i UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C- 20460
January 9, 1992
EPA-SAB-RAC-COM-92-001
OFFICE OF
, . ,,,,,,, rr ~ .,, THE ADMINISTRATOR
Honorable William K. Reilly
Administrator
401 M Street, S.W.
Washington, D.C, 20460
Subject: Status of EPA Radionuclide Models
Dear Mr. Reilly:
In recent years many Science Advisory Board reports have included
constructive criticism of the models, databases, and uncertainty analyses used by
the Agency. These criticisms have not been unique to the Office of Radiation
Programs' radiation-related activities, but the criticisms are pertinent to those
activities. Therefore, the Radiation Advisory Committee of the SAB would like to
share with you its view of the limited progress it has seen in this area and the
problems that remain. The Committee does so because outmoded or inappropriate
models, supported by inadequate data and executed to produce conservative
results, can lead to significant overestimates of impact for specific potential
hazards.
In addition, selection of regulatory limits based on overestimates may lead
to remedial actions unwarranted by actual risks and thus deprive other activities
of the resources needed for protection of public health and the environment. The
Committee strongly recomends that the EPA at" this time assign a high priority to
the development of comprehensive models and data sets for the transport of
radionuclides in the environment.
This commentary focuses on three principal topics: (1) models used for
predicting radionuclide transport, (2) data sets used as bases for prediction, and
(3) lack of uncertainly analysis,
The Radlonuclide Transport Models: The models employed by the Office of
Radiation Programs to predict the transport of radionuclides in the environment
are often inappropriate. Specific models are either outdated or are not the best
choice for the specific task. During recent years, improvements in model platform,
development, selection and peer review have been inadequate. Several examples
follow.
Printed ot Recycled Pspt
-------
In its 1988 review of the scientific documentation supporting the National
Emission Standards for Hazardous Air Pollutants (NESHAP) [SAB-RAC-89-Q03],
the reviewing Subcommittee stated:
a. The use of dynamic models is favored, rather than steady-state
models such as AIRDOS-EPA, for use with seasonal or episodic
releases.
b. Although the Gaussian dispersion portion of AIRDOS-EPA has been
subjected to validation (comparison to actual data), a great deal of
effort remains to validate as many steps in the subsequent risk-
assessment process as possible.
In its 1989 review of the Radionuclides NESHAP Background Information
Document (BID) [SAB-RAC-89-024], the Committee recommended that EPA should
quickly update and complete the CRRIS model for calculating radiation dose and
risk, and make the model and documentation available for technical review.
In its 1990 review of criteria documents supporting the proposed regulation
of radionuclides in drinking water [EPA-SAB-RAC-92-009], the Committee
concurred with a previous SAB Drinking Water Subcommittee that recommended
against use of dose calculational risk estimates for radium based on models, and
that, instead epidemiologic data should be used for radium.
Data Used for Prediction: The data sets used by the Office of Radiation
Programs as the bases for prediction of the effect of proposed regulatory actions
are not adequate. They often do not include the best and most complete
information available within the time and budget constraints posed by specific
problems, or by the general needs of the Agency. Sensitivity analysis should be
used to determine the data sets most in need of supplementation. During recent
years the Committee has found little Office of Radiation Programs support for
collection of adequate data. Several examples follow.
In its 1988 review of the scientific documentation supporting the NESHAP
[SAB-RAC-89-OQ3], the reviewing Subcommittee observed,
Despite its appreciation of modeling, the Subcommittee believes that
measured data best represent source strengths and environmental
concentrations and also near-source atmospheric and environmental
concentrations from sources subject to complex diffusion (such as near
a building complex or large gypsum or uranium tailings pile). The
use of measured source data for elemental phosphorus plants is a good
example of a case in which EPA has successfully benefttted from this
approach. Where suck data are not available or cannot be obtained on
the schedule required, it is appropriate to use assessment models.
-------
In its 1989 review of the Radlonuelid.es NESHAP Background Information
Document [SAB-RAC-89-024], the Committee recommended that:
a. The rule should permit the use of actual environmental measurements
for demonstrating compliance with the individual dose limit.
b. EPA should use measured values of radon flux near facilities and
near covered tailings piles to determine actual radon emissions, to
calculate individual risk; and
c, actual, local-dispersion data, or.best available data, should be used to
reduce uncertainties in local-dispersion calculations.
In its 1990 review of draft criteria documents supporting the proposed
regulation of radionuclides in drinking water [EPA-SAB-RAO92-0Q9], the
Committee found that the choice of model parameters was not well justified; for
example, the Fj gut-to-blood absorption factor appeared to be arbitrary and in
disagreement with recent publications and the risk factors used in the assessment
of man-made alpha emitters were ad hoc.
Inadequate Uncertainty Analysis: Office of Radiation Programs documents
using the results of modeling generally do not include detailed presentation of
uncertainty analyses. The multiple levels of conservatism often built into a
particular analysis are usually not apparent from the document. Specific results
often reflect the high end of a range of possible modeling results. Rather a
modeling result should in most cases be presented as an average (reflecting
average input data) and a range (which may include a zero health risk at the low
end). Presentation of the range of uncertainty is often helpful to the decision
maker. The Committee has observed improvement in uncertainty analysis in
specific cases in recent years; however, in general, few Office of Radiation
Programs reports present their results properly bounded. Several examples follow.
In its 1988 review of the scientific documentation supporting the NESHAP
[SAB-RAC-89-0033, the reviewing Subcommittee noted that:
1 a. without rigorously derived uncertainty estimates, the credibility of
dose and risk values cannot be judged;
b. sensitivity analyses should be used to guide the expenditure of
resources and effort for the sake of model improvement and data
development; and
c, the presentation of calculated risk data should be presented in a
format similar to that in Administrator Lee M. Thomas'
"Memorandum on 'Proposed benzene NESHAP decisions and
limitation of issue to Section 112 of the Clean Air Act"1 April 5,
1988,
-------
In its 1989 review of the Radiomiclides NESHAP Background Information
Document [SAB-EAC-89-024], the Committee recommended that:
a. with respect to risk assessment methodology > a full description of the
biases and uncertainties in the estimates used in the modeling was
not provided;
b. uncertainty analysis may have little meaning if the model itself is
incomplete and not intended to yield best estimates,
cu the Office of Radiation Program's commitment to develop state-of-the-
art models, and to apply full uncertainty analysis to its modeling
efforts (recommended in SAB reports dating back to 1984) has been
deferred previously, due to urgent situations. It is past time for EPA
to complete tasks so vital to providing a scientific basis for its action;
d, best (unbiased) estimates of dose and risk, with appropriate
uncertainty statements and ranges, should be presented in all risk
assessments and the shape of the uncertainty distribution should be
presented;
e, the range of uncertainty should be expressed for the low-LET risk
value applied here; and
f. EPA should calculate the total uncertainty from all parameter values,
compare all models for completeness, compare measurements and
model predictions (validate the models), and perform sensitivity
analyses.
In its 1990 review of criteria documents supporting the proposed regulation
of radionuclides in drinking water [EPA-SAB-RAC-92-009], the Committee found
that:
a. one of the most important flaws in the criteria documents was the
failure to address uncertainties in parameters and calculated risks and
to present them so that decision makers are made fully aware of the
quality of the estimates on which particular legal guideline values are
based;
b. reported risks for man-made radionuclides do not include
uncertainties in input 'parameters or in the results of the risk
calculation;
-------
c. uncertainties in model parameters used in estimating the risk from
radon in water must be addressed and inconsistencies removed; and
d. the basis for and uncertainty associated with the assumed values for
tap water consumption rate and 20% volatilization of Rn are not
adequately addressed.
Finally, given the eventual selection of a suite of appropriate models,
adequate supporting data sets, and development of a well-designed sensitivity and
uncertainty analysis protocol, two more topics must be considered by the Office of
Radiation Programs.
1. The selected models must be adequately validated, i.e., their
ability to predict must be tested against actual environmental measurements.
2. An ordered approach to the selection, from the above suite, of
specific models most appropriate to specific problems must be developed. For
example, single, one-dimensional models are best suited for many screening tasks.
Given a good understanding of input/output uncertainty rangest a simple model
may provide an adequate, cost-effective prediction for many cases. More complex
models, with more complex data requirements, should logically be specified only
when the increased accuracy of the resulting predictions is truly required to solve
a problem, and when an adequate input data base is available to support the
complex input requirements of the model.
In summary, many of the recommendations found in recent Radiation
Advisory Committee reports echo those in the August 1984 report of the Science
Advisory Board Subcommittee on Risk Assessment for Radionuelides and the SAB
generic resolution on modeling (SAB-EBC-89-012), The Office of Radiation
Programs has discussed these problems on numerous occasions in the interim, and
has assured the SAB it will develop the techniques and data sets to allow state-of-
the-art risk assessment as a basis for regulation, but much of the basic framework
of problems remains,
The Committee hopes that by drawing this persistent problem to your
attention, specific work, such as development of validated environmental
assessment models with integral uncertainty analysis capability, will be
emphasized. These models must be well-documented, peer-reviewed personal
computer implementations, capable of producing uncertainty-bounded best
estimates for a range of increasingly detailed input data. They must be made
-------
generally available to other researchers, and should have associated generic and
region-specific input data sets based on research programs. Development of this
comprehensive and defensible model/data set will improve the scientific basis of
impact assessment! for the next round of radiation-related regulations. We look
forward to your response.
Sincerely,
f __
Raymond CL-ieelir, Chairman
Science Advisory Board
ddvar F. Nygaard hairman
Radiation Advisory Committee
Enclosure: Committee roster
-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
SCIENCE ADVISORY BOARD
RADIATION ADVISORY COMMITTEE
FY91 ROSTER
CHAIRMAN
Dr. Qddvar F. Nygaard
Division of Biochemical Oncology
Department of Radiology
University Hospitals of Cleveland
2058 Abington Road
Cleveland, Ohio 44106
MEMBERS
Dr. Kelly H. Clifton
Department of Human Oncology and Radiology
University of Wisconsin Clinical Cancer Center
K4/330, Clinical Science Center
600 Highland Avenue
Madison, Wisconsin 53792
Dr. James £. Martin
Assistant Professor of Radiological Health
University of Michigan
School of Public Health
Ann Arbor, Michigan 48109
Dr. Genevieve M. Matanoski
Professor of Epidemiology
The Johns Hopkins University
School of Hygiene and Public Health
Department of Epidemiology
624 North Broadway, Room 280
Baltimore, Maryland 21205
Dr« H, Robert Meyer
C.N.S.I.
750 Bast Park Drive
Suite 200
Harrisburg, Pennsylvania 17111
Dr. Richard G. Sextro
Building Ventilation and
Indoor Air Quality Program
Lawrence Berkeley Laboratory
Building 90, Room 3058
Berkeley, California 94720
-------
Mr. Paul G. Voillegue
MJP Risk Assessment, Inc,
Historic Federal Building
591 Park Avenue
Idaho Falls, Idaho 83405-0430
DESIGNATED FEDERAL OFFICIAL
Mrs. Kathleen W. Conway
Science Advisory Board
U.S. Environmental Protection Agency
401 M Street, S.W., A-101F
Washington, D.C. 20460
STAFF SECRETARY
Mrs. Dorothy M. Clark
Secretary, Science Advisory Board
U.s, Environmental Protection Agency
401 M Street, S.W., A-101P
Washington, D.C. 20460
------- |