United States Science Advisory EPA-SAB-RAC-95-023
Environmental Board (1400) SEPTEMBER 1995
Protection Agency
&EPA AN SAB REPORT: REVIEW
OF RADIONUCLIDE CLEAN-
UP LEVELS FOR SOIL
REVIEW OF TECHNICAL ASPECTS
OF ORIA'S TECHNICAL SUPPORT
DOCUMENT FOR THE
DEVELOPMENT OF CLEANUP
LEVELS FOR SOIL BY THE
RADIATION ADVISORY
COMMITTEE
-------
September 29, 1995
EPA-SAB-RAC-95-023
Honorable Carol M. Browner
Administrator
U.S. Environmental Protection Agency
401 M Street, SW
Washington, DC 20460
Re: Review of Technical Aspects of the Office of Radiation and Indoor Air's
(ORIA) Technical Support Document (TSD) for the Development of
Radionuclide Cleanup Levels for Soil
Dear Ms. Browner:
This report was developed by the Radionuclide Cleanup Standards
Subcommittee (RCSS, also referred to as "the Subcommittee"), an ad-hoc
subcommittee formed by the Radiation Advisory Committee (RAC) of the Science
Advisory Board (SAB), in response to a request to review the technical aspects of the
Agency's radionuclide cleanup levels for soils. The RCSS met on October 27 and 28,
1994, January 26 and 27, 1995, March 27, 1995 (teleconference), and May 23 and 24,
1995.
The enclosed report addresses the Charge and elaborates upon those technical
aspects and issues which the Subcommittee believes would be most likely to require
the attention of the Agency in order to provide the most comprehensive and technically-
supportable basis for the radionuclide cleanup standards. This letter summarizes the
report and highlights the most significant findings and recommendations associated
with major elements of the Charge. The Charge elements and the Subcommittee's
major responses follow:
Charge T. Are the methodologies used by ORIA in the following areas
acceptable for providing a technical basis for writing a cleanup
standard: (a) methodology for evaluating source terms for
-------
radioactively contaminated sites, (b) methodology for modeling
transport to people, and (c) methodology for estimating risk to
individuals and populations?
Response: The framework for the overall approach taken by ORIA in the draft
TSD represents a creative and reasonable approach to addressing
risk reduction/cost tradeoffs in the soil cleanup standards for
radionuclides. however, a major concern of the Subcommittee is
that source term information appears weak even at the most well-
defined sites, and ORIA had to be quite inventive at some sites.
Recognizing that consistent site-wide data are limited, the
Subcommittee commends ORIA for making good use of the
available data and for its continuing efforts to work with other
agencies to improve the site-related data base and to ensure
appropriate utilization of the information collected. Still needed are
estimates of uncertainties for contaminated soil volumes. Also, for
many reference sites, it appears that the radionuclide selections
were not sufficiently inclusive.
Charge 2: Are the assumptions and modeled pathways reasonable and
suitable for assessing risk at radioactively contaminated sites: a)
for the combined residential/agricultural land use scenario, and b)
for the industrial-commercial scenario?;
Response: The Subcommittee is satisfied that ORIA's choice of an on-site
residential scenario and a commercial-industrial scenario for
thorough analysis is adequate for estimation of reasonable
maximum individual exposures and risks from sites that have been
cleaned up to a specified level of contamination. While other land-
use scenarios are also plausible, they are not likely to produce
substantially higher estimates of population risks when
subsequently applied to estimate the number of cancer cases
avoided at sites. However, the Subcommittee recommends that
ORIA consider adding qualitative discussions of the likely
magnitude of risks related to recreational and off-site residential
neighbor scenarios.
Charge 3: Is RESRAD v. 5.19 suitable for modeling radiation risks to
individuals at radioactively contaminated sites?
-------
Response: The Subcommittee finds that the initial screening and selection of
candidate models for further evaluation as reported in the TSD was
conducted in a reasonable, sound and thorough manner, using
appropriate criteria for model selection. Furthermore, the
Subcommittee concurs with ORIA's decision in its selection of
RESRAD as a reasonable transport model code for use at the
current time. However, while it incorporates some conservative
assumptions in its formulations, RESRAD itself may not necessarily
provide conservative risk estimates if inappropriate parameter
values are selected for the modeling input.
Because the Subcommittee has not evaluated the default values in
the RESRAD code, nor the full parameter set used for each
reference site, it is unable to assess fully the extent to which the
model results can be considered to be conservative or bounding
estimates of the true health effects associated with each level of
cleanup. From a cursory review, several of the transport model
parameter values appear to be inconsistent with values from the
literature or known site characteristics; hence, the Subcommittee
strongly urges EPA to obtain a thorough peer review of the default
and site-specific parameter values used in the transport modeling.
In the face of constraints of time and resources for revisiting the
pathway model definitions, ORIA should focus its efforts for any
improvement of the definitions (i.e., underlying assumptions and
adopted parameter values) on the dominant pathways, particularly
external gamma radiation, radon inhalation, crop ingestion, and
ingestion of groundwater, for selected reference sites. Ingestion of
surface water should be investigated further to determine whether
it might be an important pathway under any reasonable scenario.
Although the Subcommittee's specific Charge was to review scientific and
technical aspects of the Agency's risk assessment methodology for sites with
radioactive contamination, the Subcommittee's report also addresses some
suggestions in areas that relate to policy choices and clearly go beyond a strict reading
of the Charge. The most serious recurring problem for the Subcommittee was a
difficulty in understanding how the Agency ultimately intends to take into account the
interrelationships among its proposed regulatory actions to deal with soil cleanup,
aquifer cleanup, cleanup of structures, and waste disposal. Although recognizing that
-------
the Agency is currently constrained by statutes or regulations to address these aspects
as isolated problems, in truth, these problems are integrated, not isolated. The
Subcommittee had difficulty commenting on some of the technical aspects of ORIA's
TSD for soil cleanup, without having had the benefit of viewing it in the context of its
proposed use. The current fragmentation of the peer review process leads to
difficulties in reviewing estimates of overall exposures, risks and benefits. More
scientifically robust estimates of overall exposure and risk would be derived from an
integrated analysis.
Finally, the Subcommittee compliments the ORIA staff on its thorough
documentation and forthcoming approach during the review. Compiling the information
necessary to undertake this technical support document was obviously a formidable
task, and ORIA's ability to organize, present and make use of this scattered information
of variable quality is commendable. The Subcommittee also appreciated ORIA's
prompt and thorough responses to many of the Subcommittee's technical comments in
writing and in presentations, and that an excellent working relationship was established
and maintained throughout the review process.
The RAC and its Subcommittee appreciates the opportunity to provide this report to
you. We look forward to your response to this report, in general, and to the comments
and recommendations in this letter, in particular.
Sincerely,
Dr. Genevieve M. Matanoski, Chair
Science Advisory Board
James E. Watson, Jr., Ch
Radiation Advisory Committee
and Radionuclide Cleanup Standards Subcommittee
Science Advisory Board
-------
-------
NOTICE
This report has been written as a part of the activities of the Science Advisory Board, a
public advisory group providing extramural scientific information and advice to the Administrator
and other officials of the Environmental Protection Agency. The Board is structured to provide a
balanced, expert assessment of scientific matters related to problems facing the Agency. This
report has not been reviewed for approval by the Agency; hence, the comments of this report do
not necessarily reflect the views and policies of the Environmental Protection Agency or of other
Federal agencies. Any mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
-------
ABSTRACT
The EPA Science Advisory Board's (SAB) Radiation Advisory Committee
(RAC)/Radionuclide Cleanup Standards Subcommittee (RCSS) reviewed the Office of Radiation
and Indoor Air's (ORIA) "Technical Support Document (TSD) for the Development of
Radionuclide Cleanup Levels for Soil" (9/94). The RCSS supports ORIA's approach of defining
a generic site to compare environmental pathway models; perform sensitivity/uncertainty analyses;
and generate generic tables of cleanup soil concentrations for different land-use scenarios. ORIA
defined reference facilities to represent sites and derived site-specific risk factors to estimate soil
remediation volumes and health effects averted under each of the scenarios, for a range of cancer
incidence and radiation dose cleanup goals. The RCSS was concerned that source term
information appeared weak and the radionuclide selections were not sufficiently inclusive. The
RCSS emphasized the need to estimate uncertainties for contaminated soil volumes. The RCSS
concluded that the screening and selection of candidate transport models for risk assessment were
sound, and concurred with the use of RESRAD. The RCSS evaluated neither the default values
in RESRAD, nor the parameter set used for each reference site, and therefore was unable to
assess whether the model results are bounding estimates of the risks for each level of cleanup.
The RCSS recommended that EPA improve its definitions of the dominant pathways. The RCSS
commended the EPA's sensitivity and uncertainty analyses, but felt that the TSD did not
adequately convey the magnitude of the uncertainties in soil volumes requiring remediation and
cancers averted by remediation.
Key Words: Cleanup Standards, Environmental Radiation, Nuclear Facilities, Environmental
Quality, Radionuclide Transport, Radionuclide Cleanup
-------
SCIENCE ADVISORY BOARD
RADIATION ADVISORY COMMITTEE (RAC)
RADIONUCLIDE CLEANUP STANDARDS SUBCOMMITTEE (RCSS)
Chair
Dr. James E. Watson, Jr., Department of Environmental Sciences and Engineering, University
of North Carolina, Chapel Hill, NC
Members
Dr. Stephen L. Brown, R2C2 (Risks of Radiation and Chemical Compounds), Oakland, CA
(Also Scenarios Working Group Coordinator for RCSS)
Dr. Bernd Kahn, Nuclear Engineering and Health Physics Programs, and Environmental
Resources Center, Georgia Institute of Technology, Atlanta, GA (Also Source Terms Working
Group Coordinator for RCSS)
Dr. June Fabryka-Martin, Chemical Science and Technology Division, Los Alamos National
Laboratory, Los Alamos, NM (Also Transport/Modeling Working Group Coordinator for RCSS)
Dr. William Bair, retired, Richland, WA
Dr. Calvin C. Chien, Corporate Remediation, E.I. DuPont Company., Wilmington, DE (EEC
Liaison)
Dr. Ricardo Gonzalez, Department of Radiological Sciences, University of Puerto Rico School
of Medicine, San Juan, PR
Dr. F. Owen Hoffman, SENES Oak Ridge, Inc. Center for Risk Analysis, Oak Ridge, TN
Dr. Paul J. Merges, Hazardous Substance Regulation Division, New York State Department of
Environmental Conservation, Albany, NY
Consultants
Dr. Janet Johnson, Shepard Miller, Inc., Fort Collins, CO
Dr. Arend Meijer, GCX, Inc., Albuquerque, NM
Dr. James W. Mercer, GeoTrans, Inc., Sterling, VA (Liaison to the RCSS from the SAB's
Environmental Engineering Committee)
Dr. Mitchell J. Small, Departments of Civil and Environmental Engineering and Engineering and
Public Policy, Carnegie Mellon University, Pittsburgh, PA (Liaison to the RCSS from the SAB's
Environmental Engineering Committee)
iii
-------
Federal Experts
Colonel (Dr.) Robert N. Cherry, Jr.,, Office of the Chief of Staff, The Pentagon, Washington,
DC
Dr. Andrew Wallo, III, U.S. Department of Energy, Office of Environmental Guidance,
Washington, DC
Invited Participant
Dr. Robert A. Meek, U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory
Research, Washington, DC
Science Advisory Board Staff
Dr. K. Jack Kooyoomjian, Designated Federal Official, U.S. EPA, Science Advisory Board
(1400F), 401 M Street, SW, Washington, D.C. 20460
Mrs. Diana L. Pozun, Staff Secretary, U.S. EPA, Science Advisory Board (1400F), 401 M
Street, SW, Washington, D.C. 20460
IV
-------
TABLE OF CONTENTS
1. EXECUTIVE SUMMARY 1
1.1 General Findings 1
1.2 Response to Issue #1 2
1.2.1 Findings on Evaluation of Source Terms 2
1.2.2 Findings on Modeling Transport to People 3
1.2.3 Findings on Risk Estimation Methods 5
1.3 Response to Issue #2 (Suitability of Modeled Pathways) 7
1.4 Response to Issue #3 (Suitability of RESRAD) 8
1.5 Findings on Sensitivity and Uncertainty Analyses 8
1.6 Comments on Issues at the Science/Policy Interface 9
2. INTRODUCTION 12
2.1 Overview of EPA Modeling Objectives and Strategy 12
2.2 Charge to the SA 13
2.3 SAB Review Procedure 14
3. SOURCE TERM ISSUES 15
3.1 Introduction 15
3.2 Selection of Representative Reference Sites 17
3.3 Completeness of Source Term Information for Reference Sites 18
3.4 Modeling of Source Term Information for Reference Sites 20
3.5 Development of Distribution Functions for Soil Volume vs. Soil Specific
Activity 22
3.6 Sensitivity and Uncertainty Analysis of Source Term Parameters 26
4. SCENARIOS AND PATHWAYS USED INEPA'S ANALYSIS 29
4.1 Overview of Approach 29
4.2 Land Use 30
4.3 Selection of Pathways for Analysis 31
4.3.1 Pathways Included in the EPA Analysis 31
4.3.2 Pathways Omitted from the EPA Analysis 33
4.3.3 Duration of Exposure 35
5. METHODOLOGIES FOR MODELING TRANSPORT TO PEOPLE 37
5.1 Evaluation of Candidate Models for Pathway Modeling 37
5.1.1 Model Selection Criteria and Process 37
5.1.2 Comments on Selection of Models and Their General Limitations 39
5.1.3 Findings and Recommendations Concerning Model Selection 41
5.2 Pathway Models for Direct External Exposure 42
5.2.1 Modeling Risks to Individuals 43
5.2.1.1 Exposure correction factors 43
v
-------
5.2.1.2 Dose conversion factors and slope factors 43
5.2.1.3 Generic site assumptions and parameter values 44
5.2.1.4 Sensitivity and uncertainty analyses 45
5.2.2 Population Exposures 46
5.2.3 Cleanup Worker Exposures 46
5.2.4 Off-Site Doses Due to Cleanup Activities 46
5.2.5 Findings/Recommendations Concerning the External Radiation Pathway
Model 47
5.3 Pathway Models for Ingestion of Soil and Food 48
5.3.1 Overview Comments 48
5.3.2 Soil Ingestion Rates 49
5.3.3 Soil-to-Plant Transfer Factors for Nutrient-Poor Soils 49
5.3.4 Leaching Rates for Anions 53
5.3.5 Food Consumption Rates 53
5.3.6 Recommendations Concerning the Soil/Food Ingestion Pathway Model 54
5.4 Pathway Model for Inhalation of Particles 55
5.4.1 General Comments 55
5.4.2 Default Parameter Values Used to Model Risks to Individuals 55
5.4.3 Default Parameter Values Used to Model Risks to Populations 56
5.4.4 Updating RESRAD to Include the Revised ICRP Respiratory Tract Models?
5.4.5 Findings and Recommendations Concerning the Particle Inhalation Pathways
5.5 Pathway Model for Inhalation of Indoor Radon and Progeny 60
5.5.1 General Comments 60
5.5.2 Individual Risks 61
5.5.3 Population Risks 62
5.5.4 Technical Feasibility 62
5.5.5 Findings and Recommendations Concerning the Radon Inhalation Pathway53
5.6 Pathway Model for Groundwater Transport and Ingestion 64
5.6.1 Effect of Depth of Contamination on Relative Importance of Groundwater
Pathway 64
5.6.2 Hydrologic Parameter Values for the Generic Site Model 65
5.6.3 Transport Processes and Parameter Values 66
5.6.4 Findings and Recommendations Concerning the Groundwater Pathway 68
5.7 Pathway Model for Surface Water 69
6. METHODOLOGIES FOR ESTIMATING RISK TO INDIVIDUALS AND POPULATIONS
70
6.1 Risk Coefficients 70
6.2 Exposure-to-Dose Calculations 72
6.3 Exposure-to-Risk Calculations 74
6.4 Use of Two Methods for Relating Exposures to Risks 75
6.5 Comparison between TSD and NRC Estimates of Mortality from Exposure to
Residual Radioactive Soils at NRC Licensed Nuclear Facilities 76
VI
-------
6.6 Findings and Recommendations Concerning Risk Estimation Methodologies 77
7. SENSITIVITY AND UNCERTAINTY ANALYSES 80
7.1 Overview of EPA Approach 80
7.2 Sensitivity Analysis for the Generic Site 81
7.3 Uncertainty Analysis for the Generic Site 83
7.4 Sensitivity Analysis for Reference Sites 84
7.5 Qualitative Uncertainty Analyses 85
7.6 Generic versus Site-Specific Uncertainty Analyses 86
7.7 Findings and Recommendations Concerning Sensitivity and Uncertainty Analyses 87
8. ISSUES AT THE SCIENCE/POLICY INTERFACE 89
8.1 Radioactive Site Decontamination and Remediation Issues Excluded from the
EPA Analysis 89
8.2 Other Soil Cleanup Issues Excluded from the EPA Analysis 90
8.3 Choice of Risk Metric 91
8.4 Time Horizons 92
8.5 Choice of Target Risk Levels 93
8.6 Potential Application to NORM 95
8.7 Consistency Between Model Assumptions and Field Sampling Methods 95
8.8 Findings and Recommendations 95
APPENDIX A - COMPARISON OF SOURCE TERM INFORMATION
DEVELOPED BY THE NRC AND BY THE EPA FOR
REFERENCE NUCLEAR FACILITIES A-l
APPENDIX B - DETAILED COMMENTS ON SOURCE TERM INFORMATION USED TO
DEFINE REFERENCE SITE CHARACTERISTICS OF DOE FACILITIES B-l
APPENDIX C - GLOSSARY OF TERMS AND ACRONYMS C-l
LIST OF TEXT TABLES
Table 5-1. Total respiratory tract deposition of 1 micrometer AMAD aerosols
VII
-------
Table 6-1. Ingestion Dose Coefficients to Age 70 Years
LIST OF APPENDIX A TABLES
Table A-l. Comparison of source term parameters for the GEIS and the TSD
Table A-2. Comparison of GEIS and TSD estimates of radiogenic cancer fatalities for the
commercial nuclear power reactor reference site
Table A-3. Comparison of GEIS and TSD estimates of radiogenic cancer fatalities for the
test/research reactor reference site
Table A-4. Comparison of GEIS and TSD estimates of radiogenic cancer fatalities for the rare
earth extraction facility reference site
Table A-5. Comparison of GEIS and TSD estimates of radiogenic cancer fatalities for the
uranium fuel fabrication reference site
LIST OF TEXT FIGURES
Figure 3-1. Schematic of process used by the EPA to determine soil volumes to be remediated
Figure 5-1. Relative contributions of the external dose and ingestion dose for the population of
southern Finland
Figure 5-2. Relative contributions of several contaminated food items to the ingestion dose for
the population of southern Finland
Vlll
-------
1. EXECUTIVE SUMMARY
At the request of the Environmental Protection Agency (EPA), Office of Radiation and
Indoor Air (ORIA), the Science Advisory Board (SAB), through the Radionuclide Cleanup
Standards Subcommittee (RCSS, or the Subcommittee) of the Radiation Advisory Committee
(RAC), has reviewed the Agency's September 1994 draft report titled "Technical Support
Document for the Development of Radionuclide Cleanup Levels for Soil" (hereinafter called the
"TSD"). The Subcommittee has responded to three specific questions posed by ORIA and has
also provided additional comments and suggestions.1
The EPA ORIA staff is to be complimented on its thorough documentation and forthcoming
approach to working with the Subcommittee. Compiling the information necessary to undertake
this technical support document was obviously a formidable task, and ORIA's ability to organize,
present and make use of this scattered information of variable quality is commendable. The
Subcommittee also appreciated ORIA's prompt and thorough responses to many of the
Subcommittee's technical comments in writing and in presentations2 The Subcommittee notes
that an excellent working relationship was established and maintained throughout the review
process.
1.1 General Findings
Because the TSD presented soil cleanup and related cleanup problems separately, a recurring
problem for the Subcommittee was a difficulty in understanding how the Agency ultimately
intended to take into account the interrelationships among its proposed regulatory actions to deal
with soil cleanup, aquifer cleanup, cleanup of structures, disposal of radioactive waste generated
by the cleanup, and recycle and reuse of materials and equipment after cleanup. Interactions
among these five activities is expected to have a substantial effect on the estimated costs and
benefits of the proposed standard, and hence should be discussed in the TSD, at least in a
qualitative sense. Although recognizing that the Agency is currently constrained by statutes or
regulations to address these aspects as isolated problems, in truth, these problems are integrated,
not isolated. The Subcommittee had difficulty commenting on some of the technical aspects of
the TSD for soil cleanup, without having had the benefit of viewing it in the context of its
proposed use. The current fragmentation of the peer review process leads to difficulties in
Numbers associated with each finding or recommendation indicate the section of this review report from which it derived, i.e., discussion
supporting finding 8.8.a is found in section 8.8.
In fact, many of the issues raised in this report were adequately addressed by ORIA while the report was being prepared (e.g., through
memoranda and presentations by ORIA staff members at RCSS meetings. However, although acknowledging this fact, the Subcommittee has retained
its comments on the contents of the document under review.
-------
reviewing estimates of overall exposures, risks and benefits. More scientifically robust estimates
of overall exposure and risk would be derived from an integrated analysis. (8.8.a)
1.2 Response to Issue #1
Issue #1. Are the methodologies used by ORIA in the following areas acceptable for providing a
technical basis for writing a cleanup standard:
a) methodology for evaluating source terms for radioactively contaminated sites?
b) methodology for modeling transport to people? and
c) methodology for estimating risk to individuals and populations?
1.2.1 Findings on Evaluation of Source Terms
The overall approach taken by ORIA in the draft TSD represents a creative and reasonable
approach to addressing risk reduction/cost tradeoffs in the soil cleanup standards for
radionuclides. Specifically, the Office has defined a generic site as a basis for comparing the
behavior of several different environmental pathway models; for carrying out sensitivity and
uncertainty analyses; and for generating preliminary generic tables of cleanup soil concentrations
for different land-use scenarios. A suite of reference facilities was defined by ORIA to represent
the full spectrum of sites to be covered by the proposed rule, and site-specific risk factors were
calculated and employed to develop estimates of total soil remediation volumes and health effects
averted under each of the scenarios, for a range of cleanup goals stated in terms of lifetime cancer
incidence risk or annual radiation dose. Commendable also are ORIA's efforts to collect
information on radioactive sites, to construct the source terms for these sites, to test the
sensitivity of its assumptions, and to analyze the uncertainty of its results. (3.1 .a)
A major concern of the Subcommittee is that source term information appears weak even at
the most well-defined sites, and ORIA had to be quite inventive at some sites. Recognizing that
consistent site-wide data are limited, the Subcommittee commends ORIA for making good use of
the available data and for its continuing efforts to work with other agencies to obtain site-related
data and to ensure appropriate utilization of the information collected. Still needed are
quantitative estimates of the uncertainties for the contaminated soil volumes. Also, for many
reference sites, it appears that the radionuclide selections were not sufficiently inclusive. Most of
these sites contain multiple radionuclides, and different combinations may affect the calculated
waste volumes and allowable concentrations substantially. Specifying the chemical and physical
forms of specific radionuclides in the individual reference sites would be desirable not only for
predicting leaching rates and transport rates in groundwater, but also for those cases in which the
-------
form significantly affects human intake, as in the case of uranium at Fernald or in munitions at
Department of Defense facilities. (3.3 .a)
For example, additional discussion is needed in the TSD to justify the use of selected
reference sites to represent other sites in their categories that have different types of
radionuclides, and different hydrological, geological and meteorological conditions. The
Subcommittee was skeptical of the appropriateness of using the Oak Ridge Reservation to
represent the five major DOE facilities involved with diversified weapons research and
development activities, given that the total volume of contaminated soil in this category is
dominated by the Los Alamos National Laboratory site. (3.2.b)
1.2.2 Findings on Modeling Transport to People
With the exception of a surface water runoff and erosion pathway with subsequent potential
for drinking water and fish consumption exposures, ORIA has generally included in the RESRAD
analysis all the pathways that are likely to be important for radionuclides in inorganic forms.
(Direct and indirect dermal absorption pathways might be important for organic chemicals and
mixed wastes.) (4.3.b)
In the face of constraints of time and resources for revisiting the definitions of the pathways,
ORIA should focus its efforts for any improvement of the definitions (i.e., underlying assumptions
and adopted parameter values) on the dominant pathways, particularly external gamma radiation,
radon inhalation, crop ingestion and ingestion of groundwater (especially the population risk
assumptions). Ingestion of surface water should be investigated further to determine whether it
might be a dominant pathway under any widely prevalent conditions. (4.3.a)
Major comments on specific pathway models are as follows:
a) The external radiation pathway is relatively simple to define. With the possible
exception of exposure time, ORIA's modeling of this pathway is generally
acceptable. Of the three codes considered in detail in the generic base case study,
RESRAD is the most realistic and flexible in its capabilities to account for the
processes that govern exposure of individuals by this pathway. The assumptions and
parameter values for the rural/residential and commercial/industrial scenarios which
affect the dose from direct gamma radiation are reasonable and are, presumably,
incorporated into the RESRAD model used in the analysis. The RESRAD equations
for correcting for depth of contamination, cover thickness and source area are also
reasonable. However, the Subcommittee recommends that ORIA be more explicit
-------
regarding its assumptions about the shielding factor for indoor exposures and be sure
that the parameter value used is consistent with those assumptions. (5.2.a to 5.2.d)
b) With the generic qualification regarding non-uniformity of source term
contamination and population mobility, the soil and food ingestion pathways appear
to be consistent in form with currently available methods although the use of a larger
default value for soil ingestion by an RME individual should be considered. The
Subcommittee recommends an expansion of the discussion of the uncertainty and
variability in the risk estimates generated under the simplified assumptions of these
pathway models. Some of the default parameter values used by ORIA serve to
underpredict exposure by this pathway. For example, growth of crops on
contaminated soils will be a major pathway of concern for 90Sr, 129I, "Tc, and 137Cs
on sites with nutrient-poor and highly acidic soils. Consequently, ORIA should
revise its soil-to-plant transfer factors for those sites at which radionuclide uptake by
plants is expected to be higher than the default value. (5.3.a to 5.3.d)
c) For the particle inhalation pathway, the Subcommittee recommends that the TSD
clearly specify all parameters used in calculations of dose and risk from inhalation of
airborne radioactive dusts for the three identified exposure scenarios for cleanup
workers, and indicate ranges of values as well as those assumed or adopted for the
calculations. Although population density is addressed for all generic sites, the
characteristics of the populations themselves do not seem to be addressed with
respect to age, living behaviors or housing types. In addition, this pathway assumes
that inhalation exposure can be described by the amount of dust expected to be in
the air and that the concentration of radionuclides in the dust is the same as that in
the soil under consideration. For a small site, much of the dust in the air even at the
downwind edge of the site will come from unaffected regions upwind of the site, and
the ORIA assumption could lead to substantial overestimates of risk via this
pathway. Moreover, real atmospheric dust loadings vary substantially with site and
depend on such factors as soil particle size distributions, vegetative cover, humidity,
and precipitation patterns; such variations have not been taken into account in the
modeling of the reference sites. (5.4.a to 5.4.e)
d) The methodology used for estimating population risks from exposure to radon
indoors is reasonable, and the risk conversion factor used in this calculation
previously has been reviewed by the RAC. The total uncertainty associated with
individual and population risk estimates should be presented. Of concern is the
orders of magnitude variability in the radon entry into individual homes and the large
uncertainly that will exist in a single calculated value that is used to estimate the
-------
radon concentration in any home. The methodology used in RESRAD for
estimating individual risks does not account for advective flow of radon into a home,
and this omission can result in underestimation of indoor radon concentrations.
(5.5.ato5.5.h)
e) With regard to the groundwater pathway, the Subcommittee found aspects of some
of the reference site models to be overly conservative, using parameter values that
were internally inconsistent, or inconsistent with values from the literature or known
site characteristics. Examples include hydraulic conductivities and gradients, well
construction designs, groundwater withdrawal rates, and distribution coefficients
(i.e., Kd values) for those long-lived radionuclides for which the chemical form is
highly sensitive to the prevailing oxidation-reduction potential (e.g., "Tc, U, Pu).
Additional work in the form of sensitivity analyses and peer review of parameter
values is required before the importance of the drinking water pathway can be fully
assessed, for those sites at which the contamination does not remain at the soil
surface or within the unsaturated zone throughout the modeled period of time.
(5.6.a to 5.6.c)
f) In the surface water pathway, contributions from particulate-phase radionuclides
associated with soil that is eroded from the site as part of the rainfall-runoff process
can be a significant source of contamination for surface water bodies when large
portions of the watershed areas are contaminated. For these cases, the Agency
should encourage and seek inclusion of soil erosion and particulate-phase
radionuclides in the surface water modules of future models used for assessment of
soil cleanup standards. (5.7.a)
1.2.3 Findings on Risk Estimation Methods
The set of risk coefficients used in the TSD for risk-based standards differs from that used
for dose-based standards, and neither set incorporates the latest changes recommended by the
ICRP or NCRP. This choice can lead to considerable confusion on the part of the reader. While
the Subcommittee understands the difficulties ORIA faced in deciding how to reconcile its
internally generated risk coefficients with those in RESRAD derived from Federal Guidance
Reports Nos. 11 and 12, leaving the problem unresolved can lead the unwary user of the TSD to
erroneous conclusions. Although the Subcommittee is not specifically recommending that ORIA
undertake the magnitude of effort needed to produce a completely consistent document, it does
make the following findings and recommendations:
-------
a) The TSD should include a section specifically describing the dose conversion factors
and slope factors used for each of the types of assessments. While this information
can for the most part be gleaned from the various chapters, it is currently difficult to
sort out. Deviations from published values should be adequately explained and the
methodology for making adjustments described. (6.6.c)
b) The Subcommittee recommends that the TSD include a discussion of the proposed
annual dose limit as it applies to cleanup. The selection of appropriate assumptions
and parameter values for modeling RME exposures will depend upon whether the
standard is developed as an annual dose limit for "any" member of the public or as an
average annual measure of compliance with the long-term (lifetime) individual risk
limit. If the annual limit is intended to serve as a surrogate for the lifetime risk limit,
then the approach used in the TSD is generally acceptable and the importance of the
recommendation for age-specific factors presented below is diminished. If, however,
the goal of the standard is to assure that every individual is explicitly protected to the
annual limit, then it will be important to consider the use of age-dependent dose
factors in the TSD analysis. (6.6.a)
c) The Subcommittee recommends that EPA give consideration to adoption of the
recent recommendations of the ICRP and NCRP that provide updated metabolic and
dosimetric models and approaches for calculating age-dependent doses for the
inhalation and ingestion of radionuclides for all members of the public. Adoption of
the ICRP and NCRP approaches offers a procedural advantage to the EPA in that
they have been extensively peer reviewed and are widely accepted. Use of these
approaches would likely increase the technical acceptance of the standards by the
scientific community. If alternative approaches are used in the TSD, then ORIA
should explicitly explain its methodology and justify its reasons for departures from
ICRP and NCRP recommendations. (6.6.b)
d) The Subcommittee recognizes that, by Presidential directive in 1987, EPA should
use the exposure-to-dose conversion factors tabulated in Federal Guidance Reports
Nos. 11 and 12 and their subsequent revisions. This guidance is based on a "linear
extrapolation to zero" exposure-to-dose relationship from observed, but much
higher, dose-effect studies. As noted in the TSD, the scientific community has been
unable to come to a consensus on issues such as the possibility of threshold doses
below which no effects occur, the validity of extrapolating curves from known high
exposure effects to zero, and the possibility of hormesis (the concept that small
doses of radiation may be beneficial to humans). The Subcommittee recommends
that the uncertainties associated with extending risk analyses to very low radiation
-------
exposures in the absence of scientific consensus be reflected in the presentation of
the final results. (6.6.d)
e) A comparison was made of EPA and NRC estimates of contaminated soil volumes
and cancer risks associated with that soil, for commercial nuclear power reactors,
research reactors, rare earth extraction facilities, and uranium fuel fabrication
facilities. The EPA estimates of soil volume were significantly larger than NRC
estimates, by factors up to 100. For three of these four types of facilities, the
estimates of the number of fatalities associated with contaminated soil were generally
comparable, within a factor often of each other. The exception was the case of the
uranium fuel fabrication facility, for which mortality estimates in the TSD were
higher than those calculated by the NRC by factors ranging from 2.5 to 100. The
reasons for this difference are not clear and should be investigated by ORIA. (6.6.h)
1.3 Response to Issue #2 (Suitability of Modeled Pathways)
Issue #2. Are the assumptions and modeled pathways reasonable and suitable for assessing risk at
radioactively contaminated sites:
a) for the combined residential / agricultural land use scenario? and
b) for the industrial / commercial scenario?
The Subcommittee is satisfied that ORIA's choice of an on-site residential scenario and a
commercial-industrial scenario for thorough analysis is adequate for estimation of reasonable
maximum individual exposures and risks from sites that have been cleaned up to a specified level
of contamination. While other scenarios are also plausible, they are not likely to produce
substantially higher estimates of maximum individual risks. However, the Subcommittee
recommends that ORIA consider adding qualitative discussions of the likely magnitude of
maximum individual risks related to recreational and off-site residential neighbor scenarios.
(4.2.a)
1.4 Response to Issue #3 (Suitability of RESRAD)
Issue #3. Is RESRAD v. 5.19 suitable for modeling radiation risks to individuals at radioactively
contaminated sites?
The Subcommittee finds that the initial screening and selection of candidate models for
further evaluation as reported in the TSD was conducted in a reasonable, sound and thorough
-------
manner, using appropriate criteria for model selection. Furthermore, the Subcommittee concurs
with ORIA's decision in its selection of RESRAD as a reasonable transport model code for use at
the current time. However, while it incorporates some conservative assumptions in its
formulations, RESRAD itself may not necessarily provide conservative risk estimates if
inappropriate parameter values are selected for the modeling input. (5.1 .a, 5.1 .b)
Because the Subcommittee has not evaluated the default values in the RESRAD code, nor
the full parameter set used for each reference site, it is unable to fully assess the extent to which
the model results can be considered to be conservative or bounding estimates of the true health
effects associated with each level of cleanup. From a cursory review, several of the transport
model parameters appear to inconsistent with values from the literature or known site
characteristics; hence, the Subcommittee strongly urges ORIA to obtain a thorough peer review
of the default and site-specific parameters used in the transport modeling. As more information
becomes available through site applications and studies, the Agency should re-evaluate this issue
to determine whether RESRAD—including its default parameters—should be modified or
replaced so as to ensure the maintenance of an appropriate balance between realistic prediction
and reasonably conservative protection of public health. (5. l.c)
1.5 Findings on Sensitivity and Uncertainty Analyses
Overall, the Subcommittee commends ORIA on conducting sensitivity and uncertainty
analyses for the risk and soil volume calculations and on its thoughtful discussions of the
important assumptions and parameter value choices. However, because of limitations of the
analyses, the reader is left without a sound appreciation for the magnitude of the overall
uncertainties in soil volume requiring remediation and cancers averted via remediation. The
generic sensitivity and uncertainly analyses are limited by the choice of parameter values that were
varied, the need to select nominal values for all other values when varying one parameter, and the
lack of analysis regarding model uncertainties. Although quantitative sensitivity analyses of the
reference sites are conducted with respect to policy choices such as the target risk level, time
horizons, and land-use scenarios, only a qualitative discussion of scientific uncertainties is offered.
Thus, the true but unknown values for cancers averted and soil volumes to be remediated at each
RME risk level may be quite different from those presented in the report. Given that policy
decisions are to be made using the numerical results of the TSD models as one criterion, it is
critical that the quantitative uncertainties about those results be disclosed and emphasized in the
presentation. (7.7.a)
The importance of uncertainty analyses and communication of the results of those analyses
to environmental risk managers was underscored by the EPA Administrator in a recent
memorandum transmitting the EPA Risk Characterization Program to EPA staff (EPA, 1995).
-------
Consistent with the spirit of that guidance, the Subcommittee recommends that ORIA improve its
risk assessment and characterization in the TSD in the following areas:
a) Discussion in the TSD should clarify the purpose of the risk calculations: are these
to be screening or bounding estimates, high end estimates (e.g., above the 90th or
95th percentile), or central tendency estimates of the true value (risk)? The
objective then should guide the selection of appropriate parameter values used in
pathway modeling. (7.7.b)
b) ORIA should, at a minimum, discuss qualitatively any biases in its estimates of
cancers averted as well as biases in the soil volumes to be remediated. The positive
correlation of these biases with any in the individual risk estimates should also be
mentioned. (7.7.c)
c) To the extent possible, ORIA should provide best estimates of cancers averted total
and fatal) and soil volumes to be remediated for the various proposed standards, as
well as uncertainty ranges about each of those estimates, in addition to the nominal
values currently provided. (7.7.d)
d) In the future, when EPA evaluates the need for cleanup at a specific site in response
to the final cleanup standard, it should not only allow but also explicitly encourage
quantitative uncertainty analyses in the site assessment. The level of detail in such
uncertainly analyses should be commensurate with the stakes (potential for risk
reduction and cleanup costs) revealed in a screening analysis. (7.7.e)
1.6 Comments on Issues at the Science/Policy Interface
The Subcommittee's charge was to review scientific and technical aspects of the
methodology used by ORIA to model radiation risks to individuals at radioactively contaminated
sites. The results of the risk assessments will be used by the Agency in making policy decisions
on cleanup levels for soil. The review by the Subcommittee primarily focused on the evaluation
of source terms, environmental transport, and estimation of risks. However, in the course of this
work, some issues were identified that were outside the scope of the Subcommittee's charge. The
following comments on these issues, which involve the interface of science and policy, are
provided for consideration by the Agency.
Risk metric. EPA's decision to use lifetime risk corresponding to reasonable maximum
exposure as a risk metric for the proposed standard is appropriate. Although another metric (such
as the population risk attributable to a facility) could have been used, the formulation of the
-------
standard is consistent with other EPA waste management strategies. In its comparison of the
risks of remedial activities to those avoided through the remediation, EPA uses population risk
(cancer fatalities caused vs. cancer cases averted). While this choice is also reasonable, it
bypasses the issue of potentially higher individual risks to some remediation workers. The
Subcommittee recommends that EPA be more explicit about the reasons it chose to use an
individual risk metric for the standard and a population risk metric for describing the costs and
benefits of the standard. A brief mention of the potential for higher-than-average individual risks
to remediation workers is also recommended. (8.8.c, 8.8.d)
Time horizons. By estimating risks over time horizons of 100, 1,000, and 10,000 years,
EPA has also provided the cleanup standard decision-makers with a range of options for
evaluating the benefits of the standard. However, the Subcommittee is concerned that estimates
for risks incurred more than 100 years in the future are highly speculative given the uncertainties
about the rate and direction of change in medical science and other technological and social arenas
which could either reduce or accentuate the relative importance of cancer as a health risk in the
future. The Subcommittee therefore recommends that EPA emphasize the increased uncertainty
of its estimates for 1,000 and 10,000 years in comparison with those for 100 years. (8.8.e)
Inconsistency with existing EPA regulations. The analyses of risk associated with 226Ra
suggest that cleanup levels below 10"3 are probably not feasible because of the natural variation in
the abundance of this isotope. For comparison, the cleanup standard for 22(Ra at uranium mill
tailings sites is 5 pCi/g above background in the top 15 cm of soil and 15 pCi/g above background
in deeper layers (UMTRCA regulation, 40 CFR 192), corresponding to lifetime risks of
approximately 10"2. This illustrates a lack of consistency between the lower risk levels being
considered in the TSD and those corresponding to existing regulations dealing with radionuclide
cleanup, such as the UMTRCA regulation. (8.8.f)
Potential applicability to Naturally Occurring Radioactive Material (NORM). The
Subcommittee is not offering an opinion on the advisability of using the proposed cleanup
standard as an ARAR (Applicable or Relevant and Appropriate Regulation) or as the precedent
for any NORM regulations that might be proposed in the future. However, these possibilities
should be noted in the TSD, and the costs and benefits of any such actions should be discussed in
the Regulatory Impact Analysis. The Subcommittee notes that the TSD does not provide the
technical basis for cleanup criteria for NORM because the analyses presented in the TSD do not
address cancers averted and volumes of soil affected for sources of NORM. Also, the feasibility
issue noted in the preceding recommendation would be pertinent because 226Ra is one of the
principal NORM radionuclides. (8.8.g)
10
-------
11
-------
2. INTRODUCTION
2.1 Overview of EPA Modeling Objectives and Strategy
The EPA is proposing regulations that would set standards for radiation doses from
contaminated soil prior to release of the land for unrestricted public use. The proposed
regulations are meant to apply to contaminated soil remaining after cleanup of high-risk (> 10"2
lifetime cancer incidence risk level) contaminated soil and facilities. As such, the proposed
regulations could affect large volumes of lower-risk (< 10"6 to 10"2 risk level) contaminated soils
at sites presently under the control of a Federal Agency and sites licensed by the U.S. Nuclear
Regulatory Commission (NRC) or by an NRC Agreement State, that are to be released from
those licenses or control. The regulations also have the potential to affect large volumes of soil at
mining sites if the proposed cleanup standards extend to sites containing naturally-occurring
radioactive materials (NORM).
The major objectives of the EPA report are to: (a) estimate the volume of soil that may
require remediation at sites that fall within the scope of the proposed rule, and (b) estimate the
number of potential radiogenic cancers averted as a result of the remediation of contaminated soil.
The approach taken by EPA is to consider typical scenarios and pathways through which
individuals and populations may be exposed. Then, using mathematical models, cleanup levels are
determined that will yield acceptable predicted risks. The modeling process consisted of
development of selection criteria. Using these criteria, a model evaluation and selection was
performed which resulted in the selection of the following models: RESRAD, PRESTO, and
RAGS/HHEM. These models were then compared and tested using a hypothetical generic base
case.
As part of the generic base case study, a limited sensitivity analysis was performed using
RESRAD Version 5.19. The five parameters that were varied included area of the contaminated
zone, thickness of the contaminated zone, infiltration rate, distribution coefficients, and thickness
of the unsaturated zone. Also, as part of the generic base case study, a limited uncertainty
analysis was performed using a modified version of RAGS/HHEM Part B models and Monte
Carlo techniques. The base case study included evaluation of both individual impacts and impacts
to an aggregate population. Individual impacts were estimated using RESRAD while population
impacts were estimated using a population model developed by ORIA for the TSD analyses.
12
-------
Based on results of the model comparisons for the generic base case site assessment, the
RESRAD code was selected to analyze the set of reference sites. A result of the reference site
analysis was the development of risk factors for each radionuclide. From the risk factors, one can
estimate the volumes of soil requiring remediation in order to achieve the proposed standard's risk
level. To estimate the volume of soil that may require remediation, the approach taken is to
calculate the risk resulting from a reasonable maximum exposure (RME) to an individual either on
the site and consuming locally-produced agricultural products, or working on the site in an
industrial occupation. The risks are calculated for these individuals based on radioactive
contamination levels remaining after remediation of the site. To calculate the RME risk to an
individual at a given site, site data and generic data are used in conjunction with the risk
assessment code RESRAD.
To estimate the numbers of potential fatal radiogenic cancers averted as a result of
remediation of contaminated soil, risk levels calculated for individual exposures are combined
with various population scenarios. The resulting numbers of cancers averted are given in terms of
various individual risk levels ranging from < 10"6 to 10"2, as well as for various dose limits ranging
from 0.1 to 100 mrem/yr, together with remediated soil volumes corresponding to each of these
risk and dose levels.
2.2 Charge to the SAB
At the request of the Environmental Protection Agency (EPA), Office of Radiation and
Indoor Air (ORIA), the Science Advisory Board (SAB), through the Radionuclide Cleanup
Standards Subcommittee (RCSS, or the Subcommittee) of the Radiation Advisory Committee
(RAC), has reviewed the Agency's September 1994 draft report titled "Technical Support
Document for the Development of Radionuclide Cleanup Levels for Soil" (hereinafter called the
"TSD"). The RSCC responded to the following three questions posed by EPA and has also
provided additional comments and suggestions for its consideration.
Issue 1. Are the methodologies used by ORIA in the following areas acceptable for
providing a technical basis for writing a cleanup standard: (a) methodology
for evaluating source terms for radioactively contaminated sites, (b)
methodology for modeling transport to people, and (c) methodology for
estimating risk to individuals and populations?
Issue 2. Are the assumptions and modeled pathways reasonable and suitable for
assessing risk at radioactively contaminated sites, (a) for the combined
residential / agricultural land use scenario, and (b) for the industrial and
commercial scenario?
13
-------
Issue 3. Is RESRAD v. 5.19 suitable for modeling radiation risks to individuals at
radioactively contaminated sites?
2.3 SAB Review Procedure
The primary review document is the Agency's September 1994 draft report titled
"Technical Support Document for the Development of Radionuclide Cleanup Levels for Soil"
(TSD), including its Appendices A-O (EPA, 1994a). The Subcommittee's review also benefited
from the extensive documentation voluntarily provided by ORIA in support of the TSD.
The RCSS met on October 27-28, 1994, January 26-27, 1995, and May 23-24, 1995, at
which time it was briefed by ORIA staff on specific aspects of the TSD, including revisions in
progress since issuance of the September 1994 TSD draft. In addition, the RCSS conducted a
teleconference on March 27, 1994.
The RCSS wishes to compliment the ORIA staff on its thorough documentation and
forthcoming and candid approach to working with the RAC's Subcommittee. Compiling the
information necessary to undertake this technical support document was obviously a formidable
task, and ORIA's ability to organize, present and make use of this scattered information of
variable quality is commendable. The Subcommittee also appreciated ORIA's prompt and
thorough responses to many of the Subcommittee's technical comments in writing and in
presentations.3 The Subcommittee notes that an excellent working relationship was established
and maintained throughout the review process.
In fact, many of the issues raised in this report were addressed by ORIA while the report was being prepared (e.g., memoranda to RCSS
members and various presentations by EPA staff members at RCSS meetings. However, although acknowledging this fact, the Subcommittee has
retained its comments on the contents of the document under review.
14
-------
3. SOURCE TERM ISSUES
3.1 Introduction
To provide a technical and scientific basis for the tradeoffs between risk reduction and
cost in the proposed soil cleanup standards for radionuclides, ORIA defined a generic site,
reference sites, and several land-use scenarios. ORIA then coupled this information with a
calculational model in order to estimate the total soil volume that would require remediation
under each of the scenarios, for a range of cleanup goals stated in terms of soil specific activity
required to achieve reductions to specific dose levels and lifetime cancer risk levels. Using the
generic site , three calculational models were compared; of these, the RESRAD model was
chosen to calculate site-specific dose or risk factors from which soil specific activity and clean-up
volume were derived for each reference site and a particular dose or lifetime cancer risk level.
Additional information was generated for cancer deaths averted by the remedial action, cancer
risks to workers undertaking the cleanup activities, and radiation doses off-site using a population
effects model devised by ORIA for this purpose.
Calculations for the generic site provide dose factors in mrem/yr per pCi/g for each
radionuclide of interest so that the cleanup may be performed in terms of soil specific activity.
Furthermore, a sensitivity analysis was performed using the generic site to address some of the
issues regarding uncertainties in the choice of the dimensions, and the physicochemical
characteristics of the site that relate to geologic processes such as radionuclide transport and
leaching. Calculations for the reference sites, selected to represent the variety of facilities that
have radioactively contaminated soils, also provide dose factors in mrem/yr per pCi/g for each
radionuclide of interest. These suggest the soil volumes at specified dose or risk levels to guide
the policy choice that cleanup to such levels would be a reasonable requirement to ensure the
public's health and safety if the site were to be released for unrestricted or restricted use. The
appropriateness of the generic site is confirmed if dose or risk factors at the reference sites are
comparable. The appropriateness of the reference site approach is confirmed if the resultant
calculations capture the universe of sites to be remediated under the proposed rule.
The source term for the generic site is contaminated soil at a density of 1.5 g/cm3 in an
area of 100 m x 100 m, from a flat surface to a depth of 2.0 m. Uniform water flow rates and
distribution coefficients for radionuclides are specified. All radionuclides, natural or manmade,
that could be expected to be present at a site to be decontaminated or decommissioned are
considered if they have a half life longer than about 0.5 year or are short-lived progeny of such
radionuclides. Each radionuclide is assumed to be uniformly distributed throughout the soil at a
concentration of 1 pCi/g. The main questions about the generic site that relate to the source term
15
-------
concern the following: (a) site dimensions, (b) selection of radionuclides to be included in the
model exercise, and (c) influence of the chemical and physical form of the radionuclides on their
transport characteristics.
The source term for each individual reference site consists of a zone of specified area and
thickness contaminated with one or more radionuclides; these radionuclides are subsequently
redistributed throughout the environment over time as a function of various hydrologic parameter
values and radionuclide distribution coefficients (TSD Tables 4-5 to 4-8). The radionuclides that
have been identified and measured by soil monitoring, and that are considered to pose possible
health risks, are assumed to be distributed uniformly throughout the soil in the vertical direction to
a specified depth. The depth of contamination is either calculated, taken from the literature, or
assumed to be 5 cm. Horizontally, the radionuclides are assumed to be distributed nonuniformly
in accord with isopleths constructed on the basis of monitoring overflights or surface samples.
The reference site is assumed to be a selected portion of an actual site that excludes structures and
waste disposal areas, as well as locations considered uncontaminated. The resulting soil volumes
are multiplied (when necessary) by a weighting factor so that the reference site represents all the
facilities in that category.
The Subcommittee presents the following findings concerning the overall approach used
by ORIA to define the source term:
3.1.a Finding The overall approach taken by ORIA in the draft TSD represents a
creative and reasonable approach to addressing risk reduction/cost tradeoffs in the
soil cleanup standards for radionuclides. Specifically, ORIA has defined a generic
site as a basis for comparing the behavior of several different environmental
pathway models; for carrying out sensitivity and uncertainty analyses; and for
generating preliminary generic tables of cleanup soil concentrations for different
land-use scenarios. A suite of reference facilities was defined by ORIA to
represent the full spectrum of sites to be covered by the proposed rule, and site-
specific risk factors were calculated and employed to develop estimates of total
soil remediation volumes and health effects averted under each of the scenarios, for
a range of cleanup goals stated in terms of lifetime cancer incidence risk or annual
radiation dose. ORIA's efforts to collect information on radioactive sites, to
construct the source terms for these sites, to test the sensitivity of its assumptions
and to analyze the uncertainty of its results are also commendable.
3.1.b Finding. With regard to the generic site source term, the list of radionuclides in
Table 2.6 of the TSD appears to be complete and the associated data correct.
16
-------
Evaluation of the TSD concerning the source terms for individual reference sites is
complex due to the number of sites, the differences among them, and the assumptions made by the
EPA in its analysis. Comments are presented below in five sections: selection of representative
reference sites for each category (section 3.2), extent and completeness of the information
obtained for reference sites (3.3), acceptability of modeling of source term information for
reference sites (3.4), development of the distribution functions for soil volume vs. activity (3.5),
and adequacy of the sensitivity and uncertainty analyses of source term parameters (3.6).
3.2 Selection of Representative Reference Sites
EPA conducted additional analyses on individual reference sites in order to: (a) validate
risk factors (e.g., risk per residual pCi/g) calculated for the generic site, and (b) identify sites for
which risk factors calculated for the generic site would be inappropriate. Facilities were selected
by EPA to represent the full spectrum of sites in order to derive estimates of the soil volume that
would require remediation under each of the scenarios, for a range of cleanup goals stated in
terms of dose or lifetime cancer risk. EPA correctly recognized that analyses of actual sites
would provide more information than analyses of "representative sites," but that data and
resources would be insufficient to undertake such a major task.
Although approving of the overall approach of defining the representative sites, the
Subcommittee identified two general concerns about the selection process: a possible need to
expand the number of reference sites to include extra-territorial sites, and skepticism about the
selection of atypical sites to represent specific categories.
3.2.a Recommendation. The Subcommittee noted that there was some ambiguity
concerning the "universe" of sites to which the proposed regulations would apply.
Specifically, the TSD should state whether the regulations will apply to Territories,
trusteeships, and foreign bases of the U.S., as well as to cleanups on U.S. and
foreign soils from accidents or incidents of U.S. Government responsibility (e.g.,
see Mandelker, 1990). To the extent that the regulations will apply to such cases,
additional information is needed on the extent of radiological contamination at
U.S. Government-controlled sites outside the fifty States. The volume and activity
of the radionuclide source term can also be significantly impacted depending upon
whether or not formerly decontaminated sites are "grandfathered" from the
proposed cleanup standard.
3.2.b Recommendation. The Subcommittee was skeptical of the appropriateness of
using the Oak Ridge Reservation to represent the five major DOE facilities
involved with diversified weapons research and development activities (Reference
17
-------
Site VI), given that the total volume of contaminated soil in this category is
dominated by the Los Alamos National Laboratory (LANL) site insofar at it alone
accounts for 93% of the total volume. In fact, according to the DOE Integrated
Data Base (DDE), LANL is second only to the Nevada Test Site in terms of total
volume of contaminated soil among all DOE facilities and National Laboratories
(TSD Table 1-3, p. 1-16). In general, additional discussion is needed in the TSD
to justify the use of selected reference sites to represent other sites in their
categories that have different types of radionuclides, and different hydrological,
geological and meteorological conditions.
3.3 Completeness of Source Term Information for Reference Sites
Information regarding the types, concentrations, speciation and distributions of
radionuclides in soils appears to be incomplete for many of the reference sites (see Appendices A
and B of this report). Some specific areas of concern are as follows:
a) A significant weakness in the reference site analysis is the heavy dependence on
aerial survey data in the projections of radionuclide concentrations in soil and depth
of contamination. The aerial survey data are certainly valuable, particularly for sites
where deposition is the major pathway for contamination. The estimates of
concentrations and volumes of contaminated soil are dependent on assumptions
about the depth and extent of contamination.
b) Consideration of subsurface soil contamination at reference sites is important in
assessing cost and projected number of fatalities averted due to clean-up activities.
For many of the reference sites, the contamination is assumed to be near the surface,
in the top 5 to 15 cm. This assumption is reasonable for sites where the majority of
the contamination is due to windblown materials but is not appropriate for sites such
as Hanford where the contamination in some areas is at depths greater than 15 cm.
c) The analysis of several of the more complex reference sites such as Hanford included
only one radionuclide when, in fact, several radionuclides may contribute to the risk
associated with the site. For example, although EPA recognized that the Hanford
site included radionuclides other than 137Cs, not all were listed in the analysis (TSD,
p. 4-39).
To a large extent, incomplete source-term inventories are understandable and indeed
unavoidable, given the extensive data bases to be searched for 16 sites in a brief period, and the
18
-------
limited fraction of data collected to delineate soil cleanup volume compared to the bulk of the
data collected for the distinctly different purposes of radiation protection or compliance
monitoring at the active sites. It is recognized that a complete summary of environmental
radionuclide levels and distributions is not necessary to achieve the purpose of the TSD.
Nonetheless, efforts to improve the data set appears warranted for some of the sites, most notably
Hanford Reservation, Fernald Environmental Management Project, Savannah River Site, and Oak
Ridge Reservation (Reference Sites I, II, V, and VI, respectively).4 The following actions are
suggested to EPA regarding the TSD:
3.3.a Recommendation. A major concern of the Subcommittee is that source term
information appears weak even at the most well-defined sites, and ORIA had to be
quite inventive at some sites. Recognizing that consistent site-wide data are
limited, the Subcommittee commends ORIA for making good use of the available
data and for its continuing efforts to work with other agencies to obtain site-
related data and to ensure appropriate utilization of the information collected.
However, quantitative estimates of the uncertainties for the contaminated soil
volumes are still needed. The TSD should utilize more of the information available
for the reference sites, including aerial radiological surveys and reports describing
the presence of subsurface heterogeneities, by obtaining the cooperation of
knowledgeable site staff5 For many reference sites, it appears that the
radionuclide selections were not sufficiently inclusive, particularly for those cases
in which aerial survey data were the primary source for the source term. Most of
these sites contain multiple radionuclides, and different combinations may
drastically affect the calculated waste volumes (by affecting assumptions about
depth of contamination) and allowable concentrations. Specifying the chemical
and physical forms of specific radionuclides in the individual reference sites would
be desirable not only for predicting leaching rates and transport rates in
groundwater (see section 5.6.3 of this report), but also for those cases in which the
form significantly affects human intake, as in the case of uranium at Fernald or in
munitions at Department of Defense facilities. Some radionuclides may have been
removed from consideration because they are judged to be of minor impact, or
because insufficient information is available. The TSD should present the criteria
for such decisions, with more detailed explanations for specific reference sites
The Agency's subsequent approach to resolving this issue was described to the Subcommittee in Wolbarst (1995b;c).
Responses by Agency staff indicate that pertinent data bases are being expanded (e.g., Wolbarst, 1995b,c).
19
-------
3.3.b Recommendation. The values of remediation area and soil volume at many sites
are not explicitly calculated, but are said to be obtained based on overflight maps
for the former, and from extrapolation for the latter. More than one value is given
for some sites. Clarification of the calculation procedure and resolution of the
discrepancies are needed.
3.3.c Recommendation. The TSD should make clear for each reference site that the
available data are adequate only to achieve the purposes of the calculation, i.e.,
estimating soil volume for remediation and providing a basis for comparing the
factor of risk per radionuclide concentration in soil by various scenarios with the
generic site risk factors.
3.3.d Recommendation. The TSD should emphasize in the introduction or conclusion
that this document would be inappropriate to use as a basis for identifying priority
cleanup sites or for comparing with site decontamination and decommissioning
program proposals because of the restricted scope of this document, i.e., additional
monitoring would be needed for either of these two applications.
3.4 Modeling of Source Term Information for Reference Sites
By creating an artificial reference site, EPA weakens the reliability of their presentation,
the reader's confidence in their conclusions, and future uses of the information in the TSD.
Concerns are cited in Appendices A and B about creating composite sites and using radionuclide
data taken from other sites. Such manipulations introduce the opinions of Agency staff and
distance reference site values from reality. Nonetheless, the Subcommittee recognizes the time
and data constraints faced by EPA in developing the TSD, and understands the use of the
reference site approach as generally providing a fair representation of the source term.
The main concerns among modeling assumptions for the source term are the selected
thickness of the radioactively contaminated layer, the assumed uniformity of soil type and
hydrology, and the thickness of the contaminated layer at large sites. Specific concerns are as
follows:
a) Site-specific heterogeneities which could significantly affect the risk calculations,
such as subsurface lenses and non-homogeneous concentrations of radioisotopes,
have been ignored at some sites, notably Reference Sites II, IV, XVIII, and XX
(based on Fernald, Weldon Spring, Cintichem, and the Apollo plant, respectively).
20
-------
b) The depth of contamination in soils appeared to be arbitrarily assigned and, in some
cases such as Reference Site II, contrary to actual known situations. A heavy
reliance on radiological aerial survey results for modeling the horizontal distribution
of radionuclides is understandable because of its convenience in providing isopleths
to define soil zones of specific concentrations, but can be misleading because only
near-surface originating gamma rays are measured. There appears to have been an
over-reliance on aerial survey data to define radionuclide distributions for Reference
Sites I, III and V, for which existing data from other information sources cast
doubt on the TSD assumptions about the radionuclides present, and their aerial and
vertical distributions.
c) The TSD specifically states that site characterization was limited to areas where soil
was contaminated by spills, leaks, overflow or runoff from waste, or by windblown
deposition (TSD, p. 4-3), on the assumption that buried waste would be remediated
according to other sets of regulations (TSD, p. 4-3). Consequently, the isopleths
need to be refined at many sites to address contaminated buildings, tanks, and waste
disposal sites that are excluded from the analysis.
d) While it is recognized that the proposed cleanup standard would apply only to
cleanup of soils at sites to be released to unrestricted public use, in many cases
buildings and equipment are underlain by contaminated soils. It is unclear how the
TSD deals with this aspect of the source term inventory.
Details of these and other concerns of Subcommittee members are provided in Appendices
A and B of this report. The following recommendations are suggested for consideration by EPA:
3.4.a Recommendation. Separate modeling may be warranted for those sites with
characteristics related to radionuclides, dimensions, or transport factors that are
very different from the reference site intended to represent them. Differences
found between the reference site and these individual case studies with respect to
risk factor per radionuclide concentration or soil cleanup volume should be
reported. (See recommendation 3.2.b)
3.4.b Recommendation. Where source-term information for modeling risk factors per
soil concentration and soil cleanup volumes is speculative, notably the thickness of
contaminated soil and the variability of the radionuclide concentration, attempts
should be made to improve the data base. If development of a more defensible
21
-------
source term is not feasible, the TSD should acknowledge this weak data-base and
discuss its ramifications to the entire TSD presentation.
3.4.c Recommendation. A frequently repeated statement in the TSD is that a reference
site only partially represents the named site, without providing a detailed map that
shows waste repositories and structures. The dimensions of the reference site for
soil remediation should be specified as meticulously as possible. The possible impact
of adjoining areas that contain high- or low-level waste repositories, contaminated
structures, or contaminated groundwater should be acknowledged and discussed.
Isopleths defining soil of a specific concentration need to be better defined at many
sites that include buildings, tanks, and waste-disposal sites. The Agency should
make clear that the contaminated soil defined by the radionuclide concentration
isopleths for a given reference site is completely destined for soil cleanup, or that a
specified fraction will be removed from consideration in its analysis of the reference
site. Alternatively, EPA should indicate by estimates of uncertainty, why such
boundaries can remain vague without materially affecting the results.
3.5 Development of Distribution Functions for Soil Volume vs. Soil Specific Activity
This section examines how the source term was developed and was used by EPA to
estimate the volumes of soil that will need to be remediated. The process used by the Agency is
summarized in Figure 3-1 which shows that the source term enters the calculation in two places:
a) in the plot of soil volume as a function of the specific activity of the soil, and b) in the
description of the contaminated zone used as input for the computerized environmental pathway
exposure models. These will be discussed separately.
22
-------
The development of reference sites in chapter 4 of the TSD has one major goal: to
describe soil volume vs. soil specific activity (SSA) distributions or functions that will allow the
conversion of an SSA (estimated from risk factors and risk level) into a volume of soil to be
remediated. A second goal is to describe how the physical parameters used as input for the risk
factor calculations (TSD Table 4-6) were estimated.
The mathematical representations described above were constructed from essentially two
types of data: aerial surveys of gamma-emitting radionuclides, and soil sampling. Calculations of
contaminated soil volumes assumed a fixed depth for the contaminated zone thickness, in which
the radionuclide distribution was assumed to be continuous and uniform. Given the results
described in Chapter 7 of the RCSS report, these assumptions mean that the accuracy of the
estimated volumes of soil to be remediated is directly proportional to the accuracy with which the
contaminated zone thickness is known.
The method used to convert the aerial survey data to a planar area (as opposed to a
surface area) was a hand-drawn graphical conversion of the radioactivity isopleths to SSA
isopleths followed by manual "pixel" counting. The resolution of the "pixels" of these converted
maps is not discussed in the report, nor is the resolution of the aerial surveys. Therefore, the
quality, precision, or accuracy of the data used cannot be properly evaluated. Furthermore, it is
not possible to evaluate any degradation or loss of information , due to the methods used. It is
surprising that the area calculations were done in this way, and that digital imaging techniques
were not applied to this problem. The description of the sources of error for these analyses is
highly speculative, confusing, and brings out problems that are not necessarily relevant to the
discussion at hand. The section on errors (TSD, pp 4-12, 4-13) for the aerial surveys is neither
useful nor informative, given that it is impossible to gauge the magnitude of the uncertainties in
the data and area conversion technique from the information presented.
Several methods were then used to build the volume vs. SSA curves:
a) curve-fitting of data points with various mathematical functions, followed by pair-wise
curve-fitting with the function selected for interpolation;
b) distributions obtained from various sources;
c) creation of a distribution using averaged distribution parameters from other reference
sites; and
d) extrapolation to "zero activity above background" using curve fitting from the last pair
of data points.
23
-------
These mathematical formulae, and the methods used to derive them, are not presented in any
detail (except for reference site II) in the TSD but were evaluated based on a technical addendum
provided by Agency staff to the Subcommittee (Wolbarst, 1995a). The formula selection was
done by least-squares two-parameter curve-fitting to linear, logarithmic, exponential, and power
functions, followed by calculation of the Pearson product-moment correlation coefficient, and
selection of the function with the highest absolute value for the correlation coefficient. These
formulae, rather than the graphical representation of the data presented throughout TSD Chapter
4, were used to calculate the volumes of soil to be remediated.
The choice of exponential and power functions seems justified as providing the best fit
although the high degree of correlation obtained by the linear regression procedure employed (the
data were transformed log-log or log-linear for the curve-fitting) should not be taken to imply any
degree of accuracy for the data, but rather as an indication of the precision of the computation
(Glantz and Slinker, 1990). In general, the methods used are reasonable and sound, with the
exception of the distribution constructed for Reference Site V (which is based on the Savannah
River Site). The creation of a distribution for Reference Site V using distribution parameters from
another reference site is a guess at best, given the absence of site-specific data.
A critical parameter for the determination of the volume of soil to be remediated, and for
the reference site-specific risk factors, is the thickness of the contaminated zone. In most cases, it
is assumed to be 5 cm because this is the minimum depth that can be excavated and removed.
Other depth values are estimated based on DOE or NRC volumes divided by the total planar area
of the site. It is impossible to tell whether the assumed and/or calculated depths overestimate or
underestimate the true but unknown values based on the information presented in the report. The
Subcommittee also notes that some estimates developed by the NRC (see Appendix A) of soil
volumes to be remediated, differ from those in the TSD by as much as two orders of magnitude.
The source term input parameters used for the risk factor calculations are presented in
TSD Table 4-6. The uncertainties involved in the areas or contaminated zone thicknesses
presented cannot be evaluated. Although the TSD states that the volumes represent a
conservative estimate or upper bound, this statement is not justified by the information or data
presented. The degree of conservatism claimed for these parameters seems to be more a matter
of belief or opinion of the Agency staff, rather than substantiable scientific fact based on the data
presented.
The Subcommittee makes the following recommendation concerning errors and
uncertainties in the source term definitions:
24
-------
3.5.a Recommendation. The method used to derive the mathematical expressions
utilized to convert soil specific activity to a cleanup volume, and the formulas
themselves, should be presented explicitly in a technical Appendix to the TSD.
3.6 Sensitivity and Uncertainty Analysis of Source Term Parameters
The sensitivity analysis in the TSD for the generic site suggests that the factors for risk per
soil radionuclide concentration in various exposure scenarios are not highly sensitive to site
definition. The aspect of the generic site which leads to the greatest variability of this factor for
relatively insoluble radionuclides is the contaminated soil thickness. Only for soluble
radionuclides do other site parameters have a significant influence on the risk factor. Indeed, for
reference sites II, IV, VI, X, and XIII, where U and "Tc are dominant (that is, 5 out of 16 sites),
the uncertainties in the risk factors for these radionuclides as a result of uncertainties in the
transport parameters such as Kd value, are such that either no remediation or remediation of all
contaminated soils are possibilities, based upon the site-specific parameters. In these cases, the
reference site approach does not capture adequately the real universe of sites.
Including a section on sensitivities and uncertainties of soil cleanup volumes for reference
sites is highly beneficial, but the results focus on the impact of varying modeling assumptions,
which are only one aspect of the uncertainty. The extent to which the utilized radionuclide
concentrations and distributions may differ from actual values, and the extent to
which the created reference sites may differ from the median of each category, cause much larger
uncertainty.
No evaluation was made of the uncertainty introduced by the assumption that soil removal
is the only remediation technique to be used. Possible alternatives are radionuclide removal from
the soil, covering or stabilizing (e.g.,vitrifying) soil, or maintaining control as part of a waste
repository or elevated-contamination area. This omission is compounded further by the fact that
the draft proposed rule explicitly mandates that buildings, aquifers, and surface waters be included
in the pathway modeling for the risk calculations, to meet the determined risk level or ensure that
the Safe Drinking Water Act Maximum Contaminant Levels (MCL) are met.
A problem of circularity occurs in the uncertainty analysis presented in TSD Section 6.1.
Uncertainties in the soil volume vs. activity curves are disregarded due to lack of sufficient
information to quantify them. Yet these curves are used to estimate the area and thickness
parameters for input to RESRAD to calculate site-specific risk factors. The sensitivity analysis in
TSD Chapter 3 shows that the major sources of uncertainty in this calculation are the
contaminated zone thickness followed by the contaminated zone area. These calculated risk
factors are themselves another source of uncertainty. Finally, these risk factors are converted to
25
-------
volumes of soil to be remediated using those curves whose uncertainties were disregarded in the
first place. This discussion requires clarification, and estimates of uncertainty for all parameters
need to be presented in a logical and scientifically sound manner.
The following can be concluded:
3.6.a Finding. The generic site seems to be generally applicable to most cases under
study. Exceptions are cases where the "water-pathway dependent radionuclides"
are dominant (TSD Table 3-20, p. 3-73). For these cases, accurate site-specific
data will be necessary to describe the site and to calculate the site-specific risk
factors needed to determine the levels of remediation. In addition, the sensitivity
of modeling radionuclide migration into the groundwater to assumptions about
mobilization by complexation with non-radioactive materials, such as organic
solvents, should be examined (e.g., by varying the Kd parameter).
3.6.b Finding. For risk calculations involving insoluble radionuclides, the sensitivity
analysis indicates that the uncertainty of the result will be driven by the uncertainty
in the thickness of the contaminated zone (although this conclusion may depend
upon the dominant pathway for exposure, or on other site characteristics such as
mobilization by complexation with non-radioactive species).
3.6.c Finding. The soil volumes that need remediation for a 10"4risk level cleanup may
represent an upper bound given the fact that such a cleanup level for the case of
such nuclides as 226Ra and 232Th may require achieving soil specific activities that
cannot be distinguished from background at some sites.
3.6.d Recommendation. The discussion and presentation of sources of errors and
uncertainties, as they pertain to the source term, are limited to an acknowledgment
that they exist. It is impossible to gauge their magnitude from the information
presented in the TSD. Uncertainties should be estimated for the contaminated soil
volumes listed in the DOE Integrated Data Base and for sites under control of
Federal Agencies and sites licensed by the NRC or NRC Agreement States. A
well-defined and scientifically credible set of uncertainty ranges should be included
in TSD Tables 4-5 to 4-8 for all key input parameters used in the risk factor and
volume calculations. In addition, nonquantifiable components of the overall
uncertainty should be identified and discussed as to their likelihood of exceeding
the quantified components of the uncertainty.
26
-------
3.6.e Recommendation. The TSD analyses should make it obvious to the reader that
estimates of excavated soil quantities and associated transportation risks correctly
include the application of a bulking factor of 130% applied to in-situ volumes.
3.6.f Recommendation. The circular argument regarding uncertainties in the source
term in TSD Section 6.1 should be clarified, and these uncertainties should be
quantified in a reasonable scientifically justifiable manner.
27
-------
4. SCENARIOS AND PATHWAYS USED IN EPA'S ANALYSIS
4.1 Overview of Approach
In health risk assessments, scenarios are descriptions of the conditions under which
persons could be exposed to hazardous materials. In the case of regulatory risk assessments, the
scenarios are necessarily speculative because the risks to be reduced by the regulations have not
yet been incurred and the future conditions of exposure are in doubt. Faced with this necessity to
speculate, three courses of action are possible:
a) attempt to characterize exposures as realistically as possible, taking into account
current conditions and the reasonable persistence or predictable change in those
conditions;
b) use "conservative" assumptions about the conditions of exposure, i.e., those that will
tend to overestimate risk so that regulations based on the assessment will be biased in
the direction of the protection of health; or
c) analyze a range of scenarios to characterize the uncertainties and variabilities inherent
in assessments of future risks.
In its assessment of the costs and benefits of various proposed cleanup standards for
radioactively contaminated sites, EPA has adopted features of all three courses. For example, in
its characterization of future occupancy patterns on or near the sites, EPA has postulated the
possibility of on-site residential occupancy by persons who also consume some home-grown
produce and livestock products, a scenario that tends to maximize individual risk, but it also uses
plausible population density estimates for various scenarios, leading toward more reasonable
population risk estimates. EPA also uses variants on most of its assumptions to test the sensitivity
of the results to the assumptions and presents uncertainty analyses that have both quantitative and
qualitative components.
The Subcommittee divided its comments on EPA's scenarios into three sections, as
follows:
a) Section 4.2 on Land Use discusses the assumptions regarding who will live or work on
or near the sites in the future and how they will conduct their lives.
28
-------
b) Section 4.3 on Selection of Pathways for Analysis discusses the assumptions regarding
how people on or near the site will become exposed through air, soil, groundwater,
surface water, and direct gamma exposure.
c) Section 8.3 on Time Horizon discusses the span of years included in the analysis,
especially as it relates to the calculation of population risk (number of cancers avoided
by implementing the standards). To emphasize that this is more of a policy issue than
a technical issue, and hence outside the scope of its charge, the Subcommittee has
chosen to present its comments on this issue in a separate chapter of this report.
4.2 Land Use
EPA has chosen to develop a standard that would apply to sites with essentially
unrestricted access and therefore has focused on scenarios that feature on-site occupancy for
either residential or occupational (commercial-industrial) land use. The residential use scenarios
include an option for agricultural use by the residents. Although EPA acknowledges that some
sites (or portions of sites) may be managed with various restrictions on access or use, it has left
open the question of what standards would apply to such areas. That question may be handled
partly through EPA's standards for radioactive waste disposal, which could be used to manage the
risks of those restricted access sites by treating them as waste disposal areas. However, the waste
disposal regulations may only apply to engineered facilities, not to areas with only access or use
restrictions.
Given EPA's decision to assume unlimited access, its overall choice of scenarios is
appropriate. If its standards are designed to be protective for unlimited access, then future
residential or commercial/industrial use is possible. The actual future use will depend on the
suitability of the site for various activities, based on factors other than the past history of
radioactive contamination and cleanup. Some sites may be unsuitable for any intensive human use
even if residual radioactivity were not an issue. It is also possible that risk managers will decide
to place restrictions on access rather than require expensive and potentially ecologically damaging
soil removals at some sites. In those cases, the exposure of site neighbors to windblown dust,
migrating groundwater or surface water runoff, and "shine" (direct exposure to gamma radiation
at a distance) should be considered. While the Subcommittee is not prepared to recommend
inclusion of an off-site residential neighbor scenario in EPA's analysis, it suggests that EPA
consider adding a qualitative discussion of the influence of such a scenario.
All of the residential land-use scenarios involve food production and consumption. Food
production, treatment, and consumption patterns evolve considerably over time. For example,
29
-------
rates of beef and chicken consumption are much different now than they were 30 years ago. The
Subcommittee therefore recommends that, in transmitting its findings to the EPA group
responsible for proposing the cleanup standard, the technical support group should emphasize the
additional uncertainty of its estimates for 1,000 and 10,000 years in comparison with those for
100 years (see also Section 8.4 of this review document).
EPA has not explicitly considered a recreational scenario, in which the land would be used
mostly for outdoor activities on an occasional basis. Although such a use would result in lower
rates of occupancy (hours per lifetime), it could also introduce new pathways of exposure (e.g.,
consumption of local wild game, fish, fruit, and mushrooms). It is possible that persons living on
one part of a large site might participate in outdoor recreational activities on another part of that
site. Whether that practice would increase or decrease exposure relative to a person who spent
most of his or her time at home is difficult to predict. While the Subcommittee is not prepared to
recommend inclusion of a recreational scenario in EPA's analysis, it suggests that EPA consider
adding a qualitative discussion of the influence of such a scenario.
In summary, the Subcommittee makes the following recommendation:
4.2a Finding. The Subcommittee is satisfied that EPA's choice of an on-site residential
scenario and a commercial-industrial scenario for thorough analysis is adequate for
estimation of reasonable maximum exposures and risks from sites that have been
cleaned up to a specified level of contamination. While other scenarios are also
plausible, they are not likely to produce substantially higher estimates of
population risks.
4.2.b Recommendation. Although the Subcommittee does not propose adding a
recreational scenario to the analysis, it does recommend that EPA consider adding
qualitative discussions of the influence of recreational and off-site residential
neighbor scenarios on the risk estimates.
4.3 Selection of Pathways for Analysis
4.3.1 Pathways Included in the EPA Analysis
Pathways are mechanisms by which individuals may be exposed to radioactivity from a
contaminated site. The list of pathways considered by the EPA in its analysis are the standard
generic pathways typically considered for regulatory compliance calculations. EPA's analysis of
the reasonable maximum exposure (RME) to future users of a site subject to the cleanup rule
includes consideration of the following pathways:
30
-------
a) direct external radiation from photon-emitting radionuclides in the soil,
b) inhalation of resuspended dust containing radionuclides,
c) inhalation of radon and radon decay products via diffusion into buildings or the
atmosphere from radium-containing soils,
d) ingestion of groundwater to which radionuclides have migrated from soil
e) ingestion of soil containing radionuclides,
f) ingestion of produce grown in soil containing radionuclides,
g) ingestion of beef from cattle raised on-site,
h) ingestion of milk from milk cows raised on-site, and
i) ingestion offish from surface waters that receive runoff from soils containing
radionuclides.
Only a subset of the pathways is used for some future occupancy assumptions.
EPA's analysis of cumulative population risks (cancer fatalities avoided by reducing RME
risk from 10"2 to the target risk level) considers only the following pathways:
a) direct external radiation from photon-emitting radionuclides in the soil,
b) inhalation of resuspended dust containing radionuclides,
c) inhalation of radon and radon decay products via diffusion into buildings or the
atmosphere from radium-containing soils,
d) ingestion of groundwater to which radionuclides have migrated from soil, and
e) ingestion of produce grown in soil containing radionuclides.
In terms of risk assessments to the potentially exposed individuals and populations, the
parameter values selected for each of these pathways must be scrutinized and evaluated. The
Subcommittee recommends that this evaluation should focus on the following issues:
a) Is the purpose of the risk calculations to derive screening or bounding estimates, or to
provide the best estimate of the true value (risk)? The objective then should guide the
selection of appropriate parameter values used in pathway modeling.
b) To what extent are the parameter values selected for the reference pathways
sufficiently conservative for the real pathways that could occur at real sites?
Additional comments on these aspects of the pathway modeling are provided in Section 5 of this
report.
4.3.2 Pathways Omitted from the EPA Analysis
31
-------
Provided one accepts EPA's choice to focus on a population at risk that would occupy the
site after cleanup, the list of pathways considered for the RME risk is reasonably complete and
probably includes the pathways of most importance from the perspective of cancer risk. EPA
provides some explanation of its decision to exclude such pathways as dermal exposure to soil
containing radionuclides (i.e., the relatively low dermal absorption of radionuclide compounds in
comparison with other pathways, which is not always true for lipophilic compounds) and
volatilization of radionuclides from soil or domestic water (which is probably minor for all
radionuclides except radon, tritiated water vapor, and carbon-14 dioxide, each of which is treated
specially in the analysis).
One omission that is potentially important is the possibility of ingesting surface water that
receives runoff from soils containing radionuclides. This omission is curious because the fish
pathway is included (by assuming that groundwater infiltrates into surface water) yet drinking
water from surface sources is not. EPA dismisses this pathway by asserting that concentrations in
groundwater would be higher than those in surface water at virtually any site (with the modeling
assumptions used) and that "human ingestion of surface water is not considered a reasonable
scenario in most instances." However, given the wide use of surface water for drinking water
supplies and the better quality of surface water relative to groundwater in some environments,
these arguments are not particularly strong. Moreover, radionuclides deposited on soils can move
to surface water not only by dissolution into runoff water but also by overland transport of
particles eroded from the soil. The Chernobyl experience shows that contamination of surface-
water reservoirs occurs mainly by water runoff from the soil surface and not by groundwater
infiltration (e.g., see discussion in section 5.7 of this report); hence, it is important not to neglect
this pathway.
Groundwater and surface water pathways may also be modified by diversion of a portion
of such waters and their associated sediment loads (in the case of surface waters) into combined
sewer and storm sewer systems, the effect of which has not been addressed in the TSD analysis.
Another omission that may be important for specific sites is the recreational scenario and
related pathways. None of the modeled scenarios contain exposure pathways due to recreational
activities. People living in a residential or agricultural area (farms) will have recreational activities
in that area, such as fishing in a nearby lake, hiking, and hunting. Additional exposure pathways
such as consumption of wild game, mushrooms, wild fruit, and locally caught fish, should be
discussed and qualitatively considered in any site-specific assessment as appropriate.
With respect to the lack of concurrence between the list of pathways used for the
population impact analysis and the RME risk assessment, EPA states that the pathways omitted
from the latter do not contribute substantially to total cancer fatalities, with the possible exception
32
-------
of soil ingestion for nuclides such as239Pu that do not involve external gamma radiation or radon
and do not migrate rapidly to groundwater or crops. Given that other radiological risk
assessments have yielded similar conclusions (NRC 1994a; DOE 1994a), this focus is probably
justified, at least for the sites most important to the overall impact analysis.
Because the on-site occupancy scenarios will almost always produce higher RME risks
than a scenario in which off-site exposures are considered, the lack of off-site transport scenarios
in the RME risk analysis of reference sites is probably justified. In future assessments of real sites,
however, the following possibilities should be considered:
a) groundwater may not be used for drinking water on-site but affected aquifers may be
utilized for off-site populations; and
b) agricultural use of the site may not be contemplated, but dust transported downwind
could affect off-site agriculture, which might supply food for on-site residents.
For the population risk analysis, EPA asserts that its methods for estimating cancer
fatalities have features that compensate for the distribution of impacts on-site and off-site. In
particular, EPA uses the radionuclide inventory as its principal source term for the population
impact analysis and does not assume that the groundwater containing radionuclides would be
consumed only by on-site occupants. For example, EPA assumes that one half of all the
groundwater originating at the site will be withdrawn for use and that 1% of that will be ingested
as drinking water. With this assumption and the linear relationship between intake and risk, the
number of fatal cancers averted would not be affected by the fraction of the water used by on-site
versus off-site populations (except for decay in transit to off-site populations). Similarly, dust
containing radionuclides could be inhaled by either on-site or off-site populations and identical
cancer burdens would be calculated if the total amount of dust inhaled is assumed to be constant.
Moreover, EPA should provide the rationale for its assumptions regarding groundwater
withdrawal; the assumption that 50% of the groundwater will be withdrawn for residential use
may be reasonable but is difficult to support.
4.3.3 Duration of Exposure
In every case, the person at risk is assumed to be exposed to soil with uniform
concentrations of radionuclides for a defined period and duration of exposure, and not exposed at
all for other times. Real sites will be characterized by non-uniform distributions of radionuclides.
The lack of uniformity in some cases could lead to significantly erroneous risk calculations if the
proper statistic for the distribution of concentrations is not used. While the average concentration
over an area may be satisfactory for many pathways, it may not be if a real person's mobility were
33
-------
substantially greater than the scale of disuniformity. Generally but not always, more accurate
consideration of disuniformity and mobility would lead to lower calculated risk levels for the
reasonable maximum exposure. On the other hand, on large sites the time spent "away from
home" may be spent at other locations on site, which could lead to residual risks larger than
assessed. The net effect of these two simplifications cannot be predicted without knowing more
about the true distributions of radionuclides.
4.3.4 Findings and Recommendations Concerning Pathways
4.3.a Recommendation. In the face of constraints of time and resources for revisiting
the definitions of the pathways, EPA should focus its efforts for any improvement
of the pathway definitions (i.e., underlying assumptions and adopted parameter
values) on the dominant pathways, particularly external gamma radiation, radon
inhalation, crop ingestion and ingestion of groundwater (especially the population
risk assumptions). Ingestion of surface water should be investigated further to
determine whether this pathway might be dominant under any widely prevalent
conditions.
4.3.b Finding. With the exception of a surface water runoff and erosion pathway with
subsequent potential for drinking water and fish consumption exposures, EPA has
generally included in the RESRAD analysis all the pathways that are likely to be
important for radionuclides in inorganic forms. (Direct and indirect dermal
absorption pathways might be important for organic chemicals and mixed wastes.)
4.3.c Finding. In assessing compliance with the proposed standard at real sites,
consideration should be given to the non-uniformity of soil concentrations, even
after site cleanup, as they might affect the risk to persons who do not remain at the
same geographical point continuously.
4.3.d Recommendation. ORIA needs to identify the purpose of the risk calculations as
being either the derivation of screening or bounding estimates, or the provision of
the best estimate of the true value (risk ?). The chosen objective should then guide
the selection of the appropriate parameter values to be used in pathway modeling.
4.3.e Recommendation. ORIA should determine if the parameter values selected for
the reference pathways are sufficiently conservative for the real pathways that
could occur at real sites.
34
-------
4.3.f Recommendation. In future assessments of real sites, ORIA should take into
account the possibility that groundwater on-site may not be used as a source of
drinking water, but that affected aquifers may be utilized by off-site populations.
4.3.g Recommendation. Although agricultural use of a site may not be contemplated in
the current scenario future analyses should consider the possibility that dust,
transported downwind, could affect off-site agriculture which might supply food
for on-site residents.
35
-------
5. METHODOLOGIES FOR MODELING TRANSPORT TO PEOPLE
Section 5.1 reviews the Agency's rationale for selecting RESRAD for modeling exposure
to on-site individuals, and a a population model developed by EPA for the TSD for modeling
exposure to populations. Sections 5.2 through 5.7 provide Subcommittee comments on details of
the model formulations used as part of these codes to simulate transport by the selected pathways,
including the choice of variables to be included and appropriateness of selected parameter values
that were used by EPA.
5.1 Evaluation of Candidate Models for Pathway Modeling
5.1.1 Model Selection Criteria and Process
Once the desired exposure scenarios and pathways were determined for the assessment,
EPA sought to identify a set of candidate models for predicting exposure and risk, and to evaluate
which model or models were most appropriate for the task. The models considered needed to be
multimedia, considering the full range of direct, soil, air, surface water, ground water and food
ingestion pathways. In particular, the models needed to address the following exposure
pathways:
a) external radiation exposure;
b) soil ingestion;
c) plant, meat and milk ingestion;
d) inhalation of volatiles and fugitive dusts;
e) migration of radionuclides to ground water; and
f) ingestion of contaminated drinking water.
In addition, it was desired that the models be capable of assessing the exposure and risk from
radon.
A set of 21 potential models was identified by EPA through review of scientific literature,
EPA reports and databases, and discussions with project staff. The models were evaluated
according to their ability to address the multiple exposure pathways identified above, and based
on the following model performance criteria:
a) level of validation and peer-review, outside and within EPA;
b) availability and accessibility of the computer code;
c) extent to which the computer code was user friendly; and
d) the amount of site data required to implement the model.
36
-------
Based on this evaluation, the DOE RESRAD code was chosen for the modeling of
individual risks at reference radiation sites, and two additional codes, EPA's PRESTO-CPG and a
modified version of EPA's RAGS/HHEM Part B, were chosen for further comparative studies.
The RESRAD code covers all exposure pathways of interest, has been subject to validation and
peer review studies, has an available and user-friendly computer code with a moderate amount of
required site data, and is being used by DOE for site evaluation studies. The PRESTO-EPA-CPG
model has similar capabilities (although it does not compute exposure and risk due to radon) and
has undergone previous review by the EPA SAB for use in the evaluation of low-level radioactive
waste disposal rules (Federal Guidance Report No. 11, (EPA, 1988)) (EPA, 1987). The
RAGS/HHEM Part B model is a simple, conservative screening model based on unit transport
and exposure factors. While it is not computerized, and has not undergone validation and peer
review by the SAB, it is believed to provide a conservative baseline for comparison of the more
realistic process models. The Subcommittee finds that the initial screening and selection of
candidate models for further evaluation as reported in the TSD was conducted in a reasonable,
sound and thorough manner, using appropriate criteria for model selection.
The comparative evaluations of RESRAD, PRESTO and RAGS/HHEM included a
detailed comparison of model formulations and assumptions for the various exposure pathways,
and application of each to a generic site for calculation of unit risk factors for the radionuclides of
interest. The latter comparisons indicated significant differences between the models for many of
the radionuclides. Differences were identified both in the overall unit risks and in the allocation of
this risk to the different exposure pathways. A number of reasons for these differences were
identified; principal among these are the different treatment of source-term depletion and progeny
ingrowth by the models, and differences in the methods for computing leachate flows and
concentrations to the ground water. The RAGS/HHEM model assumes an infinite, nondepleting
source, and PRESTO is limited in its treatment of radioactive ingrowth when performing source
calculations for a series of ground water sources. RESRAD is more conservative than PRESTO
in assigning radionuclide concentrations to leachate; the former assumes equilibrium while the
latter considers the contact time between the infiltrating water and the waste.
As a result of these differences, when the models are applied to a generic site at which the
ground water exposure pathway is important, RAGS/HHEM estimates the highest risk for a given
waste concentration, PRESTO the lowest, and RESRAD an intermediate value, although
sometimes much more conservative than PRESTO. Differences between the models are greatest
for radionuclides with low Kd values (low adsorption onto soils) and short half lives, such as
tritium; the models provide similar estimates for highly adsorbing, slowly decaying radionuclides.
However, significant differences in unit risk are present even for some radionuclides with very
long half-lives, such as 232Th. Other differences in model predictions are ascribed to process and
exposure assumptions such as inclusion of erosion for surface waters (included in PRESTO, but
37
-------
not in RESRAD) and corrections for time spent indoors (included in RESRAD, but not in the
other models).
The significant differences in the predicted unit risk for the three models evaluated are
somewhat disturbing, though not surprising. Similar, order-of-magnitude differences between
RESRAD and six other radiation exposure pathway models were found in a benchmarking study
conducted by Argonne National Laboratory (Faillace et a/., 1994). These arise from fundamental
differences in pathway representations and assumptions, as well as possible errors in input
databases. Presentation and acknowledgment of such differences for selected radionuclides
provides an honest appraisal of the current state-of-the-art for radiation exposure modeling.
5.1.2 Comments on Selection of Models and Their General Limitations
RESRAD and the other models considered by the Agency are reflective of the
state-of-the-art for comprehensive integrated multimedia/multipathway models for exposure and
risk. These models include simplified modules for many of the pathways and exposure scenarios
they consider. In some cases, these modules are similar to models that could be used in
site-specific applications focusing on a specific pathway. In other cases, they are more idealized
and simplified.
RESRAD is particularly well-suited for the application to which it is applied by
EPA—determining on-site risks from residual radionuclides in soils—because this is the sole
focus of the model. The Agency summarizes its basis for selection of RESRAD for the individual
unit risk calculations as follows (TSD, bottom of p 3-29):
"The results calculated by the three models are similar for many of the radionuclides. The
most significant changes are caused by the decay and ingrowth corrections, which at this
time only RESRAD applies to all of the radionuclides. The most significant change for a
single radionuclide is H-3, and this is caused by a combination of slow leach rate and short
half-life for the PRESTO calculation. RESRAD was selected for performing the
calculations for the reference sites because it calculates a more conservative result for H-3
and includes corrections for ingrowth and decay of principle (sic) radionuclides."
While the Subcommittee believes the reasons for selecting RESRAD are somewhat more
extensive than those described above (including its use by DOE, user friendliness, and full
coverage of on-site exposure pathways), the comparative evaluation appears to result in a
reasonable and appropriate selection of the RESRAD model for use at the present time.
However, because of the highly simplified and idealized conceptual models for the specific
pathways, many of the important processes which can affect radionuclide fate and transport are
38
-------
neglected. The generic processes included in these codes may not always represent the most
important processes that will dominate the actual dose and risk to real individuals exposed at a
specific location (e.g., see discussion of Chernobyl studies in section 5.3.3 and 5.7 of this report).
Even in the case when all real processes are included, the generic (default) parameter values might
not be appropriate for a real site. Consequently, RESRAD, as configured for the TSD analyses, is
appropriate only to provide an approximate generic assessment for highly aggregated studies in
which results from several individual studies are pooled together and in which it is understood
that only the results taken as a whole are intended to represent the universe of real sites. Thus,
RESRAD may not be appropriate for analyzing a specific site's compliance with the standard for
complex site conditions. Furthermore, from a cursory review, several of the transport model
parameters appear to be internally inconsistent, or inconsistent with literature values or with
known site characteristics (e.g., see section 5.6); hence, the Subcommittee strongly urges EPA to
obtain a thorough peer review of the default and site-specific parameters used in the transport
modeling. As more information becomes available through site applications and studies, the
Agency should revisit this issue to determine whether RESRAD should be modified or replaced
so as to ensure that the appropriate balance is maintained between realistic prediction and
reasonably conservative protection of public health.
RESRAD is user-friendly and quite flexible for input data. However, the current version
of this model is limited by the absence of a transport module to on-site analysis; other models
may have to be used to determine risk at off-site locations. In some cases, the calculated on-site
risk may not be significantly different from the risk at an off-site location if, for example, the half-
life of the radionuclide is very long, the location under consideration is not very far from the site,
and/or the transport process is not very rapid. The difference in risks can be very significant for
other cases. Because the Agency does not rule out "institutional control" (i.e., imposition of
"restricted access" to a contaminated site) as a realistic remediation measure for some sites, the
need for the model to be able to assess off-site risk becomes obvious. It is unlikely that people
who live near, but not on, the soil with residual radioactivity would be more at risk than
equivalent on-site residents, but the ability to calculate off-site risks for such a residential neighbor
scenario would allow the analysis of a strategy where affected soils might be allowed to have
higher concentrations if restricted from residential use. For example, a site with commercial /
industrial use might still pose more risk to residential neighbors than to on-site workers. The
Subcommittee recommends
that EPA supplement RESRAD to include the capability for calculation of off-site risks, if the
Agency chooses to continue using this model for this task.
The basis for the selection of a methodology for the population dose model was not
discussed extensively in the TSD text. The population model used for the TSD analyses,
39
-------
(hereafter called the CU-POP model, for "cleanup population" model, as requested by the
Agency), was developed by the Agency and is based loosely on the RAGS/HHEM model in that
it employs simple bounding algorithms and assumptions. However, the CU-POP model differs
fundamentally from RAGS/HHEM in a number of respects. It explicitly accounts for radioactive
decay, progeny ingrowth, and source depletion by leaching of the soil as a function of time, as
well as decay and ingrowth during transport through the unsaturated zone; in contrast,
RAGS/HHEM assumes a non-depleting and non-decaying source term. Appendix C of the TSD
describes RAGS/HHEM, and Appendix E describes the CU-POP model. The TSD does not
discuss the advantages and limitations of alternative methodologies for calculating population
effects.
5.1.3 Findings and Recommendations Concerning Model Selection
5.1.a Finding. The Subcommittee finds that the initial screening and selection of
candidate models for further evaluation as reported in the TSD was conducted in a
reasonable, sound and thorough manner, using appropriate criteria for model
selection.
5.1.b Finding. The Subcommittee concurs with the Agency's decision in its selection of
RESRAD as a reasonable transport model for use at the current time. However,
while it incorporates some conservative assumptions in its formulations, RESRAD
itself may not necessarily provide conservative risk estimates if inappropriate
parameter values are selected for the modeling input.
5.1.c Recommendation. Because the Subcommittee has not evaluated the default
values in the RESRAD code, nor the full parameter set used for each reference
site, it is unable to fully assess the extent to which the model results can be
considered to be conservative or bounding estimates of the true health effects
associated with each level of cleanup. From a cursory review, several of the
transport model parameters appear to be inappropriate (e.g., see section 5.6 of this
report); hence, the Subcommittee strongly urges EPA to obtain a thorough peer
review of the default and site-specific parameters used in the transport modeling.
5.1.d Recommendation. Recognizing the generic and idealized nature of the RESRAD
model and its limitation to on-site exposure scenarios, the code may not be
appropriate to use in many of the site-specific applications where specific
processes or exposure scenarios not included in the model are present and
important. The Agency should make readers aware that, for these site-specific
applications, there should not be an attempt to force sites to fit the model (as there
may be a tendency to do simply because RESRAD is the model upon which the
40
-------
national regulations are based); rather, the Agency should emphasize that models
and estimation procedures appropriate for site conditions are needed.
5.1.e Recommendation. The Subcommittee believes that the population dose model
developed by the Agency for the TSD is generally adequate but could be improved
through further consideration of off-site risks and the use of better site-specific
transport parameters.
S.l.f Recommendation. The Agency should provide a more extensive summary of the
validation and model comparison studies that have been conducted for RESRAD.
These studies may suggest that very significant differences often occur between
model predictions and observed data, or between the predictions of alternative
models, when they are applied to specific sites or contamination incidences. This,
however, provides an honest appraisal of the state-of-the-art of such modeling and
a good sense of the significant uncertainty associated with the application of
general models to specific cases with varying conditions and highly uncertain input
parameters.
5.1.g Recommendation. As more information becomes available through site
applications and studies, the Agency should revisit this issue to determine whether
RESRAD should be modified or replaced so as to ensure that the appropriate
balance is maintained between realistic prediction and reasonably conservative
protection of public health.
5.2 Pathway Models for Direct External Exposure
Models, assumptions, parameter values, and results of model runs related to direct gamma
radiation exposure were reviewed. In general, direct gamma radiation doses and risks were
appropriately determined for both the Reasonable Maximum Exposure (RME) for individuals and
the Population Exposures. Problems inherent in the application of the results of the generic site
assessment involving simple, convenient contamination patterns to real
sites with complex contamination profiles are not discussed in this subcommittee review report.
5.2.1 Modeling Risks to Individuals
5.2.1.1 Exposure correction factors
41
-------
The three principal exposure scenarios for the Reasonable Maximum Exposure (RME) to
individuals include a direct external gamma exposure pathway in which assumptions regarding
exposure duration, time and shielding factors differ for each scenario. Two of the three pathway
models (RESRAD and RAGS/HHEM) include correction factors for shielding for indoor
exposure vs. outdoor exposure as well as for exposure time, frequency and duration. The third
model (PRESTO) corrects only for exposure duration. RESRAD also takes into account other
factors which affect external exposure from contaminated soil such as radioactive decay and
progeny ingrowth, geometry of the contaminated soil, depletion of the contaminated soil by
environmental processes and shielding by clean cover. RAGS/HHEM does not account for these
processes; PRESTO corrects for depletion, geometry and decay (as well as ingrowth in the
version used). Therefore, among these three models, RESRAD is the most realistic in its
mathematical formulation of this pathway.
One issue regards how the shielding factor for indoor exposure should be applied. In the
case of deposition of radioactive materials around a structure after it has been built, much of the
"shielding" effect comes from the simple fact that a person in the middle of a house is several
meters away from any radioactive materials, independent of any attenuation of radiation by
building materials. If a one-story house is built on undisturbed soil that has already been affected
by radionuclides, then the distance effect largely disappears and is replaced by whatever
attenuation is provided by any slab or flooring. The shielding factor may thus depend on the
depth of soil removed in constructing the foundation and basement, if any, and the depth of
radioactive concentration assumed.
5.2.1.2 Dose conversion factors and slope factors
All three codes use EPA-approved dose conversion factors (DCFs) taken from Federal
Guidance Report No. 12 (EPA, 1993a). The TSD is very confusing with regard to the slope
factors used to estimate risks. Several versions of the Health Effects Assessment Summary
(HEAS) Tables are referenced, but these references are apparently incorrect. In fact, the values
used and tabulated in TSD Appendix Table B-l are from a personal communication with Jerry
Puskin of EPA/ORIA. This creates confusion. The source and derivation of slope factors should
be clarified.
Because the dose conversion factors and slope factors used in the models are EPA-
approved values, they were not reviewed in depth. However, Federal Guidance Report No. 12
(EPA, 1993a) values were similar to the DOE values used in RESRAD. Selected slope factors
from the 1993 HEAS Tables (EPA, 1993b) were compared to the slope factors in Federal
Guidance Report No. 12 (EPA, 1993a), and were found to be similar after appropriate conversion
factors were applied. As noted in the TSD, the uncertainty in the slope factors for external
42
-------
gamma radiation supports a lower bound of zero and an upper bound of approximately three
times the nominal value.
All three codes use the same slope and dose conversion factors, and thus the differences in
results should be due only to the way the models treat the other factors. For the generic site,
under the rural/residential scenario, the calculated unit risk estimates for most nuclides for which
external exposure accounts for 95% or more of the total risk show little difference among the
models. This indicates that the processes neglected by PRESTO and RAGS/HHEM are not
significant for the generic site. However, Table 3-10 is somewhat confusing in this regard. For
example, the risk factors (presumably in risk per pCi/g although units are not given in the table)
for "Co for RESRAD, PRESTO, and RAGS/HHEM are 6.25 x 10'6, 2.44 xlO'6, and 6.40 x 10'6
respectively. External gamma radiation accounts for 98% of the dose from 57Co which has a half-
life of 270 days. Because both RESRAD and PRESTO account for decay, the risk factors would
be expected to be similar. According to the technical document, RAGS/HHEM does not correct
for decay; therefore, the risk factor would be expected to be higher. This is obviously not the
case. The risk factors calculated by RAGS/HHEM and RESRAD are similar, and the risk factor
calculated by PRESTO is lower by a factor of 2.5. The comment in the table notes that the risk
factor calculated by PRESTO is "higher" (sic, table entry should read "lower"; this erroneous
wording is common throughout Table 3-10) due to the short half-life of 57Co. In this case, the
lower risk factor is due to the fact that PRESTO does not calculate the risk at the beginning of the
simulation but only at the end of the first year. This would result in underestimating the risk
because only 40% of the original 57Co remains after the first year.
5.2.1.3 Generic site assumptions and parameter values
The principal assumptions for the generic site which affect direct external gamma radiation
include a homogeneous, infinite depth of contamination and 10,000 m2 area.6 These are obviously
base case values and are not applicable for most contaminated sites. For most real sites, the depth
of contamination would be variable and the concentrations not homogeneous. RESRAD has the
capability of taking into account variability in depth and areal extent of contamination and thus
can be used to derive more realistic risk estimates. RAGS/HHEM does not have this flexibility,
making it less useful for analysis of real situations.
The assumptions and parameter values for the rural/residential and commercial/industrial
scenarios which affect the dose from direct gamma radiation are reasonable and are, presumably,
The generic site is described as uniformly contaminated to a depth of 2 m. For the purpose of estimating direct gamma dose, this depth is
equivalent to "infinite depth" as used in Federal Guidance Report No. 12 (EPA, 1993a).
43
-------
incorporated into the RESRAD model used in the analysis. The RESRAD equations for
correcting for depth of contamination, cover thickness and source area are also reasonable.
The TSD apparently has not considered the major impacts on direct radiation exposures
by the site uses assumed in the exposure scenarios. Excavation for construction may move the
contaminated soil layer, paving and buildings will attenuate gamma rays, and plowing may either
increase or decrease the gamma-ray flux into air. The TSD should call attention to the site-
specific nature of direct radiation exposure for each scenario.
5.2.1.4 Sensitivity and uncertainty analyses
The factors with significant impact on direct gamma radiation exposure which were
considered in the RESRAD sensitivity analysis include the area of the contaminated zone and the
thickness of the contaminated zone. Infiltration rate, which was also considered in the sensitivity
analysis, affects depletion of the contaminated zone which also impacts direct gamma radiation
exposure. Other site or scenario factors affecting direct gamma radiation risk such as duration of
exposure were not considered. As would be expected, for nuclides with greater than 95% of the
risk due to direct gamma exposure, the area of the contamination affected the risk by less than a
factor of two. The contaminated zone thickness affected the risk by up to a factor of five for a 2
cm thickness or less than two for a thickness of 10 cm.
The uncertainty analysis for the RAGS/HHEM model considered a greater number of
variables using distribution functions rather than discrete values. However, from the output, it
was not obvious which parameters had the greatest influence on the final results.
Additional concerns relating to sensitivity results are found in section 7 of this report.
5.2.2 Population Exposures
Population exposure estimates for all sites were determined using the CU-POP model,
which assumes that the population impact is proportional to the total radioactivity at the site.
The equations for calculating dose due to direct gamma radiation are based on the soil
concentration, not total activity. The equations given in the appendix of the document include a
correction for depth of contamination and depletion due to physical decay and leaching but not for
44
-------
ingrowth of progeny. The tables in the TSD indicate that, at least in the case of 230Th and 232Th
where the principal health effect is due to the progeny, ingrowth is included.
The equations and assumptions for risk due to direct radiation exposure are reasonable
and contain the appropriate corrections for depth. However, the implication in the TSD, with
regard to population dose, is that the total inventory of the radionuclide under consideration at the
site is the basis for the population dose calculations (TSD, p. 5-11). In addition, the TSD states
that the direct gamma population dose calculation is based on the dose conversion factors from
Federal Guidance Report No. 12 for an "infinitely thick" contamination depth (TSD p. 2-52). In
fact, while this assumption was appropriately applied in the generic site calculation, the population
doses for the reference sites were calculated using dose conversion factors from Federal Guidance
Report No. 12 which more accurately reflected the assumed depth of contamination for each
specific site. This is a reasonable approach to the calculation, but the method used is not obvious
from the information presented in the TSD.
5.2.3 Cleanup Worker Exposures
The estimated direct radiation doses to workers for the reference sites are based on the
volume of material to be handled and the radionuclide concentration. This is a reasonable
approach.
5.2.4 Off-Site Doses Due to Cleanup Activities
Because direct radiation exposure decreases rapidly with distance from the source, off-site
doses due to cleanup activities would be confined to individuals in proximity to transport vehicles
such as rail cars. The total amount of time any individual or population would be exposed under
these conditions is very small, and hence this pathway is not significant, as noted in the TSD.
5.2.5 Findings/Recommendations Concerning the External Radiation Pathway Model
5.2.a Finding. The external radiation pathway is relatively simple to define. With the
possible exception of exposure time (see section 4.3.3), the EPA's modeling of this
pathway is generally acceptable. Of the three codes considered in detail in the
generic base case study, RESRAD is the most realistic and flexible in its
capabilities for accounting for the processes that govern exposure of individuals by
this pathway. The assumptions and parameter values for the rural/residential and
commercial/industrial scenarios which affect the dose from direct gamma radiation
are reasonable and are, presumably, incorporated into the RESRAD model used in
45
-------
the analysis. The RESRAD equations for correcting for depth of contamination,
cover thickness and source area are also reasonable.
Areas in which the EPA could improve its analysis and presentation of results are as follows:
5.2.b Recommendation. EPA be more explicit regarding its assumptions about the
shielding factor for indoor exposures and be sure that the parameter value used is
consistent with those assumptions. The assumed shielding factor for indoor
exposure to external gamma radiation does not apply to buildings which have no
contamination underneath the structure.
5.2.c Recommendation. The sensitivity analysis for contaminated zone thickness
showed a significant anomaly which should be further investigated. The ratio of
calculated soil concentration to base case soil concentration at a 1 x 10"4risk level,
as a function of contaminated zone thickness, shows a value two orders of
magnitude higher for 230Th than for any other nuclide. The reasons for such a
drastic difference in risk for the same total radioactivity concentration are not
obvious and should be explained.
5.2.d Finding. The uncertainty analysis for the RAGS/HHEM model considered a
greater number of variables using distribution functions rather than discrete values.
However, from the output, it was not obvious which parameters had the greatest
influence on the final results.
5.3 Pathway Models for Ingestion of Soil and Food
5.3.1 Overview Comments
The produce ingestion pathway is somewhat simplified with respect to other multipathway
risk assessment models currently in use. It employs two generic groups of produce: fruits,
vegetables, and grain are lumped together while leafy vegetables are treated differently. In other
models, root vegetables may be treated separately, or a list of specific food crops may be
analyzed. The EPA document states that certain corrections were applied for fruits and non-leafy
vegetables, but does not make it clear whether that correction was applied to grains as well. The
EPA pathway assumptions share with many other systems the need to make many simplifying
assumptions for very complex phenomena. The most critical issues, however, probably have
46
-------
more to do with the details of the occupancy scenarios as discussed above. In particular, the
amount of locally grown foods as a percentage of the total diet will be highly variable from site to
site, from person to person for a given site, and among different specific commodities. This
variability has generally been handled by making conservative assumptions, which may lead to
substantial overestimates of population risk via this pathway and perhaps to overestimates of
RME risks.
The meat and milk pathways are similar in strengths and weaknesses to the produce
pathways. As with many other multipathway models, meat is represented by beef. Some
multipathway models attempt to consider chicken and eggs. Again, this pathway is unlikely to be
important for many sites and, if so, any conservatism is correspondingly less important. It is
possible that omission of poultry-related pathways could underestimate risks, although it seems
unlikely that including them would move the livestock pathway into prominence. Note that
dietary patterns have changed remarkably in the last 30 years and projections of food pathway
risks even to 100 years should be viewed with skepticism.
The fish ingestion pathway is based on transfer factors from surface water to fmfish and
shellfish, after applying a groundwater to surface water seepage model to estimate concentrations
of radionuclides in water from soil sources. A significant shortcoming of this pathway is the
omission of a runoff pathway, which will probably be more important than the seepage pathway
for any radionuclide that binds to soil yet can be liberated to the food chain from sediments. The
runoff pathway is clearly more important than the seepage pathway for certain pesticides and
highly lipophilic organic chemicals (Kellogg et a/., 1992); EPA should probably conduct a
sensitivity analysis to determine whether any radionuclides behave similarly. However, the overall
importance of this omission will likely also be small because of the negligible contribution of the
fish ingestion pathway in the current risk assessment.
In the sections that follow, comments are provided on specific parameter values used in
the EPA analysis of the soil and food ingestion pathway.
5.3.2 Soil Ingestion Rates
Table 2-1 of the TSD gives the standard default parameter values used for each land-use
scenario. The soil ingestion rate of 50 mg/day for commercial/industrial usage of the land is
suitable only for office industrial activities. Other industrial activities (e.g., construction, surface
mining) may require outside work or can produce high rates of resuspension of surface soil in the
atmosphere, potentially leading to more deposition on skin and food with the opportunity for
greater rates of soil ingestion. Therefore, a higher rate of soil ingestion is recommended as a more
appropriate default value for this scenario. The exact value to be used, however, depends upon
47
-------
the overall intent or objective of the proposed annual dose limit as it applies to cleanup (see
recommendation 6.6.a).
For children and adults affected with pica, the soil ingestion rate may be as high as 1 to 5
g/day. The highest values to be found in the literature are at 10 g/day. These are extreme values,
which are not reasonable default values for the RME individual. Also, the population average soil
ingestion rate is inappropriate for the RME individual who is defined as a maximally exposed
individual. The EPA standard soil ingestion rate assumed for children is 200 mg/day.
5.3.3 Soil-to-Plant Transfer Factors for Nutrient-Poor Soils
The Savannah River Site is an example of a site with nutrient-poor soils that exhibit very
high soil-to-plant transfer rates for 137Cs and 90Sr (Whicker and Hinton, 1993). Therefore, plant
contamination will be unusually high at this site and the ingestion pathway is expected to
dominate over other pathways of exposure. Chapter 4 of the TSD describes Reference Site I,
based on a description of Hanford Reservation contamination, and Reference Site V, based on
contamination reported for Savannah River Site. Both of these sites have 137Cs as a major
contaminant. The difference between these two sites, as stated in the reference site descriptions,
is found mainly in the fractional distribution of 137Cs in soil. No consideration is given to
differences in the soil properties between these two sites. These properties would influence the
uptake of 137Cs by vegetation. For the Savannah River Site as well as for the Hanford
Reservation, it is assumed in the TSD that the external exposure pathway would be dominant for
any of the modeled scenarios. This assumption would not be valid for these sites if agricultural
activities were to exist.
For a given radionuclide, the relative contribution to human health risk from different
exposure pathways can vary considerably from site to site. Therefore, a model which incorporates
only a limited number of generic pathways may be useful for one site, but may be inappropriate
for another. For 137Cs+D, the results presented in TSD Tables 2-8, 3-1, and 3-4 (for the rural
residential exposure scenario) suggest that external radiation has a much larger contribution to the
total dose than does the ingestion pathway. To the contrary, however, the ingestion pathway
should dominate for any site with nutrient-poor surface soils because 137Cs is rapidly recycled
from such soils to plants grown on those soils (NCRP, 1993). Growth of crops on contaminated
nutrient-poor or highly acidic soils will also be a major pathway of concern for sites at which 90Sr,
129I, and "Tc are present for the same reason. Eight of the sixteen reference sites described in the
TSD contain one of these four radionuclides.
48
-------
The distribution of fallout following the 1987 Chernobyl accident illustrates this point.
Estimates made for southern Finland after the Chernobyl accident showed that exposure from
ingestion of contaminated foodstuffs is equal to or greater than external exposure from
contaminated surface soil (IAEA, 1995; Suomela et a/., 1991; Rahola et a/., 1991; Rantvaara,
1991). Figure 5-1 shows the relative contribution of ingestion and external exposure to the total
dose for southern Finland. For one year and five years after the accident, the ingestion dose
exceeds the external dose. When exposure is extended to a human lifetime, the contribution from
external exposure becomes equal in importance to exposure from the ingestion of contaminated
foods.
Furthermore, results in TSD Tables 3-1, 3-4, and 3-7 suggest that meat provides the
largest contribution to the ingestion risk, and fish provides no contribution. The results for Finland
showed a different situation, as can be seen in Figure 5-2. The figure shows the top three
contributors to the ingestion dose delivered in the first year after the accident, in the first five
years after the accident, and the lifetime ingestion dose. In the first year, milk was the major
contributor, followed by fish, and beef. For the lifetime ingestion dose, fish is as important as
milk, and beef is no longer one of the top three contributors to dose. Moreover, mushroom
consumption becomes more important than beef, grain, fruits or leafy vegetables. None of the
codes selected for use in the TSD include this pathway. Although fish and mushrooms may not
be important at many U.S. sites, the possibility of atypical food sources dominating the exposure
to humans should not be ignored when assessing exposures to members of critical population
subgroups.
5.3.4 Leaching Rates for Anions
Table 3-1 of the TSD lists the ingestion of drinking water as the most important exposure
pathway for 129I and "Tc. At least for 129I, this result is probably a consequence of the
unrealistically low default value used by RESRAD for the 129I distribution coefficient (Kd) used to
model its rate of leaching from surface soil (a value of 1 is listed in Table 3-13, p. 3-43, of the
TSD). Iodine-129 is expected to complex with organic matter at the soil surface, and thus be
retained in this material; hence a higher Kj value is warranted for some sites. In this case, the risks
associated with the ingestion pathways, particularly milk and meat, would probably dominate the
total risk associated with this nuclide. Similarly, the risk associated with "Tc in surface soil
should also be highly dependent on the chemical form of the radionuclide. Differences in the
chemical form of "Tc were not taken into account in RESRAD. For both 129I and "Tc, the food
ingestion pathway should be more important than the groundwater pathway for a rural-residential
scenario where the source of contamination is the surface soil.
5.3.5 Food Consumption Rates
49
-------
In Chapter 3 of the TSD, a base case analysis was made of a Generic Test Site using the
RESRAD code. The parameter values considered are given in Table 3-11 of the document.
Explanation for the parameter values is provided on pages 3-48 to 3-50. However, the values
used in the TSD calculations differ by a factor of two from those in the cited references (EPA
1989a; ANL 1993a) for the milk, meat, fish, and fruit-vegetable-grain groups of food. On page 3-
49 of the TSD, EPA explains that it divides by 0.5 to account for the RESRAD contamination
factor described on page 3-48. It then seems to apply its own contamination factor (0.25 for leafy
vegetables). Our understanding is that RESRAD calculates contamination factors for plants, milk
and meat internally (within the model structure). For the particular case of a contaminated area of
10,000 m2, these factors are equal to 0.5. The intake rates are corrected (doubled) in order to
eliminate the effect of the contamination factor. It is not clear why such a roundabout approach
was taken; a more straightforward approach would have been to set the contamination factors to
1 (overriding the internal calculation) and then use the normal intake rates. In any case, better
explanations are necessary.
The Agency may want to consider increasing the assumed intake value for freshwater fish.
Superfund human health risk assessments typically assume a freshwater fish ingestion of about
two fish meals a week, equivalent to a daily ingestion rate of about 54 g/day or approximately 20
kg/yr. This value is much higher than the 4.6 kg/yr value used for the generic calculations, which
is based on a population average that includes people who do not eat fish. The fish intake rate for
an RME individual should be increased to a reasonable maximum value.
5.3.6 Recommendations Concerning the Soil/Food Ingestion Pathway Model
5.3.a Recommendation. With the generic qualification regarding non-uniformity of
source term and population mobility (see recommendation 4.3.c), the soil ingestion
pathway appears to be consistent in form with currently available methods.
However, the rate of soil ingestion assumed for the industrial/commercial land-use
scenario appears appropriate only for office (indoor) industrial activities; higher
soil ingestion rates would be expected to occur for outdoor construction activities.
Hence, the Subcommittee recommends the use of a larger default value for the
industrial/commercial land-use scenario.
5.3.b Recommendation. The produce, meat, milk and fish consumption pathways are
simplified in comparison with state-of-the-art pathway analyses. Although the
Subcommittee does not recommend that they be refined at this point, the
uncertainty and variability in the risk estimates generated under the simplified
assumptions should be discussed in greater detail.
50
-------
5.3.c Recommendation. The EPA needs to clarify whether the corrections applied to
fruits and non-leafy vegetables were also applied to grains. A better explanation is
also needed for the apparent discrepancies between recommended and default food
ingestion values.
5.3.d Recommendation. Some of the default parameter values used by the EPA in its
analysis of the food ingestion pathway lead to under-predictions of the exposure
by this pathway. Specifically, the EPA should increase the fish intake value for
RME individual, revise its soil-to-plant transfer factors for those sites at which
radionuclide uptake by plants is expected to be higher than the default value, and
consider using lower leaching rates (i.e., higher Kd values) for such nuclides as 129I
and "Tc for sites in which the contaminated soil contains a high organic content.
5.4 Pathway Model for Inhalation of Particles
5.4.1 General Comments
The inhalation of resuspended soil and dust is appropriately included as a potential
exposure pathway for all three exposure scenarios (rural residential, commercial/industrial, and
suburban). The methodology used to calculate risks from inhalation of radioactive dust at the
generic site is not readily apparent from the report. Part of the confusion might be attributed to
the inclusion of the comparison of the three computer codes, RESRAD, RAGS/HHEM and
PRESTO-CPG, and from the use of different values for a number of the inhalation parameters.
Supporting documents provide information about the models and the parameters used in the
calculations, but justification for application to the cleanup conditions is lacking. Information
provided about the three computer codes indicates that all use dosimetric models that have since
been updated or revised. Relevant to the inhalation of radioactive particles is the fact that
respiratory tract doses are calculated using the 1979 ICRP model (ICRP, 1979) which has
recently been replaced (ICRP, 1994). The new ICRP model generally gives lower effective doses,
such as by a factor of three, for the respiratory tract for many radionuclides. For other
radionuclides, the new model gives about the same or somewhat higher effective dose values.
5.4.2 Default Parameter Values Used to Model Risks to Individuals
RESRAD (version 5.19) is used to model risks to on site individuals from inhaled
particles. Selection of RESRAD followed comparative calculations with PRESTO-CPG and
RAGS/HHEM codes. The following inhalation parameters are used in the risk calculations for
individual members of the population:
51
-------
Inhalation Rate (volume of air inhaled by an individual in one year): 7300 m3, from
HHEM Supplemental Guidance (EPA, 1991) (per Appendix C of the TSD). The rate is
given as 20 m3/day, 365 days per year. A value of 8000 m3/yr was used in the example
calculations comparing the three codes (page 2-55 of the TSD). Appendix C of the TSD
also gives fractions of time spent indoors and outdoors for the three scenarios for use with
the RAGS/HHEM code but there is no indication whether these were used in the
RESRAD code for this report.
Mass Loading (quantity of contaminated dust contained in each m3 of air): 1.5 xlO"6 g/m3
for suburban exposures (EPA, 1994a); 2xlO"4g/m3for rural residential and
commercial/industrial exposures (ANL, 1993b). The value given for suburban exposures
appears to be low by at least an order of magnitude and should be verified. The residential
release is in the range expected for dust particles less than about 10 //m.
Particle Size. Dose conversion factors used in RESRAD apply to dusts with an aerosol
mean activity diameter (AMAD) of 1 micrometer.
Physiological Characteristics of Exposed Individual: RESRAD code is applicable to
ICRP Reference Man (worker) (ICRP, 1975). It is not apparent that the model
accommodates adjustments for members of the public.
Shielding Factor (ratio of quantity of dust in indoor air to the quantity of dust in outdoor
air): 0.4 (ANL, 1993b)
5.4.3 Default Parameter Values Used to Model Risks to Populations
The following inhalation parameters are used in the risk calculations for populations: (see
Table 3-24, page 3-82, of the TSD):
Inhalation Rate: 2400 m3 (note: Inhalation dose conversion factors from Federal
Guidance Report No. 11 (EPA, 1988) are used. These are based on calculations for
Reference Man who works 2000 hours per year and breathes at a rate of 0.02 m3/min.
However, Appendix E gives the breathing rate used in the calculations as 8400 m3. It is
not specified whether committed dose equivalents per unit intake for lung or effective
committed dose equivalents were calculated.)
Mass Loading: SOxlO"6 g/m3 (reference is given in Appendix E, page E-5, as NRC 92. It
is probably NUREG/CR-5512 (NRC, 1992 in this report), but is not identified in the TSD
52
-------
reference list). However, TSD p. 2-55 gives a value of 100 |ig/m3, also referenced to
NRC (1992); it is unclear which value was used in the analysis.
Particle Size: Inhalation dose conversion factors from Federal Guidance Report No. 11
(EPA, 1988) are for aerosols with an AMAD of 1 micrometer.
Physiological Characteristics: Federal Guidance Report No. 11 (EPA, 1988), page 11,
says the derived guides in the report apply to Reference Man and application to other than
normal occupational exposures should consider the effect of the different conditions of
physiology, age, and sex on uptake and retention of radionuclides. This should also
include different exposure conditions such as aerosol properties.
Slope Factors. Age-specific cancer risks for intake of radionuclides from EPA (1994b)
were used.
The following inhalation parameters were used in the risk calculations for cleanup workers
(see pages 5-17 of the TSD): inhalation rate, not given; mass loading, 400 |ig/m3; physiological
characteristics, not given; slope factors, none used.
5.4.4 Updating RESRAD to Include the Revised ICRP Respiratory Tract Model
Although inhalation of radionuclides does not to appear to be the dominant exposure
pathway at the reference sites using the methods apparently used in the draft report, it seems
prudent to consider whether this would be the case if the inhalation risk calculations were more
realistically based on the exposure conditions postulated for the three scenarios at the reference
sites and if the risks were calculated for the general population and individuals in the populations
rather than for Reference Man (a worker). There are some suggestions in the report that this was
considered, but it is not clear whether parameters were adjusted for such factors as sex, age, level
of physical activity, and body size. It appears that age-specific cancer risks were applied to non-
age-specific estimates of radionuclide intakes. Calculations of risks to cleanup workers probably
do not merit this criticism.
It was undoubtedly the intent of the draft report to calculate risks that could reasonably be
applied to the general public, but the models available to EPA predated recent efforts to address
exposures of the general public by the ICRP (ICRP, 1993a, 1994). Since the cleanup standard is
to apply to the "reasonable maximum exposure" (RME) individual (page 2-3 of the TSD) , it
should be ensured that the pathway model used to support the standard and eventually used to
measure compliance with the standard, is the best available to calculate the RME. With respect to
the inhalation of particles, the recently revised human respiratory tract model of the ICRP
53
-------
provides for calculating doses to different members of the population under various exposure
conditions. Thus, using this model, the "maximum exposure that any individual is reasonably
expected to receive at a site" (RME) can be identified.
For the above reasons, the Subcommittee recommends that RESRAD be updated to
include the new ICRP human respiratory tract model (ICRP, 1994) which has been adopted by
both the ICRP and the NCRP. If this would result in an unacceptable delay in completing the
cleanup document, an attempt should be made to estimate the uncertainties resulting from
applying models developed for Reference Man (workers) to the general population and to justify
the approach. It may well be that the current effort results in conservative estimates of dose, but
the impacts on cleanup should be estimated to determine just how conservative they are and the
magnitude of associated cleanup costs. Table 5-1 that follows demonstrates that deposition of
particles differs significantly among individuals in a population, depending upon age, sex and level
of activity. The data are taken from ICRP Publication 66 (ICRP, 1994) which gives fractional
deposition in the different regions of the respiratory tract. Only total deposition values are given
in this table, although differences in regional deposition are frequently greater among members of
the population than differences in total deposition and may have a greater impact on the doses
received by sensitive tissues.
Depending upon the level of activity, deposition ranges from a low of 0.38 for a sleeping
adult female to a high of 0.67 for a 1 yr old infant undertaking light exercise. This difference in
deposition does not translate directly into a difference in dose that the two individuals would
receive because of differences in the actual amounts inspired, organ and tissue sizes, etc.
5.4.5 Findings and Recommendations Concerning the Particle Inhalation Pathway
5.4.a Recommendation. The TSD should clearly specify all parameters used in
calculations of dose and risk from inhalation of airborne radioactive dusts for the
three identified exposure scenarios and indicate ranges of values as well as those
assumed or adopted for the calculations. Although population density is addressed
for all generic sites, the characteristics of the populations themselves do not seem
to be addressed with respect to age, living behaviors or housing types. Some of
these may be accounted for by the different scenarios, but this issue should be
clarified. Analysis may conclude that none of these factors significantly affect the
final results, but it should be documented.
5.4.b Recommendation. Details of calculations of risk estimates given in TSD Table
5.9 should be described in the TSD (or, if already described, should be more
clearly identified).
54
-------
5.4.c Finding. The dust resuspension pathway as described in the EPA document
assumes that the inhalation exposure can be described by the amount of dust
expected to be in the air and that the concentration of radionuclides in the dust is
the same as that in the soil under consideration. For a small site, much of the dust
in the air even at the downwind edge of the site will come from unaffected regions
upwind of the site, and the EPA assumption could lead to substantial overestimates
of risk via this pathway. This difficulty could be
Table 5-1. Total respiratory tract deposition of 1 micrometer AMAD aerosols (Note 1)
Adult worker
Nose breather
Mouth breather
Sleeping, nose breathing
Adult male
Adult female
1 5 yr male
1 5 yr female
lOyr child
5 yr child
1 yr infant
3 mo infant
Sitting, nose breathing
Adult male
Adult female
1 5 yr male
1 5 yr female
lOyr child
5 yr child
1 yr infant
Light exercise, nose breathing
Adult male
Adult female
1 5 yr male
1 5 yr female
lOyr child
5 yr child
1 yr infant
3 mo infant
Breathing rate
(m3 / hr)
1.2
1.2
0.45
0.32
0.42
0.35
0.31
0.24
0.15
0.09
0.54
0.39
0.48
0.40
0.38
0.32
0.22
1.50
1.25
1.38
1.30
1.12
0.57
0.35
0.19
Fraction
Deposited
(Note 2)
0.51
0.34
0.39
0.38
0.39
0.38
0.42
0.46
0.54
0.54
0.42
0.39
0.40
0.39
0.44
0.50
0.61
0.53
0.53
0.53
0.54
0.58
0.57
0.67
0.66
55
-------
Heavy exercise, nose breathing
Adult male
Adult female
15 yr male
15 yr female
lOyr child
3.00
2.70
2.92
2.57
2.03
0.42
0.42
0.41
0.41
0.43
Note 1. Values from ICRP 66 (ICRP, 1994)
Note 2. "Fraction deposited" is the fraction of activity present in a volume of ambient air before it is inspired, that
is deposited.
corrected by adjusting either the dust loading or the concentrations downward
depending on the size of the area under consideration. Moreover, real atmospheric
dust loadings vary substantially with site and depend on such factors as soil particle
size distributions, vegetative cover, humidity, and precipitation patterns.
5.4.d Finding. The dust resuspension pathway does not provide for differences in dust
generation that might be expected in dry climates, where much of the soil may be
exposed to wind and easily lofted, as opposed to moist climates, where vegetative
cover and soil moisture may be effective in suppressing dust lofting over much of
the year.
5.4.e Recommendation. The TSD is unclear whether the assumed dust loading is all
respirable. EPA has stated orally that it is; this point should be clarified in the text
and the appropriateness of the loading assumptions verified. In particular, the
rationale for using different mass loading assumptions for individual and
population risk calculations should be given.
5.4.f Recommendation. The risk models available to EPA predated recent efforts to
address exposures of the general public by the ICRP (ICRP, 1993 a, 1994).
Consequently, the Subcommittee recommends that RESRAD be updated to
include the new ICRP human respiratory tract model (ICRP, 1994) which has been
adopted by both the ICRP and the NCRP. If this would result in an unacceptable
delay in completing the cleanup document, an attempt should be made to estimate
the uncertainties resulting from applying models developed for Reference Man
(workers) to the general population and to justify the approach.
56
-------
5.5 Pathway Model for Inhalation of Indoor Radon and Progeny
5.5.1 General Comments
The document emphasizes that "The inhalation of radon and its decay products is a major
contributor to total exposure when radium isotopes are present in the soil." Methods used to
estimate doses and risks from inhalation of indoor radon involve modeling the buildup of radon
and radon progeny indoors and then multiplying these concentrations by a dose or risk conversion
factor which relates airborne concentrations of radon progeny to dose and risk. The uncertainty
in the estimated risks from indoor radon is associated with these components, that is the indoor
radon concentration and the risk conversion factor. RESRAD is used to model indoor radon and
radon progeny concentrations for calculating individual doses and risks, and CU-POP is used to
calculate population doses and risks. PRESTO-CPG does not include a pathway for inhalation of
radon.
5.5.2 Individual Risks
RESRAD, which is used to calculate individual risks, employs a diffusion model based on
empirically derived constants to estimate the flux of radon into a home. As noted in the
document, RESRAD does not account for advective flow. This omission can result in
underestimation of indoor radon concentrations. However, of greater concern is the large
variability in the radon entry into individual homes. As noted in the document, the ratio of the
indoor radon concentration (pCi/L) to the soil 22(Ra concentration (pCi/g) for individual homes
can range over several orders of magnitude. The document also appropriately notes that "The
buildup of indoor radon is highly site-specific and cannot be reliably predicted for individual
homes." Even when site-specific parameters are used in the model, there is likely to be a large
uncertainty in the calculation of the indoor radon concentration in an individual home, and there
would be variability among homes on the same site. A single value of the risk estimate does not
reflect the uncertainty and variability in indoor radon concentrations. (It is also stated in the
document that the variability in the ratio of indoor radon concentration to the soil 226Ra
concentration is likely to be relatively small when the parameter of interest is the ratio for large
populations and long periods of time, which seems reasonable.)
The methodology presented in the RESRAD manual (DOE, 1993a) for calculating radon
risk for a specified indoor radon concentration is different from the methodology normally used by
the EPA (also used in the CU-POP model to calculate the population risk), which previously has
been reviewed by the RAC (SAB, 1992). Exactly what was done using RESRAD is not clearly
presented, but it appears that an effective dose equivalent was first calculated, and that this dose
was multiplied by a risk conversion factor, which was determined largely from Japanese A-bomb
57
-------
survivors. This is not the recommended approach for estimating risks from radon exposures
(ICRP, 1977)
The analysis of indoor radon appears to depend solely on the diffusion of radon from soil
into a building. If so, the accounting for all significant transport mechanisms is incomplete. The
dose contribution from buildings contaminated with radium is not addressed. Further, it is unclear
how indoor radon can be modeled for all future structures in that assumptions about building
characteristics of housing in the future can substantially influence the radon risk estimates.
5.5.3 Population Risks
The document points out the fact that CU-POP does not explicitly model the transport
and buildup of indoor radon. Rather, the model uses an empirically determined relationship
between indoor 222Rn concentrations and 22(Ra concentrations in soil. A similarly derived
relationship is used to estimate the 222Rn concentration in outdoor air. This approach seems
reasonable for estimating population exposures to radon because individual variations would tend
to average out. The risk associated with the indoor radon concentration is computed assuming an
equilibrium fraction of 0.5, an indoor occupancy factor of 0.6, and a risk conversion factor of
2.36xlO"4 per Working Level Month (WLM). Depletion of 226Ra in the soil by decay and leaching
is considered when calculating future risks.
The risk conversion factor of 2.36xlO"4per WLM previously has been reviewed by the
RAC (SAB, 1991). This value was determined by the EPA using the model recommended by the
National Academy of Sciences Committee on the Biological Effects of Ionizing Radiation in their
1988 BEIRIV report (NAS, 1988), along with the adjustments for homes recommended by the
National Academy of Sciences in their 1991 report titled "Comparative Dosimetry of Radon in
Mines and Homes" (NAS, 1991). The International Commission on Radiological Protection has
recommended use of a similar risk factor of 3xlO"4 per WLM (ICRP, 1993b).
The methodology for calculating the risk corresponding to outdoor radon concentrations
is not presented. The equilibrium fraction and risk coefficient for this application would differ
from those applied to indoor radon concentrations.
5.5.4 Technical Feasibility
The TSD shows that, for the rural residential and suburban land use scenarios, when radon
is considered, the radium soil concentration corresponding to a risk level of 10"4 or a dose of 15
mrem/yr is about an order of magnitude lower than the typical natural background soil
58
-------
concentrations of 0.2 to 4.2 pCi/g (presented on TSD p. 7-18). Consistent with this fact, the
document notes that "As a general rule, notwithstanding the uncertainties in the risk factors, if the
cleanup for a site containing 226Ra is set at IxlO"4 to IxlO"3 or less and includes the potential for
the buildup of indoor radon, virtually all of the soil contaminated above background will need to
be remediated." Thus, some method would be needed in the proposed regulations to determine
whether soil contamination exceeds background concentrations.
5.5.5 Findings and Recommendations Concerning the Radon Inhalation Pathway
5.5.a Finding. The methodology used for estimating population risks from exposure to
radon indoors (for a specified radium soil concentration and population density) is
reasonable, and the risk conversion factor used in this calculation previously has
been reviewed by the RAC (SAB, 1991).
5.5.b Recommendation. The total uncertainty associated with individual and population
risk estimates should be presented. Of great concern is the orders of magnitude
variability in the radon entry into individual homes (as noted in the document), and
the large uncertainty that will exist in a single calculated value that is used to
estimate the radon concentration in any home. The implications of the expected
large uncertainty in the individual risk estimate should be explored.
5.5.c Recommendation. The methodology used in RESRAD for estimating individual
risks should (but does not) account for advective flow of radon into a home. This
omission can result in underestimation of indoor radon concentrations.
5.5.d Finding. The document shows that for the rural residential and suburban land use
scenarios, when radon is considered, the radium soil concentration corresponding
to a risk level of 10"4 or a dose of 15 mrem/yr is about an order of magnitude lower
than the typical natural background soil concentrations of 0.2 to 4 pCi/g. Thus, a
key aspect of the proposed rule would be a method for determining whether soil
contamination exceeds background concentrations.
5.5.e Recommendation. The methodology used in RESRAD for calculating radon risk
for a specified indoor radon concentration is different from the methodology
normally used by the EPA (also used in the CU-POP model to calculate the
population risk), which previously has been reviewed by the RAC (SAB, 1991).
The implications of these differences should be investigated.
59
-------
5.5.f Recommendation. Consideration should be given to the implications of the fact,
as noted in the TSD, that the buildup of indoor radon can be greatly diminished if a
home is designed to preclude radon entry.
5.5.g Finding. For a large site, site-related radon exposures may occur even when not
"at home" or "at work." In the population risk analysis, the radon risks depend on
a ratio method for estimating radon in indoor and outdoor air from radium
concentrations in soil, which effectively assumes that a person is only affected by
locally generated radon. This assumption may lead to a slight underestimation of
impacts because it ignores the transport of site-related radon off-site. However,
the off-site radon impacts would likely be much smaller than those of background
radon.
5.5.h Recommendation. The TSD produced estimates of soil remediation volumes and
cancers avoided with and without consideration of radon at every site with radium
in soil. Although not stated in the document, this option was apparently selected
to enable the drafters of the cleanup rule to understand the difference in results if
the radon issue was handled separately, for example by requiring radon
remediation in any residences or places of employment built on the sites (or by
assuming that radon-resistant construction will become standard everywhere).
Although policy considerations may also have influenced this option, the
differential treatment of radon seems to require further explanation in the TSD.
5.6 Pathway Model for Groundwater Transport and Ingestion
5.6.1 Effect of Depth of Contamination on Relative Importance of Groundwater
Pathway
A critical issue not addressed in the TSD is the depth to which soil cleanup may be feasible
and reasonable, i.e., at what depth does contamination become an aquifer cleanup problem instead
of a soil cleanup problem? At various sites including Fernald, contamination has clearly
penetrated below soil levels (e.g., see p. 4-44, first paragraph, of the TSD). If the surface soils at
Fernald were "remediated" (i.e., removed), the main radiological risk would be from the deeper
soils not considered in the calculations. One could conceive of situations in which this risk would
exceed the risk posed by the original "unremediated" soils. Long-lived radionuclides with small
distribution coefficient values (Kjs) would be the most likely candidates for this sort of behavior.
By using unreal!stically large Kd values (e.g., 1600 for uranium at Reference Site II based on
Fernald), the potential risks associated with deeper contamination are ignored. The danger is that
60
-------
adequate risk reduction may not be achieved after remediation of shallow soils if deeper soil
remains contaminated.
The Subcommittee recommends that the EPA conduct sensitivity analyses on Kd values,
focusing on long-lived redox-sensitive radionuclides such as uranium and plutonium, in order to
evaluate the extent to which the relative importance of the groundwater pathway is dependent
upon this variable. A similar concern is expressed for the effect of non-radioactive mobilizing
agents, such as organic solvents, which could also enhance the migration of otherwise relatively
immobile radionuclides.
5.6.2 Hydrologic Parameter Values for the Generic Site Model
According to the RESRAD documentation (pp 198-200), the volumetric water content is
determined by estimating a yearly average. However, recharge events are intermittent and
transient phenomena, particularly in arid environments. What errors are introduced to estimation
of leaching using this average approach? Also, as indicated in the RESRAD documentation (p.
290), dispersion is not considered in the RESRAD code. As a result, breakthrough
concentrations will be overestimated by the model.
As stated in the TSD (p. 3-30), "The parameter values used in the calculations were
selected as realistic but conservative estimates of the conditions at the generic test site for each of
the three scenarios [suburban, rural residential, and commercial / industrial]." However, Table 3-
11 of the TSD shows that the hydrological data are identical for all three scenarios. Effectively,
the groundwater / leaching model is simulated using only one scenario. This scenario is illustrated
in Figure 3-4 of the TSD. Given the scenario in this figure and the data in Table 3-11, this generic
test site is very conservative with inconsistencies in the hydrological data, as follows.
The depth to the water table for the generic site is only 4 m (2 m of contaminated zone
plus 2 m of vadose zone). For many sites in the western United States, the depth to the water
table is greater by one to two orders of magnitude. The given parameters of hydraulic
conductivity K = 5,550 m/yr, hydraulic gradient i = 0.02, and porosity <&= 0.2 yield a
groundwater velocity of 555 m/yr (1821 ft/yr). Very few aquifers have velocities that are this
high, and this velocity is overly conservative. Hydrologic parameter values reported in TSD
Table 4-5 (p. 4-24) for Reference Sites -A, -B, and -C (subsets of Reference Sites XIII and XVI-
XXI) lead to even more extreme and unrealistic groundwater velocities of 4.5 to 6.7 km/yr. In
these cases, the large hydraulic gradients assumed for these cases appear to be inconsistent with
the associated large hydraulic conductivities. Velocities calculated for the remaining reference
61
-------
sites range from 2 mm/yr (Reference Site IX, based on Rocky Flats) up to 240 m/yr (Reference
Site II, based on Fernald).
The pumping for all scenarios is 250 m3/yr or 0.13 gal/min. This pumping rate is low.
The well intake depth is at 3 m below the water table, which is too shallow. Such well
construction for many types of wells is probably not allowed in many States. Because of seasonal
water-level fluctuations and for more realistic pumpage and groundwater flow conditions, a pump
set so shallow would go dry many times during the year. Such a depth is overly conservative and
unrealistic.
For the generic base case, the precipitation is 1 m/yr and infiltration is 50% or 0.5 m/yr
(Table 3-12 of the TSD). These values are very high and overly conservative. In the RESRAD
documentation (p. 198), this infiltration rate is based on an irrigation rate of 0.2 m/yr. What
about sites where there is no irrigation? What is the justification for selecting a hydraulic
conductivity above the water table that is 25 times less than the hydraulic conductivity below the
water table?
The infiltration rate and thickness of the vadose zone are addressed in the sensitivity
analysis. Other parameters varied in the sensitivity analysis include area of the contaminated zone,
thickness of the contaminated zone, and distribution coefficients. No sensitivity analysis was
performed on any aquifer parameters.
In summary, several aspects of the generic site model are overly conservative and
unrealistic. The importance of these parameters was not assessed in a sensitivity analysis.
Additional work is required before the importance of the drinking water pathway can be fully
assessed.
5.6.3 Transport Processes and Parameter Values
A major assumption in the source-term definition is that the soil volume containing the
highest concentration of the dominant nuclide under consideration also contains the highest
concentration for each nuclide at the site. This assumption implies that all radionuclides migrate
together at roughly the same velocity, which is not generally the case. There is no discussion of
how this assumption may or may not bias the result of the calculations.
Radionuclides such as 137Cs, 226Ra, 232Th, and 239Pu are generally highly attenuated in soils
due to their high affinity for various mineral surfaces. Radium-226 and 137Cs are somewhat of a
special case because they are sorbed primarily by ion exchange processes. If ion-exchanging
62
-------
minerals (e.g., clays) are deficient in the soil, 226Ra and 137Cs may not be attenuated as much as is
assumed in the calculations. For this and other reasons, it is clear
that site-specific data (e.g., actual depth of radionuclide penetration) are essential for realistic risk
calculations.
Existing data in the scientific literature suggest that colloidal species can enhance the
subsurface transport of radionuclides, such as 232Th and 239Pu, that otherwise would be fairly
immobile (e.g., see reviews of this subject by McCarthy and Zachara (1989) and Triay et al,
(1995)). Similarly, organic solvents may increase the mobility of radionuclides. Yet these
transport mechanisms are completely ignored in the TSD modeling.
The RESRAD groundwater module assumes perfect mixing in an idealized homogenous
aquifer with linear equilibrium adsorption and radioactive decay. In reality, the processes
determining radionuclide speciation are much more complex than given by the linear equilibrium
adsorption model, with adsorption and phase transformations dependent on pH,
oxidation-reduction (redox) state and multicomponent interactions. A geochemical model is
required to represent such interactions. At many sites, deviations from the ideal groundwater
model in RESRAD may also occur due to the presence of non-aqueous phase liquids (which may
result in facilitated transport of radionuclides), fractured media, and nonequilibrium sorption. The
EPA should bound the potential impacts on radionuclide transport of differences in local
geochemical conditions at reference sites (e.g., redox state of uranium).
The equation used by RAGS/HHEM in section 2.1.7.2 (bottom of p 2-33 of the TSD)
relates the groundwater (soil-water) concentration (Cw) to the soil concentration (Cs). Ignoring
the dilution factor (DF, which transforms groundwater concentration in the contaminated soil
zone to that at the receiving well), the equation can be rearranged to express the soil
concentration Cs as a function of the groundwater Cw concentration in the contaminated zone:
cs = [Kd + es/P)]cw
The first part of the equation (CwKd) represents the radionuclide that is actually bound to the soil
matrix. The second part [(6 S/p)Cw] represents the radionuclide that is in the interstitial water
(6=soil porosity; S=fraction of porespace saturated with water; and p=bulk soil density). This is
an appropriate way to represent Cs It implies, however, that such a relationship is in fact how Cs
will be measured when implementing the regulation, i.e., take the total mass of radionuclide in the
sample (soil+interstitial water) and divide by the dry weight of the soil. Because the definition of
"soil concentration" is a matter of controversy, this needs to be explicitly addressed-and the issue
does not appear to be addressed anywhere
63
-------
in the review document. Because this is an issue related to the implementation of the proposed
regulations, it is again addressed in Chapter 8.
Underlying assumptions and model formulations for the individual transport processes are
very important in deriving reasonable risk levels. For example, for the infiltration / leaching
module, the simple linear basic equation used in RESRAD is logical and easy to understand:
I = (1 - CE) [(1 - CR) P + IR]
where I = infiltration rate, CE = evapotranspiration, CR = runoff, P = precipitation rate and IR =
irrigation rate. However, the parameter values often have wide ranges and, in many cases, are not
available for a specific site. When the parameter values are selected from the literature or when
heavy reliance is placed on the default values in the code, compound errors can be very large,
even a couple orders of magnitude. This point emphasizes the need for uncertainty analyses, in
which the effect of ranges for each parameter can be appropriately taken into account.
5.6.4 Findings and Recommendations Concerning the Groundwater Pathway
5.6.a Recommendation. With respect to the groundwater pathway, several aspects of
the generic site model are overly conservative and unrealistic. The importance of
these parameters was not assessed in a sensitivity analysis. Additional work is
required before the importance of the drinking water pathway can be fully
assessed.
5.6.b Finding. The population risk model (CU-POP) incorporates the assumption that
population exposures and risks can be directly related to the inventory of
radionuclides at a site through the assumption that a constant fraction of the
groundwater originating at the site is consumed for drinking water by some
person, independent of the distribution of on-site and off-site residents. In reality,
groundwater use is probably extremely dependent on such site-specific factors as
the quality of the water before being affected by radionuclides, details of the
regional hydrogeology, and discharge rates to surface water.
5.6.c Recommendation. EPA should make a greater effort to identify and use site-
specific IQ values for long-lived radionuclide species sensitive to
oxidation/reduction (redox) conditions. Sensitivity analyses should be conducted
on Kd values for specific reference sites where these nuclides are present (i.e., not
just for the generic base case) in order to assess the sensitivity of the model results
to this parameter. The objective should be to bound the potential impacts on
64
-------
radionuclide transport of differences in local geochemical conditions at reference
sites (e.g., redox state of uranium).
5.7 Pathway Model for Surface Water
The surface water models for RESRAD and RAGS/HHEM are very simplified,
considering a pond receiving runoff from the site diluted according to the fraction of the
contributing watershed which is occupied by the site. The runoff concentration associated with
the site is the leachate concentration of the interstitial soil water in the contaminated zone. This
approach ignores contributions from particulate-phase radionuclides associated with soil that is
eroded from the site as part of the rainfall-runoff process. For sites having small contaminated
areas, dilution due to runoff from the uncontaminated parts of the watershed may be sufficient to
negate the importance of this exposure pathway. However, the aftermath of the Chernobyl
accident is relevant in that it shows that erosion can be a significant source of contamination for
surface water bodies when large portions of the watershed areas are contaminated. Radionuclides
can be transported by runoff water either in solution or attached to particles, according to the
chemical and physical properties of the contaminant. In the case of the Chernobyl nuclides, both
kinds of transport were identified. However, the main point here is that surface water
contamination by water runoff must not be neglected because there are conditions when it is more
important than contamination by ground-water transport (Konoplev et a/., 1995; Bulgakov and
Konoplev, 1992; Saxen and Rantvaara, 1987; Broberg and Andersson, 1991; Hammar et a/.,
1991). Consideration should be given to adding a soil erosion/runoff component (with associated
radionuclide concentrations) to the surface water model for such sites.
5.7.a. Recommendation. The Agency should encourage and seek inclusion of soil
erosion and particulate-phase radionuclides in the surface water modules of future
models used for assessment of soil cleanup standards.
65
-------
6. METHODOLOGIES FOR ESTIMATING RISK TO INDIVIDUALS AND
POPULATIONS
6.1 Risk Coefficients
The outputs of the pathway models are radionuclide intake rates for exposed individuals
by inhalation and ingestion (or, in the case of external gamma radiation, the whole body dose
rate). In order to estimate the risks from these exposures, intake rates must be converted to
equivalent radiation doses and then to cancer risk or, in some cases, directly to cancer risks.
These conversions are accomplished through risk coefficients based on observations of radiogenic
cancer in humans with supporting data from experiments with laboratory animals.
In its risk assessment supporting the cleanup standard, EPA used two different methods
for expressing risk coefficients. EPA apparently used the methods of the ICRP (although not the
most recent ICRP recommendations) to calculate an effective dose7 through models of organ
doses from ingested or inhaled radionuclides. This method was used to derive a dose rate
standard (e.g., 15 mrem per year).
EPA also calculated risks directly from intake rates using slope factors which appear to
have been derived usually through models that convert intake to organ dose and then organ dose
to risk. For some radionuclides, the models appear to differ from the ICRP models, and for
others, risk per unit intake is expressed directly, as for radon. The slope factor method was used
to derive a standard expressed in risk terms (e.g., 1 in ten thousand lifetime cancer risk).
Nine references were identified in the TSD as sources for dose conversion coefficients and
slope factors for both external and internal exposures to the RME individual and the population:
a) Dose conversion factors:
EPA(1993a): Federal Guidance Report No. 12
EPA (1988): Federal Guidance Report No. 11
DOE (1993a): RESRAD manual
b) Slope factors:
"Effective dose" is the currently preferred terminology replacing the term "committed effective dose equivalent (CEDE).'
66
-------
NAS (1988): BEIRIV, with adjustments recommended by the
National Academy of Sciences (radon only), as
presented in EPA (1992b)
EPA (1992a): Health Effects Assessment Summary (HEAS) Tables
EPA (1993b): Health Effects Assessment Summary (HEAS) Tables
EPA (1994b): Estimating Radiogenic Cancer Risks
EPA (1994a): TSD Appendix Table B-l (nominally based on EPA, 1993b;
but actually from Puskin, pers. commun.; see discussion
below)
c) Dose-to-risk conversion factor:
EPA (1989a): Risk Assessment Guidance for Superfund, Vol. 1: Human
Health Evaluation Manual
Dose conversion factors for internal exposures for both population impacts and RME
doses were, according to the TSD, taken from RESRAD. Dose conversion factors for external
exposures were taken from Federal Guidance Report No. 12 (EPA, 1993a).
The TSD states that slope factors used to derive RME risks were taken from 1993
HEAST (EPA, 1993b). This statement in the TSD is incorrect and leads to confusion with regard
to the risk analyses. In fact, the values used, as given in Appendix Table B-l of the TSD, are
from a personal communication with Jerry Puskin of EPA, not from the cited reference. To
model population risks, both "total" as well as "fatal" cancer values were taken from the EPA
report, Estimating Radiogenic Cancer Risks (EPA, 1994b); these values are also the most recent
HEAST numbers. The confusion in the TSD with regard to the source of slope factors used in
deriving RME risks makes it difficult to verify the accuracy of the calculated risks.
RESRAD dose conversion factors are presumably based on the ICRP methodology, which
weights equivalent dose according to the risk of fatality. The slope factors used in the TSD
reflect total risk of cancer, not risk of fatal cancer. Therefore, comparing effective dose based on
the ICRP methodology to RME risk is misleading and not valid. See also Section 6.4, below.
Some confusion also exists with respect to the exposure time periods used in the
document. For the RME, the exposure is assumed to occur over a thirty-year period, from birth
to age 30. For the population impacts, the exposures are assumed to continue for a lifetime, 70
years. EPA has stated orally that it is reasonable to assume that the most exposed person will
remain so at most 30 years, but that such a person could be born just as exposure begins, whereas
for population risk, people will move in and out of high exposure areas, but overall they will be
exposed 70 years. This rationale needs to be more clearly stated in the TSD.
67
-------
The TSD also refers to a 50-year dose commitment. This phrasing may be a source of
confusion for readers unfamiliar with health physics terminology and should be explained in the
TSD. The 50-year dose commitment is the total effective dose from intake of a radionuclide, i.e.,
the integrated dose from the time of intake to 50 years post intake. This is the period over which
an exposure received at age 20 (the beginning of an occupational exposure) is committed. For
radionuclides that reside in the body for long periods of time, this dose may be spread over 50
years, but for shorter lived radionuclides, the 50-year dose commitment may be incurred mostly in
a much shorter period.
In the remainder of this chapter, the Subcommittee discusses first the calculations that
derive doses from intake estimates and, second, the conversion of dose to risk. Finally, the
difficulties of using two different systems are discussed.
6.2 Exposure-to-Dose Calculations
It appears that not even the calculations of dose from risk in the RESRAD code are based
upon the latest metabolic and dosimetric models that have been published by the ICRP and the
National Council on Radiation Protection and Measurements (NCRP) in recent years. In the past,
the overall policy of EPA (as well as those of DOE, NRC and DOD) has been to use the models
recommended by the ICRP and NCRP. For example, Federal Guidance Report No. 11 (EPA,
1988) gives limiting values of radionuclide intake and air concentration and dose conversion
factors for inhalation, submersion and ingestion derived directly from ICRP publications. The
values are appropriate for a "reference man," specifically, a male worker. Because Federal
Guidance Report No. 11 was published in September 1988, it does not reflect the work of the
ICRP over the past seven years in developing models for all members of the public. EPA does
not seem to have implemented the more recent ICRP 60 (ICRP, 1990) recommendations or
NCRP (1993) recommendations, as evidenced by the use of the obsolete term, committed
effective [annual] dose equivalent (CEDE) (see TSD page 3-2) instead of "effective dose," which
is in current use world wide. It is noted, too, that EPA (as well as other government agencies)
have not yet adopted SI units, although Federal Guidance Report No. 11 gives values in both SI
and conventional units.
In its January 1995 presentation to the Subcommittee, ORIA stated that EPA is required
to use Federal Guidance Report No. 11 (EPA, 1988) in developing cleanup standards. Federal
Guidance Report No. 11 was prepared to implement "Radiation Protection Guidance to Federal
Agencies for Occupational Exposure," signed into law by President Reagan on January 20, 1987.
Federal Guidance Report No. 11 thus was intended to provide protection against the intake of
radionuclides in the workplace and, as such, its application to members of the public may not be
68
-------
appropriate. At best, EPA's use of Federal Guidance Report No. 11 makes its projection of the
"maximum exposure that [any individual] is reasonably expected to [receive] at a site" (RME)
quite uncertain. Recent publications of the ICRP (ICRP, 1989, 1993a) show that the adult is
more frequently the least exposed and the 3-month old infant is the most exposed individual in a
population following ingestion of radionuclides. Examples are given in Table 6-1 on the
following page, which compares ingestion dose coefficients to age 70 yrs for adults and 3-month
old children (ICRP, 1993a). The values of dose coefficients for other members of the public are
generally between those for the adult and 3-month old child.
Except for the transuranic elements, the dose coefficients for adults are similar to those
given in Federal Guidance Report No. 11 (EPA, 1988). Metabolic models for the transuranics
were updated after publication of Report No. 11. The ingestion dose coefficients for 3-month old
infants range up to almost 50 times those for the adult. It is to be noted, however, that the
effective dose coefficients tabulated in ICRP (1993a) and illustrated in the above table are for
acute intakes and that coefficients for chronic intakes (30 years in the case of the EPA scenarios)
may be less where growth is substantial during the exposure period. Nevertheless, the lifetime
effective dose will be greater for a person whose exposure begins as a child than for a person
whose exposure begins as an adult, even though the duration of exposure is the same.
Whether or not EPA's conclusions about soil volumes requiring cleanup or cancers
averted by cleanup would be significantly affected by the differences in Federal Guidance Report
No. 11 and the current ICRP recommendations is unknown. Ideally, EPA should update its
analysis using the latter. At a minimum, it should state the existence of more recent and age-
specific dose coefficients and explain what the effect might be on the results.
69
-------
Table 6-1. Ingestion Dose Coefficients to Age 70 Years (Sv/Bq) (from ICRP, 1993a)
Radionuclide
H3 water
H3 organic
C14
Ru106
Cs137
Ce144
pu238
pu239
pu240
pu241
Am241
Np237
Ra224
Ra226
Ra228
Sr89
Sr90
Adult
1.8E-11
4.2E-11
5.8E-10
7.0 E-09
1.4E-08
5.2 E-09
2.3 E-07
2.5 E-07
2.5 E-07
4.8 E-09
2.1 E-07
1.1 E-07
6.3 E-08
2.8 E-07
6.6 E-07
2.6 E-09
2.8 E-08
3 -Month
Old Child
6.3E-11
1.2E-10
1.6 E-09
8.4 E-08
2.1 E-08
6.6 E-08
4.0 E-06
4.2 E-06
4.2 E-06
5. 7 E-08
3. 7 E-06
2.0 E-06
2.8 E-06
4.7 E-06
3.1E-05
3. 6 E-08
2.3 E-07
6.3 Exposure-to-Risk Calculations
The calculations of risk directly from exposure, both in the RESRAD code and in the EPA
model for estimating cancers averted from the radionuclide inventory at a site, are based on
coefficients developed independently by EPA (1994b) from information provided by the National
Academy of Sciences (NAS, 1990) and other sources. In developing those estimates, EPA
requested the RAC to review the methods for estimating risk from estimated doses to the whole
body or to various organ systems. The RAC did not review EPA's methods for estimating dose
from exposure to specific radionuclides. While the methods for estimating risk from dose were
found acceptable by the RAC (SAB, 1992), they differ in detail from those used by the ICRP.
The methods for estimating dose from exposure apparently are also different from those used by
the ICRP. Together, these differences lead to sometimes markedly different results for various
nuclides.
70
-------
Lacking a clear statement of the above history, it is easy to be skeptical of the EPA
approach, which uses the slope factor tables in EPA (1994b) to translate radioactivity intakes by
ingestion and inhalation directly to risks for each radionuclide and for each body tissue and organ.
The EPA methods for translating exposure to dose and dose to risk must be understood from the
original document (EPA, 1994b), not from the TSD.
The Subcommittee cautions EPA to be consistent when it specifies how compliance with
its standard would be verified. Perhaps the final TSD will dictate how doses are to be calculated
at the cleanup sites to assess compliance with the cleanup standard (15 mrem/yr is being
suggested). In any case, the dosimetric methods used for the compliance calculations should be
the same as those which led to the establishment of the risk associated with a dose rate of 15
mrem/yr. There is a potential for confusion if the risk limits are established with the EPA risk
factors, while the doses to determine compliance are calculated with the methods of the ICRP and
NCRP (Effective Dose using quality factors and organ weighting factors) (ICRP, 1989).
In the NRC Generic Environmental Impact Statement (GEIS) in Support of Rulemaking
on Radiological Criteria for Decommissioning of NRC-Licensed Nuclear Facilities (NRC, 1994),
it is noted that the EPA risk factor published in the EPA NESHAPS Background Information
Document (EPA, 1989b) was used. Because the EPA has recently revised its risk factor (EPA,
1994b), the application of this risk factor to estimate the "maximum exposure (effective dose) that
any individual is reasonably expected to receive at a site" for the TSD would not seem to be a
departure from EPA-approved methodology.
6.4 Use of Two Methods for Relating Exposures to Risks
As explained above, EPA uses two methods for translating estimates of exposure to
estimates of cancer risk, one that passes through an explicit estimate of dose and one that jumps
directly from exposure to risk. If EPA chooses to continue calculating risks in both ways, it
should still recognize the difficulties inherent in such an approach.
Because the two methods have different endpoints (fatal cancer vs. total cancer incidence)
and are derived using different methodologies, the results should not be directly compared.
Nevertheless, the EPA has effectively compared them, calculating values of "risk per mrem" by
dividing the ICRP values for dose per unit intake into the EPA values for risk per unit intake. If
the two methods were essentially identical, then the ratios for all radionuclides should be the same
and equal to the risk per mrem value derived from the Japanese data. In fact, the ratios vary
substantially: in Tables 3-1 to 3-3 of the TSD, using the RESRAD assessment code, they range
from 3.66 x 10'8 per mrem to 2.96 x 10'6 per mrem, a factor over 80. With the PRESTO code
(not used for the actual assessment), the range is a factor of 300. Because the dose per unit
71
-------
intake varies by organ system and therefore with the route of exposure, the ratios also differ for
different exposure scenarios.
The risk per mrem values for the more important radionuclides tend to cluster near the
mean and the importance of the greater variations is less than it might first seem. Nevertheless,
the large variation is clear evidence for the difference in the two approaches. The inclusion of risk
per mrem in the tables is at best confusing and may adversely affect the credibility of the analysis.
6.5 Comparison between TSD and NRC Estimates of Mortality from Exposure to
Residual Radioactive Soils at NRC Licensed Nuclear Facilities
The Nuclear Regulatory Commission (NRC) and the Environmental Protection Agency
(EPA) have both conducted analyses of the impacts of various levels of cleanup of radionuclide
contamination at nuclear facilities in support of cleanup standards. The objective of the NRC
analyses was to assess the environmental impact of the proposed amendment to the regulations in
10 CFR Part 20 to include radiological criteria for decommissioning of land and structures at
nuclear facilities licensed by the NRC [Generic Environmental Impact Statement (GEIS); NRC,
1994]. The objective of the EPA analysis was considerably broader, in that it was to assess the
environmental impact of the proposed regulations in 40 CFR Part 196 that would specify
radiological criteria for soil, ground water, surface water, and structures at Federal facilities to be
released for public use.
The EPA's Technical Support Document (TSD) (EPA, 1994a) and the NRC's GEIS used
similar types of information, such as estimated source terms, to derive estimates of impacts,
including potential fatal cancers, for various cleanup and occupancy scenarios. Although the
methods used by the two Agencies differ significantly, a comparison of the two documents was
judged worthwhile in that it could provide a degree of confidence in the TSD estimates to the
extent that there were cases for which the results of the two analyses were similar.
The TSD includes 22 reference facilities in the analysis of impacts; the GEIS uses ten
reference facilities. Four of the reference facilities are common to both documents: commercial
nuclear power reactors, research reactors, rare earth extraction facilities, and uranium fuel
fabrication facilities. Appendix B (TSD) provides details of the comparison of the TSD and GEIS
source term assumptions, including radionuclide concentrations in soil and areal extent and
volume of soil contamination, the use of the source terms in calculating impacts, and the results of
the analyses in terms of fatal cancers projected at various residual dose values. The comparison
shows that EPA estimates of soil volume at these reference sites were significantly larger than
NRC estimates, by factors up to 100. However, despite the difference in methods and in source
term characterization, the two agencies derived similar estimates of the number of fatalities
72
-------
associated with a population residing on the contaminated soil, well within a factor often of each
other, for three of the four reference facilities. The exception was the case of the uranium fuel
fabrication facility, for which mortality estimates in the TSD were higher than those calculated by
the NRC by factors ranging from 2.5 to 100. The reasons for this difference are not clear and
should be investigated by the EPA.
6.6 Findings and Recommendations Concerning Risk Estimation Methodologies
The set of risk coefficients used in the TSD for risk-based standards differs from that used
for dose-based standards, and neither set incorporates the latest changes recommended by the
ICRP or NCRP. This choice can lead to considerable confusion on the part of the reader. While
the Subcommittee understands the difficulties EPA faced in deciding how to reconcile its
internally generated risk coefficients with those in RESRAD derived from Federal Guidance
Reports Nos. 11 and 12, leaving the problem unresolved can lead the unwary user of the TSD
into erroneous conclusions. Although the Subcommittee is not specifically recommending that
the Agency undertake the magnitude of effort needed to produce a completely consistent
document, it does make the following recommendations:
6.6.a Recommendation. The TSD should include a discussion of the overall intent or
objective of the proposed annual dose limit as it applies to cleanup (e.g., 15
mrem/yr). The selection of appropriate assumptions and parameter values for
modeling RME exposures will depend upon whether the standard is developed as
an annual dose limit for "any" member of the public or as an annual measure of
compliance with the long-term (lifetime) individual risk limit. The assumptions
used to define the RME depend on the length of the exposure period that is
relevant to assess compliance with the protective limit. If the annual limit is
intended to serve as a surrogate for the lifetime risk objective, then the approach is
generally acceptable and the importance of the age-specific factors presented
below is diminished. If, however, the goal of the standard is to assure that every
individual is explicitly protected to the annual limit, then the Subcommittee
recommends that EPA consider revising the TSD analysis to incorporate the age-
dependent dose factors developed by the ICRP and NCRP. This cannot be
achieved using the exposure-to-dose conversion factors tabulated in Federal
Guidance Report No. 11 which are based on metabolic and dosimetric models
developed for workers. However, proposed Federal Protection Guidance for
Exposure of the General Public permits the use of ICRP dosimetric models and
conventions.
73
-------
6.6.b Recommendation. The Subcommittee recommends that EPA give consideration
to adoption of the recent recommendations of the ICRP and NCRP that provide
updated metabolic and dosimetric models and approaches for calculating age-
dependent doses for the inhalation and ingestion of radionuclides for all members
of the public. Adoption of the ICRP and NCRP approaches offers a procedural
advantage to the EPA in that they have been extensively peer reviewed and are
widely accepted. Use of these approaches would likely increase the technical
acceptance of the standards by the scientific community. If alternative approaches
are used in the TSD, then EPA should explicitly explain its methodology and
justify its reasons for departures from ICRP and NCRP recommendations.
6.6.c Recommendation. The TSD should include a section specifically describing the
dose conversion factors and the slope factors used for each of the types of
assessments. While this information can for the most part be gleaned from the
various chapters, it is currently difficult to sort out. In addition, deviations from
published values should be adequately explained and the methodology for making
adjustments described.
6.6.d Recommendation. The Subcommittee recognizes that, by Presidential directive in
1987, EPA should use the exposure-to-dose conversion factors tabulated in
Federal Guidance Reports Nos. 11 and 12 and their subsequent revisions. This
guidance is based on a "linear extrapolation to zero" exposure-to-dose relationship
from observed, but much higher, dose-effect studies. As noted in the TSD, the
scientific community has been unable to come to a consensus on issues such as the
possibility of threshold doses below which no effects occur, the validity of
extrapolating curves from known high exposure effects to zero, and the possibility
of hormesis (the concept that small doses of radiation may be beneficial to
humans). The Subcommittee recommends that the uncertainties associated with
extending risk analyses to very low radiation exposures in the absence of scientific
consensus be reflected in the presentation of the final results.
6.6.e Recommendation. The TSD is not entirely clear on the distinction between the
assumed exposure periods for the various scenarios and the period over which the
dose commitment should be calculated. The understanding of the Subcommittee is
that the TSD assumes exposure periods of 30 years for the on-site residential
scenario and 25 years for the commercial/industrial scenario, with dose
commitments of 70 years for a person who begins residential exposure at birth vs.
50 years for a person who begins occupational exposure at age 20. The document
should include in one place a clear and explicit explanation of these distinctions
74
-------
and the rationale for choosing the values for exposure time and commitment
period.
6.6.f Recommendation. At a minimum, EPA should explain the variability in the risk
per mrem results shown in TSD Tables 3-1 to 3-3 when discussing its overall
approach to dose and risk conversions; the Subcommittee recommends that EPA
remove the risk per mrem column from the tables so that the casual user will not
be confused. Ideally, EPA should fix on a method for converting intakes to risks
and use it regardless of whether the standard will be based on risk or dose.
6.6.g Recommendation. Several risk metrics are used in the TSD to describe health
risks associated with radionuclides in soils. These include cancer fatalities, cancer
incidences, individual and population doses. It is recommended that these metrics
be fully described and used in an internally consistent manner throughout the TSD.
6.6.h Recommendation. A comparison was made of EPA and NRC estimates of
contaminated soil volumes and cancer risks associated with that soil, for
commercial nuclear power reactors, research reactors, rare earth extraction
facilities, and uranium fuel fabrication facilities. The EPA estimates of soil volume
were significantly larger than NRC estimates, by factors up to 100. Nonetheless,
for three of these four types of facilities, the estimates of the number of fatalities
associated with contaminated soil were generally comparable, within a factor of
ten of each other, thereby confirming that the TSD results based on the available
site data are not unreasonable. The exception was the case of the uranium fuel
fabrication facility, for which mortality estimates in the TSD were higher than
those calculated by the NRC by factors ranging from 2.5 to 100. The reasons for
this difference are not clear and should be investigated by the EPA.
75
-------
7. SENSITIVITY AND UNCERTAINTY ANALYSES
7.1 Overview of EPA Approach
Given that policy decisions are to be made using the numerical results of the TSD models
as one criterion, it is critical that the quantitative uncertainties about those results be disclosed and
emphasized in the presentation. In support of this objective, EPA's Technical Support Document
includes
a) a quantitative sensitivity analysis for selected assumptions and parameters using the
generic site characteristics as a touchstone,
b) a quantitative uncertainty analysis of the risk factors for the generic site,
c) a quantitative sensitivity analysis of the results for the reference sites for four major
assumptions, and
d) a qualitative discussion of other uncertainties in the reference site analyses.
The TSD does not attempt a quantitative uncertainty analysis in the sense of portraying
the confidence with which risks, soil volumes, and cancers averted can be stated to be less or
greater than various possible values. Consequently, while EPA states that volumes of soil
requiring remediation at various target risk levels and population impacts in terms of cancers
averted are probably both overestimated due to conservative assumptions in the analysis, the
degree of potential overestimation cannot be assessed very well with the information provided.
This characteristic of the assessment is important because the selection of a target cleanup risk
level will presumably be based in part on a balancing of the benefits (cancers averted) and costs
(soil volume requiring remediation). When both are estimated conservatively, but with a degree
of conservatism that is not quantified, the actual balance achieved with a given target cleanup
level may differ substantially from that estimated. Recognizing this limitation, the sensitivity and
uncertainty analyses conducted by EPA for this study do provide a number of useful insights. The
remainder of this section thus focuses on the strengths and weaknesses of the analyses that were
conducted.
Chapter 3 of the TSD includes a number of sensitivity studies of the analysis used to
determine soil concentration standards at the generic sites, including intercomparison of
predictions for three models (RESRAD, RAGS/HHEM, and PRESTO), a simple parametric
sensitivity analysis (varying parameters one at a time) for the risk factors generated with
RESRAD, a hypothetical uncertainty analysis for the RME risks at the generic sites (the
uncertainty analysis is perforce hypothetical because the sites themselves are hypothetical), and
evaluation of risk factor predictions (which should ostensively be the same) made by five different,
independent modelers. The results of the latter comparison indicated four modelers with similar
76
-------
results (within a factor of four) and one outlier due to some fundamental errors in the application.
The quantitative results are accompanied by a thoughtful qualitative discussion of the fundamental
model assumptions that result in the different predictions (although it does not address processes
omitted from all of the models, such as those identified for groundwater in section 5.6 of the
Subcommittee's report) including tradeoffs among different pathways (e.g., more mobile
radionuclides result in less direct exposure, but more groundwater exposure). This discussion
helps provide insight into the important assumptions underlying the models and the relative
importance of the different routes of exposure.
In TSD Chapter 6, sensitivity and uncertainty analyses are conducted on the expected
costs (soil volumes remediated) and benefits (fatal cancer cases avoided) of the proposed
regulation, based on analysis of the reference sites. The results are relatively sensitive to the
alternative risk level chosen for the standard (from 10"6 to 10"2 RME risk), insensitive to the time
horizon of the analysis (from 100 to 10,000 years), moderately sensitive to the land use assumed
for future use of the reference sites (rural residential land use results in factors of 2 to 5 higher
fatal cancer cases avoided than does commercial land use), and insensitive to whether indoor
radon is considered in the analysis. The analysis is informative and helps to bound the range of
impacts that can be expected, albeit given the highly idealized conceptual representations for the
physical and regulatory world. Again, the quantitative uncertainty bands are accompanied by
thoughtful qualitative discussion of the key assumptions and limitations of the analysis.
7.2 Sensitivity Analysis for the Generic Site
The sensitivity analysis of the generic site included one-by-one consideration of the
following assumptions and parameter values:
a) choice of overall risk assessment model,
b) dimensions of the contaminated soil region (area, long dimension, depth),
c) infiltration rate for precipitation,
d) distribution coefficients for radionuclides (soil vs. water), and
e) thickness of the unsaturated zone
The sensitivity analysis with respect to the model used was quite informative in that it
demonstrated that significant differences did exist among the models, but that the differences were
generally much less for the radionuclides that appear to dominate the overall need for cleanup. In
some respects, this conclusion is weakened in that all the models chosen were very similar with
respect to the treatment of the key radionuclides and pathways, but it is consistent with the idea
that we know somewhat more about the important pathways and that knowledge is captured
acceptably well in all the models.
77
-------
The sensitivity analyses with respect to the parameter values used was well documented
and effectively demonstrates the sensitivity of the RESRAD model to these parameters for the
touchstone conditions assumed for the generic site. However, the generic site was designed to
have a high potential for the groundwater pathway to be important. The sensitivity of the results
to changes in single parameter values therefore might have been different if a site less subject to
groundwater transport had been selected. Because many of the reference sites do not appear to
be dominated by groundwater concerns, the conclusions about sensitivity from the analysis of the
generic site may not be robust for the analysis as a whole. The parameters chosen for the
sensitivity analysis are called "key parameters." The EPA should provide in the TSD a short
explanation about the reasons that these particular parameters were selected for the sensitivity
analysis.
The sensitivity analysis varied a single parameter over a predetermined range while
keeping the others at the default values (TSD Tables 3-12 and 3-13). The Agency was correct in
varying each parameter across the whole range of values comprised by the 16 reference sites
(except for the contaminated zone area, where the value ranged from 30 times lower than the
minimum value of the reference sites up to a value that approximates the median value for the
area of the reference sites). However, the results of a sensitivity analysis will depend to a great
extent upon the nominal or default values adopted for the model parameters that are not varied
and upon the structural form of the equations used to represent actual processes. RESRAD
provides only approximations of the real processes. Therefore, the applicability of the sensitivity
analysis is limited. Furthermore, a similar analysis performed with the parameters in the PRESTO
or RAGS/HHEM code may show different results.
Finally, the parameter values analyzed in the generic sensitivity analysis are limited to
those of interest to a groundwater modeler, even though they do affect the results for other
pathways. For example, for 137Cs, parameters affecting the transfer from soil to terrestrial and
aquatic foodstuffs should be more important than the hydrologic parameters that were varied in
the analysis. However, parameters that affect soil ingestion and resuspension rates (such as
vegetative cover or particle size distribution) were not investigated. The sensitivity to
assumptions regarding dust loading in the atmosphere is not treated in the same fashion as the
groundwater parameters. Factors that would affect the transport of radon are not systematically
investigated. While the EPA correctly states that many of these sensitivities are linear in the
choice of parameter values, an equivalent level of presentation for them would enhance the utility
of the report. It may be prudent for the EPA to conduct a sensitivity analysis for a generic site
with characteristics similar to those of the reference sites with the greatest risk per unit soil
concentration.
78
-------
The sensitivity analysis for contaminated zone thickness showed a significant anomaly
which should be further investigated. The ratio of calculated soil concentration to base case soil
concentration at a 1 x 10"4risk level, as a function of contaminated zone thickness (Table 3-16, p.
3-58 of the TSD), shows a value two orders of magnitude higher for 230Th (5183 for a depth of 2
cm) than for any other nuclide. This indicates that the risk from 230Th in the first 2 cm of soil is a
factor of 50 less than the risk from the same total activity distributed through 2 meters of soil.
This anomaly may be due to the ingrowth of 226Ra and 222Rn over time because 70% of the risk
from 230Th is due to radon inhalation or direct radiation from the short-lived radon daughters.
The tables show that the time period of maximum risk increases with increase in depth of
contamination. However, the reasons for such a drastic difference in risk for the same total
radioactivity concentration are not obvious and should be explained.
The text describing the effect of contaminated zone thickness indicates that, at the 10"4 risk
level, the "RSC (Radionuclide Soil Concentration} will increase by a factor of approximately 100
when the thickness decreases by a factor of 100 from the base case (i.e., from 2 to 0.02 m)." This
statement is misleading because, in fact, if230Th is taken out of the average, the RSC increase is a
factor of approximately 16. For a ten-fold decrease in thickness, the RSC increase is 3.5, not
6.11, if 230Th is deleted. The inclusion of 230Th distorts the effect of increasing contaminated zone
thickness on risk from other nuclides.
7.3 Uncertainty Analysis for the Generic Site
The uncertainty analysis of the risk factors for the generic site was conducted by the
Monte Carlo method, the type of method that might be appropriate for the whole analysis if
resource limitations were not important. However, it was conducted with the RAGS/HHEM
model, which makes its findings difficult to interpret in relation to the sensitivity analyses
conducted with the RESRAD model. The choice is understandable in that RAGS/HHEM is
implemented by a spreadsheet computer program, which is especially convenient for off-the-shelf
Monte Carlo simulation programs, and it does provide some valuable insights. The Subcommittee
notes that a Monte-Carlo version of RESRAD has been developed by Argonne National
Laboratory and has been applied in international model testing programs.
The parameters evaluated statistically in the TSD were mostly those that relate media
concentrations to human exposures: water and foodstuff ingestion rates, daily time of exposure to
the site conditions, gamma shielding factor, and so on. The parameters evaluated in the generic
sensitivity analysis were almost always kept constant at their nominal (generic site) values for the
Monte Carlo analysis. Certain "policy" assumptions, such as the duration of residency in the same
location, were also maintained at their nominal values. The cancer slope factors for both isotopes
of radon were varied, but those for other radionuclides were not. Because many of the
79
-------
parameters were kept constant at the value for the generic site, the conclusions of the analysis
suffer from the same doubts as for the generic sensitivity analysis. EPA states that the Monte
Carlo type of analysis is more appropriate for a specific real site than for a regulatory analysis and
presents the results only as a "proof of concept." It does not rely on the results for any
conclusions regarding the robustness of the analysis of the reference sites. The Subcommittee
believes that the choice of parameters for which distributions were used was limited and therefore
limits the interpretation of the results. This problem is common in Monte Carlo assessments and
can give the erroneous impression that the confidence limits are well understood, when in fact
model uncertainties are not incorporated to any significant degree.
7.4 Sensitivity Analysis for Reference Sites
The sensitivity analysis for the reference sites analyzes the sensitivity of the results for soil
volume and cancers averted to the following choices:
a) the target cleanup level (10"6 to 10"2 RME individual risk),
b) the time horizon (100 to 10,000 years),
c) future use scenario (rural residential and commercial/industrial, with further
consideration of agricultural use and population density); and
d) with and without indoor radon.
All of these choices are beyond scientific resolution and are policy choices to be made by the
drafters of the cleanup rule. They therefore are extremely important for understanding the
implications of the policy choices and whether or not the choice affects the results greatly or only
marginally. In general, the choices appear to be considerably more important to the estimates of
cancers averted than they are to the estimates of soil volumes requiring remediation. However,
some of the lack of sensitivity of soil volumes is due to the manner in which the analysis was
structured. The lack of sensitivity of soil volumes to the time horizon is related to the lack of
importance of the groundwater pathway, the only situation in which maximum individual risks
would occur later than 100 years in the future. If a substantially different characterization of
source terms and site conditions were to arise in later analyses, this conclusion would need to be
revisited.
Because the above parameters are not ones that could be subjected to a probabilistic
uncertainty analysis, they do not tell us anything about the robustness of the analysis with respect
to the assumptions and parameter values chosen for phenomena that demonstrate real variation or
are subject to uncertainty about their true but unknown values. That issue is addressed with the
qualitative (or perhaps more accurately semi-quantitative) uncertainty analyses presented
80
-------
alongside the sensitivity analyses for each representative site. Those uncertainty analyses include
separate overviews of the uncertainties in soil volumes and cancers averted and site-by-site
discussion of uncertainties for each reference site.
7.5 Qualitative Uncertainty Analyses
The soil volume uncertainty analyses include a qualitative discussion of uncertainties in
site contamination patterns, an "independent" derivation of the risk factors for each site, and a
qualitative discussion of the parameter uncertainties in the risk factors. Both qualitative
discussions are thoughtful and touch on most of the important uncertainties. They cannot, of
course, provide a very good sense of the overall magnitude of the uncertainties in the estimated
soil volume needing remediation at the various target cleanup levels, but the overall conclusion
that the soil volumes could be significantly overestimated but probably not substantially
underestimated seems justified. The discussions lack much detail about how the various types of
uncertainty could interact. For example, the uncertainties in the source term could markedly
affect the importance of various uncertainties in the risk factors. The "independent" derivation of
the risk factors is an important verification that RESRAD was operating approximately as
designed. Because the secondary calculations used essentially the same assumptions and
parameter values as the RESRAD model, the conformance of results is not fully independent and
does not indicate much about the uncertainty of the final estimates.
The cancers averted uncertainty analysis includes a verification of the computer
calculations and a qualitative discussion of the uncertainties in the key calculational parameters
and assumptions for the population risk analysis. As with the soil volume analysis, the verification
step is valuable but is not truly an uncertainty analysis. Again, as with the soil volume analysis,
the qualitative discussions of uncertainty are thoughtful and relatively complete. For the cancers
averted, the TSD lacks a summary statement about the overall direction and significance of any
bias in the assumptions and parameter selections. Such a statement would be valuable.
The Subcommittee notes that any biases in the estimates of individual RME risks
associated with specific soil concentrations of radionuclides will tend to be reflected as biases in
the estimates of both soil volumes to be remediated as well as cancers averted by soil remediation.
For example, if individual risks tend to be overestimated, then the concentrations needed to
reduce risks to the target level will be underestimated, and both soil volumes and cancers averted
will tend to be overestimated. A statement to that effect is recommended.
81
-------
7.6 Generic versus Site-Specific Uncertainty Analyses
In many regulatory analyses, generic calculations of risk are best made in the form of
screening estimates in order to identify situations in which site-specific risk assessments are
warranted. The output of such screening calculations is a bounding estimate of risk that should
have a high probability of containing the true, but unknown, risk to a maximally exposed
individual. Site-specific assessments then are warranted when generic screening calculations
indicate the need for cleanup but the cost of remediation is potentially high. When the cost of
cleanup is low, the need for a detailed site-specific analysis should be marginal, at best.
The TSD is different in the sense that both its estimates of cancers averted and its
estimates of soil volumes requiring remediation are influenced by the uncertainties in the risks of
specified concentrations of radionuclides in soil. The higher the estimate of risk per pCi/g, the
more soil must be remediated to achieve a desired RME risk level and the greater number of
cancers will be averted by remediating to that level. If the risk managers will use the balance
between cost and cancers averted as one measure of the suitability of the various possible RME
risk targets, they may be misled by an analysis that consistently overestimates RME risk.
Therefore, the meaning of a screening calculation is less clear in this situation than it may be in
others.
The Subcommittee understands from oral comments by EPA that site-specific assessments
would be performed in order to determine how much cleanup, if any, is required to meet the risk-
based goals implicit in the proposed cleanup standard. The Subcommittee supports that policy.
Such site-specific assessments should use data from the site to quantify model parameters and
include all realistic pathways of exposure. Site-specific assessments that are carried beyond the
screening phase should be as realistic as possible and include quantitative estimates of uncertainty
for exposure, risk, and the cost of cleanup. The most important contributors to uncertainty
should be identified and additional effort spent to reduce the uncertainty in these important
components of the calculation of the human health risk. If EPA's generic analyses prove as
expected to be conservative in most cases, less soil will need remediation than anticipated in the
reference site analyses of the TSD. In some cases, however, EPA's assumptions may prove less
than conservative, for example in environments where uptake of radionuclides by crops and
livestock is greater than that modeled.
7.7 Findings and Recommendations Concerning Sensitivity and Uncertainty Analyses
7.7.a Finding. Given that policy decisions are to be made using the numerical results of
the TSD models as one criterion, it is critical that the quantitative uncertainties
about those results be disclosed and emphasized in the presentation. Overall, the
82
-------
Subcommittee commends the EPA on conducting sensitivity and uncertainty
analyses for the risk and soil volume calculations and on its thoughtful discussions
of the important assumptions and parameter value choices. However, because of
limitations of the analyses, the reader is left without a sound appreciation for the
magnitude of the overall uncertainties in soil volume requiring remediation and
cancers averted via remediation. The generic sensitivity and uncertainty analyses
are limited by the choice of parameter values that were varied, the need to select
nominal values for all other values when varying one parameter, and the lack of
analysis regarding model uncertainties. Although quantitative sensitivity analyses
of the reference sites are conducted with respect to policy choices such as the
target risk level, time horizons, and land-use scenarios, only a qualitative
discussion of scientific uncertainties is offered. Thus, the true but unknown values
for cancers averted and soil volumes to be remediated at each RME risk level may
be quite different from those presented in the report.
The importance of uncertainty analyses and communication of the results of those analyses
to environmental risk managers was underscored by the EPA Administrator in a recent
memorandum transmitting the EPA Risk Characterization Program to EPA staff (EPA, 1995).
Consistent with the spirit of that guidance, the Subcommittee recommends that ORIA improve its
risk assessment and characterization in the TSD in the following areas:
7.7.b Recommendation. Discussion in the TSD should clarify the purpose of the risk
calculations: are these to be screening or bounding estimates, high end estimates
(e.g., above the 90th or 95th percentile), or central tendency estimates of the true
value (risk)? The objective then should guide the selection of appropriate
parameter values used in pathway modeling.
7.7.c Recommendation. EPA should, at a minimum, discuss qualitatively any biases in
its estimates of cancers averted as well as biases in the soil volumes to be
remediated. The positive correlation of these biases with any in the individual risk
estimates should also be mentioned.
7.7.d Recommendation. To the extent possible, EPA should provide best estimates of
cancers averted (total and fatal) and soil volumes to be remediated for the various
proposed standards, as well as uncertainty ranges about each of those estimates, in
addition to the nominal values currently provided. For example, such statements
could be of the form: "The number of fatal cancers averted is likely to fall between
83
-------
x andy. EPA's conservative nominal estimate is w fatal cancers averted, while its
central estimate is z fatal cancers averted."
7.7.e Recommendation. In the future, when EPA evaluates the need for cleanup at a
specific site in response to the final cleanup standard, it should not only allow but
also explicitly encourage quantitative uncertainty analyses in the site assessment.
The level of detail in such uncertainty analyses should be commensurate with the
stakes (potential for risk reduction and cleanup costs) revealed in a screening
analysis. Such an approach is consistent with current industry standards such as
the Risk-Based Corrective Action (RBCA) guidance in ASTM ESS80-94.
7.7.f Recommendation. The TSD lists the calculated clean-up volumes required to
achieve specific risk levels from 10"6 to 10"2 in Table 6-7 of the report (TSD, p. 6-
74). It would be useful to have the collective dose or risk averted at each clean-up
level also provided in the main part of the document. This information is available
in the TSD Appendices but should be given greater emphasis.
84
-------
8. ISSUES AT THE SCIENCE/POLICY INTERFACE
The Subcommittee's charge was to review scientific and technical aspects of the
methodology used by the Agency to model radiation risks to individuals at radioactively
contaminated sites. The results of the risk assessments will be used by the Agency in making
policy decisions on cleanup levels for soil. The review by the Subcommittee primarily focused on
the evaluation of source terms, environmental transport, and estimation of risks. However, in the
course of this work, some issues were identified that were outside the scope of the
Subcommittee's charge. The following comments on these issues, which involve the interface of
science and policy, are provided for consideration by the Agency.
8.1 Radioactive Site Decontamination and Remediation Issues Excluded from the EPA
Analysis
One major concern is the separation presented in the TSD of (a) soil cleanup from (b)
cleanup of structures, (c) cleanup of aquifers, (d) disposal of radioactive waste, and (e) recycle
and reuse of materials and equipment after cleanup. This distinction is necessary for focusing on
the various topics, and simultaneous treatment of all five topics may be beyond staffing limits of
the Agency. Nevertheless, interactions among these five activities are expected to have a
substantial effect on the estimated costs and benefits of the proposed standard, and hence should
be discussed in the TSD, at least in a qualitative sense. The amounts of soil to be remediated, for
example, will depend on the extent to which structures and waste areas are also remediated for
public access or retained as controlled areas. At sites where cleanup for complete public access
seems unreasonable, it may be relatively simple either to move soil to the restricted area or to
extend the fence to include the more highly contaminated soil. Similar suggestions apply to
locations where groundwater contamination will prohibit water use for the indefinite future.
Another consequence of segregating the contaminated features of a site into discrete
categories to be treated under separate regulations is the added cost and implications which a
facility would have in complying with the regulations and these measures are likely to be vastly
inefficient and unduly expensive if these regulatory actions are separated from one another. For
example, if one area which is to remain restricted (e.g., a surface waste repository) is next to a site
to be cleaned up for unrestricted public use, it may be most cost-effective to include the more
radioactive portion of soil with the restricted area. Other potential areas of regulatory ambiguity
include contaminated pipes beneath structures and contaminated groundwater passing underneath
soil to be removed for decontamination.
85
-------
8.2 Other Soil Cleanup Issues Excluded from the EPA Analysis
The Subcommittee notes that a full risk characterization for the proposed radionuclide soil
cleanup standards will include not only human cancer risks, but also other potential health
impacts, ecological risks, risks to natural resources, and risks to the economic and social welfare
of affected individuals and communities.8 This will require evaluations that go well beyond the
individual and population cancer risk calculations which form the basis for the report, including
estimates of the distribution of risk to different subpopulations and communities (with possible
implications for environmental equity), and careful review of stakeholder concerns and priorities.
The Subcommittee recognizes that such studies are being conducted in parallel with the Technical
Support document reviewed by the Subcommittee, and that the Agency does have plans to
synthesize these studies. The Subcommittee strongly encourages these synthesis efforts, so that
no-one outside the Agency develops the mistaken impression that the cancer risk/soil volume
tradeoff calculations presented in this report are the sole driving factor for development and
evaluation of the proposed standards.
The TSD document examines only the extent of soil cleanup and the volumes of soil
removed. To evaluate fully these aspects, the EPA must consider the costs of soil removal, the
non-radioactive adverse aspects, and the disposal requirement for removed soil. The costs include
soil removal, containment, transportation, storage, and disposal, as well as monitoring to control
removal and assure compliance with transportation and disposal regulations, and landscaping the
cleared areas. Adverse impacts (beyond radiation exposure of workers and the public due to
cleanup, which has been considered in the TSD) are labor and transportation accidents and
incidents with associated injury and death, damage to the ecosystem due to soil removal, and the
withdrawal of sites from other uses for soil storage and disposal.
Additional consideration must be given to comparisons with the no-action alternative and
with in-situ soil remediation. The former consists generally of maintaining controlled areas with
the associated costs of maintenance and security and of radiation doses to on-site workers and to
the public due to effluents. In-situ remediation can be achieved by various forms of solidification
or by covering the soil surface to limit infiltration, radioactive particle suspension, and direct
radiation exposure.
Although direct radiation exposures to cleanup workers are estimated, it is recognized that
other worker risks exist. Accidents, exposures to nonradiological hazards, mixed waste issues
Vis-a-vis ecological considerations, ORIA should take into consideration the fact that other ecosystem components and non-human
animal species could have utility for exposure assessment and development of the cleanup standard. Other species could serve as useful biomonitors
and, in some cases, might serve as the focus for development of the site standard.
86
-------
and criticality incidents are of concern and probably represent collectively a greater risk to the
worker than direct radiation exposures.
8.3 Choice of Risk Metric
The importance of individual risk versus population risk is a philosophical and policy
question, not a technical matter, and the Subcommittee is not asking EPA to reconsider its
approach. The following discussion is intended to clarify the distinction between the two metrics.
Our comments are in support of the policy issued recently by the EPA Administrator on risk
characterization (EPA, 1995), which called for both estimates of individual risk and population
risk. Although the decision as to which of these should form the basis of a regulation is clearly a
risk management decision, the new policy states that both types of risk estimates must be
presented.
The TSD reports population risks (which EPA will use to estimate the benefits of the
standard in terms of cancers avoided) as well as risks to a hypothetical person with reasonable
maximum exposure (RME, which EPA will use to define the standard). The lower EPA sets the
risk-based standard for an individual, the more cancers can be avoided, but the benefits of that
more stringent standard depend on how many people will incur lower doses and risks as a result
of the standard. In the TSD, cases are analyzed for individual risk limits ranging from one in a
million (10"6) to one in a hundred (10"2). EPA has generally assumed that risks greater than one in
a hundred will be abated regardless of the soil cleanup standard EPA eventually sets. Not evident
in the TSD is whether or not EPA will set a maximum risk level regardless of the number of
cancers avoided at that level. (The option to regulate based on an annual dose to the RME
person is in principle nearly the same as setting an individual risk limitation.)
The Subcommittee notes that EPA could have chosen to base its regulation on some other
risk metric, for example on the residual population risk after cleanup for any one facility. Under
such an option, cleanup would be considered adequate if the facility-wide population risk did not
exceed a certain number (e.g., 10 cancers in 1,000 years), regardless of how high the individual
RME risk was calculated to be.
Population density has traditionally been a consideration in siting facilities for such
potentially hazardous activities as radioactive waste disposal. When population density is low, the
population risk for a given pattern of exposure to the hazards of the facility is minimized. At
contaminated sites, however, the population is already at risk and the purpose of the cleanup is to
reduce that risk. Here the question becomes whether it is sufficient to establish an acceptable
level of RME risk and set the standard accordingly, or whether EPA needs to keep population
risk below a fixed level as well. If the standard is based on individual risk, then arguments about
87
-------
future growth of population on or near the facility are less important. However, a linear no-
threshold dose-response relationship implies that population risks might be limiting in terms of
public health detriment even when individual radiation exposures are small fractions of the
variability in natural background exposures.
A more stringent standard involves moving greater volumes of material, which in turn has
environmental impacts and increases the risks of radiation exposures and conventional accidents
to remediation workers. EPA has assessed the radiation risks to such workers and has concluded
that the cumulative cancer impacts attributable to any reasonable cleanup standard are very small
compared to the potential cumulative impacts averted in post-cleanup, on-site populations. That
analysis is based on aggregate person-rem calculations, which relate to population risk, not to
individual risks. The maximum level of individual worker risk is not stated, but could be higher
than the individual risk level used for the standard9 While the Subcommittee does not assert that
such decisions are inappropriate, some mention of this difference in risk management philosophy
may be worth discussion. EPA is apparently also conducting an analysis of the risks from
conventional accidents, but the Subcommittee has not reviewed that analysis.
8.4 Time Horizons
As is well-recognized by the EPA, extrapolations in risk assessment are necessary
procedures that are fraught with inherent variability and uncertainty (EPA, 1995). In the case of
temporal extrapolations with long-lived radioactive contaminants, however, the extrapolations
carry the added uncertainties associated with an inability to factor in social, economic, and
scientific changes that are bound to occur in the long run. While these added uncertainties cannot
be assessed effectively either qualitatively or quantitatively, the risk manager should be made
aware of their existence and their implications when reaching a decision.
EPA conducted analyses of the number of cancers averted for three time periods of
exposure: 100, 1,000, and 10,000 years. The number of cancers calculated to be averted will be
one of the factors for consideration in policy decisions on a standard, and this number increases as
the time period of the exposure increases. The following comments are offered with respect to
the time period over which to consider exposure.
When the standard is implemented, large expenditures of present-day dollars will be
required to reduce predicted cancers over some future time period. As noted in the EPA's
9
The maximum risk to individual workers is not presented in the TSD because it is not considered to be pertinent to the assessment. Workers will
be covered under a radiation protection program which limits their exposures to the occupational radiation protection standard, which is considerably
higher than the 15 mrem/yr level considered in the Rule for post-cleanup on-site populations of residents or of commercial/industrial (but non-
radiation) workers (A. Wolbarst, memo to J. Watson dated May 18, 1995).
88
-------
document, there is uncertainty in predicting the number of cancers averted, and, at low doses, it is
also noted that the possibility of zero risk can not be ruled out. This uncertainly may be reduced
in the future. Extrapolating today's medical, industrial, and scientific technology into the future
adds further uncertainty. For example, cancer was not a major health concern in the 1800's,
principally because other causes of death were so prevalent. Will it still be a major concern 100,
1,000, or 10,000 years in the future? It is possible that medical researchers will discover how to
prevent or cure cancer within the next 100 years. Also, it is possible that better, more efficient
cleanup methods than are currently available will be discovered in the future. On the other hand,
it is conceivable that society may slip backwards, in terms of its ability to deal with cancer as a
health problem. For all these reasons, as the exposure time period increases, the uncertainty in the
number of cancers averted increases.
By estimating risks (cancers avertable) over time horizons of 100, 1,000, and 10,000
years, EPA has also provided the cleanup standard decision-makers with a range of options for
evaluating the benefits of the standard. However, the Subcommittee is concerned that risks
incurred more than 100 years in the future are highly speculative given the uncertainly about the
rate and direction of change in medical science and other technological and social arenas, which
could either reduce or accentuate the relative importance of cancer as a health risk in the future.
The Subcommittee therefore recommends that, in transmitting its findings to the EPA group
responsible for proposing the cleanup standard, the technical support group should emphasize the
additional uncertainly of its estimates for 1,000 and 10,000 years in comparison with those for
100 years.
8.5 Choice of Target Risk Levels
The analysis in the TSD includes calculations of the risks and soil cleanup volumes
associated with 10"2 to 10"6 risk levels. The analyses of risk associated with 226Ra suggest that
cleanup levels below 10"3 are probably not feasible because of the natural variation in the
abundance of this isotope (see p. 6-20 of the TSD). For comparison, the cleanup standard for
226Ra at uranium mill tailings sites is 5 pCi/g above background levels in the top 15 cm of soil and
15 pCi/g above background in deeper layers (40 CFR 192), corresponding to lifetime risks of
approximately 10"2 (based upon extrapolation of results in TSD Table 3-1, p. 3-4). This illustrates
a lack of consistency between the lower risk levels being considered in the TSD and those
corresponding to existing regulations dealing with radionuclide cleanup, such as the Uranium Mill
Tailings Radiation Control Act (UMTRCA) regulation (40 CFR 192).
In some cases, large variations in the local background concentrations of naturally-
occurring radionuclides such as potassium, radium, thorium and uranium make it difficult to
identify the presence of residual radioactive material from natural sources. Even on individual
89
-------
residential properties, radium concentrations can be several times higher than the concentration
value equivalent to 15 mrem/yr or 10"4 risk-based limits. Areas which have been used for disposal
of furnace or fireplace ash can contain several pCi/g of radium while other locations on the
property have background concentrations of 1 pCi/g or less. In another example, survey teams in
Wayne, New Jersey (another thorium processing facility near the Maywood site), found road-bed
asphalt/ballast material and rock gravel used in landscaping to have above-normal gamma dose
rates. The rock material came from a local quarry and had uranium and thorium concentrations of
3 and 1.7 pCi/g, respectively, and exposure rates of 60 to 80 |iR/hr (DOE, 1995)10. Identifying
concentrations of 0.08 or 0.4 pCi/g above such concentrations would be problematic. Some of
the landscaping rocks had natural concentrations that were higher by an order of magnitude. If
not appropriately qualified, a 10"4 risk-based standard could require cleanup of typical residential
material (as noted above) or enhanced monitoring costs to differentiate residual and naturally-
occurring materials. Other sources of natural interference include certain slag refuse from metal
production, phosphogypsum (fertilizer), certain ceramic building materials, and cinder blocks. It
is questionable whether concentrations that are a fraction of background have sufficient risk
associated with them to address. Furthermore, it is not clear, given that the variations in
background could exceed these levels, how one can be certain that the "clean fill" which could
have elevated natural background is not increasing risk over and above those being mitigated by
removal of the contaminated material.
A related concern has to do with the technical feasibility of implementation of the
proposed standard (p. 7-14 of the TSD). With regard to measurements, EPA concluded that a
combination of laboratory analyses and field monitoring could detect all radionuclides at
concentrations corresponding to 15 mrem/yr (i.e., 3 x 10"4risk limits). For some radionuclides,
however, only the more costly laboratory analyses have the requisite low detection limit. Costs of
monitoring have been considered a valid consideration in establishing concentration limits for
other regulatory activities, such as regulations promulgating the Safe Drinking Water Act.
8.6 Potential Application to NORM
In comments to the Subcommittee, members of the public have expressed concern that
criteria in the site cleanup regulations being developed might be applied to sources of Naturally
Occurring Radioactive Material (NORM). The Subcommittee is not offering an opinion on the
advisability of using the proposed cleanup standard as an ARAR (Applicable or Relevant and
Appropriate Regulation) or as the precedent for any NORM regulations that might be proposed in
the future. However, these possibilities should be noted in the TSD, and the costs and benefits of
Regarding the DOE reported exposure rates, the Subcommittee observes that these appear questionable for the soil concentrations given. It
should not be construed by the reader that the Subcommittee endorses these numbers.
90
-------
any such actions should be discussed in the Regulatory Impact Analysis. The Subcommittee notes
that the TSD does not provide the technical basis for cleanup criteria for NORM because the
analyses presented in the TSD do not address cancers averted and volumes of soil affected for
sources of NORM. Also, the feasibility issue noted in the preceding section would be pertinent
because 226Ra is one of the principal NORM radionuclides.
The cost/benefit analysis of the proposed regulation would be enhanced if results with and
without NORM sites were presented. Because the source term for NORM is even less well
understood than for the reference sites already analyzed, this may be an unrealistic suggestion.
EPA could, however, discuss qualitatively how the results might change if the standard were to be
applied to NORM sites.
8.7 Consistency Between Model Assumptions and Field Sampling Methods
The leachate equation used in the TSD to relate the groundwater concentration (Cw) of a
radionuclide to the soil concentration (Cs) assumes that the soil concentration is determined by
taking the total mass of radionuclide in the original sample (soil + interstitial water) and dividing
by the dry weight of the soil (see section 5.6.3 of this report). Because soil concentrations can be
measured in a variety of ways, it is important that the model assumptions in the TSD be consistent
with the methods specified for measuring soil radionuclide concentrations in the regulations. This
is an important issue affecting all soil cleanup standards, and it is important that a consistent
approach be taken both within the EPA and across Federal Agencies. The TSD should include
some discussion of standard and approved methods for measuring radionuclide concentrations in
soils and how these relate to the assumptions of the modeling exercises in the TSD.
8.8 Findings and Recommendations
8.8.a Recommendation. Because the TSD presented the issue of soil cleanup separately
from related cleanup problems, the Subcommittee had difficulty in understanding
how the Agency intended to take into account the interrelationships among its
proposed regulatory actions to deal with soil cleanup, aquifer cleanup, cleanup of
structures, cleanup of aquifers, disposal of radioactive waste generated by the
cleanup, and recycle and reuse of materials and equipment after cleanup.
Interactions among these five activities are expected to have a substantial effect on
the estimated costs and benefits of the proposed standard, and hence should be
discussed in the TSD, at least in a qualitative sense. The amounts of soil to be
remediated, for example, will depend on the extent to which structures and waste
areas are remediated for public access or retained as controlled areas. While
recognizing that the Agency is currently constrained by statutes or regulations to
91
-------
address these aspects as isolated problems, in truth, these problems are integrated,
not isolated. The Subcommittee feels strongly that the current fragmentation of
regulatory impact analyses leads to unrealistic estimates of risks and benefits.
More scientifically robust estimates of overall exposure and risk would be derived
from an integrated analysis.
8.8.b Finding. An estimate of the volume of soil to be remediated to achieve a specified
level of acceptable risk and dose is only one aspect contributing to the technical
and scientific judgment for site cleanup. Other pertinent aspects include soil
removal costs, adverse impacts of removal (including non-radioactive impacts),
availability of repositories for the contaminated soil, and the feasibility of remedial
alternatives such as soil stabilization, soil covers, and no action (continued control
and surveillance). The Subcommittee strongly endorses EPA's intentions to
include these aspects in the Regulatory Impact Analysis.
8.8.c Finding. EPA's decision to use lifetime risk corresponding to reasonable
maximum exposure as a risk metric for the proposed standard is appropriate.
Although another metric (such as the population risk attributable to a facility)
could have been used, the formulation of the standard is consistent with other EPA
waste management strategies. In its comparison of the risks of remedial activities
to those avoided through remedial action, EPA uses population risk (cancer
fatalities caused vs. cancer fatalities averted). While this choice is also reasonable,
it bypasses the issue of potentially higher individual risks to some remediation
workers.
8.8.d Recommendation. EPA should be more explicit about the reasons it chose to use
an individual risk or dose metric for the standard and a population risk metric for
describing the costs and benefits of the standard. A brief mention of the potential
for higher-than-average individual risks to remediation workers is also
recommended.
8.8.e Recommendation. By estimating risks (cancers avertable) over time horizons of
100, 1,000, and 10,000 years, EPA has also provided the cleanup standard
decision-makers with a range of options for evaluating the benefits of the standard.
However, the Subcommittee is concerned that risks incurred more than 100 years
in the future are highly speculative given the uncertainties about the rate and
direction of change in medical science and other technological and social arenas,
which could either reduce or accentuate the relative importance of cancer as a
health risk in the future. The Subcommittee therefore recommends that EPA
92
-------
should emphasize the additional uncertainty of its estimates for 1,000 and 10,000
years in comparison with those for 100 years.
8.8.f Finding. The analyses of risk associated with 226Ra suggest that cleanup levels
below 10"3 are probably not feasible because of the natural variation in the
abundance of this isotope. For comparison, the cleanup standard for 226Ra at
uranium mill tailings sites is 5 pCi/g above background in the top 15 cm of soil and
15 pCi/g above background in deeper layers (UMTRCA regulation, 40 CFR 192),
corresponding to lifetime risks of approximately 10"2. This illustrates a lack of
consistency between the lower risk levels being considered in the TSD and those
corresponding to existing regulations dealing with radionuclide cleanup, such as
the UMTRCA regulation.
8.8.g Recommendation. The Subcommittee is not offering an opinion on the
advisability of using the proposed cleanup standard as an ARAR (Applicable or
Relevant and Appropriate Regulation) or as the precedent for any NORM
regulations that might be proposed in the future. However, these possibilities
should be noted in the TSD, and the costs and benefits of any such actions should
be discussed in the Regulatory Impact Analysis. The Subcommittee notes that the
TSD does not provide the technical basis for cleanup criteria for Naturally
Occurring Radioactive Material (NORM) because the analyses presented in the
TSD do not address cancers averted and volumes of soil affected for sources of
NORM. Also, the feasibility issue noted in 8.8.e would be pertinent since 226Ra is
one of the principal NORM nuclides.
8.8.h Recommendation. ORIA should take into consideration the fact that other
ecosystem components and non-human animal species could have utility for
exposure assessment and development of the cleanup standard. For example,
other species could serve as useful biomonitors.
93
-------
REFERENCES
Abel, K.H., et al., 1986. Residual radionuclide contamination within and around commercial
nuclear power plants. NUREG/CR-4289. U.S. Nuclear Regulatory Commission,
Washington, D.C.
ANL. 1993 a. A compilation of radionuclide transfer factors for the plant, meat, milk, and aquatic
food pathways and the suggested default values for RESRAD. Argonne National
Laboratory. Report ANL/EAIS/TM-103. August 1993.
ANL. 1993b. Data collection handbook to support modeling of impacts of radioactive material in
soil. Argonne National Laboratory. ANL/EAIS-8. April 1993.
Broberg, A. and E. Andersson. 1991. Distribution and circulation of Cs-137 in lake ecosystems.
In: The Chernobyl Fallout in Sweden. The Swedish Radiation Protection Institute,
Stockholm.
Bulgakov, A. A., and A.V. Konoplev. 1992. The role of chemical speciation of radionuclides in
their transfer to surface runoff. Proc. of the Intl. Symp. on Radioecology. Chemical
Speciation - Hot Particles. Znojmo, October 12-16, 1992.
DOE. 1993a. Manual for Implementing Residual Radioactive Material Guidelines using
RESRAD. Version 5.0. U.S. Department of Energy. Argonne National Laboratory
Report ANL/EAD/LD-2, September 1993.
DOE. 1993b. Remedial investigation and feasibility study for the environmental restoration
disposal facility.U.S. Department of Energy. Report DOE/RL-93-99.
DOE. 1994a. Letter from R.F. Pelletier to D.A. Cool (U.S. NRC), transmitting U.S. Department
of Energy Comments on the NRC Draft Radiological Criteria for Decommissioning, dated
March 23, 1994. U.S. Department of Energy.
DOE. 1995. U.S. Department of Energy Comments to NRC on proposed rule on radiological
criteria for decommissioning,. From R. Berube, Deputy Assistant Secretary for
Environment to Docket for 59 FR 43200. March 10, 1995.
DOE. 1994b. Preliminary analysis of the technical feasibility of achieving various residual risk
standards for cleanup of radionuclides in soil. U.S. Department of Energy. Revision 1.
R-l
-------
EPA. 1987. Low-level and NARM radioactive wastes, model documentation PRESTO-EPA-
POP: Volume 1 - Methodology Manual, EPA 520/1-87-024-1, 1987.
EPA. 1988. Limiting values of radionuclide intake and air concentration and dose conversion
factors for inhalation, submersion, and ingestion. U.S. Environmental Protection Agency.
Federal Guidance Report. No. 11, EPA 520/1-88-020.
EPA. 1989a U.S. Environmental Protection Agency. Exposure factors handbook. Report
EPA/600/8-89/043.
EPA. 1989b. U.S. Environmental Protection Agency. Risk assessment methodology,
background information document. NESHAPS for Radionuclides - Volume I.
EPA/520/1-89-006-1.
EPA. 1989c U.S. Environmental Protection Agency. HHEM Manual, Vol. 1.
EPA. 1991. U.S. Environmental Protection Agency. Human health evaluation manual,
supplemental guidance: standard default exposure factors. OSWER Directive 9285.6-03.
EPA. 1992a. U.S. Environmental Protection Agency. Health effects assessment summary tables.
EPA. 1992b. U.S. Environmental Protection Agency. Technical support document for the 1992
citizen's guide to radon. May 20, 1992.
EPA. 1993a. U.S. Environmental Protection Agency. Federal guidance report No. 12.
EPA. 1993b. U.S. Environmental Protection Agency. Health effects assessment summary tables.
EPA. 1994a. U.S. Environmental Protection Agency. Radiation site cleanup regulations:
Technical Support Document for the Development of Radionuclide Cleanup Level for Soil
(Review Draft). September 26, 1994.
EPA. 1994b. U.S. Environmental Protection Agency. Estimating radiogenic cancer risks. EPA
402-R-93-076, June 1994.
EPA. 1994c. U.S. Environmental Protection Agency. Draft guidance for soil screening level
framework. Quick reference fact sheet. Review draft.
R-2
-------
EPA. 1995. U.S. Environmental Protection Agency. Memorandum entitled EPA Risk
Characterization Program. March 21, 1995, from Administrator Browner.
Faillace, E.R., Cheng, JJ,. and C. Yu. 1994. RESRAD benchmarking against six radiation
exposure pathway models, report ANL/EAD/TM-24. Environmental Assessment
Division, Argonne National Laboratory. 44 pages + appendix.
Glantz, S.A., and B. K. Slinker. 1990. Primer of applied regression and analysis of variance.
Chapter 2, pp. 10-49. McGraw-Hill, Inc., NY,
Hammar, J., M. Notte, and G. Neumann. 1991. Northern reservoirs as sinks for Chernobyl
Cesium: sustained accumulation via introduced Mysis Relicta in arctic char and brown
trout. In: The Chernobyl Fallout in Sweden. The Swedish Radiation Protection Institute,
Stockholm.
IAEA. 1995. Validation of models using Chernobyl fallout data from southern Finland -
Scenario S. Second report of the VAMP Multiple Pathways Working Group.
International Atomic Energy Agency. IAEA-TECDOC-XXX (number not yet assigned),
IAEA, Vienna, March 1995 (in press).
ICRP. 1975. Report of the task group on reference man. International Commission on
Radiological Protection. ICRP Publication 23, Pergamon Press, Inc., Oxford.
ICRP. 1977. Recommendations of the International Commission on Radiological Protection.
International Commission on Radiological Protection. ICRP Publication 26, 1977,
Pergamon Press.
ICRP. 1979. Limits for intakes of radionuclides by workers. International Commission on
Radiological Protection. ICRP Publication 30, Part 1, Pergamon Press, Inc., Oxford.
ICRP. 1989. Age-dependent doses to members of the public from intake of radionuclides: Part
1, International Commission on Radiological Protection. ICRP Publication 56, Pergamon
Press, Inc., Oxford.
ICRP. 1990. Recommendations of the International Commission On Radiological Protection.
International Commission on Radiological Protection. ICRP Publication 60, Pergamon
Press, Inc., Oxford.
R-3
-------
ICRP. 1993a. Age dependent Doses to Members of the Public from Intake of Radionuclides:
Part 2 Ingestion Dose Coefficients, International Commission on Radiological Protection.
ICRP Publication No. 67, Pergamon Press, Inc., Oxford.
ICRP. 1993b. Protection Against Radon-222 at Home and at Work. ICRP Publication No. 65,
Annals of the ICRP 23. Pergamon Press, Inc., Oxford.
ICRP. 1994. Human respiratory tract model for radiological protection. International
Commission on Radiological Protection. ICRP Publication 66, Pergamon Press, Inc.,
Oxford.
Kellogg, R.L., Maizel, M.S., andD.W. Goss. 1992. Agricultural chemical use and ground water
quality: where are the potential problem areas? U.S. Department of Agriculture,
December.
Konoplev, A.V., Bulgakov, A.A., Popov, V.ZE., Popov, O.F., Scherbak, A.V., and Y.V.
Shveikin. 1995. Wash-off of Sr-90 and Cs-137 from two experimental plots established
in the vicinity of the Chernobyl reactor (part I), submitted to Health Physics Journal, (in
press)
Mandelker, D.R. 1990. NEPA Law and Litigation: The National Environmental Policy Act,
Callaghan & Company, Deerfield, IL.
McCarthy, J.F., and J.M. Zachara. 1989. Subsurface transport of radionuclides, Environmental
Science and Technology. Vol. 23, No. 5, pp 496-502.
NAS. 1988. Committee on the biological effects of ionizing radiation. Health Risks of Radon
and Other Internally Deposited Alpha-Emitters; BEIRIV. BEIRIV, National Academy
of Sciences/National Research Council. National Academy Press, Washington, D.C.
NAS. 1990. Committee on the biological effects of ionizing radiation. Health Effects of
Exposure to Low Levels of Ionizing Radiation. BEIR V. National Research Council,
National Academy Press, Washington, DC.
NAS. 1991. Committee on the biological effects of ionizing radiation. Comparative Dosimetry
of Radon in Mines and Homes. National Academy Press, Washington, DC.
R-4
-------
NCRP. 1993. NCRP commentary no.8, uncertainty in NCRP screening models relating to
atmospheric transport, deposition and uptake by humans. National Council on Radiation
Protection and Measurements. September.
NRC. 1982 Technology, safety and costs of decommissioning reference nuclear research and test
reactors. NUREG/CR-1756, Vol. 1. page 3-4. U.S. Nuclear Regulatory Commission,
February 1982.
NRC. 1992. Residual radioactive contamination from decommissioning; technical basis for
translating contamination levels to annual total effective dose equivalent. U.S. Nuclear
Regulatory Commission. NUREG/CR-5512. October 1992.
NRC. 1993. Site decommissioning management plan. NUREG-1444. Division of low-level
waste management and decommissioning. U.S. Nuclear Regulatory Commission.
October 1993.
NRC. 1994. Generic environmental impact statement support of rulemaking on radiological
criteria for decommissioning of NRC-licenses nuclear facilities. U.S. Nuclear Regulatory
Commission. NUREG-1496. Vol. 1, page 5-4, column 1.
Rahola, T., Suomela, M., Illuka, E., and S. Pusa. 1991. Radioactivity of people in Finland,
fmnish centre for radiation and nuclear safety. STUK-A91. Supplement 2 to Annual
Report STUK-A89. Helsinki, March 1991.
Rantvaara, A., 1991. Radioactivity of foodstuffs in Finland in 1987-1988, Finnish Centre for
Radiation and Nuclear Safety. STUK-A78, Supplement 4 to Annual Reports STUK-A74
and STUK-A89, Helsinki, June 1991.
SAB. 1991. Letter report on radon risk estimates for general population and smokers, non-
smokers, and children. U.S. Environmental Protection Agency, Science Advisory Board.
EPA-SAB-RAC-LTR-91-001. January 22, 1991.
SAB. 1992. Evaluation of EPA's proposed methodology for estimating radiogenic cancer risks.
U.S. Environmental Protection Agency, Science Advisory Board. EPA-SAB-RAC-LTR-
93-004. December 1992.
Saxen, R., and A. Rantvaara. 1987. Radioactivity of fresh water fish in Finland after the
Chernobyl accident in 1986. Report STUK-A61, Supplement 6 to Annual Report STUK-
A55, Helsinki, Finland, Finnish Centre for Radiation and Nuclear Safety.
R-5
-------
Suomela, M., Blomquist, L., Rahola, T., and A. Rantvaara. 1991. Studies on environmental
radioactivity in Finland, Finnish Centre for Radiation and Nuclear Safety, Annual Report
STUK-A74, Helsinki, March 1991.
Triay, I, Simmons, A., Levy, S., Nelson, S., Nuttall, H., Robinson, B., Steinkampf, W. andB.
Viani. 1995. Colloid-facilitated radionuclide transport at Yucca Mountain, Report LA-
12779-MS, Los Alamos National Laboratory, Los Alamos, NM, April 1995, 24 p.
Whicker W., and T. Hinton. 1993. Health risk to hypothetical residents of a radioactively
contaminated lake, Proc., ER '93 Conference, Augusta GA, October 24-28, 1993.
Wolbarst, A.B. 1995a. Memorandum on characterizing the soil contamination at the reference
sites, to Jack Kooyoomjian and various RSCC members, dated February 9, 1995, 13
pages.
Wolbarst, A.B. 1995b. Memorandum on responses to comments and questions in Appendix A of
draft SAB report on technical support document for radioactive site cleanup rule, to Jack
Kooyoomjian, dated March 9, 1995, 23 pages.
Wolbarst, A.B. 1995c. Memorandum on March 27, 1995, Science Advisory Board Conference
Call, to Jack Kooyoomjian, dated April 7, 1995, 4 pages.
R-6
-------
APPENDIX A -COMPARISON OF SOURCE TERM INFORMATION
DEVELOPED BY THE NRC AND BY THE EPA FOR
REFERENCE NUCLEAR FACILITIES
1. INTRODUCTION
The Nuclear Regulatory Commission (NRC) and the Environmental Protection Agency
(EPA) have both conducted analyses of the impacts of various levels of cleanup of radionuclide
contamination at nuclear facilities in support of cleanup standards. The objective of the NRC
analyses was to assess the environmental impact of the proposed amendment to the regulations in
10 CFR Part 20 to include radiological criteria for decommissioning of land and structures at
nuclear facilities licensed by the NRC [Generic Environmental Impact Statement (GEIS); NRC,
1994]. The objective of the EPA analysis was considerably broader, in that it was to assess the
environmental impact of the proposed regulations in 40 CFR Part 196 that would specify
radiological criteria for soil, ground water, surface water, and structures at Federal facilities to be
released for public use.
The EPA's Technical Support Document (TSD) (EPA, 1994a) and the NRC's GEIS used
similar types of information, such as estimated source terms, to derive estimates of impacts,
including potential fatal cancers, for various cleanup and occupancy scenarios. Although the
methods used by the two Agencies differ significantly, a comparison of the two documents was
judged worthwhile in that it could provide a degree of confidence in the TSD estimates to the
extent that there were cases for which the results of the two analyses were similar.
The TSD includes 22 reference facilities in the analysis of impacts; the GEIS uses ten
reference facilities. Four of the reference facilities are common to both documents. The TSD and
GEIS source term assumptions, including radionuclide concentrations in soil and areal extent and
volume of soil contamination, are compared in Section B.2 of this Appendix. In Section B.3, the
use of the source terms in calculating impacts and the results of the analyses in terms of fatal
cancers projected at various residual dose values are examined. Conclusions concerning the
comparison of source term parameter values and results of risk calculations are presented in
SectionB.4.
2. SOURCE TERM INFORMATION AT REFERENCE SITES
Both the GEIS and the TSD address the difficulty of obtaining sufficient contamination
data to model the reference sites. The methods and sources used to estimate the soil radionuclide
activity concentrations and the areal extent and volume of contaminated soil for the four reference
facilities common to both documents are summarized below and in Table B-l.
A-l
-------
2.1 Commercial Nuclear Power Reactors (TSD Reference Site XVI)
Soil contamination levels: The level of soil contamination used in the TSD to represent
commercial nuclear power reactors is based on NUREG/CR-4289 (Abel et al., 1986), the same
reference as is used in the GEIS. In both analyses, Co-60 is considered to be the nuclide of
greatest concern. In the analysis, the TSD assumes a peak concentration of Co-60 of 300 pCi/g
(TSD, Figure 4-21, p. 4-93). The GEIS gives low, medium and high values for soil contamination
for each of the nuclides of concern.11 The "high" value for Co-60 for power reactors given in the
GEIS is 60 pCi/g (GEIS, Appendix C, Table 5.5.1, p. C 5.14)
Areal Extent and Volume: The areal extent and volume of soil contamination used in the TSD
is based on information contained in requests to the NRC for on-site disposal of soil at nine
reactors, in compliance with 10 CFR 20.302 (TSD, p. 4-92). The on-site disposal volumes for
these nine reactors average approximately 580 m3. The TSD adjusts this value to 1060 m3 to
account for soils which may be contaminated at levels above background but below the NRC
release limits for Co-60 and Cs-137. These volumes are translated in Table 4-6 of the TSD (p. 4-
27) to a surface area contamination around a reactor facility of 7,000 m2 or approximately 1.7
acres. The level of contamination is assumed to be uniform and limited to the first 15 cm of soil,
the maximum depth of the soil samples (TSD, Table 4-6, p. 4-27). The analysis in the TSD for all
reference sites assumes that radionuclide contamination is limited to a specific depth and is
uniform to that depth (TSD, p. 4-2).
The GEIS uses an areal extent of contamination of 2,000 ft2 (186 m2) (GEIS, Table 4-1, p. 4-14)
based on Abel et al. (1986) which indicates that "radionuclide contamination of soils around
nuclear power plants was limited to small patches of very low concentrations of radionuclides,"
and that "the volumes in most cases would be very small, e.g., tens of cubic meters," (Abel et al.,
(1986) p iv, 25, 26). The depth profile assumed for contamination in soil in the GEIS is based on
diffusion theory and the soil models of NUREG/CR-5512 (NRC, 1992). The model, as used in
the GEIS, assumes that the contaminant was deposited at the soil surface at time zero and has
penetrated the soil for a known period of time. The dispersion of the initial radionuclide inventory
is calculated taking into account the Kd for the particular soil and radionuclide and the soil
porosity. The assumed contamination depths, comparable for those assumed in the TSD, are not
specifically stated in the GEIS but the approximate contamination depths can be inferred from
Figures 5.5.1 - 5.5.6 in Appendix C of the GEIS (p. 5.10-5.13). In contrast to the TSD
Note that the peak levels of contamination, as reported in the TSD, are not directly comparable to the "high" level given in the GEIS. The
"high," "medium," and "low" levels given in the GEIS are intended to give some flexibility to the analysis and are average levels over the site. The
"peak" levels from the TSD may exist in only a small portion of the site. The NRC values are both alternate estimates of average concentrations, based
on the real data available at the time, and some measure of sensitivity. With these few test points, it could not qualify as an analysis.
A-2
-------
assumptions, the contamination levels assumed in the GEIS analysis are not constant with depth.
Therefore, a direct comparison between the two methods is not feasible.
2.2 Test/Research Reactors (TSD Reference Site XVIII)
Soil contamination levels: The level of radionuclide contamination used in the TSD for the
reference case is based on data from the Cintichem reactor, a 5 MWt reactor (TSD, p 4-94). Cs-
137 and Sr-90 were assumed to be the nuclides of concern (TSD, p. 4-95). The GEIS uses two
separate references cases: (a) the 5 MWt Plum Brook reactor is used as the reference test reactor,
and (b) the 1 MWt Oregon State University reactor is used as the reference research reactor, a
category that includes a number of research reactors in the kWt range. There are about four test
reactors and 59 research reactors in the United States. The TSD analysis assumes a peak
concentration of Cs-137 of 10,000 pCi/g (TSD, p 4-94 and Figure 4-22 p. 4-96). The GEIS
assumes a "high" general concentration of Cs-137 in soil equal to 20 pCi/g (GEIS, Appendix C,
Table 5.5.1, p. C.5.15).
Areal extent and volume: The areal extent and volume of contamination assumed in the TSD
analysis is also based on the Cintichem reactor. The TSD acknowledges that the pattern of soil
contamination at Cintichem may not be typical of that at a research reactor (TSD, p. 4-94). The
area of contamination at the Cintichem reactor is estimated to be 3,300 m2 (approximately 0.8
acres) around the facility and the soil volume is approximately 500 m3 (TSD, p. 4-94). The power
level of Cintichem is 5 MWt which is larger than nearly all research reactors which are generally
TRIGAs (Training, Research, and Instruction — General Atomics, which is a small reactor
manufactured by General Atomics.) or AGNs (Atomic General Nuclear — another line of small
reactors often found in universities.) of power level equal to or less than one MWt (NRC, 1982;
Table 3.1-1). NRR (NRR stands for U.S. Nuclear Regulatory Commission, Office of Nuclear
Reactor Regulation) input indicates that the Cintichem reactor is the only currently licensed
research reactor with significant soil contamination. Experience would indicate that only one
other decommissioned research reactor site has had significant soil contamination and that
contamination was primarily from byproduct material work that was performed under a separate
license at the site.
The GEIS uses a contaminated area of 5,000 ft2 (465 m2) for the reference test reactor based on
conditions at the Plum Brook reactor (GEIS, Table 4-1, p. 4-14). For the reference research
reactor, the GEIS assumes a contaminated area of 500 ft2 (47 m2).
2.3 Rare Earth Extraction Facilities (TSD Reference Site XXI)
A-3
-------
Soil contamination levels: The level of contamination used in the TSD is based on data from the
Molycorp facility at Washington, PA, a rare earth extraction facility listed in the NRC's Site
Decommissioning Management Program (SDMP) (NRC, 1993). The radionuclide of concern is
Th-232 (and its decay products). The peak soil concentration in the TSD analysis is 300 pCi/g
(TSD, Figure 4-24, p. 4-101). The "high" concentration used in the GEIS analysis is 200 pCi/g
(GEIS, Appendix C, Table C.5.5.1, p. C.5.15).
Areal extent and volume: The areal extent and volume of contamination used in the TSD
analysis is based on the Molycorp site in Washington, PA. The volume of contaminated soil is
estimated based on an areal extent of contamination of 13,800 m2 and a vertical depth of
contamination over this area of 2.5 m (TSD, Table 4-6, p. 4-27). The depth of soil contamination
over the site is estimated on the basis of borehole information gathered at the site; however, it was
noted that the concentration data from this information should be regarded as provisional (TSD,
p. 4-102). NRC (1993) notes that operations at Molycorp resulted in thorium slag which was
used as fill over portions of the site (NRC, 1993, p. A-65). The GEIS analysis uses an areal
extent of contamination of 100,000 ft2 (9,300 m2) also based on operations at the Molycorp site
(GEIS, Table 4-1, p. 4-14).
2.4 Uranium Fuel Fabrication Facilities (TSD Reference Site XX)
Soil contamination levels: The level of contamination assumed in the TSD is based on data
from the Babcock and Wilcox facility at Apollo, PA, a uranium fuel fabrication facility listed in the
NRC's SDMP (NRC, 1993). The GEIS uses both the Wilmington NC facility and the Apollo
facility. In the analysis, the TSD assumes a peak concentration of approximately 2,000 pCi/g for
U-234; 300 pCi/g, U-238; and 70 pCi/g, U-235 (TSD, Figure 4-23, p. 4-98). The GEIS,
Appendix C, Table 5.5.1 (p. C.5.15) and Figure 5.5.5 (p. C.5.12) refers only to concentrations of
U-235 and shows a "high" value of 1,000 pCi/g.
Areal extent and volume: The areal extent and volume of contaminated soil used in the TSD
are based on data from Apollo and on extrapolation of the available data. The amount of
contamination at >30 pCi/g was taken from recent Apollo information and is given in Table 4-6 of
the TSD (p. 4-27) as approximately 20,000 m2. Two values are reported for contamination depth
in TSD Appendix Table B-l (p. B-9). A depth of 36 cm was used to calculate population
impacts; the more conservative value of 100 cm was used to calculate RME risks (TSD, p. 4-99).
This information was extrapolated to estimate the total volume of soil contaminated to levels
greater than 1 pCi, approximately 490,000 m3. It is not clear what the corresponding
contaminated area is assumed to be. The GEIS assumes 50,000 ft2 (4,600 m2) as the areal extent
of contamination based on Wilmington and Apollo information (GEIS, Table 4-1, p. 4-14). As
A-4
-------
with the other reference site, the depth profile is based on diffusion theory and soil parameters.
For this site the depth of contamination is over 35 cm (GEIS, Appendix C, Figure 5.5.5, p.
C.5.12).
2.5 Summary
The "peak" and "high" estimated soil concentration levels, the areal extent of
contamination, and the estimated depth of contamination for the four common reference sites are
summarized in Table B.I. The following observations are made:
a) The estimated levels of contamination at the reference sites common to the EPA and
NRC analysis are generally of the same order of magnitude. However, the peak levels
of contamination, as given in the TSD, are not directly comparable to the "high" level
given in the GEIS. The NRC estimated soil surface contamination on the basis of
reports on the level of contamination in nuclear facilities (GEIS, p. 4-11). The "high,"
"medium," and "low" levels used in the GEIS represent average soil concentrations
over the reference site and take into account the uncertainties and potential ranges in
soil contamination. They are intended to "bound the problem" and allow the range of
potential impacts to be estimated (GEIS, p. 5-13). In contrast, the "peak" levels from
the TSD may exist in only a small portion of the site. The "peak" levels from the TSD
may exist in only a small portion of the site.
b) The bases for estimating the areal extent of contamination are similar for the two sets
of analyses. However, the values differ by orders of magnitude in some cases.
c) The methods for estimating depths of contamination are significantly different for the
two sets of analyses. The TSD assumes uniform contamination to a specific depth.
For the TSD risk analysis, discrete contaminated segments are assumed to be removed
to a uniform depth to achieve alternative dose criteria. The GEIS assumes that
contamination occurs on the surface and has dispersed vertically through the soil over
time according to characteristics of the soil and radionuclide. Dose criteria are
achieved by removing layers of contaminated soil. The concentration of radionuclides
is not uniform over depth; however, the average concentration of contaminant
remaining in the soil as layers are removed was used in the dose calculations.
3. ESTIMATES OF MORTALITY FROM EXPOSURE TO RESIDUAL
RADIOACTIVITY IN SOIL AT FOUR REFERENCE SITES
A-5
-------
The approach to calculating total cancer mortality for populations residing on sites
remediated to a specific residual dose criterion in the GEIS is significantly different from the
method used in the TSD. However, a comparison of the results of the two independent analyses,
as shown in Tables B-2 to B-5 of this Appendix, may be useful as a check on the reasonableness
of the TSD results.
For this analysis, only the residential scenario is considered. The EPA method assumes
that the contaminated structures on site are not occupied and that the dose is derived only from
contaminated soil. The residential scenario for the NRC method also assumes no occupancy of
contaminated buildings.
The NRC analysis does not consider the fatalities averted by actions taken to reduce the
dose at the unremediated site to 100 mrem/year because the NRC assumes that 100 mrem/year is
the maximum reasonable value for the residual dose criterion. Therefore, the comparison in this
appendix only considers the number of cancers averted for residual dose criteria ranging from 100
mrem/year to 0.1 mrem/year.
The EPA method used in the TSD is based on the site being remediated to a residual soil
contamination level such that the reasonable maximum exposure (RME) to any individual does
not exceed the criterion level. The population doses and risks are then calculated based on the
residual soil contamination, population density, and appropriate dose conversion factors.
Fatalities averted are calculated based on the estimated number of fatal cancers expected for the
unremediated site. The number of population fatalities averted is calculated for RME annual dose
levels ranging from 0.1 mrem/year to 100 mrem/year. This information is contained in Appendix
M to the TSD.
The NRC method, as described in the GEIS, simply calculates the total number of cancer
deaths which would be averted in a population receiving radiation doses at the residual dose
criteria levels. The NRC assumes that the criterion would not be set higher than the current 100
mrem/year limit for the general public. Therefore, the fatalities averted in remediating the site to a
condition where the annual dose would be 100 mrem, are not estimated. The number of cancer
fatalities is directly proportional to the number of individuals exposed. The population exposed
for the residential scenario is assumed to be the contaminated site area for the reference facility
multiplied by a population density factor of 0.0004 persons/m2 (400 persons/km2) (GEIS,
Appendix B, p. B-7). The exposure duration is 70 years and the risk conversion factor, 3.92 x 10"
4 per rem (GEIS, Appendix B, p. B-7). The following is a sample calculation for the power
reactor at the 100 mrem residual dose criterion using the data from Table B-l:
A-6
-------
Fatalities = Area x 0.0004/m2 x 0.1 rem/yr x 70 yr x 3.92 x 10-4/rem
= 186m2 x 0.0004m-2 x 0.1 rem/yr x 70 yr x 3.92 x lO'Vrem =
2.04 x ID'4
The EPA method uses a minimum integration period of 100 years for population effects as
compared to the NRC method which assumes a 70-year lifetime. The residual dose criteria levels
are also somewhat different for the EPA method as compared to the NRC method. In addition,
the TSD reports cumulative fatalities averted for cleanup to a dose criterion, whereas the GEIS
reports residual expected population cancer fatalities at each residual dose criteria levels.
Comparisons between the number of cancer fatalities expected for onsite residents for the
TSD and the GEIS are shown in Tables B-2 through B-5. In order to compare the two analyses,
the risk estimates from the TSD on fatal cancers averted were converted to total fatal cancers at
the residual dose criteria levels by subtracting the cancers averted at each RME dose level from
the number of cancers averted at a dose level of 0.1 mrem/year. In the EPA analysis, the majority
of the cancers averted result from remedial actions taken to reduce the RME to 100 mrem/year
except for the uranium fuel fabrication reference facility. The differences which result from
further reductions are small compared to the total fatalities averted and, for the lower dose
criteria, are outside the range of significant figures for total cancers averted. In the EPA analysis,
the significant figures reported are not sufficient to calculate the residual cancer burden for the
lower dose criteria.
It should be noted that the EPA analysis shows that the great majority of effect, in terms
of potential cancers averted, for reactor reference sites occurs due to remediation of the site to an
RME level of 100 mrem/year. The effect of reducing the RME below 100 mrem per year is much
smaller in comparison. In contrast, for the uranium fabrication facility reference site, there is a
reduction in projected total fatalities by a factor of 25 in remediating the site from a residual RME
dose level of 100 mrem/year to 0.1 mrem/year (Table B-5).
Data from TSD Appendix Tables M-89 and M-90 (TSD, p. M-89 and M-90) were used in
the comparison. These tables list calculated potential cancer deaths averted, including radon in
the analysis, for the "Reasonable Occupancy Scenario." Data for the NRC analyses were obtained
from GEIS Tables 5-1, 5-2, 5-3, 5-4 and 5-10 (GEIS, p. 5-20 to 5-25).
4. CONCLUSIONS
The following observations are made from the comparison of the EPA and NRC risk
analyses:
A-7
-------
a) The volumes and aerial extent of contaminated soil assumed by EPA in the TSD are
generally greater than those assumed by the NRC in the GEIS.
b) The estimated peak soil concentrations reported in the TSD are higher than the "high"
average concentrations used in the GEIS, as would be expected.
c) The estimated number of fatalities averted for specific dose criteria are comparable.
The NRC (GEIS) method simply multiplies the residual dose criterion level by the
population density and a dose conversion factor. In contrast, the residual dose criteria
for the EPA analysis in the TSD were based on the dose derived for the RME.
Therefore, it is not a simple conversion from the dose criterion level to potential
cancer deaths. The RME doses were calculated using RESRAD, while the population
risks were calculated using CU-POP. In addition, the integration time for population
exposure in the TSD is a minimum of 100 years while the GEIS uses the average
lifetime, 70 years, for the calculation of potential fatalities.
While the TSD assumes a much greater contaminated soil area for the commercial
nuclear power reactor reference site, the calculated number onsite resident fatalities is
lower than that reported in the GEIS for a similar reference site. The reason for this
difference is not immediately apparent but may be due to the fact that the critical
nuclide for this reference site is Co-60. The NRC method does not allow for decay
whereas the EPA method does. (NRC-1994A, Vol. 1, page 5-4, column 1, last full
sentence, which states... "Radiological decay over this period is not included in the
analysis of these impacts." Because Cs-137 was assumed to be much more mobile
than Co-60 in both soil and concrete, removal of the first layers removed virtually all
of the Co-60 leaving Cs-137 to deliver the remaining dose. The resulting error for the
assumption over a 70-year residency is thus less than a factor of 4. The half-life of
Co-60 is short enough (5.2 years) that not taking it into account would significantly
affect the results of the analysis.
The estimates for the number of onsite resident fatalities for the test/research reactor
reference sites are similar for the EPA and NRC analyses. The estimated onsite
resident fatalities for the rare earth extraction facility reference site are also similar for
the two methods but slightly lower for the TSD.
The estimates for the uranium fuel fabrication reference sites ranged from a factor of
2.5 to a factor of 100 higher for the TSD analysis compared to the values reported in
the GEIS. The reasons for this difference are not clear.
A-8
-------
d) The methods used by NRC and EPA differ significantly; however, the results in terms
of potential cancer fatalities in a population residing on the remediated site are similar,
except for the case of the uranium fuel fabrication reference site.
e) In summary, the results of the analyses of fatal cancers averted for a specific residual
dose criterion are reasonably consistent between the GEIS and the TSD. However,
because the methods differ significantly in their assumptions, it is not clear whether the
consistency is evidence for the robustness of the analysis, or simply fortuitous. In any
case, the residual cancer estimates below 100 mrem/yr are all below 0.01 cancers,
which suggests that the differences would not be very significant for regulatory policy.
A-9
-------
Table A. 1: Comparison of Source Term Parameters for the GEIS and the TSD
Reference
Site (Nuclide of
concern)
Power Reactors
(Co-60)
Test/Research
Reactors
(Cs-137)
Rare Earth Facilities
(Th-232)
Uranium Fuel
Fabrication Facilities
(U-tot)
Est. Soil Cone.1
(pCi/g)
TSD
300
10,000
300
2,300
GEIS
60
20
200
1,0003
Areal Extent (m2)
of contamination
TSD
7,000
3,300
13,800
20,000
GEIS
186
46S/464
9,300
4,600
Depth (cm) of
contamination2
TSD
15
15
250
36/1005
GEIS
<20
<40
<6
<45
1. Estimated soil concentrations are "peak" concentrations for the TSD and "high" average
concentrations for the GEIS.
2. The depths of contamination for the GEIS are estimated from Figures 5.5.1 - 5.5.5 (GEIS,
Appendix C, p. C.5.10 to C.5.12).
3. The GEIS refers specifically to the concentration of U-235 in Table 5.5.1 and Figure 5.5.4. It
is not clear whether that value is intended to reflect concentrations of U-235 or total uranium.
4. Assumed areal extent is 470 m2 for the reference test reactor (Plum Brook) and 47 m2 for the
reference research reactor (Oregon State University).
5. Two values are reported for contamination depth in TSD Appendix Table B-l (p. B-9). A
depth of 36 cm was used to calculate population impacts; the more conservative value of 100
cm was used to calculate RME risks (TSD, p. 4-99).
A-10
-------
Table A.2:
Comparison of GEIS and TSD Estimates of Radiogenic Cancer Fatalities for the
Commercial Nuclear Power Reactor Reference Site (TSD Reference Site XVI)
Residual
Dose
Criterion
(mrem)
> 100
100
75
60
30
25
15
10
3
1
0.1
On-Site Residential
Fatalities1
GEIS2
2.0 E-4
1.2E-4
6.1E-5
3.1E-5
2.0 E-5
6.1E-6
2.0 E-6
2.0 E-7
TSD3
2.18E-3
4 E-5
3 E-5
1E-5
1E-5
-------
Table A.3: Comparison of GEIS and TSD Estimates of Radiogenic Cancer Fatalities for the
Test/Research Reactor Reference Site (TSD Reference Site XVIII)
Residual
Dose
Criterion
(mrem)
> 100
100
75
60
30
25
15
10
3
1
0.1
On-Site Residential
Fatalities1
GEIS2
5.1E-4
3.1E-4
1.5E-5
7.6 E-5
5.1E-5
1.5 E-5
5.1E-6
5.1E-7
TSD3
6E-4
6E-4
4E-4
-------
Table A.4: Comparison of GEIS and TSD Estimates of Radiogenic Cancer Fatalities for the
Rare Earth Extraction Facility Reference Site (TSD Reference Site XXI)
Residual
Dose
Criterion
(mrem)
> 100
100
75
60
30
25
15
10
3
1
0.1
On-Site Residential
Fatalities1
GEIS2
1.0 E-2
6.1E-3
3.1E-3
1.5E-3
1.0 E-3
3.1E-4
1.0 E-4
1.0 E-5
TSD3
7.47 E-2
5.1 E-3
3.1 E-3
1E-3
5 E-4
3 E-4
1E-4
-------
Table A.5: Comparison of GEIS and TSD Estimates of Radiogenic Cancer Fatalities for the
Uranium Fuel Fabrication Reference Site (TSD Reference Site XX)
Residual
Dose
Criterion
(mrem)
> 100
100
75
60
30
25
15
10
3
1
0.1
On-Site Residential
Fatalities1
GEIS2
5.1E-3
3.1E-3
1.5E-3
7.6 E-4
5.1E-4
1.5 E-4
5.1E-5
5.1E-6
TSD3
1.28E-2
1.24E-2
1.24E-2
1.23E-2
1.20E-2
1.08E-2
7.1E-3
4.6 E-3
-------
APPENDIX B - DETAILED COMMENTS ON SOURCE TERM
INFORMATION USED TO DEFINE REFERENCE SITE
CHARACTERISTICS OF DOE FACILITIES
1. Reference Site I
For this reference site, EPA used aerial survey data and assumed Cs-137 concentrations to
estimate source terms and contamination volume/concentration relationships. While the analysis
itself is reasonable, the extent to which this analysis is representative of the actual Hanford site is
questionable. EPA recognized that the Hanford site included many more radionuclides than just
Cs-137. However, those listed are not the complete list, although EPA's intentions to consider
the most recent Remedial Investigation/Feasibility Study (RI/FS) documents (TSD, p. 4-39)
should address this concern. Some of these other radionuclides may significantly impact waste
volumes, depending on cleanup levels selected. The assumption that Cs-137 is the only important
radionuclide needs further consideration. For example, EPA indicated that concentrations of Pu
are well below 1 pCi/g. This may be true if concentrations are averaged over the entire site;
however, many areas have concentrations well above this level. The Remedial Investigation
Feasibility Study (RI/FS) for the Environmental Restoration Disposal Facility (ERDF) reported
maximum concentrations for Pu-239/240 of 2,800 pCi/g (DOE, 1993b). Similarly, the statement
that the U-238 concentrations are within the range of natural background is only true for offsite
regions and those unused areas of the site (which do, however, make up most of the site). DOE
(1993b) reported a maximum uranium-238 concentration of 20,000 pCi/g.
These data should help develop a reference site that is more representative than a
reference site based on aerial survey data. While aerial data may be useful, the concentrations
resulting from such data depend greatly on the assumed depth and extent of the contamination.
Hence, the concentration isopleths and associated volumes at concentration intervals are totally
dependent on the assumptions. All of the volume data are an artifact of the 5 cm depth
assumption. While this assumption is reasonable for surface deposited material (Cs-137 typically
moves very slowly through the surface and even in wet eastern sites, more than 90% remains in
the top 15 cm), many of the most contaminated areas of Hanford are not the result of surface
deposition. In addition, other radionuclides such as Sr-90 and Tc-99, migrate differently than
Cs-137 and, therefore, the 5 cm assumption may not be appropriate. Much of the soil
contamination at Hanford is at depths of 1.5 to 5 meters or more. Without incorporating
consideration for these volumes of material, the data for Reference Site I can only be considered
representative of a site contaminated with wind-blown Cs-137. This reference site requires
further analysis using the more recently released Comprehensive Environmental Response,
Compensation, and Liability Act (CERCLA) RI/FS and Focused Feasibility Study (FFS) reports
B-l
-------
for the 100, 200 and 300 areas of Hanford before it can be considered representative of the
Hanford site.
2. Reference Site II
In this reference site, the modeling considered only the upper 18 inches of soil from
several subunits of Operable Unit 5 at FEMP (Fernald Environmental Management Project). EPA
recognizes that the contamination at the FEMP site exists at depths greater than 18" but the
subsurface contamination was not considered because it did not fit well in the simplified modeling
approach used to develop the reference sites. EPA also notes that the subsurface data are difficult
to interpret.
The acceptability of this approach and use of the data are dependent on the desired use of
the data. It is possible to complete multiple runs with the models used by EPA and analyze the
more complex contamination (e.g., subsurface lenses). Furthermore, analysis of the complex data
with multiple radionuclides at low concentrations would provide EPA with insight into the
difficulty associated with designing remedial actions at low dose-based cleanup concentrations.
Therefore, if the goal of the reference site is to assess the viability of establishing cleanup criteria
at concentrations that are based on doses of 15 mrem/year and below, analysis of the more
realistic data set would seem worthwhile and appropriate. This would be more beneficial than
assuming the contamination depth was limited to 5 cm (TSD, page 4-45).
EPA should also review the waste volume/cost to concentration data provided in the DOE
March 23, 1994, comments to NRC related to their cleanup standards. While most of the sites
analyzed in these DOE comments were relatively small, there is much to learn from the ratios of
volume to concentration. Some of these sites conducted operations that were similar to those
conducted at Fernald.
3. Reference Site III
EPA chose to rely primarily on aerial survey data for this site. The volume to
concentration ratios are largely an artifact of the assumptions. Many of the comments relating to
Reference Site I also relate to this reference site. While the general shape of the concentration to
volume curve (Figure 4-12) is reasonable, this analysis appears to provide little new insight into
the relationships. It is not likely to produce good cost to cleanup concentration relationships
because it is limited to Cs-137 alone and again assumes that contamination is restricted to the top
5 cm of soil.
B-2
-------
Rather than conducting multiple reference sites that are largely controlled by the
assumptions of the depth of the contamination, it might be more useful to concentrate on one site
and conduct a more complex analysis. Such an analysis would also be beneficial in determining
how representative the simplified sites are to the actual sites.
4. Reference Site IV
Estimates of contaminated volumes from the Weldon Spring site were made by DOE
(from NEPA and CERCLA documents supporting the Record of Decision) and provided to NRC
and EPA in a March 23, 1994, Comment letter (DOE, 1994a). The estimates of contaminated
volume (rounded to 2 significant digits) for the Weldon site were:
Sludge
Sediment
Soil
Structural Mat.
Process Chemicals
Vegetation
Total
220,000
yd3
120,000
yd3
340,000
yd3
170,000
yd3
4,000 yd3
31,000yd3
885,000
yd3
170,000 m3
92,000 m3
260,000 m3
130,000 m3
3,000 m3
23,000 m3
678,000 m3
As can be seen in this example, soil represents about 40% of the waste volume from the
remedial action. (Vegetation represents about 10% of the soil volumes and, while it is typically
part of the soil restoration, it does not add significantly to the volume.) The site is
non-homogeneous. Excluding the highly contaminated raffmate pits, the soil includes a few highly
concentrated areas extending tens of centimeters in depth, with the bulk of the soil contamination
being slightly elevated surface contamination. Building and equipment debris constitute the
largest fraction of the waste (excluding the waste pit area, regardless of the soil limit selected, the
soil will comprise only a small fraction of the total). Therefore, the analysis conducted by EPA
represents only a portion of the residues that result from the action.
The EPA analysis suggests that the relationship (Figure 4-12) between the volume of
contaminated soil at a uranium standard in pCi/g is:
B-3
-------
60pCi/g 36,000 m3 (approximate from Fig. 4-12)
30 pCi/g 48,000 m3
15pCi/g 60,000m3
For comparison, DOE's analysis provided in the March 23, 1994, comment package suggested
remediated soil concentrations to volume relations as follows:
120 pCi/g 8,400 m3
60 pCi/g 20,000 m3
30 pCi/g 28,000 m3
15pCi/g 38,000m3
These volumes excluded the waste pits and quarry. In this case, the volumes differ by about 70%,
i.e., EPA estimates are between 1.6 to 1.8 times higher than the DOE estimates. Although this
difference is significant, a factor of 2 agreement is not unreasonable for volume estimates from
different modeling analyses. In any case, the log-linear relationship between soil volume and
cleanup limits seems to be a reasonable approximation for the soil at this site. The difference in
the actual volume estimates is likely due to EPA's assumptions of uniform depth of 11 cm for the
15 pCi/g volume of the waste. Contamination at the site was non-homogeneous. Many areas
consisted of very shallow surface contamination, with some localized highly contaminated ares
that extended a few tens of centimeters in depth.
In general, if the goal of this reference site is to collect data that will be used to assess the
cost of soil cleanup at a site absent of waste disposal activities and structures, this analysis of soil
limits may be adequate. However, if it is to represent a site like the Wei don Spring Site, it is not
clear how an adequate analysis of costs, risks and benefits can be completed without considering
at least some of the other wastes. For example, the statement on page 4-61 of the TSD that little,
if any, of the quarry waste falls within the scope of this report is not clear. If the quarry site were
not already being decontaminated under a site-specific CERCLA Record of Decision (ROD), it
would seem appropriate to consider the standard being developed in 40 CFR Part 196 in the
action.
The comparisons above were based on data collected for DOE's March 23, 1994,
comments to the Nuclear Regulatory Commission on their draft cleanup standards for 10 CFR
Part 20. In the March 1995 Baseline Environmental Management Report ("Estimating
the Cold War Mortgage"), DOE provided the following estimates of waste from the various parts
of the Weldon Spring site:
B-4
-------
Quarry Site (9 acres)
Four Waste lagoons
Chemical Plant area (44 buildings and
structures)
Vicinity Properties (offsite locations near the
former processing site)
- 127,000 yd3 of soil and rubble.
- 3 million gallons of contaminated water.
- 408,000 yd3 of sludge and soil.
- 57 million gallons of contaminated water
- 348,000 yd3 of soil and building material
-25,300 yd3 of soil
These data are not broken out sufficiently to compare to previous estimates. Total soil, sludge,
rubble and building wastes comprise about 908,000 yd3, which is only about 3% greater than the
885,000 yd3 used in the above analysis.
5. Reference Site V
As noted in comments on the other reference sites, aerial survey data do not necessarily
give a true representation of volume to concentration ratios. At this reference site, it is assumed
that the distribution of contaminants in the soil is the same as the arithmetic mean of the fraction
of soil found at other sites. Little real site-specific data have been used in the development of the
source term for this site. As a result, it is not clear if there is any real benefit in including this
reference site in the EPA analysis.
In any case, use of the average values from all five sites would result in 90% of the volume
from this site being less that 100 pCi/g. However, if one only considered the eastern sites (West
Valley and Oak Ridge Reservation), about 84% of the volume is less that 100 pCi/g and 50%
rather than 58% is less than 10 pCi/g. While these are not significant differences given the large
uncertainty of this approach, these data could result in differences of a few million cubic meters at
the 5 pCi/g range.
6. Reference Site IX
In general, the estimates of this reference site provide a reasonable estimate of the soil
contamination source term at the Rocky Flats site. It of course does not include building rubble
or decontamination residues. However, as noted, these wastes are not within the scope of the
study.
B-5
-------
7. Reference Site XXII
Reference site XXII was developed to represent the Maywood Chemical site in particular
and, in turn, other Formerly Utilized Sites Remedial Action Program sites (FUSRAP sites). Two
general comments relate to these assumptions. First, the Maywood site is a former thorium
processing site. While several FUSRAP sites are former thorium, rare earths or uranium process
locations, most are not. Many sites are former uranium metal working sites (e.g., uranium metal
machining, rolling or casting), while others are storage or conversion sites. Contamination at
many FUSRAP sites consists mostly of building contamination. Therefore, while the Maywood
site is an example of one of the larger soil contamination sites, it is not representative of many
FUSRAP sites. It may, however, be representative of many similar licensed fuel cycle or rare
earths processing facilities. The results should be compared to such sites.
The second comment relates to the development or data to support a standard for
Maywood-like sites. In, general, the background documents do not appear to provide sufficient
information in this regard. For example, Table 6-6 of the TSD indicates that over a 1000-year
integration time, a cleanup to the 10"4 risk-based level will avert 53 cancers. However, the
analysis does not provide the number of cancers averted at other levels (e.g., 10"3 or 10"2). Such
comparisons are needed to determine the incremental collective dose or risk averted by the 1 in
10,000 (lO'4) risk-based standard. Page 6-94 of the TSD indicates that the 1 in 10,000 standard
would be met by a cleanup standard of 0.09 pCi/g for radium-226 and 0.37 pCi/g for
thorium-232. These are extremely low concentrations and are not practically differentiable from
background. (A copy of this analysis was provided to the SAB. See DOE, 1994b).
However, it is also not clear that the incremental protection afforded by a 0.09 pCi/g
radium-226 standard is sufficient to justify the costs and risk resulting from the standard. In
analyzing and selecting a standard, DOE compared several potential cleanup standards for the
site. These are discussed in some detailed in the DOE comments to the Nuclear Regulatory
Commission (DOE, 1994a). Background was not specifically considered (0.09 pCi/g cannot be
distinguished from background); hence, the results do not overlap with EPA's analysis. However,
useful comparisons can be made.
Under the no-action alternative, DOE estimated potential doses from continued use of
contaminated properties to range from 12 to 2,800 mrem in a year. Over a 200-year period (DOE
selected 200 years as a practical limit over which collective doses could be reasonably estimated;
see Subcommittee comments on time period in section 8.4 of this report, the collective dose for
continued use of the property was estimated to be about 12,000 person-rem. Assuming a risk
factor of 5 in 10,000 cancers per person-rem, indicates a potential 6 fatal cancers. (Note: EPA
estimated 53 cancers averted in 1,000 years which at background is sufficiently close to 6 in 200
B-6
-------
years times 5 or 30 cancers in 1,000 years.) A cleanup standard of 30 pCi/g with backfill of clean
soil resulted in a residual collective dose of 880 person-rem over 200 years or about 11,000
person-rems averted (5.5 cancers) over 200 years. The residual individual dose was estimated to
have been reduced to 3.6 mrem in a year for residential properties and 8.2 mrem per year for
commercial properties. A 15 pCi/g standard would reduce individual doses to one half that of the
30 pCi/g standard and would have a residual collective dose of 440 person-rem in 200 years. The
incremental reduction in collective dose and risk between a 30 pCi/g and 15 pCi/g standard was
estimated to be 440 person-rems or 0.2 fatal cancers averted. At 5 pCi/g, the incremental
reduction was 280 person-rem in 200 years or about 0.1 cancers averted. The residual risk from
the site following a 5 pCi/g cleanup was estimated to be 160 person-rems in 200 years or 0.08
cancers. Therefore, the incremental reduction of the 1 in 10,000 risk-based standard of 0.09
pCi/g would be less than 0.08 cancers per 200 years.
At a 5 pCi/g standard, the collective dose to remedial workers was estimated to be 30
person-rems or about 0.015 cancers. Transportation risks were estimated to range from about
0.002 to 0.2 fatalities depending on mode. Construction accidents were estimated to range from
0.001 to 0.01 fatalities. Therefore, at 5 pCi/g, worker risk begins to be comparable to risks
averted by the action. While the DOE has not estimated risks at a cleanup to background, it is
possible, given the data, that worker risks will exceed the incremental benefit of the cleanup.
These could be estimated using EPA's volumes which are comparable to DOE estimates. DOE
has estimated that at the cleanup standard of 5 pCi/g for the surface and 15 pCi/g for the
subsurface that there is about 300,000 cubic meters (this is an in-situ volume estimate; excavated
volumes would be about 30% greater). A 5 pCi/g overall standard was estimated to increase the
volumes between 20 and 100% (the difficulty in measuring 5 pCi/g and less resulted in these large
uncertainties). The 300,000 cubic meter volume is equivalent to the EPA 10"2 risk-based level in
Table 6-7 of the TSD. Cleanup to the 10"4risk level would increase the EPA estimated volumes
to 1,300,000 cubic meters and, hence, would increase the transportation and accident fatalities
proportionally while decreasing the public risk by less than 0.08 hypothetical cancers.
Appendices L and M of the TSD allow assessment of the incremental numbers of cancers
throughout the range of 10"2to 10"6lifetime cancer risk, and for dose rates between 0.1 to 100
mrem/yr.
B-7
-------
APPENDIX C - GLOSSARY OF TERMS AND ACRONYMS
AGN Atomic General Nuclear (A line of small, research reactors found at many
universities)
Am Americium, as an element or as an isotope of thorium or uranium alpha-decay
chains. . The isotope of most importance is Am-241, produced by the decay of
Pu-241 present in fallout.
AMAD Activity Median Aerodynamic Diameter (e.g., aerosol)
ARAR Applicable or Relevant and Appropriate Requirement
BEIR Biological Effects of Ionizing Radiation
C Carbon, as an element or isotope. Isotope C-14 is radioactive with a half-life of
5,730 years
Ce Cerium, as an element or isotope (e.g., Ce-144)
CE evapotranspiration rate
CEDE Committed Effective Dose Equivalent
CERCLA Comprehensive Environmental Response, Compensation and Liability Act of 1980
(Public Law PL 96-510, also known as "Superfund")
CFR Code of Federal Regulations
Ci Curies , nuclear transformations (disintegrations). The special unit of activity.
One curie equals 3.7xl010 disintegrations per second)
cm centimeter
Co Cobalt, as an element or isotope (e.g., Co-57, Co-60)
CR runoff rate
Cs Soil Concentration (Concentration in soil)
Cs Cesium, as an element or isotope (e.g., Cs-137)
CU-POP Cleanup Population Model
Cw Porewater concentration
D Decay product(s) or progeny of radioactive species, which may itself be
radioactive
DCF Dose Conversion Factor(s). The absorbed dose in rad per working-level month.
DF Dilution Factor
DOD U.S. Department of Defense
DOE U.S. Department of Energy
E A factor of 10 to an exponent (powers of 10)
EEC Environmental Engineering Committee (U.S. EPA/SAB)
EG&G Edgarton Germhausen & Grier, Inc.
EPA U.S. Environmental Protection Agency (Also known as U.S. EPA, or "the
Agency")
ERDF Environmental Restoration Disposal Facility
C-l
-------
FEMP Fernald Environmental Management Project
FFS Focused Feasibility Study
ft feet
FR Federal Register
FUSRAP Formerly Utilized Sites Remedial Action Program
g gram
gal gallon
GEIS Generic Environmental Impact Statement
H Hydrogen, as an element or isotope (e.g., H-3)
HEAST Health Effects Assessment Summary Table
hr hour
i Hydraulic gradient
I Infiltration rate
I Iodine (e.g., 1-129)
IAEA International Atomic Energy Agency
ICRP International Commission on Radiological Protection
IDE Integrated Data Base
INEL Idaho National Engineering Laboratory
IR Irrigation Rate
K Hydraulic Conductivity (permeability). The volume of water that will move per
unit time in the aquifer under a unit gradient through a unit cross sectional area
perpendicular to the direction of flow.
Kd Distribution coefficient
Kg Kilogram
L Liter
LANL Los Alamos National Laboratory
m Milli-, [10"3] in combination with specific units
m meters
M Mega (e.g.., Mwt). Also MW(t), Mega Watts (thermal) or Million Watts (thermal
m2 square meters)
m3 cubic meters
min minute
mrem millirem (milli roentren equivalent man)
NACEPT National Advisory Committee on Environmental Policy and Technology
NAS National Academy of Sciences (National Research Council)
NCRP National Council on Radiation Protection and Measurements
NEPA National Environmental Policy Act (Public Law P.L. 91 -190)
NESHAP National Emission Standards for Hazardous Air Pollutants
NESHAPS National Emission Standards for Hazardous Air Pollutants
C-2
-------
NRR U.S. Nuclear Regulatory Commission, Office of Nuclear Reactor Regulation
NORM Naturally Occurring Radioactive Material
Np Neptunium, as an element or isotope (e.g., Np-237)
NRC U. S. Nuclear Regulatory Commission
ORIA Office of Radiation and Indoor Air (U. S. EPA)
ORR Oak Ridge Reservation
P Precipitation rate
pCi pico-Curies
p pico-, [10"12] in combination with specific units
Pb Lead, as an element or an isotope of thorium or uranium alpha-decay chains (e.g.,
Pb-210)
pH Negative log of hydrogen ion concentration
Po P_olonium, as an element or as an isotope of thorium or uranium alpha decay chains
(e.g., Po-210)
PRESTO A family of codes developed by the EPA to evaluate doses resulting from the
disposal of low-level radioactive waste. These codes include PRESTO-EPA-CPG
(assesses annual effective dose equivalents to a critical population group), and
PRESTO-EPA-POP (assesses cumulative population health effects to the general
population residing in the downstream regional basin on a low-level waste site)
Pu Plutonium, as an element or as an isotope (e.g., Pu-239, Pu-240)
Ra Radium, as an element or as an isotope of thorium or uranium alpha decay chains
(e.g., Ra-223, Ra-224, Ra-226)
RAC Radiation Advisory Committee (U. S. EPA/SAB)
RAD Radiation Absorbed Dose
RAGS/ Risk Assessment Guidance for Superfund/Human Health Evaluation Manual
HHEM
RCF Risk Conversion Factor
RCRA Resource Conservation and Recovery Act, as amended by the Solid Waste
Disposal Act
RCSS Radionuclide Cleanup Standards Subcommittee (U.S. EPA/SAB/RAC)
redox reduction-oxidation condition
rem roentgen equivalent man. A special unit of absorbed dose equivalent numerically
equal to the absorbed dose in rads multiplied by the quality factor, the
distributional factor and any other necessary modifying factors.
RESRAD Residual Radiation pathway model computer code developed by DOE
RI/FS Remedial Investigation/Feasibility Study
RME Reasonable Maximum Exposure (Also Reasonably Maximum Exposed) -
Individual
Rn Radon, as an element, or as an isotope (e.g., Rn-222)
C-3
-------
RSC Radionuclide Soil Concentration
Ru Ruthenium, as an element or as an isotope (e.g., Ru-106)
S fraction of porespace that is water-filled
SAB Science Advisory Board (U.S. EPA)
SDMP Site Decommissioning Management Program
SF Slope Factors
SI International System of units (invert S & I)
SSA Soil Specific Activity
STUK A model developed by the Finnish Centre for Radiation and Nuclear Safety
Sr Strontium (e.g., Sr-90)
t1/2 Half-life (of a radionuclide)
Tc Technetium, as an element or isotope (Tc-99)
Th Thorium, as an element or as an isotope (e.g., Th-228, Th-230, Th-232, Th-234)
TRIGA Training, Research, and Instruction — General Atomics (A small reactor
manufactured by General Atomics.)
TSD Technical Support Document
U Uranium, as an element or as an isotope (e.g., U-234, U-235, U-238)
Uf Uptake factor for food ingestion
UMTRCA Uranium Mill Tailings Radiation Control Act
VAMP An international model validation study, organized by IAEA using data sets
provided by STUK
WLM Working Level Month
Wt Watt
yd yard
yr year
C-4
-------
DISTRIBUTION LIST
Deputy Administrator
Assistant Administrators
EPA Regional Administrators
EPA Laboratory Directors
Deputy Assistant Administrator for Office of Policy, Planning and Evaluation
Director, Office of Strategic Planning and Environmental Data
Director, Office of Policy Analysis
Director, Office of Regulatory Management and Evaluation
Deputy Assistant Administration for Air and Radiation
Director, Office of Radiation and Indoor Air
Director, Office of Radiation Programs
EPA Headquarters Libraries
EPA Regional Libraries
EPA Laboratory Libraries
Library of Congress
National Technical Information Service
------- |