sets after the outliers were removed. This
regression showed considerable improvements in the
data correlation (Table 5-1), however, none of the target
analytes exhibited an acceptable correlation.
Regressions based on the benzene, ethylbenzene,
naphthalene, and pyrene exhibited coefficients of
determination of between 0.6 and 0.7. With regard to
these compounds, and based on the slopes and
y-intercepts of the regression equations, the ROST™
appears to be most sensitive to PAH compounds relative
to the single ring aromatics. The slopes for the PAH
compounds were all less than 1.0, indicating that
changes in PAH concentration would cause relatively
larger changes in ROST™ readings. Conversely, the
slopes of the single ring aromatic compounds were
greater than one, indicating that a given change in
contaminant concentration would cause a relatively
smaller change in ROST™ measurements.
The lack of correlation for the quantitative
evaluation cannot be solely attributed to the technology.
Rather, it is likely due to the combined effect of matrix
heterogeneity, lack of technology calibration,
uncertainties regarding the exact contaminants being
measured, and the age and constituents in the waste.
Based on the data from this demonstration, it is not
possible to conclude that the technology can or cannot be
quantitative in its current configuration.
Similar conclusions are drawn if the data from the
two ROST™ pushes are averaged and used in the
regression analysis (Table 5-2). However, the
correlations are generally improved. Under this scenario
of data evaluation, the naphthalene regression produced
an acceptable coefficient of determination value,
46
-------
TABLE 5-2. REGRESSION ANALYSIS RESULTS FOR AVERAGED ROST™ PUSH AND THE
REFERENCE METHODS—ALL SITES
Compound
TPH
VPH
Benzene
Toluene
Ethylbenzene
Xylene
Naphthalene
1 -Methylnaphthalene
2-Methylnaphthalene
Acenaphthene
Fluoranthene
Phenanthrene
Pyrene
Benzo-a-Pyrene
Total Naphthalene
Total PAH
Total BTEX
n
21
20
16
19
20
20
21
18
20
12
19
21
17
19
21
21
20
Initial
r2
0.01
0.05
0.03
0.08
0.02
0.08
0.01
0.00
0.03
0.08
0.01
0.06
0.04
0.06
0.04
0.03
0.08
Regression
slope
3.2
0.68
7.5
45.9
13.3
83.1
-0.001
0.00
-0.04
-0.09
-0.00
-0.03
-0.01
-0.00
-0.31
-0.15
160
y-intercept
fopm)
3,315
236
4,658
10,224
10,475
23,518
41.5
57.5
36.9
52.5
3.53
19.8
8.92
1.92
221
127
45,381
n
8
8
5
10
10
10
8
12
9
4
11
9
8
11
11
9
9
Final Post-
r2
0.07
0.67
0.79
0.38
0.76
0.62
0.83
0.05
0.17
0.23
0.02
0.10
0.62
0.04
0.28
0.23
0.66
Hoc Data Reduction
slope
19.8
2.1
50.0
249
181
308
0.06
0.11
0.16
0.71
-0.004
0.07
0.13
-0.003
0.69
0.75
949
y-intercept
(ppm)
2,241
-28.0
-164
3,056
-2,119
2,821
5.14
42.5
17.2
18.0
1.71
9.43
-1.55
1.04
44.9
63.3
-9,542
0.83. The slope of this regression equation was
0.06 supporting the conclusion that the ROST™ is most
sensitive to the PAH compounds. The y-intercept for
this regression was 5.14 percent of the standard. This
suggests that ROST™ readings of greater than
5.14 percent of the standard are required before the
predictive model for naphthalene should be applied. The
coefficient of determination values for benzene and
ethylbenzene were between 0.75 and 0.80. These
regressions produced slopes much greater than
1.0, supporting the conclusion that the ROST™, in its
configuration during the demonstration was less
responsive to single ring aromatics than to PAHs.
Pyrene, xylene, total BTEX, and VPH produced
coefficients of determination between 0.6 and 0.7. The
slopes for the pyrene, xylene, and total BTEX were
consistent with the trends already discussed for PAHs
and single ring aromatics. The VPH regression
produced a slope of 2.1. The slope data cannot be used
to assess data quality since the LIF data was not in the
same units as the reference data. However, the slope
data can be used to indicate general trends in the relative
fluorescence, as discussed above.
The qualitative determination of a detection limit or
threshold for the ROST™ technology was not possible
given the data produced from this demonstration.
Review of the data presented in Table 5-2 shows that the
relationship between the ROST™ data and the compounds
exhibiting the best correlation based on the averaged
push data was not consistent, even when only the data
used in the final regression is considered.
Qualitative observations regarding the detection
limits of this technology can be made with the data
produced from this demonstration. Measurable relative
fluorescence was reported for benzene concentrations as
low as 264 mg/kg, ethylbenzene as low as 807 mg/kg,
naphthalene as low as 1.1 mg/kg, and pyrene as low as
0.03 mg/kg.
47
-------
ROST™ reported negative relative fluorescence for
two quantitative samples. For these samples, ROST™
reported a relative fluorescence of -1.37 and
-0.20 (Table 5-3). These readings are equivalent to zero
fluorescence. The reference methods indicated that
these intervals contained measurable concentrations of
these four compounds (Appendix A). The ROST™
reading of -1.37 occurred at a sample interval that
contained, at its minimum, total PAH, TPH, and VPH
concentrations of 799 ppm, 1,878 ppm, and 175 ppm,
respectively. The ROST™ reading of -0.20 occurred at
a sample interval that contained, at its minimum,
concentrations of total PAH, TPH, and VPH of 31 ppm,
1,416 ppm, and 1,102 ppm, respectively. This could be
interpreted as false correlation based on the averaged
push data was not consistent, even when only the data
used in the final regression is considered. Increases in
relative negative readings. Two false negatives out of
21 sample points equates to a false negative rate of
10 percent.
Another qualitative method for assigning a detection
threshold is to determine the relative fluorescence at the
x-intercept for the benzene, ethylbenzene, naphthalene,
(based on averaged pushes) and pyrene regression
models discussed above. The x-intercept represents the
0 rag/kg point on the fluorescence versus concentration
graph. For the benzene and ethylbenzene the relative
fluorescence intensity at the x-intercept was -0.17 and
21, respectively. For the naphthalene and pyrene the
relative fluorescence at the x-intercept was 85 and
4, respectively.
To examine the potential for site-induced affects on
the data evaluation, the data was divided by site and
regression analyses were run on the resultant three data
sets. This regression analysis began with data sets
whose gross outliers had been removed. These outliers
were defined as data points where the maximum and
minimum values varied by over one order of magnitude.
This site-specific regression showed the following:
xylene, naphthalene, pyrene, and total PAH regressions
exhibited acceptable correlations (r2 greater than
0.80) at the Atlantic site; no compounds showed
acceptable correlations at the York site; and acceptable
correlations were exhibited for VPH, naphthalene, and
total BTEX at the Fort Riley site. The number of
samples resulting in these acceptable correlations ranged
from 3 to 5, out of a maximum of 6 to 8. This small
sample set greatly limits the use of this data to form
predictive models. However, these regressions exhibited
the same trends hi their slopes, as exhibited in the data
set as a whole. The slopes for the PAHs were all less
than 1.0, and the single ring aromatic compounds
produced regression equations with slopes greater than
1.0.
The potential causes other than sample matrix
effects such as heterogeneity, of the relatively poor
correlations of ROST™ data to reference data seen
during this demonstration were examined. The FVD
wavelengths used for emission wavelength monitoring
may have affected the comparability of the ROST™ data
to the reference method data. ROST™ utilized an FVD
wavelength of 400 nm for the Atlantic and York sites,
and 360 nm for the Fort Riley site. ROST™ explained
that 400 nm would provide the best FVD-monitoring
wavelength for coal tar-contaminated sites, and that
360 nm would provide the best FVD-monitoring
wavelength for petroleum-contaminated sites. To assess
the use of these wavelengths, WTMs were compared to
the emission monitoring wavelengths used at each site.
It was observed that the WTMs from the Atlantic and
York sites produced maximum fluorescence intensities at
greater wavelengths than the ones used for the emission
wavelength monitoring. This means that the ROST™
was not monitoring the wavelength which exhibited the
greatest fluorescence intensity. For fluorescence
methods the wavelengths of the greatest fluorescence
intensity provides the most accurate and linear results.
By not monitoring at the wavelength of maximum
fluorescence intensity the ROST™ was monitoring on the
rising or falling limbs of the fluorescence wave forms of
the primary fluorescing contaminant. In this area,
similar to a side slope of a hill or mountain, large shifts
in the fluorescence intensity of the contaminant matrix
could result in relatively smaller changes hi measured
fluorescence.
According to the developers, saturation effects for
the detection system used hi the ROST™ do not appear
until soil contamination exceeds approximately
1,000 ppm. These effects could result hi inaccurate LIF
data. This type of error would only impact the most
contaminated areas seen during this demonstration, and
it would result hi underestimates of contamination. It is
likely that the areas of contamination potentially causing
this effect would be identified as grossly contaminated
even with underestimated LIF results. ROST™ was
monitoring the rising and falling arm of the fluorescence
wave form, and therefore, was not monitoring the
wavelength of maximum fluorescence intensity that
should produce the most accurate and linear results.
A related problem was noted concerning the
calibration procedures employed by ROST™. ROST™
performed calibration before each CP push by placing a
solution of 10,000 /zg/mL gasoline over the sapphire
48
-------
TABLE 5-3. DATA FOR MEAN
Site
Atlantic
York
Fort Riley
Node
2
2
3
4
4
4
5
5
1
2
2
3
4
4
5
1
1
2
2
5
5
Depth
fleet)
21-22
24-25
16-17
6.5 - 7.5
10-11
27.5 - 28.5
16-17
23.5 - 24.5
15-16
13.5-14.5
17-18
17-18
14-15
18-19
1.5-2.5
2-3
13-14
6-7
17-18
10.5-11.5
16-17
ROST™ —ALL SITES
ROST™ Relative
Fluorescence
Intensity
71.99
2.94
38.78
21.56
174.6
498.0
12.1
67.30
2.16
132.43
20.72
101.5
97.14
-1.37
23.59
16.94
-0.20
34.88
120.2
188.4
171.1
Benzene
(mg/kg)
2,200
8,128
1,100
573
1,334
6,300
3,420
3,190
ND
ND
ND
ND
1,148
1,150
ND
264
1,437
811
16,800
5,114
5,100
Ethylbenzene
(mg/kq)
34,000
14,960
3,900
1,410
31,600
10,520
1,198
1,174
807
4,760
4,860
1,558
9,900
21,811
ND
1,076
8,595
763
39,600
13,680
31,400
Naphthalene
(mg/kg)
151
65
2.0a-b
1.6a*b
19
33a
1.1
1.4a'b
23
56a
35a
17a,b
99a
249
No data
8.1a'b
5.0a-b
No data
1ta,b
21a,b
5.0
Pyrene
26
13
0.2
0.7
2.4
8.3
No data
0.03
12
9.0
5.7
8.1
12
31
0.2
0.3
ND
ND
ND
ND
No data
Notes:
mg/kg Milligrams per kilogram.
ND Not detected.
a Data points remaining after the initial removal of outliers based on maximum and minimum comparisons.
Data point used in the final regression analysis.
window and standardizing the ROST™ system. The
response of the standard is then measured at the
FVD-monitoring wavelength. The ROST™ system then
normalizes the response of the gasoline to equal
100 percent fluorescence. At the Atlantic and York
sites, 400 nm was used as the FVD-monitoring
wavelength. WTMs performed for the gasoline standard
showed that the wavelength of maximum fluorescence
intensity was 340 nm. Again, this means that ROST™
was not monitoring the wavelength of maximum
fluorescence intensity, where accuracy and linearity is
greatest.
Calibration procedures were evaluated to determine
a cause of the relatively poor correlation between
ROST™ and the reference methods. ROST™ performs
a calibration by analyzing a single level of gasoline.
This procedure does not provide any information
concerning the linearity of the ROST™ system.
Typically, fluorescence methods are extremely sensitive
49
-------
and linear over a wide range, a 1,000-fold or greater
linear range is not uncommon. However, fluorometry
is limited to low concentrations.
Inherent instrument precision for the ROST™
measurements at the York, Atlantic, and Fort Riley sites
was evaluated by calculating the percent relative
standard deviation (%RSD) of each set of
11 measurements for each depth sampled. Replicate
measurements were taken at three separate nodes at the
Atlantic and Fort Riley sites and at only one node at the
York site. The precision measurements taken at the
Atlantic and Fort Riley sites were at a total of five and
four different depths, respectively. The York site only
had one depth sampled. The %RSD was calculated by
dividing the standard deviation from the 11 replicate
measurements by the mean. This number is multiplied
by 100. The maximum inherent instrument precision at
the York, Atlantic, and Fort Riley sites was 4.1, 3.5,
and 1.7, respectively. Based on this data, the standard
deviations are most likely due to heterogeneity in
contaminant distribution.
50
-------
Section 6
Applications Assessment
The ROST™ technology is only available as a
service from the developers and it is designed to be
operated by trained technicians. The ROST™ does
require some type of platform to advance the sensor into
the ground. Currently, this is done with a cone
penetrometer truck either provided by the developers or
by the customer. The ROST™ is designed to be
compatible with most standard cone penetrometer trucks.
This technology has been tested at a variety of military
and industrial installations, and as of the date of this
report, it has been used for several commercial site
characterizations. As demonstrated, this technology can
rapidly acquire data defining zones of general
contamination, as long as the contamination has a
fluorescent signature. This data can greatly facilitate site
characterization activities.
During this demonstration, the cost of this
technology, including a CP sensor, was less than the
reference methods used to produce the qualitative data.
The LIF sensor and CP sensor produced far more data
in real-tune, resulting hi improved physical and chemical
resolution relative to the reference method. However,
this unproved resolution and real-time data was at a
lower level of data quality (screening versus definitive).
The qualitative assessment portion of this demon-
stration showed that this technology is comparable to
reference methods hi its ability to map subsurface
contaminant plumes at petroleum fuel and coal tar
contamination sites. This demonstration showed that
both ROST™ and the reference methods identified
similar zones of subsurface petroleum and coal tar
contamination at each of the three demonstration sites.
Many of the differences between the ROST™ and the
reference methods can be explained by their respective
methods of data collection. The ROST™ produces a
continuous profile, while the reference methods take a
few selective samples targeting boundaries and zones of
contamination. In addition, the reference methods had
difficulty retrieving samples hi running sands, adding
potential data gaps. The ROST™ technology produced
relatively continuous data on petroleum and coal tar
contaminant distribution over a 30-foot depth in
approximately 1.5 hours during the demonstration. The
reference methods would be able to collect samples over
this interval, however, definitive analytical services
would require, at best, several days. Even if the
reference methods used on-site analysis, and produced
only screening level data, it would take several hours to
provide data on the samples. Therefore, on-time critical
projects that can use screening level data, or on projects
where it is more critical to cover large areas hi greater
detail, the ROST™ technology seems appropriate.
Another powerful aspect of this technology is that it
can be advanced with a standard CP to provide
continuous descriptions of the subsurface soil
concurrently with the chemical data. This demonstration
found that the subsurface logging capabilities of the CP
were comparable to the reference methods. Similarities
between the two data sets were observed even without
site-specific borings to calibrate the CP logging tool
prior to deployment. The major limitation of the CP
logging was its apparent inability to detect slight shifts hi
particle size distribution. It is questionable if this
limitation would greatly impact the use of the data. For
example, a silt and clayey silt would probably be dealt
with similarly hi a hydrogeological perspective.
The quantitative data assessment for this technology
indicated that the ROST™ data was most correlated to
naphthalene concentrations. The small data set sizes
limit the use of the predictive model (created from the
regression analysis). Any predictive model should be
based on site-specific calibration and confirmation. The
regression analysis showed some correlation between the
technology's results and individual compounds. The
technology's data was not well correlated to TPH and
VPH measurements (.07 and .67, respectively). The
generally poor correlation with the quantitative data from
classes of compounds may be partially due to matrix
heterogeneity, analytical limitations of the reference
methods, ROST™ calibration procedures, and the
emission monitoring wavelengths used by the ROST™
during the demonstration. Basically, the calibration and
51
-------
emission wavelength selection were not optimum for the
site-specific classes of contaminants. This apparently
did not greatly affect the qualitative evaluation, only the
quantitative evaluation. This is consistent with the
current evolutionary stage of this technology at the time
of the demonstration in August 1994. The identification
of individual compounds is not a current application of
this technology. However, the collection of WTM data
allows the ROST™ to identify changes hi constituent
compostion, and in many cases, allows ROST™ to
identify types of contaminants such as diesel fuel,
gasoline, jet fuel, and coal tar. This capability can assist
in the identification and attribution of contaminant
sources.
Based on the results of this demonstration, the use
of site-specific calibration samples for the application of
ROST™ may be of most value in determining emission
monitoring wavelengths and possibly in calibration
activities. Future development of this technology may
allow this type of calibration for quantitation (see Section
7). The effects of site-specific calibration samples for
the ROST™ was not evaluated during this demonstration.
Site-specific calibration may allow a more representative
estimation of contaminant concentration distribution. In
other words, with this type of calibration, perhaps the
ROST™ data would exhibit more similar trends to the
reference method data, for example, as the contaminant
concentrations increase or decrease, the ROST™ data
would show a more consistent relationship relative to
increases or decreases in contaminant concentration.
Based on this demonstration, this technology appears
to produce screening level data. The lack of better
correlations may not be wholly attributable to the
technology performance, rather it may have been due to
the relatively small reference method data set size, the
lack of a reference method that measures the same suite
of compounds as the ROST™ sensor monitors, the
complex interactions between the fluorescing
compounds, and the soil matrix which resulted in the
observed heterogeneity. The first two factors can be
addressed with changes in experimental design and
innovations in analytical methods, however, the final
factors will require more research to isolate specific
matrix interactions and to resolve the heterogeneity
issue.
If this technology, in the configuration demon-
strated, is to be evaluated in the field, this demonstration
has shown that on a point-by-point quantitative basis, it
is unlikely that significant correlation to reference data
will be observed. This is due to a combination of
heterogeneity effects, limitations in conventional
sampling and analysis, and the complex interaction of
waste aging and constituent distribution of relative
fluorescence. It is possible that site-specific calibrations,
different reference analytical methods, alternative
sampling methods, or a less heterogeneous matrix would
improve the potential for significant correlation between
ROST™ and the reference method data. Therefore,
based on the results of this demonstration, field
evaluations of this technology should be restricted to
qualitative evaluations consisting of cross section
comparisons of simply verifying that LIF highs
correspond to higher levels of contamination. This latter
comparison will also be affected by effects listed above.
Although there are many advantages to this
technology, a potential user should be aware of potential
disadvantages. This technology has a sampling volume
several thousand times smaller than conventional
sampling analysis. This makes the technology very
sensitive to matrix heterogeneity. Some of this
sensitivity is reduced (vertically) by the averaging of
fifty data points every 2 cm. This affect can also be
minimized by the sampling of more push locations to
reduce the sensitivity hi a horizontal orientation. At a
data collection rate of approximately 300 linear feet per
day (4,572 data points), additional pushes can be
conducted without greatly increasing project duration.
The LIF results can be influenced by the age and
constituent distribution of wastes. This fact coupled with
heterogeneity effects, and a lack of site-specific
instrument calibration, makes field verification of LIF
results difficult. The use of the LIF and CP sensors is
restricted to the maximum push depth of the cone
penetrometer truck. This depth can be as much as
300 feet, or hi the case of the demonstration, 30 to
70 feet. These shallow depths were realized when
deeper strata exhibited increased cone tip resistance and
sleeve friction, and at location where strata at shallower
depths would not provide adequate lateral support for the
push rod. These conditions greatly increase the chance
for push rod breakage and sensor loss.
This technology can currently provide rapid
assessment of the distribution of fluorescent material in
the subsurface. When these materials are PAHs or
petroleum fuels, the technology can be used to map the
extent of subsurface contamination. The WTM data can
also be used to provide qualitative identification of waste
type or at a minimum, it can identify changes hi
constituent distributions. All of this data can be used to
guide critical conventional soil sampling, and the
placement of groundwater monitoring wells. All of this
data can be produced and interpreted hi the field. This
real-time sampling and analysis allows the use of
contingency based sampling, which assists hi
characterizing a site with a single mobilization. These
aspects coupled with its low waste production during
decontamination make this technology a powerful and
effective ske characterization tool.
52
-------
Section 7
Developer Comments and Technology Update
The developers of the ROST™ technology submitted
both editorial and technical comments on the draft ITER.
Only comments not resulting in changes to the report are
presented below. The technical comments are presented
verbatim in italics. The response to the comments is
presented below each developer comment in plain type.
Loral Comments (April 1995)
1. Complications caused by spatial inhomogeneity.
The statistics of the quantitative sampling, shown in
Table 7-1, emphasize how large are the variances in the
chemical concentrations obtained by the reference
methods. The RSD is greater than 50% for 17 of the
21 (81%) quantitative evaluation sampling intervals, and
greater than 100% for 12 of the 21 (57%) intervals. At
the 99-percent confidence level only three of the
21 locations for quantitative evaluation give TPH
concentrations different from zero.
We feel that the draft report minimizes the
difficulties posed by the heterogeneity problem and, more
significantly, fails properly to take it into account in the
quantitative evaluation of ROST™s performance. From
page 4-10 of the draft ITER; "An examination of the
maximum and minimum concentration for each set of
collocated samples indicated that several locations at
each site exhibited considerable heterogeneity. This was
expected and is normal for environmental sampling.
Using data points from confirmatory sampling depths
that exhibit wide ranges in contaminant concentration
introduces additional uncertainty into the data
evaluation. In these cases, it is hard to define a
representative mean concentration. Concentrations are
highly location dependent (sic). In an effort to reduce
the impact of the heterogeneity on the data evaluation,
all data points exhibiting greater than 1 order of
magnitude range between the maximum and minimum
were eliminated. This range was selected after
consultation with ROST™ operators. In most cases, this
resulted in almost a 50 percent reduction in usable data.
For this reason the subsequent data analyses should be
considered indicators of trends in correlation, and not
well defined predictive models. "
We are troubled by several things. In the first
place, the "ROST™ operations" were not consulted
about eliminating data points on the basis or range
between minimum and maximum. To state that we
agreed with this procedure is untrue. We don't, believe
that this method of rejecting data is statistically valid. It
is also unclear whether the procedure was to reject all
five values (for the set of five collocated samples) or just
those that led to more than an order of magnitude
variation. Note that the rejection criterion eliminates
Node 4, 6.5-7.5 feet at the Atlantic site with a RSD of
65 percent (one of the lower values) and retains Node
4, 14-15 feet at York with a RSD of 104 percent.
PRC consulted both ROST™ and SCAPS
representatives regarding the elimination of data points.
PRC conducted the initial data review for the
quantitative evaluation based on the entire data set. This
initial data review showed no correlation between the
technology's results and the reference method data. In
addition, PRC noted that several data points exhibited
considerable spatial variation. To reduce the impact of
the observed spatial variability, PRC conducted a post-
hoc review of the raw data in an attempt to identify
points that showed the greatest spatial variability. PRC
reviewed the 99 percent confidence intervals around the
mean concentrations for the quantitative data. In many
cases, the confidence intervals ranged into negative
concentrations on the low ends, and at both the high and
low endii were far more extreme than actually measured.
PRC then considered the concentration ranges within a
replicate sampling interval. These ranges represented
actual measured concentrations of contaminants within a
sampling interval. This nonparametric approach is more
representative of reality than statistically generated
confidence intervals that span negative concentrations for
intervals where the low concentration was in the 100's or
1,000's of ppm. When sample intervals were identified
as outliers based on this criteria, all the replicate data for
the interval was eliminated.
53
-------
TABLE 7-1. SUMMARY OF TPH RESULTS FOR QUANTITATIVE EVALUATION
Site Node
Atlantic Node 2
Node3
Node 4
NodeS
York Node 1
Node 2
Node3
Node 4
Node 5
FortRiley Node 1
Node 2
NodeS
Depth
21-22
24-25***
16-17***
6.5-7.5***
10-11
27.5-28.5***
16-17
23.5-24.5***
15-16***
13.5-14.5**8
17-18***
17-18
14-15
18-19***
1.5-2.5
2.3***
13-14***
6-7***
17-18
10.5-11.5
16-17
TPH-mean
11,090
4,004
425
255
2,436
1,094
201
239
773
1,539
497
778
2,281
1,878
60
5,728
1,416
2,169
13,150
22,480
3,926
TPH-SD
2,690
4,592
577
166
1,130
1,157
177
367
905
1,036
535
513
2,361
2,155
68
6,699
1,608
3,187
4.18
4,935
3,471
RSD (%)
24
114
136
55
46
106
88
154
117
67
108
66
104
115
113
117
114
147
32
26
88
Notes:
Data rejected by PRC on the basis of range criterion.
TPH varies in mg/kg.
The report then goes on to apply a correlation
coefficient criterion even though the independent
variables (the reference data) are so imprecise. The
standard regression analysis is inappropriate when there
is a high degree of imprecision in the independent
variable. Replacement of the -widely varying values by
the mean does not improve matters. Moreover,
allowance was not made for any similar variability's
in the ROST™ values, even though they will also be
affected by the spatial inhomogeneity.
As defined in the final demonstration plan (PRC
1994), the reference data was considered accurate and
precise. PRC concedes that the reference data exhibited
some variation within each interval, however, this
impact was reduced when PRC eliminated over
50 percent of the data based on the criteria discussed
above. In addition, PRC also evaluated the ROST™
technology based on the mean of its replicate pushes
over all quantitative sampling intervals. This showed
little unproved correlation. The ITER has been revised
to include a detailed discussion of the impact of matrix
heterogeneity and it will explain that the lack of
correlation could be due to matrix, instrument, and
analytical factors in any combination. PRC feels that
this data must remain in the report to give potential users
some frame of reference if they plan to conduct
confirmatory sampling. In addition, the trends identified
by the slope data from the regression data also seems to
present consistent and valuable data.
We all knew going into the demonstration that
sampling could turn out to be the Achilles heel of the
evaluation. Note the following quote from the Final
Demonstration Plan: "Total precision is controlled by
two sources: analytical error and spatial soil variability.
These sources cannot be readily separated and the most
serious concern for this demonstration is the total
54
-------
precision, not its sources." The QA/QC checks address
the analytical error, which is negligible. The spatial soil
variability was to have been address through the
establishment of 99-percent confidence intervals: "To
create (these) confidence intervals, PRC will collect five
replicate samples from 6-inch intervals, collocated to the
Unisys and Tri-Services technologies quantitative
sampling intervals." (Page 6-8)
The final demonstration plan (PRC 1994) explains
that the 99 percent confidence intervals were intended to
create pseudo PE samples. If the resultant technology
data fell within this confidence interval, the technology
would be determined to have acceptable accuracy.
However, since the technology did not produce data that
represented contaminant concentrations in ppm or any
other comparable unit of measure, this comparison could
not be made. This is explained in the ITER.
Obviously it was expected that analytical error
would be of much less significance than the uncertainties
cause by inhomogeneity in the contaminant distribution
of the soil. The quality control/quality assurance
procedures described in Chapter 2 verified that good
laboratory practice was followed in the chemical
analyses. Thus, we accept the stipulation on page
2-8 "Based on a review of this data, the reference
method accuracy requirements were met: if it is
understood that the terms reference method accuracy and
precision refers to the analytical measurements, but not
the overall measurement. The total precision (dominated
by the sampling problem) is poor and the reference
methods cannot be considered as a "gold standard" for
purposes of assessing ROST™s capabilities. Consider,
for example, a precision test to the "reference" data.
How many of the individual reference values fall within
20% of the mean value for each co-located sampling
interval? In the case of Atlantic, only 6 of the 38 total
determinations satisfy a 20% precision criterion. In fact,
less than half (16/38) even satisfy a 50% precision
criterion. To use the mean values and treat them as
errorless is highly misleading.
Matrix heterogeneity is an inherent condition in the
environment, and not an artifact of sampling as implied
in the comment above. In the case of this
demonstration, and possibly any other attempt to
quantitatively evaluate the ROST™ technology, the
smaller size of the ROST™ sampling volume may
magnify the effect of matrix heterogeneity. The
sampling volume of the ROST™ technology is 100's to
1000's of times smaller than the sample volume used to
drive regulatory decisions.
PRC agrees that the reference methods used may not
even measure the same compounds measured by the
technology, however, these methods were chosen and
presented in the final demonstration plan (PRC 1994) as
being the closest matches in the suite of regulatory
analytical methods. The ITER has been revised to
clarify that the analytical methods are merely attempts to
identify potential correlation between the technology's
data and analytical data used to drive regulatory
decisions. In addition, the ITER has been revised to
clarify that the quantitative evaluation is only intended to
provide a baseline look at the quantitative potential of the
technology.
The developer's use of the precision criteria as a
comparative tool to assess the heterogeneity of the
matrix is a misuse of the criteria. The 20 percent
criteria was intended to evaluate inherent method
precision. The reference analytical methods produced
data that met or exceeded this criteria.
PRC will clarify that the reference data is affected
by heterogeneity, and discuss the potential impacts of
this heterogeneity on the results and data interpretation
for this ITER.
Given the above facts, we are particularly dismayed
that the, confidence intervals were deleted from the data
interpretation yet language such as "The quantitative
assessment found that the ROST™data was [sic] poorly
correlated to any of the concentrations of the target
analytes. The data produced from this demonstration did
not identify a consistent relationship between ROST™
fluorescence data and the reference data." These
statements are unjustified and too easily taken out of
context.
The ITER has been revised to clarify that given the
configuration of the technology, as used in this
demonstration, it cannot produce quantitative data. This
is based on a lack of calibration and the inability to
evaluate the technology's performance in an in situ
mode. However, the regression data will remain in the
report to provide potential technology users with
information regarding attempted quantitative evaluation.
We also feel that the quantitative evaluation is
emphasized out of proportion in the draft report. We
discussed with PRC on several occasions that the
quantity directly measured by ROST™ (as well as
SCAPS) is a fluorescence intensity that depends
(predictably) on instrumental parameters (laser power,
gain settings on detectors, etc.) And is proportional to
concentration. However, the proportionality factor can
be approximated only be calibrations on spiked samples.
The calibrations require using soil and petroleum
products as close as possible to what is encountered in
the ground and have to be repeated if there are large
55
-------
variations in the soil lithology or product composition
(gasoline vs. Coal tar, for example). Since ROST™ is a
screening tool and was being evaluated as such, the
parties agreed not to include such calibrations. We were
agreeable to a comparison of the ROST™ data and the
quantitative evaluation results to see what sort of
correlations might emerge, but were never claiming the
ROST™ in the tested configuration is a "definitive"
method.
PRC agrees that the ITER focused too heavily on
the quantitative aspect of the technology. The ITER has
been revised to focus on the qualitative application and
performance of the technology.
PRC collected two sets of site-specific samples for
each site and presented them to the developers, including
analytical data. The developers were aware that PRC
expected them to produce some type of concentration
data, as explained in the developer-approved
demonstration plan. The developers opted not to
calibrate the technology. Therefore, any calibration
claims not supported by data collected during the
demonstration will be written by the developers, and
included in Section 7 Developer Comments and
Technology Update. In addition, the ITER has been
revised to clarify that this technology, in the
configuration demonstrated, cannot produce quantitative
data, and produces screening level data that cannot be
confirmed by existing analytical methods.
Using the mean values without accounting for the
variability in the reference values renders the whole
"f >0.80" exercise meaningless. In fact, if one applies
regression criteria to the reference data themselves, they
fail miserably! Let's pretend that five different
combinations of drilling companies and testing
laboratories of their choice were involved in the
quantitative test. Assign the square that come first in
alphabetical order at each unique node-depth
combination to the results obtained by drilling company
"alpha" and the laboratory they use. Assign the results
from each node for the next letter in alphabetical order
to the "beta" and so forth. Then use the mean value of
the results for each node/depth as the "right" answer
(ignoring the standard deviations) and compare against
the results of the alpha test, the beta test, etc. What do
we find? Most of the determinations fail at the r2 >
0.80 criterion.
The replicate sampling was conducted to define the
spatial variability of the matrix. This allowed the
elimination of quantitative evaluation samples that
exhibited excessive spatial variability. If the above
comparison is conducted on the reduced data set, the
reference method would not perform as described above.
In addition, the correlation coefficient criteria stated
above was only used to determine if the technology was
producing definitive or screening level data. All
regression data is included in the ITER to allow potential
users to determine the value of this approach of data
evaluation, relative to the technology. The ITER has not
been revised based on this developer comment.
DTI Comments (May 1995)
On May 31, 1995, DTI submitted additional
comments on the draft ITER. A total of eight comments
were submitted, in addition to a ROST™ executive
summary. Of these comments, five were revisitations of
comments made in Loral's April 1995 comment letter.
Two of the remaining three comments were new and are
presented and responded to below. The last remaining
comment was an update of the technology application
and it will be included in the technology update that
follows the response comments.
We strongly urge that you eliminate all attempts to
correlate ROST™fluorescence data with concentrations
of individual compounds (e.g., xylene orpyrene). This
is comparable to trying to correlate a TPH-gasoline
measurement with the concentration of a single
component, say isooctane, from a complex mixture. The
nature of the fluorescence measurements is such that any
such apparent correlation is an illusion since the spectra
of the different compounds overlap with each other. It is
virtually impossible to choose a set of excitation and
emission conditions such that only a compound is
excited. We never claimed this sort of capability.
You make the statement in the draft ITER "For the
purpose of this demonstration, the lack of approved EPA
methods did not preclude ROST™from being considered
a definitive level technology." This is setting up a brick
wall for us. As you note, "Definitive data are generated
using rigorous analytical methods, such as approved
EPA reference methods. Data are analyte specific, with
confirmation of analyte identity and concentration." We
never stated that ROST™ could be a definitive level
technology and were always careful to point out that the
fluorescence intensity is compound class sensitive, not
individual compound sensitive. On the other hand, it
does make sense to examine correlations of ROST™
fluorescence data with concentrations of classes of
compounds, such as naphthalenes, PAHs, BTEX or
measurements such as 418.1 or OA-1.
The ITER has been revised to reflect the developer-
intended use of the technology and the complexities
associated with individual compound quantitation.
56
-------
Did you check on the TRPH values from continuous
sampling at Atlantic? The TPH values for 8-8.5feet and
8.5-9 feet seem very low in comparison to the trends in
the naphthalene concentrations.
The concentrations reported in the data tables match
the concentrations reported by the laboratory.
Technology Update
This update is based on on direct correspondence
from DTI and Loral Corp.
As of June 1, 1995, ROST™ can be used to gen-
erate semiquantitative estimates of TPH concentration.
This involves post-data collection processing and waiting
for confirmatory analysis on select, site-specific
calibration samples.
Some type of calibration is necessary to convert
from percentages to actual units, such as ppm. Choosing
calibration factors is not simple in this case. One thing
learned from this demonstration is that the sampling
should be as close as possible to the ROST™
measurement to minimize heterogeneity. The procedure
by the Navy SCAPS for validation reduces this distance
to only a few inches. They have not quantitated what
sort of variation is observed in this case. Overboring the
same sample (true overboring) might be even better and
the Army SCAPS is exploring this alternative. Another
option is to place recovered material on the window.
This removes the sample variability problem, but leaves
the sample disruption problem.
By observation, DTI found that a scale factor of 10
to 12 works best for the two coal tar sites (Atlantic and
York) and a factor of 100 works best for the Fort Riley
site. Multiplying the fluorescence percentages by these
scale factors gives an estimate of the TPH values. This
type of post field work data processing is now a standard
ROST™ practice.
The following technology update was supplied by
the developers on June 5, 1995.
Since the time of the SITE demonstration
measurements (August 1994), the ROST™instrumentation
and procedures have been improved in several ways:
1. As before, the fluorescence intensity is reported as a
percentage relative to the fluorescence intensity of a
reference solution which is acquired prior to each push.
We formerly referred to this as a "percent of standard."
We now refer to it as "percent of reference. The
reference solution provides an end-to-end system check
and normalizes the data for any variation in the power of
the laser light used to excite the contaminant, length of
cable carrying the excitation and emission light,
background noise, and other instrument settings such as
monochromator slitwidth. We have now gone to a single
reference solution, referred to as "M-J", to provide
better comparability from one site to another. The
composition of M-l has been chosen so that its
fluorescence emission covers the wavelength range of
commonly occurring POLs.
2. The voltage distribution on the photomulitplier tube
has been modified. Two benefits have resulted. The
linear response range extends to higher signal levels and
the noise has been reduced.
3. Soft\vare procedures for subtracting the instrumental
baseline from the fluorescence vs. depth plots have been
elaborated. These procedures, in conjunction with the
reduced electrical noise mentioned above, have
drastically improved the ROST™ability to detect small,
but real fluorescence signals.
4. The ROST™module that fits between the cone and the
push rods and contains the sapphire window has been
redesigned. The opportunity for contamination to get on
the mirrors has been eliminated. The collection
efficiency of the optics has been increased.
5. A multiple wavelength capability, such that the
fluorescence vs. depth measurements can be made at up
to four wavelengths simultaneously, is under
development.
6. Several calibration procedures for converting the
ROST™fluorescence data from the percent of reference
format to concentration units have been developed and
are under evaluation. These procedures are outlined in
the accompanying document.
Converting Rapid Optical Screening Tool
(ROST™) Fluorescence Intensities to Con-
centration Equivalents
ROST™ senses the aromatic hydrocarbon con-
stituents of petroleum, oil, and lubricants (POL) by their
fluorescence response to ultraviolet wavelength laser
excitation. All chemical analysis methods (including
laboratory-based ones) measure variables related to
concentration, not concentration itself. Conversion from
the measured variable to concentration requires a
calibration curve. Several calibration options for
relating the raw ROST™ fluorescence data to
contaminant concentrations are under evaluation, but the
best option (or options) have not been determined yet.
The SUE demonstration and similar ROST™ studies
show how difficult it is to validate in situ measurements.
57
-------
The reference percentage data format is well-suited
for field screening applications, in which the goal is to
delineate contaminant plume boundaries and to define
the relative distribution of contamination over the site.
The fluorescence intensity is proportional to POL
concentration over a wide range of concentration. The
reliability of LIF-CPT for screening sites in this fashion,
i.e., without any formal calibration procedure, has been
demonstrated on many occasions. This form of data
presentation, in which the instrument response is
expressed relative to some reference compound (hat may
or may not be actually present at the site) is similar to
other site assessment methods, e.g., organic vapor
monitors. However, since ROST™ is much closer to
conventional analytical methodology in its amendability
to QA/QC criteria, its flexibility, and the level of detail
it provides, a client may want to perform an instrument
calibration, which allows POL concentrations to be
reported in concentration units such as mg/kg.
The following factors must be considered in the
selection and preparation of calibration standards:
1. The measured fluorescence is a composite of the
contributions from all fluorescence chemical components
in the sample. Thus, aging and weathering processes
that affect chemical composition must be considered.
2. The fluorescence response of a petroleum-impacted
soil sample is affected by the soil composition and
physical properties. For example, a contaminant
exhibits higher fluorescence intensity on sand than it
does on a soil matrix with high clay content. The effect
is believed to be related to available surface area
considerations.
3. Calibration standards can be difficult to prepare for
low concentrations of volatile fuels, such as gasoline and
jetfuel.
Two basic strategies are available: (1) Obtaining the
calibration "standards" by actual sampling at the site;
(2) Making calibration standards based on assumptions
on the soil type and POL that exists in the ground. Each
of these, and the relative advantages and disadvantages,
are discussed below. Note that the former is the
procedure used during the SITE demonstration and
similar validation studies to date.
Calibration Derived from Site Materials with
In Situ Fluorescence Measurements
Method validation studies for LIF and other in situ
sensors have generally used calibration standards
obtained directly from the ground by soil borings, which
are then submitted for analysis by approved laboratory
methods. The major advantage of this approach is that
the influence of confounding variables such as
weathering, soil moisture, soil matrix, and other
changes, are eliminated. The concentrations are
established with the standard analytical methods so
regulatory agencies readily accept the results. The
primary disadvantage of in situ calibration or validation
methods is that it can be extremely difficult to obtain a
sample for the conventional analysis from the same
location as surveyed by ROST. The more homogeneously
the contamination is distributed, the less this is a
concern.
An option well worth considering is to subject any
sample provided to a laboratory for analysis to an
"above ground " ROSTmfluorescence measurement. This
second reading provides a quality check on the soil
sample to ensure it is representative of the soil measured
in situ by the ROST™ system. Note, however, that the
"uphole" measurement doesn't precisely replicate the
conditions of the in situ push data, since the degree of
soil compression is not the same.
Calibration Derived from Synthetic
Standards with "Above Ground"
Fluorescence Measurements
Another method, which closely resembles
traditionally analytical methodology, uses standards for
purposes of constructing the calibration curve. The
synthetic calibration standard approach eliminates the
need for any soil borings. The laboratory methods
generally use solution standards, which are highly
homogeneous and easily made by conventional weighing
and volumetric procedures. The calibration samples for
ROST™ are prepared by quantitatively spiking the
petroleum produces) of interest on to soil. Thus, for the
most exacting requirements, the choice of the POL and
soil for the calibration phase is crucial since the
proportionality factor between fluorescence response and
concentration depends on both the POL type and the soil
matrix itself. The added material should match the
target POL analyte as closely as possible. One of the
difficulties in establishing the target POL analyte is that
often many different petroleum products are present at a
particular site. The actual contaminant may represent a
combination of POL products. In addition, the
contaminant will have weathered from long-term
exposure at the site. Many other quantifying analytical
methods also encounter this problem.
58
-------
Depending on the objectives of the investigation,
there are several approaches to designing a calibration
procedure -with synthetic standards:
Approach 1: Designation of POL and
Soil Type
The client can assume that their contaminant is
similar to a common product (gasoline, dieselfuel, coal
tar, etc.) and their soil is similar to typical soil types
(sand, silt, clay). If these assumptions are made, the
conversion from the raw percent of fluorescence format
to concentration units can be made using standard tables
determined from laboratory studies. Note: These tables
exist currently for common fuels on sand and are under
construction for common fuels on clay and silt. This
approach assumes a linear response of the instrument to
the various contaminant concentrations.
Approach 2: Specific POL Material, Designated
Soil Type
The client provides contaminant from the site that
can be spiked in the laboratory onto reference soils and
then analyzed by ROST. A set of standards is prepared
by inoculating the soil samples -with a series of
increasing amounts of the target analyte. The spiked
samples are tumbled for 24-48 hours to ensure uniform
distribution of the fuel.
Approach 3: Specific POL Material and Soil
from the Site
The client provides contaminant from the site and
clean soil samples from the site. Contaminants from the
site can then be spiked onto these soils and then analyzed
by ROSL The soil is gathered from below the surface at
a depth of 1-2 feet, to reduce hydrocarbon contamination
from aerosols and other airborne particulates. This
option is the most specific of the synthetic calibration
standard approaches. However, please note that the
calibration still assumes that the soil and product used in
the calibration is representative of the site.
59
-------
Section 8
References
American Society for Testing and Materials (ASTM). 1990. "Standard Test Method for Particle-Size Analysis of
Soils."
Environmental Protection Agency (EPA). 1986. "Laboratory Manual Physical/Chemical Methods." In Test Methods
for Evaluating Solid Waste. Volume IB.
EPA. 1991. National Functional Guidelines for Organic Data Review. Contract Laboratory Program. June.
EPA. 1993. Data Quality Objectives Process for Superfund - Interim Final Guidance.
Page, A. L., ed. 1982. "Chemical and Microbiological Properties." In Methods of 'Soil Analysis. 2nd Edition.
Number 9. Part 2.
PRC Environmental Management, Inc. 1994. "Final Demonstration Plan for the Evaluation of Cone Penetrometer- and
Geoprobe®-Mounted Sensors." August.
University Hygienic Laboratory. 1991. "Method OA-1 for Determination of Volatile Petroleum Hydrocarbons
(Gasoline)." University of Iowa. Iowa City, Iowa.
60
-------
APPENDIX A
Qualitative, Quantitative, Geotechnical, and TOC Data
Table
A-1.
A-2.
A-3.
A-4.
A-5.
A-6.
A-7.
A-8.
A-9.
Page
Qualitative Reference Laboratory Data for TPH and PAH - Atlantic Site 62
Qualitative Reference Laboratory Data for TPH and PAH - York Site 63
Qualitative Reference Laboratory Data for TPH and PAH - Fort Riley Site "....'.' 64
Quantitative Reference Laboratory Data - Atlantic Site '' 65
Quantitative Reference Laboratory Data - York Site 66
Quantitative Reference Laboratory Data - Fort Riley Site 67
Geotechnical and TOC Data - Atlantic Site '.'.'.'.'.'.".'.'.'. 68
Geotechnical and TOC Data -York Site '.'.'.'.'.'. 69
Geotechnical and TOC Data - Fort Riley Site '.'.'.'.'.'.'.'.'.'.'.'.'.'. 69
61
-------
TABLE A-1. QUALITATIVE REFERENCE LABORATORY DATA
FOR TPH AND PAH— ATLANTIC SITE
Node
Number
2
2
2
2
2
3
3
3
3.
4
4
4
4
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
Notes:
PAH
ppm
TPH
NS
ND
Depth
ffeett
1-1.5
8-9
13-14
25-26
35-36
1-2
10-11
20.5-21.5
33.5 - 34.5
2-2.5
6-6.5
6.5-7
7-7.5
7.5-8
8-8.5
8.5-9
9-9.5
9-10
9.5-10
10-10.5
10.5-11
15-16
27-28
1-2
5-6
7-8
27-28
33.5-34.5
TPH
(com)
1,680
15
24.7
NS
19.3
55.4
1,130
222
54.2
149
330
614
1,650
4,170
541
73.7
1,680
897
2,880
2,960
3,820
1,170
118
399
ND
275
146
ND
PAH
toom)
88.19
3.59
0
0
.06
11.40
158.37
6.06
.99
13.39
.258
5.267
21.494
44.205
128.811
71.760
55.482
40.644
80.999
104.487
107.437
62.091
48.879
7.007
0.020
18.496
3.481
0.030
Polynuclear aromatic hydrocarbon.
Part per million.
Total petroleum hydrocarbon.
Not sampled.
Not detected.
62
-------
TABLE A-2. QUALITATIVE REFERENCE LABORATORY DATA
FOR TPH AND PAH— YORK SITE
Node Depth
Number (feet)
1
1
1
1
2
2
2
2
3
3
3
3
3
3
3
3
4
4
4
4
4
4
5
5
5
5
Notes:
PAH
ppm
TPH
ND
12-13
14-15
17-18
22-22.5
8.5-9
10.5-11
14-14.5
20-21
10-11
12-13
16-16.5 2
16.5-17 3
17-17.5
17.5-18
18-18.5
21.5-22.5
8-9
11-12
14-15 8
17-18
18-18.5 13,
21.5-22
10-11
12-13
17-18
21-22
Polynuclear aromatic hydrocarbon.
Part per million.
Total petroleum hydrocarbon.
Not detected.
TPH
(pom)
26.1
345
13.7
ND
ND
417
855
10.2
10
259
,570
,650
57.5
12.7
27.8
ND
115
174
,150
137
,100
74.2
23.7
66
377
ND
PAH
(com)
1.09
48.81
.88
.01
0
7.72
127.62
.060
0
37.62
134.67
313.97
1.90
0.23
0.20
0.01
0.66
182.09
1,412.16
10.33
1,130.18
14.35
0
9.31
128.09
0.165
63
-------
TABLE A-3. QUALITATIVE REFERENCE LABORATORY DATA
FOR TPH AND PAH—FORT RILEY SITE
Node
Number
1
1
1
2
2
2
2
2
3
3
3
3
4
5
5
5
5
5
5
5
5
5
5
5
Depth
ffeett
1-1.5
18-19
28.5 - 30
5-6
6-7
'15-16
23.5 - 25
28.5 - 30
1.5-2.5
5.5-6.5
15-16
23.5 - 24
15-16
2.5-3.5
5-6
10-10.5
10.5-11
11-11.5
11.5-12
12-12.5
12.5-13
13-13.5
24-25
29-30
TPH
(DDm)
47.1
482
ND
37.4
233
6,720
89.3
96.9
112
2,670
1,850
ND
37
ND
1,280
6,730
32,800
19,300
9,360
12,700
2,830
2,550
9.94
ND
PAH
(DDrrrt
0.667
12.964
0.036
0.108
0.142
137.885
1.707
2.713
1.358
118.471
18.730
0
0
0
1.655
143.622
338.566
344.984
206.888
190.693
87.639
64.711
0
0
Notes:
PAH Polynuclear aromatic hydrocarbon.
ppm Part per million.
TPH Total petroleum hydrocarbon.
ND Not detected.
64
-------
TABLE A-4. QUANTITATIVE REFERENCE LABORATORY DATA - ATLANTIC SITE
Chemical
Node 2 (21
Ethyl-
benzene
TPH
VPH
Total
PAH
Total
BTEX
Chemical
Minimum
to 22 feet)
25,000.00
8,850.00
910.00
70.42
15,400.00
Minimum
Maximum
42,000.00
15,400.00
2,000.00
918.32
293,000.00
Maximum
Mean
34,oQao0.
11,09QJ3Q
1,402,00
672.69
221,200.00
Mean
Standard
Deviation
7,035.62
2,689.89
427.22
354.08
53,049.03
Standard
Deviation
Node 3 (16 to 17 feet)
Ethyl-
benzene
TPH
VPH
Total
PAH
Total
BTEX
Chemical
3,600.00
104.00
88.00
4.43
25,590.00
Minimum
4,600.00
1 ,290.00
130.00
6.84
33,550.00
Maximum
425.25
112,00
&?$
2S;330,QQ
Mean
469.04
577.00
21.35
1.05
3,728.07
Standard
Deviation
Node 4 (10 to 11 feet)
Ethyl-
benzene
TPH
VPH
Total
PAH
Total
BTEX
Chemical
29,000.00
959.00
1 ,200.00
12.26
218,700.00
Minimum
35,000.00
3,780.00
1 ,400.00
148.10
307,000.00
Maximum
31.,609,00.
2,435,80
1,320,00
77.91
2W&M3Q
Mean
2,408.32
1,129.72
83.67
60.07
34,028.49
Standard
Deviation
Node 5 (16 to 17 feet)
Ethyl-
benzene
TPH
VPH
Total
PAH
Total
BTEX
390.00
87.80
36.00
0.48
5,490.00
2,100.00
516.00
160.00
4.72
28,700.00
1<198,0Q
201.36.
86,00
2.39
15,798,00
811.80
177.36
59.36
1.79
11,794.98
Minimum
Maximum
Node 2 (24 to 25 feet)
250.00
36.00
7.90
3.99
1 ,250.00
Minimum
39,000.00
9,880.00
1,400.00
691 .99
260,000.00
Maximum
Mean
14,690.00
4,004.40
537.58
290.96
93,950.00
Mean
Node 4 (6.5 to 7.5 feet)
250.00
20.70
10.00
3.91
320.00
Minimum
3,600.00
412.00
110.00
13.73
37,300.00
Maximum
1,410.00
254.93
43.33
i!$
10,435.00
Mean
Node 4 (27.5 to 28.5 feet)
1,300.00
117.00
42.00
29.51
13,700.00
Minimum
23,000.00
3,030.00
970.00
270.81
193,000.00
Maximum
Node 5 (23.5 to 24.5 feet)
940.00
48.20
52.00
1.96
14,330.00
1 ,700.00
893.00
160.00
5.11
22,200.00
10,520.00
1 ,093.60
452.40
1-23,15
96,740.00
1,174,-Q0
238.50
77.20
2S&
17,754^0
Standard
Deviation
16,663.45
4,592.73
579.81
324.69
109,210.58
Standard
1 ,897.71
166.39
57.74
4.63
17,931.46
Standard
Deviation
10,146.77
1,156.83
461.91
98.83
85,721.28
Standard
301.30
366.76
46.38
1.26
3,340.21
Notes:
Values used in the final regression equations.
65
-------
TABLE A-5. QUANTITATIVE REFERENCE LABORATORY DATA - YORK SITE
Minimum
Node 1 (15 to 16 feet)
TPH
VPH
Total
PAH
Total
BTEX
Chemical
53.00
a
64.65
580.00
Minimum
Node 2 (17 to 18 feet)
Ethyl-
benzene
TPH
VPH
Total
PAH
Total
BTEX
Chemical
1,600.00
88.70
5.40
40.41
1,900.00
Minimum
Node 4 (14 to 15 feet)
Ethyl-
benzene
TPH
VPH
Total
PAH
Total
BTEX
Chemical
2,200.00
647.00
37.00
166.06
4,540.00
Minimum
Node 6 (1.5 to 2.5 feet)
Ethyl-
benzene
TPH
VPH
Total
PAH
Total
BTEX
ND
15.20
ND
0.01
ND
Maximum
2,270.00
*
755.70
4,700.00
Maximum
7,200.00
1,380.00
33.00
252.47
14,100.00
Maximum
19,000.00
6,450.00
97.00
1,043.82
36,300.00
Maximum
ND
138.00
ND
0.95
ND
Mean
773.20
"
260.31
Mean
4,$6Q.C0
496.94
l&M
1&X42
8,848,00
Mean
9,800,00
2^+48
S4.2S
51462
19,542.00
Mean
ND
ND
0.45
ND
Standard
Deviation
905.13
284.20
2,317.79
Standard
Deviation
2,846.58
535.84
12.51
98.64
5,131.92
Standard
Deviation
6,412.49
2,361.49
24.87
342.68
11,891.11
Standard
Deviation
ND
67.83
ND
0.40
ND
Minimum
Maximum Mean
Standard
Deviation
Node 2 (13.5 to 14.5 feet)
156.00
14.00
159.85
1,800.00
Minimum
Node 3 (17 to 18 feet)
290.00
261.00
7.50
97.45
1 ,090.00
Minimum
Node 4 (18 to 19 feet)
92.00
18.20
6.00
1.40
274.00
2,710.00 1,539.20
45.00
466.41 iHlli
23,280.00 7,556.00
Maximum Mean
2,700.00 1$j&QO
1,450.00 77B.40
30.00 W^SO
359.67 £530,34.
4,900.00 2.S4B.40
Maximum Mean
57,000.00 21,810.50
4,000.00 1,878.15
280.00 175.33
2,332.21 798.91
128,800.00 42,528.80
1 ,035.82
11.90
129.85
8,853.57
Standard
Deviation
874.25
513.04
9.30
116.76
1,641.76
Standard
Deviation
27,363.23
2,155.00
148.01
1,116.53
59,970.23
Notes:
ND
No data.
Not detected.
Values used in the final regression equations.
66
-------
TABLE A-6. QUANTITATIVE REFERENCE LABORATORY DATA - FORT RILEY SITE
Chemical Minimum
Node 1 (2 to 3 feet)
Ethyl-
benzene 79.00
TPH 27.50
VPH 6.00
Total
PAH 0.97
Total
BTEX 339.00
Chemical Minimum
Node 2 (6 to 7 feet)
Ethyl-
benzene 107.50
TPH 48.60
VPH 9.00
Total
PAH 0.05
Total
BTEX 89.00
Chemical Minimum
Node 5 (10.5 to 11.5 feet)
Ethyl-
benzene 1 ,400.00
TPH 17,700.00
VPH 250.00
Total
PAH 157.14
Total
BTEX 23,270.00
Maximum Mean
3,700.00 1,075.80
15,800.00 5,727.98
110.00 47.50
260.60 89.15
20,610.00 6,412.40
Maximum Mean
2,000.00 762.50
7,720.00 2,169.32
98.00 41.83
42.05 10.98
10,070.00 2,956.63
Maximum Mean
29,000.0013,680.00
32,800.00 miSmtO.
430.00 HtH
340.26 i$iip
96,300.00 SHiH
Standard
Deviation
1,502.51
6,698.52
44.90
105.57
8,295.04
Standard
Deviation
1,072.32
3,186.75
48.87
18.15
4,756.48
Standard
Deviation
11,238.86
5,934.81
80.81
67.41
27,580.73
Minimum
Maximum
Nodii 1 (13 to 14 feet)
100.00
32.90
9.20
0.02
230.00
Minimum
20,000.00
3,110.00
320.00
96.62
70,000.00
Maximum
Node 2 (17 to 18 feet)
28,000.00
7,050.00
530.00
60.66
147,000.00
Minimum
60,000.00
16,900.00
1,200.00
224.76
254,000.00 1
Maximum
8,595.00
1,416.00
183.55
31.44
26,497.60
Mean
39,600X10
13,1SCHJ0
TSOm
1?,&SS
mm*®
Mean
Node 5 (16 to 17 feet)
20,000.00
1,090.00
170.00
18.23
63,100.00
55,000.00
9,630.00
930.00
, 162.28
219,700.00 |
31,405 0$
3.928,00
44.&QO
5$ 43
Standard
Deviation
10,009.87
1 ,607.95
152.78
44.47
35,204.28
Standard
13,464.77
4,182.11
259.71
72.14
47,704.30
Standard
13,612.49
3,470.96
289.34
58.45
62,975.05
Notes:
Values used in the final regression equations.
67
-------
TABLE A-7. GEOTECHNICAL AND TOG DATA - ATLANTIC SITE
Node/
Grid
1/F
1/F
1/F
1/F
1/F
4/C
4/C
Notes:
mg/kg
mm
nvn
USDA
uses
ND
Depth TOC
(feet) (ma/kcrt
2-3 4,000
10-11 ND
20.5-21 600
30.5-31 200
35-35.5 400
9-10 3,800
15-16 3,200
% Sand
% > 2 mm (0.5-2 mrm
.03
0
1
28.38
0
0
0
12.43
36
50.84
62.78
24.72
19.34
44.79
% Silt % Clay
(2-50 urn) «2 urn)
58.33 29.21
43.78 20.22
34.17 13.99
4.72 4.12
44.73 30.55
51.24 29.42
30.57 24.64
USDA USCS
Classification Classification
Silty clay loam Sandy lean clay (CL)
Loam Silt or clay (CL or ML)
Loam Silt or clay (CL or ML)
Sand Well to poorly graded
sand (SW or SP)
Clay loam Sandy lean clay or sandy
silt (CL or ML)
Silty clay loam Sandy lean silt or sandy
lean clay (CL or ML)
Loam Silt or clay (CL or ML
Milligram per kilogram.
Millimeter.
Micrometer.
United States Department of Agriculture.
Unified Soil Classification System, ( ) two-letter classification code.
Not detected.
68
-------
TABLE A-8. GEOTECHNICAL AND TOC DATA - YORK SITE
Node/
Grid
1/G
1/G
1/G
1/G
3/C
3/C
Notes:
mg/kg
mm
/urn
USDA
uses
ND
TABLE
Node/
Grid
4/H
4/H
4/H
4/H
2/E
3/G
Notes:
mg/kg
mm
Depth TOC %Sand
(feet) (ma/ka) %>2mm (0.5-2 mm)
5-6 ND 0.00 13.66
7-8 2,800 0.05 26.08
15-15.5 1,400 0.23 60.24
18.5-19 490 30.54 46.38
12-13 3,200 0.00 8.90
16.5-17 2,600 6.69 52.48
% Silt
(2-50 urn)
58.94
51.05
20.93
12.09
60.43
17.92
%Clay
<<2 urn)
27.40
22.82
18.60
10.99
30.67
22.91
USDA
Classification
Silty clay loam
Silt loam
Sandy loam
Sandy loam
Silty clay loam
Sandy clay loam
uses
Classification
Clay or silt with sand
(CLorML)
Clay or silt with sand
(CL or ML)
Silty to Clayey sand
(SM or SC)
Poorly graded sand
with silt or clay
(SW-SC or SP-SC)
Silt or lean clay with
sand (CL or ML)
Clayey or silty sand
(SM or SC)
Milligram per kilogram.
Millimeter. ' .
Micrometer.
United States Department of Agriculture.
Unified Soil Classification System, ( ) two-letter classification code.
Not detected.
A-9. GEOTECHNICAL AND TOC DATA -
Depth TOC
(feet) (ma/ka) %>2 mm
2 - 3 3,400 0.00
7.5-8.5 600 .16
15-16 800 0.00
29 - 30 300 20.36
15-16 4,600 .11
5.5-6.5 9,000 .10
Milligram per kilogram.
Millimeter.
FORT RILEY
% Sand
(0.5-2 mm)
31.32
60.76
62.44
57.48
55.13
47.61
SITE
% Silt
(2-50 urn)
43.48
22.08
19.16
10.46
25.87
36.02
% Clay
25.20
17.00
18.40
11.70
18.89
16.27
USDA
Classification
Loam
Sandy loam
Sandy loam
Sandy loam
Sandy loam ,
Loam
/j.m Micrometer.
USDA United States Department of Agriculture.
USCS Unified Soil Classification System, ( ) two-letter classification code.
it U.S. GOVERNMENT PRINTING OFFICE: 1995-653-447
69
-------
------- |