UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. S0460
' "4B-EC-88-QQ8
January 14, 1988
OFFICE OF
THE ADMINISTRATOR
Honorable1Lee M. Thomas
Administrator
U.Si Environmental Protection Agency
401 M Street, s.W.
Washington, D.C. 20460
Dear Mr* Thomas;
The Hazard Ranking System Review Subcommittee of Che Science Advisory
Board has completed its review of a number of issues related to the
Superfund Hazard Ranking System (HRS). The Hazard Ranking System is the
principal mechanism used by EPA to determine whether an uncontrolled
waste site should be placed on the National Priorities List*
Besides evaluating the questions referred by the Office of Emergency
and Remedial Response (OERR), the Subcommittee has also chosen to address:
the overall algorithm for the HRS, the inclusion of exposure in the HRS,
how the HRS could be evaluated in the future, and work which could be
done to provide better documentation for the next revision of the HRS.
The Subcommittee has suggested changes that will allow the HRS to
provide a more accurate and scientifically based estimate of the relative
risk of candidate uncontrolled waste sites. Ideally, the HRS scores
should accurately assess the relative degree of risk at a site. However,
we recognize this is not always feasible due to scientific and data
limitations and to value and policy decisions implicit when considering -*
and balancing human health and environmental impacts* A revised HRS,
better designed to evaluate sites by relative risk, will provide an
improved mechanism for determining which sites should be included on the
National Priorities List (NPL), and can potentially provide useful Input
to the subsequent priorttization of NFL si-tes. However, the Agency must
continue to base this prioritization on many factors in addition to the
HRS. Most of the changes needed to improve the current HRS are changes
in the risk variables assessed and in the overall algorithm, not changes
with vast new data requirements*
The Office of Emergency and Remedial Response referred three issues
to the Subcommittees the types of toxlcity the HRS should address and
how it should' do so; distances front an uncontrolled hazardous waste
-------
-------
site that are relevant when considering air pollutants from the site;
the feasibility of including waste concentration in the HRS and whether
large volume waste sites had been treated differently than other sites
by the HRS.
The Subcommittee finds that all the options OEM proposed for revising
the toxleity factor are improvements over the current approach. The
Subcommittee believes, given the difficulty of ambient air monitoring,
that the inclusion of potential release via the air pathway is an important
improvement to the HRS. The Subcommittee also believes that improved
characterization of the source and of exposure/mobility will improve the
functioning of the HIS for all sites, not just large volume waste sites.
With respect to the three issues posed to the SAB,, the Subcommittee
recommends the following:
1, In order to improve the discriminating power of the HRS, the
currently used Sax rating scale should be replaced by multiple measures
of toxlcity. In addition, exposure measures need to be improved.
2. Modification of the HRS to include the potential for air release
seems both appropriate and possible.- A scoring system that weights
population exposure in concentric rings is recommended.
3, EPA's experience in applying the HRS to mining sites has not
proven biased against such sites, that is, it has not treated such sites
with systematic error; however, it has that potential. Our recommendation
for improvement includes incorporating hazardous-constituent concentration
in a tiered system, modifying the toxicity factor to reflect metal speeiation,
incorporating a mobility factor, and adding additional transformation
factors.
The Subcommittee also recommends that more attempts be made to learn'
from subsequent experience so that when the HRS is next revised, a better
basis for those revisions will be available. These recommendations are
detailed in Appetidlx. 6, . •
The Subcommittee's evaluation of these and other issues, and its
conclusions and recommendations, are discussed in greater detail in the
attached report. Because of their cross cutting nature, certain issues,
such as exposure and the overall algorithm, appear in more than one area.
The Subcommittee would like to call to your attention the high
quality efforts of the staff from the Office of Emergency and Remedial
Response that briefed and worked with the Subcommittee. 'They were
professional, well-prepared, and responsive to the Subcommittee's
requests for information.
-------
-------
We appreciate this opportunity to present our scientific views and
look forward to an official written response from the Agency concerning
the Garments and reccumendations in the attached report.
cc; W. Porter
H. Longest
T. Yosie
Sincerely,
Nelson, Chairman
Executive Ccnroittee
Science Advisory Board
C.
loehr, Chairman
Hazard Ranking System Review Subcommittee
Science Advisory Board
-------
-------
United Statis Office of the Administrator SAB-EC-SS-ooa
Environmental Protection Science Advisory Board January 1988
Agency Washington, DC -20460
Review of the Superfund
Hazard Ranking System
Review by the Hazard
Ranking System Review
Subcommittee of the
Science Advisory Board
-------
-------
SAB-EO88-OQ8
SCIENCE ADVISORY BOARD
HAZARD RANKING SYSTEM REVIEW SUBCOMMITTEE
REVIEW
of the
SUFEEFUND HAZARD HANKING SYSTEM
January 1988
-------
-------
II
NOTICE
This report has been written as a pare 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 issues related to
problems facing the Agency. This report has not been reviewed for
approval by the Agency and, hence, the contents of the report do not
necessarily represent the views and policies of the Environmental
Protection Agency, nor of other agencies in the Executive Branch of
the Federal government, nor does mention of trade names or coanercial
products constitute endorsement or recomaetidatioa for use.
-------
ill
U, S. ENVIRONMENTAL PROTECTION AGENCY
SCIENCE ADVISORY BOARD
HAZARD RANKING SYSTEM REVIEW SUBCOMMITTEE
ROSTER
CHAIRMAN
Dr. Raymond C. Loehr, Professor, Civil Engineering Department,
8.614 ECJ Hall, University of Texas, Austin, Texas 78712
VICE CHAIRMAN
Or. Warner North, Principal, Decision Focus Inc., Los Altos
Office Center, Suite 200, 4984 El Camino Real, Los Altos,
California 94022
MEMBERS AND CONSULTANTS
Dr. Stephen L* Brown, Project Manager, Environ Corporation,
1000 Potomac Street, N.W., Washington, DC 20007
Dr. Thomas Burke, Deputy Commissioner» New Jersey Department
of Health, Room 805, State Health, CN360, Trenton,
New Jersey 08625
Dr. Shepherd Burton, Systems Applications, 101 Lucas Valley load,
San Rafael, California 94903
Mr. George F, Carpenter, Michigan Department of Natural Resources,
Environmental Response Division, Post Office Box 30028,
Lansing, Michigan 48909
Mr. Richard Conway, Corporate Development Fellow,
Union Carbide Corporation, Post Office Box 8361 (770/342),
South Charleston, West Virginia 25303
Dr. Paul Delsler, 11215 Wilding Lane, Houston, Texas 77025
Dr. John Doull, Professor of Pharmacology and Toxicology,
University of Kansas Medical Center, Kansas City, Kansas 66103
Dr. Naihua Duan, Statistician, Rand Corporation, 1700 Main Street,
Santa Monica, California 90406
-------
iv
Dr. Keith Ferguson, Environmental Protection* Environment Canada,
Kapilaao 100, Park Royal, West Vancouver, British Columbia
CANADA V7T1A2
Dr. C, Daniel Kealy, U, S. Bureau of Mines, Spokane Research Center,
East 315 Montgomery Avenue, Spokane, Washington 99207
Dr. James 0. Leckie, Department of Civil Engineering,
Stanford University, Stanford, California 94305-4020
Dr. Donald Mackay, Department of Chemical Engineering and Applied Chemistry
and Institute for Environmental Studies, University of Toronto,
Toronto, Canada M5S 1A4
Mr. Bruce Napier, Battelle NW, Post Office Box 999,
Hichland, Washington 99352
Dr. David Pollock, U.S. Geological Sutvey, National Center, Mall Stop 411,
Reston, Virginia 22092
Dr. Joseph ¥. Rodrlcks, Environ Corporation, 1000 Potomac Street, N.W.,
Washington, DC 20007
Dr. Ellen Stlbergeld, Chief Toxics Scientist, Toxic Chemicals Program,
Environmental Defense Fund, 1616 P Street, N.W., Room 150,
Washington, D,C, 20036
Dr. Mitchell Small, Department of Civil Engineering, Carnegie Mellon
University* Sehenley Park, Pittsburgh, Pennsylvania 15213
Dr, Rebecca T, Zagraniski, Division of Birth Defects and Developmental
Disabilities, Center for Environmental Health and Injury Control,
Centers for Disease Control, Koger Center Room 2008 (F-37)
1600 Clifton Road, N.E.* Atlanta, Georgia 30333
Executive Secretary
Mrs. Kathleen W. Conway, Deputy Director, Science Advisory Board (A-101F),
U.S. Environmental Protection Agency, 401 M Street, S.W»»
Washington, D.C. 20460
Staff Secretary
Mrs. Dorothy M. 'Clark, Secretary, Science Advisory Board (A-101F),
U.S. Environmental Protection Agency, 401 M Street, S.W.
Washington, D.C. 20460
-------
TABLE OF CONTENTS
PAGE
EXECUTIVE SUMMARY 1
Toxicity Factor 1
Air Target Distance 2
Large Volume Waste 3
Exposure 4
Algorithm 5
Recommendations To Evaluate and Improve the HRS 5
INTRODUCTION 7
The Current HRS: Purpose and Prior Reviews 7
Required Revisions to the HRS 7
Science Advisory Board Review 8
Structure and Practice of the Current HRS 9
SUMMARY OF THE ISSUES " U
y Factor It
Air Target Distance 13
Large Volume Waste 15
Exposure 18
Algorithm 19
Recommendations to Evaluate and Improve the HRS 21
APPENDICES
I. Report of the Toxicity Factor Work Group
2. Report of the Air Target Distance Work Group
3» Report of the Large Volume .Waste Work Group
4 Connnents on Exposure
5. Comments on Algorithm
6. Recommendations to Evaluate and Improve the HRS
7t Requests from the Office of Emergency and Remedial
Response (SUPERFUND)
8. References Cited
-------
EXECUTIVE SUMMARY
The Hazard Ranking System (HRS) is the principal mechanism used by
EPA to determine whether an uncontrolled waste site should be placed on
the National Priorities List (NPL). The Office of Emergency and Remedial
Response (OERR) requested that the Science Advisory Board (SAB) review
certain aspects of the technical basis for revising the HRS. The major
issues identified by OERR.dealt with the toxicity factor, air target
distance, and large volume wastes. (See Appendix 7 for the1full text of
this request).
Overall, the Subcommittee is generally supportive of the changes
OERR presented and found no fatal flaws in the approach to the HIS that
OE1R presented; however, the Subcommittee has made recommendations in
this report that would further improve the HIS.
Toxicity Factor
Toxicity should be an important component of a site ranking scheme*
Because the toxicity factor in the current HRS makes virtually no distinc-
tions among sites listed or proposed to be on the NFL, and because EFA
did not present evidence that it had examined the ability of the various
factors to discriminate between sites- —including sites not on the EPL—the
Subcommittee questions whether the existing toxicity factor has the power
to discriminate "very toxic" from "not so toxic" sites. The Sax chemical
toxicity ratings are a crude basis for setting priorities since they
address only the acute toxicity of one constituent of the waste. The
Subcommittee recommends replacing Sax ratings and using multiple measures
of toxicity far ranking sites. All of the options presented by QERB. are
substantial improvements. These options are described in OERR's
Discussion of Options for Revising the Hazard Ranking System (HRS) Toxicity
Factor (12).
Scientific techniques exist which would permit the HRS to consider
several endpoints that are revelant to human health and the environment.
Multiple measures of toxicity -should be considered, any one of which, if
sufficiently severe, could, by placing the site on the NPL, trigger a
more detailed evaluation. By multiple measures, the Subcommittee means
acute human health effects, human cancer, non-cancer chronic disease in
humans, and impact on the non-human natural environment. However, a
system of weighting the severity of different health endpoints, as is
done in the Eeportable Quantities (RQ) approach, is not encouraged because
a single composite score obscures the value judgments implicit in such an
approach.
It is possible-that sites containing hazardous chemicals whose
toxicities are not well or widely known might be omitted from the NPLf
thus creating false negatives. Measures to counter this problem include
-------
the continued development of toxicity profiles on additional chemicals and
the keeping of good records so that sites can later be re-evaluated when
additional toxicity data become available,
Air Target Dis_tance_
The Subcommittee supports efforts to modify the alt pathway so that
it considers the potential for release instead of relying only on a score
for observed releases* Alt emissions are often episodic and/or narrowly
focused along a particular wind direction and, therefore, are difficult
to detect. Furthermore, time and site activities (such as excavation in
the course of site remediation) can markedly change the potential for air
emissions»
Methods are currently available which can be simplified to calculate
approximate, orcler-of-magnitude emission rates for impoundments and
landfills. Using them requires information on the chemical identities
at the site. When such information Is totally lacking, it is impossible
to even estimate the quantity of the chemical present, and volatile air
emissions can not be predicted. The subsequent consideration of air
exposure routes Is very weak, A minimum data requirement which mandates
contaminant identification by record.review or direct sampling (either
within containment structure or of emissions) would greatly improve the
MRS" validity in all pathways. In the absence of an inventory, direct
."sampling methods such as soil gas analysis or contained surface sampling
complement estimated emissions and provide a more effective estimate than
the estimated emissions alone, Such methods normally will yield samples
of much higher concentration than ambient air samples, and chemical
analyses will'be more Informative.
The Subcommittee recommends that the Agency derive a scoring system
which weights the number of exposed people in a "ring" according to the
distance from the site at which they live, (This approach Is described in •"
greater detail In Appendix 2 and illustrated by the figure on page A2-12.)
The widths of the rings can be determined to account for the decrease in
air concentrations with distance.
This ring-weighting method does not account for risks to the
maximally exposed Individual (MET), but it does explicitly account for the
higher exposures that will occur closest to the site. The Subcommittee
does not consider a separate or additional evaluation of MEI air risks
as part of the HRS to be necessary. The recommended ring-weighting
method will essentially serve the sarae purpose by weighting most heavily
the nearest and most exposed population.
-------
- 3 -
The ring-weighting method should be calibrated on the basis of the
variation of concentration with distance. The influence of pollutants
borne on particles will likely be concentrated at lesser distances which
would change the ring weights, but would not often lead to markedly
different scores.
Large Volumefaste
The issues before the Subcommittee were the "applicability of the
HRS in scoring mining waste sites" and the "feasibility of using waste
concentration data in a revised HRS."
The Subcommittee finds that the studies conducted thus far regarding
the adequacy of the HRS listing process-for mining sites have been United
and inconclusive. Sufficient evidence based on experience has not been
evaluated and presented to show an inadequacy of the HRS in regard to
mining sites. However, the present scoring system has a potential to
treat mining wastes with systematic error.
The Subcommittee believes the present HRS Is not well suited for
scoring potential releases from mining- site wastes because mobility is
not included. Improved ways of considering the concentration (often
low), toxieity release (mobility in various matrices), and information
on the transport and transformation of chemicals would make the HRS acre
accurate.
Mobility is a more discriminating concept for both inorganic and
organic substances than is persistence in the subsurface, especially
with respect to inorganic compounds. The Subcommittee recommends that,
during the development of a structured-value representation of the mobility
concept, a means of incorporating Important matrix characteristics be
explored. Significant matrix characteristics are extreme acidity or f
alkalinity (as expressed by both high and low pH), crystalline phase
modifications of raining wastes that differ from native geologic materials,
and the sorptive .capacity of surrounding geologic materials (influencing
migration tendency).
In regard to the surface water route, EPA. is considering keeping
persistence as the parameter to include with toxtcity in the waste
characteristics portion of the HRS, but would consider other transformation
parameters along with the current use of blodegradatlon. Inclusion
of these additional parameters would improve the existing HRS because
it would more closely correspond with what happens in the real world.
It appears from 'the presentations that QERR would also include
mobility In the air 'route score. Mobility and waste quantity are important
factors in determining exposure. Because of this importance, the basic
approach of Including mobility in the HRS should be pursued. (See also
Appendix,2,)
-------
- 4 -
Modifying the current HRS to incorporate factors which capture some
measure of both the physical-chemical characteristics of the hazardous
constituents and the waste matrix, as well as those site characteristics
responsible for risk, is clearly an improvement,
EPA presented two approaches foe incorporating concentration into
the HRS. While the direct measurement approach provides the soundest
scientific basis for the HRS, in most cases, it will not be* practical™
because of safety and cost considerations—to require waste constituent
concentration data for every site at the site Inspection stage. In
contrast, the tiered approach used hazardous constituent concentration
data where available and default values where they were not. The Subcommittee
supports the use of the tiered approach because it encourages the gathering
and use of concentration data and provides for the use of indirect estimates
where data gathering is impractical.
Both approaches translate waste constituent concentration data into
an estimate of the total mass of a hazardous constituent at the site,
which is then used to compute the waste quantity score for the HRS. The
approach that uses the "total mass" of hazardous constituent, without
relevant information on metal speciation and mobility, would produce
false positives for some large volume wastes and pathways.
For the ground water pathway, the total concentration or mass
of a constituent is of much less importance than the leachate concentration
produced at the site. Consequently, more emphasis should be placed on
the partitioning of hazardous constituents between the solid waste and
the leachate., Actual or estimated leachate concentration should then be
factored into the waste quantity term in the HRS. For wastes that ate
in liquid form, total contaminant mass quantity should continue to be
used.
Exposure
The Subcommittee believes that toxic!ty issues are integrally linked
to dose, i.e., migration and exposure (see Appendix 4).
Consideration of exposure in the evaluation of relative risk is
vital. If there is no exposure, there is no risk and where there is
exposure, risk varies with exposure. While the current HRS gives some
consideration to exposure, often through the use of proxies such as
persistence, distance to nearest well and target populations, the
Subcommittee concludes that additional chemical-specific consideration
of.routes of exposure, particularly with respect to the Toxlcity Factor,
would strengthen the HRS.
The Subcommittee is mindful of the difficulties in characterizing
exposures to chemicals at waste sites that have undergone only preliminary
investigation. It does not propose the implementlon of expensive and
-------
- 5 -
time-consuming sampling programs to measure actual concentrations in
ambient media where pollutant levels may be near or below detection
limits• However, additional surrogates for exposure—e.g., a calculation of
quantity estimates with mobility'considerations'—win provide better
.rankings than does the current system. Without appropriate estimates
of mobility, consideration of toxieity in the algorithm is tantamount to
assuming that the exposures to every detected chemical at every waste
site are equal. The Subcommittee recommends adoption of a tiered approach
for exposure similar to Option W3 EPA developed for concentration
(discussed in Appendix 3 of this report).
Algorithm
Revisions to the HRS should begin with the development of a chain of
logic, without regard for the ease or difficulty of collecting data, that
would lead to a risk assessment for each site. This framework, but not
the underlying logic, would be simplified to account for the very real
difficulties of data collection.
This chain of logic, which is termed the algorithm in this report,
should lead to a situation in which an increased score reflects an increased
risk presented by a site* Without this consistency in the meaning of a
score it is impossible to assess the relative degree of risk posed by a
site.
The Subcommittee places special emphasis on the algorithm issue
because it is, impossible to review the components of the HRS without
considering how the components fit together. If the HRS is to evaluate
sites by relative risk, then the HRS must be consistent with the underlying,
quantitative relationships of the factors involved. Specific recommendations
concerning this issue are presented in Appendix 5.
Consistency in terms of relative risk and score does not exist
between sites in the current HRS, although for a single site a larger
score represents*a larger risk, 'Changes in the HRS algorithm mean that a
score of 28,5 ia a revised HRS will almost certainly mean something
different than a score of 28.5 in the current HRS.
RecommendationsTo Evaluate and Improve the HRS
Grappling with a difficult problem like the HRS clarifies the need
for additional data, studies,1 and evaluations. The Subcommittee assumes
that the HRS will again be revised. Results from the following would be
most helpful at t~ha,t time. Peer review of the plans for such studies is
recommended.
o The evaluation of the relative risk of sites as estimated by a
HRS score should be compared with the results of risk assessments
based on the Remedial Investigation/ Feasibility Studies (RI/FS).
-------
- 6 -
Advance planning Is needed so that additional Rl/FS~llke data
is gathered on some sices with scores below 28*5; these can be
selected using an a PjrlojrA sampling scheme. Similarly, since
the success of the HRS at ranking by relative risk depends, In
part, on the consistency and completeness of Che scoring, at
least some sites scoring above 28-5 may require additional
data gathering. Such work would provide the basis for a meaningful
retrospective study o£ the HRS and a better understanding of
the basic parameters that are important in the use of the HRS
scoring model,
EPA should determine, based on results of RI/FS studies, whether
the health risk associated with a site was likely to be dominated
by one or a few chemicals (for each pathway) or by a larger
number of chemicals. Such a review would provide an objective
basis for recommending how many chemicals should be selected
for toxicity (or site) ranking. The two Agency studies of this
Issue presented to the Subcommittee were limited in the number
and types of sites evaluated, and the sites studied were not
selected for representativeness, but for convenience. A better
evaluation is warranted.
EPA should consider, how the scores through the various pathways
relate to relative risk. With this information, the algorithm
can be calibrated appropriately.
A vigorous effort should be made to Improve the overall quality of
analytical data collected at sites. Standardized collection and
laboratory methods currently exist for only a small fraction of
substances potentially present. Expanded chemical characterization
of all media, coupled with a strong laboratory certification program
will Improve not only the HRS but all aspects of the Superfund .*
process.
Screening models like the HRS must be simple. They do not have
much resolving power and therefore, some false positives and
false negatives are Inevitable. Because of this limitation,
HRS scores should not be overemphasized. A process should be
established either to review sites subject to scoring or to
review HRS scores in an attempt to spot false positives and
negatives. Such a scientific review process could involve the
use of additional models; in many cases, however, obvious false
negatives or positives might be best handled by a "manual" review.
The Important point is that the system be flexible enough to
allow for a variety of approaches to be used, as needed, during
the scientific review process.
-------
M "7 _
INTRODUCTION
The Current^ Hazard Ranking System; Purpose and Prior Reviews
The Haaard Ranking System (HIS) is the principal mechanism used by
EPA to determine whether to place sites on the National Priorities List
(NFL). The HRS was made available for public comment on two occassions.
On March 12, 1982 a draft of the current version was placed in the Superfund
Docket and a notice was published in the Federal Register* EPA summarized
the public comments and addressed them in the preamble to the final HRS
rule on July 16, 1982. On April 9, 1987 EPA published an Advance Notice
of Proposed Rulemaking in the Federal Register requesting comments and
information related to revising the current HRS. The Agency subsequently
held a two-day public meeting ,to .solicit public testimony. The issues
referred to the Science Advisory Board (toxicity factor, air target
distance, and concentration) are ones that the public had previously
identified.
Required Revisions to the Hazard Ranjcing System
Section 105(c)(l) of the Superfund Amendments and Reauthorization,
Act of 1986 (SARA) requires that the-Agency modify the Hazard Ranking
System so that, "to the maximum extent feasible, it accurately assesses
the relative degree of risk to human health and the environment posed by
sites and facilities subject to review." The amendments require an:
o Assessment of the human health risks associated with contamination
or potential contamination of .surface waters, either directly or as
the result of run-off from a site.
o Evaluation of the damage to natural resources which may elfect
the human food chain and which is 'associated with any release or f
threatened release,
o Assessment of the contamination or potential contamination
of the ambient air' which is associated with a release or
threatened release.
Section 125 of SARA also requires that for certain wastes generated
primarily from combustion of coal, EPA consider the following:
"(I) The quantity, toxicity, and concentrations of
hazardous constituents which are present in such
waste and a comparison with other wastes;
(2) The extent of, and potential for, release of such
hazardous5constituents into the environment; and
(3) The degree of risk to human health and the
environment posed by such constituents."
-------
Section 1187 requires that EPA give a high priority to facilities where
the release of hazardous substances has resulted in the closing of drinking
water wells or has contaminated a principal drinking water supply,
Science Advisory Board Review
Three memoranda from the Office of Emergency and Remedial Response
constitute the request for the review, (Appendix 7 contains copies
• of these memoranda,) In response to the first memorandum requesting an
independent outside scientific review of the technical basis supporting
revisions of the HRS the Executive Committee of the Science Advisory
Board formed the Hazard Ranking System Review Subcommittee. (The Subcommittee
roster appears at the beginning of this report.) The members were chosen
for their special expertise and experience in the hazardous waste area.
The Subcommittee first met May 19-20, 1987 in Washington, DC for
background briefings on the Superfund program and on the current HRS,
OERR provided the Subcommittee and Work Group with briefings on the
toxicity factor and air target distance issues, issue papers, and key
references. At the July 16-17, 1987 Subcommittee meeting in Washington
the Office of Policy Planning and Evaluation (OPPE) orally presented its
Site Ranking Panel Study, four ranking systems and expert panel to rank
20 hazardous waste sites.
The Subcommittee formed three work groups to address issues
referred by the OERR. These were the Toxicity Factor Work Group, the
Air Target Distance Work Group and the Large Volume Waste Work Group.
(Rosters for these Work Groups can be found in the front of Appendices 1,
2, and 3, respectively). Each work group reviewed documents on that
issue prepared by OERR and reported to the full Subcommittee. (Review
and important background documents are listed in Appendix 8.)
.*•
The Toxicity Factor Work Group met June 29-30, 1987 in Washington, DC
to review Discussion of Options for Revising the Hazard_ Ranking System
(HRS) Toxicity Factor (12). The Work Group reported its conclusions
and recommendations to the full Subcommittee July 16-17, 1987.
The Air Target Distance Work Group met July 27-28, 1987 in Washington
to review Analysis of the Air Target Distance_ _Limit in__th_e_ Hazard Ranking
S_y_gjt_em (9).
The Large Volume Waste Group met August 20-21, 1987 in Denver
to review two documents, The Supejrfu_nd_ Hazard Ranking System (HRS);
Applicability _to Miniug_ Waste Sites, and The Superfund Hazard Ranking
System (HRS); Feasibility of Using Concentration Data ina Revised HRS (13,
14). OERR briefed the Large Volume Waste Work Group, Two members of the
public were invited to give presentations to the Work Group. Dr. Ishwar
Murarka of the Electric Power Research Institute presented EPRI"s utility
-------
_ 9 -
waste research program (1, 2, 3, 4, 5, 6, 7). Mr. William Bluke of CH2M
Hill spoke on his experiences at Superfund mining waste sites. Two
members of the public, Dr. Brian Murphy of Gradient Corporation and Ms*
Susan Sawtelle representing the Edison Electric Institute and other
utility organizations, provided oral comment.
The latter two work groups reported their conclusions and
recommendations at the final, full Subcommittee meeting held on September
14-15, 1987 in Washington, DC.
Materials given to the Subcommittee were made available to the
public through the Superfund Bocket, All meetings were announced in c"he
Federal Register and opportunity for both written and oral public comment
was provided at each meeting.
The Subcommittee's report was approved by the Executive Committee of
the Science Advisory Board on January 14, 1988 prior to transmittal to
the Administrator.
Structure and Practice of the Current HRS
The current HRS uses a technique called structured value analysis
(or scoring) to rank sites. In the HRS structured value analysis, a set
of rules is developed which parallel what is thought to be occurring in
the real world. The rules address what factors should be considered, and
how they, should be scored and combined. The combination rules for the
current HRS have been adjusted on the basis of professional judgment—-that
is, weights have been introduced to make the HRS output match a subjectively
ranked test set of sites.
The current HRS» described in Uncontrolled Hazardous Waste Site
Ranking System; A Users Manual (8), assigns three scores to a site. Two
are used primarily to identify facilities requiring emergency attention,
These are the score for Fire and Explosion and for Direct Contact. The
.third, or Migration score, is-the primary basis for inclusion on the
NPL. " '
Three pathways (air, surface water» and ground water) are considered
in developing the Migration score. For each pathway, various factors are
considered. These fall into three broad categories; likelihood of release,
waste characteristics, and targets. The actual factors considered can
vary with the pathway. For example, distance to the nearest well applies to
Che ground water pathway but not to the air pathway.
-------
- 10 -
After a numerical value is assigned to each factor, it is multiplied
by a weight to obtain a factor score. Factor scores within the same
category are added. Scores for the categories are multiplied together*
This procedure yields a. score for Che pathway. The pathways are then
combined through a method called quadratic averaging. Each step in this
process affects the final score and, therefore, how well the HRS discriminates
between sites of greater and lesser risk to human health and the environment.
EPA and the states use the HRS to calculate a site score, from 0 to
100, based on the actual or potential release of hazardous substances
from a site through air, surface water or ground water that may affect
people. This score is the primary technical factor used to decide if a
hazardous waste site should be placed on the National Priorities List.
EPA, the states and other public agencies have used the HRS to
evaluate several thousand sites. The scores and supporting documentation
of these sites are submitted to EPA Headquarters where they are reviewed
to ensure correct and consistent application of the system. Sites having
scores above 28»5 are proposed in the Federal Register for the National
Priorities List. Based on public comment and all available Information
EPA develops a final score for each site and places it oti the NPL if the
score is still above 28.5.
-------
- 11 "
SUMMARY OF THE ISSUES
Toxicity Factor
The Subcommittee believes that toxlclty to human health and the
environment is an important consideration in site ranking. To be useful,
the toxlcity factor should discriinate between "very toxic" and "not so
toxic" sites. If the objective of the HE.S is to generate estimates of
relative risk, it must .'address both toxleity and exposure. Although
toxicity is part of the current HRS, the toxieity factor makes distinctions
among only a few of the 951 sites listed at proposed for the NFL.
The Sax chemical toxicity ratings have provided the current HRS
with a crude basis for prioritizing sites* The Sax system is insufficient
because (a) it focuses on acute adverse outcomes instead of the range of
toxieologlc endpoints that need to be addressed in the HRS, and (b) it
does not provide citations so that toxicologic ratings can be reevaluated
and verified for the purposes of the HRS, All of the options (11, E2, £3)
discussed in Discussion of Options for Revising the Hazard RankingSystem .
(HRS) ToxicJLty_ Factor (12) represent improvements over the Sax rating
method.
The Subcommittee recommends the Sax rating method be replaced with
a toxicity factor which addresses multiple measures of toxtclty because
most known toxic chemicals are associated with a range of health and
environmental effects. Therefore, to protect human health and the
environment, the HIS needs to incorporate the potential for a wide range
of adverse effects* A high score for any of these endpolnts could motivate
a more detailed evaluation of a site at the RI/FS stage,
To the extent that scientifically legitimate techniques exist for
the consideration of various toxic effects In humans and in nature, the
decision of which endpolnts to address is more of a policy rather than
a scientific choice, Nevertheless, because there are so many endpolnts
relevant to human health and the environment for which good methods exist,
the Subcommittee encourages EPA to address multiple endpoints as part of
the toxtcity factor. The multiple measures of human toxicity SPA should
consider addressing with the revised toxicity factor include carcinogenlcity»
other chronic health endpoints used to establish reference doses, and
acute toxictty.
The three options presented in EPA's Discussion of Options for Revising
the Hazard Ranking System (HRS) Toxiclty^Factor (12) make use of essentially
the same data bases for toxicity, and none of the options is overwhelmingly
superior to the others. Howevert because there is no scientifically
credible basis for weighting the severity, of different health endpoints,
as Is currently done for calculating Reportable Quantities (RQs), the
Subcommittee discourages use of the RQ method. Such weightings are subjective;
-------
- 12 -
subsequently, the subjectivity may be obscured by the final numerical
score. The strengths and weaknesses of the options are summarized here
(see Appendix i for more detail).
For acute toxielty, the Subcommittee prefers use of the Registry of
Toxic Effects of Chemical Substances (RTECS) method rather than the
Reportable Quantity (RQ) method, because RTECS facilitates route-specific
toxieity ratings and closer attention to the raw toxicity data.
For non-cancer chronic effects, the Subcommittee prefers use of the
Reference Dose (RfD) method because the RQ method involves weighting the
severity of various endpoints, and the modified ADI method has not received
extensive review and may differ from the RfD method in rating individual
chemicals.
For carcinogenic potency the Subcommittee sees little difference
between the two options presented (q^* vs. ED10) and suggests that other
measures, e.g., ED(t)and ED(Q.l), be investigated.
EPA addressed the question, "How Many Chemicals Should Be Evaluated?"
"Based on information presented ta the Subcommittee orally and in writing (10)
it appears that the studies of this topic, while generally indicating
only a, few chemicals would dominate the health risk associated with a
site, were limited in the number and type of sites evaluated and the
sites were selected for convenience rather than representativeness*
Because of this, and the fact that toxlcological information is missing
or scanty for many chemicals» the Subcommittee cannot recommend a fixed
number of chemicals for OERR to examine at each site. A study to address
this question'is one of the Subcommittee's recommendations (see page 6).
The Subcommittee recommends that EPA consider the risk represented
by the numbers of chemicals known to be present. All things being equal,
and because the toxicological effects of so many chemicals are unknown, a
site with more chemicals should receive a higher priority in the Preliminary
Assessment/Site Investigation (PA/SI) stages for further evaluation.
Another approach is the assignment of discretionary points (see Appendix 5).
The Subcommittee could not formulate an absolute recommendation on
the question of whether separate toxicity ratings for a chemical should
be developed for various exposure routes. Such a route-specific approach
has appeal because toxicity is known to vary by exposure route for some
chemicals. However, this toxicity question cannot be addressed independently*
it is inextricably tied to how the MRS deals with migration and exposure
Issues, If the revised HRS aggregates route-specific exposure scores,
this total exposure score could be multiplied by a net toxicity score
which would not need' to be route-specific* (The cruder the scheme, the
less need for route-specific factors.)
-------
- 13 -
Alternatively, a more accurate analysis would independently evaluate
risks by route before aggregating. Route-specific toxicity factors would
improve the accuracy of such ati approach. Although such toxicity factors
are not yet available for a large number of compounds, differences in
toxicity between routes is potentially important. SPA may find consideration
of route-specific toxicity on an exception basis to be a practical solution
especially applicable to compounds such as asbestos where a particular
route, in this case inhalation, is well known to be of great importance.
Whether or not route-specific toxleity is addressed in the HRS, it is of
potential "utility in the subsequent RI/FS stage because the more detailed
information supports more realistic analyses of risks,
The Subcommittee has discussed the need for two additional steps In
the NPL process. First, some type of consistent methodology or guidelines
Is needed to select chemicals during the preliminary assessment and site
investigation for use in the site ranking process. Second, the option
of using more detailed risk Information, if available, could be used as
a supplement to the HRS process (see Appendix 5).
Aif Target Distance
The basis for this section of the review was Analysis of the Air
Target Distance Limit in the Hazard Ranking System (9>*
The .Subcommittee is very supportive of OESS's efforts to modify
the air pathway so that it considers the potential for releases, instead
of relying solely on a score for observed releases.
The air pathway in the current Hazard Ranking System evaluates
population risks to derive a score for "air targets" using a matrix of
population versus distance for four concentric circles with radtl of
1/4, 1/2, 1, and 4 miles.
The Subcommittee recommends that EPA consider alternatives to the *
use of a single air target distance limit for all sites and derive a
scoring system which uses site-specific population rings, fhe widths of
these rings may vary by size and type of site to account for the decrease
in air concentration with distance. (See Appendix 2, especially the
illustrations in Attachment B,) The Subcommittee favors such a refinement
because the ring-weighting method relates more directly to risk at a
site than does the current air target approach. The ring weighting
method should be able to address risk more accurately because It considers
both the population exposure experienced near the site and the toxicity
factor.
In the ring-weighting approach, information on the chemicals present
is used to estimate emissions factors. These emissions factors and information
about the health effects that can be expected to occur at various levels of
exposure can be used to develop ring-widths that relate to the exposure
people living in those locations can expect to experience.
-------
- 14 -
Although this method does not evaluate maximally exposed individual
(MEI) riska per _se» it does explicitly account for the higher exposures
chat will occur within the ring closest to the site. These higher exposures
would receive an appropriately weighted score.
The Subcommittee was asked to address approaches to assessing health
risks from airborne contaminants and has suggested a variety of approaches
for EPA's consideration. All of these approaches employ the following
assumptions; carcinogens are non-threshold contaminants; there are thresholds
for other toxic effects; concentrations of contaminants decrease with distance
from the site; this decrease is more marked for partlculates than for
volatile chemicals; and acute toxicity and toxic effects from direct contact
with partlculates are the over-riding health concerns in the area closest to
an uncontrolled waste site.
Acute effects are addressed by Superfund's emergency program and by the
current HRS toxicity factor which is based on the Sax ratings (see Appendix 1).
The Subcommittee favors a toxicity factor which Includes multiple endpoints
of toxicity including acute effects because some nearby populations
around uncontrolled waste sites report symptoms consistent with acute
toxicity (headaches, nausea, irritation, and respiratory effects) and
because of the possibility of acute effects occurlng In the event of a
catastrophic release. Because of the weight given the inner-ring such
acute effects could be well addressed with the ring-weighting approach
or by a ring-weighting approach modified to account for the existence of
a threshold concentration below which no acute effects should occur.
However, other scientifically supportable options include; considering
acute toxicity in a separate assessment outside the HIS (such as under
the emergency program); assigning discretionary points to the HRS score
for the confirmed presence of local odors or health effects reasonably
associated with materials found at the site; a separate assessnent of
che probability of a release of materials from a catastrophic event; and
treatment of acute effects in a direct-*contact pathway.
Chronic non-cancer effects could also be addressed with the ring-weighting
approach or by one aodified to account for the existence of a threshold
concentration below which no chronic effects should occur. In the modified
ring-weighting approach this threshold concentration can be converted to
a distance from the site beyond which no adverse chronic effects would
be expected. Chronic effects, particularly of particulate pollutants,
could also be treated in a direct contact pathway.
Carcinogens are dealt with in a reasonably precise manner by the
ring-weighting approach, given the usual assumption that risk is directly
related to average lifetime concentration in air, While the Subcommittee
favors this approach, it is complex and EPA may want to "evaluate the
effects of different diameters to see if the use of a single diameter for
all sites is reasonable*
-------
- 15*-
The ring-weighting approach described in Appendix 2 is best suited
to carcinogenic effects of volatile chemicals. While concentration of
particulates decreases more sharply with distance than does the concentration
of volatile chemicals, the Subcommittee believes that this difference is
not large enough to require separate estimation methods for volatile and
particulate emissions in the HRS.
In assessing exposure, consideration of each media pathway is
recommended, including air, water, food and direct contact* Measurement
of exposure should also consider the route of exposure through which
substances exert their effects. While the relationship between exposure
level and the biologically effective dose is different for different
pathways, not all compounds are toxieologieally well characterized. As
a result, EPA may only be able to deal with route-specific toxieity on
an exception basis; e.g., in the case of asbestos, inhalation is more
important than ingestion.
Many of the Subcommittee's recommendations hinge on knowing the
identities of the chemicals at the site. A minimum data requirement
which mandates contaminant identification by record review or direct
sampling would greatly improve the validity of the HRS in all pathways
and is particularly helpful in estimating exposures to air pollutants
and in addressing toxicity. For example, when the chemical identities
are known, it should be possible to modify and simplify the Treatment
Storage and Disposal Facilities (TiBF) procedures to calculate an
approximate, order-of-magnitude emission rate for impoundments and
landfills.
Both the identity of chemicals at the site and emission rates can
be confirmed using other investigative techniques, the most obvious of
which is air concentration measurement. Because ambient monitoring is
subject to the problem of dilution, to detect the presence and amount of
emissions it is desirable to analyze at the source, or as close to the ..
source as possible. There Is, thus, an incentive to use methods of
"probing" sources using such techniques as soil gas analysis.
Large Volume Waste
The large volume waste Issues before the Subcommittee were the
applicability of the HRS in scoring mining waste sites and the
feasibility of using waste concentration data in a revised HIS.
QERR provided The Superfund Hazard Ranking System (HRS);
Applicability to Mining Waste Sites (13) and The Superfund Hazard Ranking
System (HRS); Feaslblllty^pf tlsingConcentration Data in a Revised HRS (14)
as the basis for the review,.
-------
- 16 -
There are two ways to evaluate the applicability of the HRS Co
mining sites. The Superfund Hagard_Ranklng_System _(HRS); Applicability
cp_Mining Waste Sites (13) presents EPA"s experience in ranking mining sites
with the HRS system in terras of false positives and false negatives and
then distinguishes between the sites according to potential danger (determined
by other means). In its evaluation, the Subcommittee coupled this approach
with an examination of the scientific issues independent of the HRS
experience.
In addressing large volume waste (LVW) issues, It is useful
to note that clear qualitative differences exist between many LVW and
the more numerous hazardous waste sites containing mixed synthetic
organic/inorganic hazardous constituents in both liquid and solid form.
LVW monofills are materials of a single type. Specific sites may have
multiple types of wastes (e.g., mill tailingSj smelting slags, etc.).
Sites where these wastes are spatially separate are more easily characterized
than sites where they are mixed. LVW are commonly contain low
concentrations of hazardous trace elements (HTE) in the waste matrix
which are released by very specific geochemlcal processes and subject to
migration constraints that are increasingly better understood,
With respect to the question of the jj^licability of the current
HRS to mining waste sites, the Subcommittee finds that the studies conducted
regarding the adequacy of the HRS listing process for mining sites have
been limited and inconclusive. The reports reviewed included the TRC (27, 28,
29) and MITRE (24, 25) reports contained in The Superfund Hazard _Ran_king
System (HRS): Applicability to Mining Waste Sites (l'3T"as well as EPA's
own studies.' Sufficient experiential evidence has not been presented to
show an Inadequacy of the HRS in regard to mining sites. The present
system has possibly rated mining sites no worse than other waste sites.
Some of the findings supporting this conclusion follow.
(U While the TRC (27, 28, 29) and MITRE reports (24, 25) reported
that mining waste sites scored higher in waste quantity compared
to noa-inining sites (as expected given the large volume of
wastes generated by the ruining industry), TRC did not conclude
that this bias caused false positives or false negatives for
mining sites. Both sets of reports found that waste quantity
contributed relatively little to the final HRS score.
(2) Although the TRC reports (27, 28, 29) suggested the degree of
information available was a major factor in NPL listing of
mining sites, they presented no data showing that the relatively
larger.data bases for certain mining sites resulted in unfair
scoring,
(.3) The TRC reports (27, 28, 29) also suggested the final HRS score
for mining sites could be predicted by population alone. Since
the TRC reports only addressed" mining sites, they could not
prove a bias against mining waste sites.
-------
- 17 -
Nevertheless, the Subcommittee believes the present HRS is not well
suited for scoring potential releases from mining sice wastes because
nobility and concentration ate not included. Therefore, the present
scoring system has a potential to treat mining wastes with systematic
error.
For the ground water pathway, the concentration of a constituent in
a waste Is of much less importance than the concentration of that constituent
in the leachate produced at the site. Consequently, more emphasis should
be placed on the partitioning of hazardous constituents between the
waste and the leachate. Leachate concentration or Interstitial pore
water concentration, either known ot estimated, should then be factored
into the waste quantity term in the HRS, For washes which are not In
solid form, total contaminant mass quantity should continue to be used,
Improved ways of considering the concentration, toxicity (inorganics
have special characteristics), release (mobility in various matrixes),
and information on the trans port/transformation of chemicals would make
the HRS more accurate* The area of contact of the site with the environmental
compartments (air, water, humans, etc.) Is also an important factor.
The Issue of mobility Is especially Important at mining sites because
the large-volume of mining wastes often means large masses of the contaminant
are available for release. • However, certain conditions must exist for these
releases to occur. The current HRS does not address specific ionic form
of inorganic metals, and ion form affects mobility.
Because'mobility is a more discriminating concept than persistence
in the subsurface, the Subcommittee suggests that OIRR explore means of
incorporating important matrix characteristics In a revised HRS. Such
characteristics include extreme acidity or alkalinity (as expressed by
both high and low pH), any crystalline phase modifications of mining
wastes that differ from native geologic materials, and the sorptive
capacity of surrounding geologic materials. The latter influences
migration tendency*
With respect to the feasibility of using concentration data in the
revised HIS, the Subeotimittee finds that including hazardous constituent
concentration is desirable because both the severity of observed releases
and the risk o£ potential releases are related to concentration.
EPA has proposed two options for modifying the HRS to account for
the concentration of hazardous constituents. Both approaches translate
waste constituent concentration data into an estimate of" the total mass
of a hazardous constituent at the site, which is then used to compute
the waste quantity score for the HRS. Even the approach which uses
"total mass" of hazardous constituent could contribute to produce false
positives for some large volume wastes and pathways.
-------
- 18 -
The Subcommittee's preferred approach (Option 3) encourages the use
of concentration data and provides the flexibility to substitute Indirect
estimates such as default concentrations to estimate a constituent's
total mass when direct measurements of concentrations are not available,
While the direct measurement approach (Option W2) provides the soundest
scientific basis for Che HRS Is most cases, it may not be practical, due
to safety 'and cost considerations for every site at the site inspection
stage. The Subcommittee suggests the direct use of concentration rather
than conversion to a mass value,
A method of determining a "representative concentration" for a site
based on a stratified sampling strategy would be useful. In this approach,
a complex site could be subdivided into a set of more honogeneous regions
or strata, and a representative value for each stratum determined through
limited sampling. Statistical techniques exist for manipulating stratified
data, which could then be recomblned with appropriate weighting into a
final single HRS score for the site.
Exposure
To the extent that the HRS is intended to assess relative risk, it
must address both toxlcity and exposure sufficiently.
Exposure occurs when there is contact between pollutants and receptors.
It is Important to address exposure when there are toxic chemicals present
In order to determine whether there is any actual or potential risk to
health or the environment. Where there are no toxic substances or where
there Is no exposure, there is no risk, even though there may be release
or contamination. (However, determining the existence of a release or of
environmental contamination is often the first step in evaluating actual
exposure,) Where there is Incomplete information on either toxlcity or
exposure, risk cannot be fully assessed. Risk can be evaluated by f
assessing four parameters: (a) the presence of chemicals, (b) their
potential for release and mobility, (c) the probability of contact with
humans, plants or animals and (d) their intrinsic toxteity.
Much could be gained by adding a chemical-specific exposure score
that could be combined with chemical-specific toxiclty scores to evaluate
which chemicals dominate the risk at a single site and how the risk
might compare across sites that are otherwise similart
Given some minimal information about a site and its chemical inventory,
it may be possible to make crude estimates of partitioning among environmental
media and, therefore, of emissions by route which could be used to generate
part of a chemical-^specific exposure score* The following information would
support development of a chemical-specific exposure score;
1. An estimate of the total mass or source concentration
of important chemicals at the site.
-------
2» A description of the site-specific factors influencing
release, e.g.; impoundment vs. landfill.
3. Partition factors applicable to the chemical, e.g. Henry's
Law constant and organic carbon partition coefficient.
Some- characterization of the potential for migration also allows
better consideration of time as a factor in site management.
Environmental pathways include air, water (both ground and surface
water), soil, and plants and animals that become part of the human food
chain. The potential for release and mobility of chemicals through
these pathways affects risk ""estimates* Chemicals- behave "differently in
different environmental pathways, so all pathways need to be evaluated
to estimate risk effectively*
The main difficulty of dealing with exposure directly is that we
are dealing with potential exposure as well as actual exposure. There
are two aspects of exposure to be considered—the concentration of contaminants
in the environment around the site and the number of people potentially
exposed to these concentrations. Actual release way be demonstrated by
environmental sampling. Potential release may be addressed by considering
properties of the site and characteristics of the chemicals present,
such as distance to ground water and volatility. Similarly, the number
of people (receptors) present and actually exposed today can be counted,
while the number of people who might be present at the location in the
future can only be estimated from .studies of population dynamics.
Measurement of exposure may be person*-based or common!ty-based.
While estimates of individual person-based exposures can be developed
from biological samples, fro«t personal household samples, from self-reported
exposures, and from self-reported symptoms, each approach has its limitations.
Community-based exposures can be developed from ambient pathway
measurements, from site measurements, and Iron modeled exposures. Numerous
assumptions about targets and behavior are needed to impute exposure
from ambient pathway measurements. However, even more assumptions are
needed to inputs exposure from site measurements.
Incorporating concentration and mobility could produce a more
comprehensive site assessment procedure. One way to do this is by using
concentration as a weighting factor for the waste quantity. The resulting
effective waste quantities for low volumes of high concentration waste
might be similar to those for large volumes of low concentration wastes.
Similarly, the scores for toxiclty/perslstenee and for effective waste
quantity could be adjusted by considerations of mobility.
-------
- 20 -
Algorithm
Improving the algorithm could potentially do more to improve the
HIS Chan fine-tuning individual components.
The Subcommittee has recommended changes that will allow the MRS to
provide a more accurate and scientifically based estimate of the relative
risk of candidate sites. To the extent possible, the HRS scores should
correspond to an objective analysis of relative risk at sites. However,
this is not always feasible due to both scientific and data limitations,
as well as Che value and policy decisions implicit when considering and
balancing human health and environmental impacts.
A revised HIS, better designed to evaluate sites by relative risk,
will provide an improved nechanlsm for determining which sites should be
included on the NPL, and can potentially provide meaningful Input to the
subsequent prioritlzatlon of NPL sites. However, the Agency must continue
Co base this prioritlzatlon on many factors in addition to the HRS. Most
of the changes needed to transform the current HRS into a system more
reliably related to risk are changes In the overall algorithm and not
changes with vast new data requirements,
Internal consistency is important. Making rules that are consistent
requires little up-front Investment and no additional costs ia the scoring
of specific sites. These changes can be acre cost-effective than some
improvements to portions of the HRS alone, especially ones requiring
additional data collection.
The way'that various components of the score are combined should
reflect how their real world counterparts Interrelate, The approach for
accomplishing this is Co begin with a physically-based exposure assessment
model for each exposure pathway, structured to properly translate expected
or potential releases into environmental concentrations and subsequent
exposures and effects. These models would, of necessity, be highly
simplified for a screening assessment. Even, a simplified exposure assessment
model may not be, feasible given the time, resource and data limitations
associated with the HRS process. However, the manner in which the pathway
scores are estimated and combined should be consistent with the fundamental
material balance and exposure principles of such an underlying model. This
can be accomplished in the context of the structured value approach used
in the current HRS system.
When there isn't "enough" data for the HRS, it is necessary to have
a default value that encourages data collection (also termed missing
value replacement). Lacking complete information on chemical inventory
and concentration at a site may well be the rule rather than the exception.
Therefore, a minimum data requirement will reduce the inconsistencies
that are otherwise inevitable when different people perform the scoring.
-------
- 21 -
One concern with Che use of default data is that the fact of data
limitations could Indicate that the site might be different from otherwise
comparable sites. The definition of comparable sites is crucial and «ay
depend on the scoring component.
When data are available that exceed what is needed for the HRS,
there is the question of how to use all of it. The default or tiered
approach OEM presented on concentration is one mechanism for using
additional data in the revised HRS, Another approach is the assignment
of discretionary points. The Subcommittee discussed several scenarios
where discretionary points could be assigned including sites where:
a very large number'of chemicals were identified, the presence of the
toxic materials in the surrounding population had been demonstrated, and
toKic particulate air emissions were anticipated. For the HRS score to
reflect relative risk, the extra points assigned ttust be proportional to
the extra risk the situation presents. (See Appendix 5.)
Recommendations to Evaluate and Improve the HRS
The current HRS is appropriate for the original purposes QSRR has
•described because most of the factors scored are related to risk, and
higher scores for those factors reflect higher risks. Thus, the current
HIS is plausible. However, it is both possible and desirable to revise
the HIS to relate more closely to the relative risk posed by uncontrolled
waste sites.
The- HRS performance should be judged by an" empirical retrospective
evaluation of how successfully the HRS predicts risk or on how successfully
Its components predict phenomena (such as release) which contribute to
risk, this evaluation should be based on an in-depth technical review.
Whatever the definition of risk, the HRS should be judged on how well it
approximates that definition, not on how well it matches a subjective
notion of the relative importance of the sites in a test set,
The Subcommittee recognises the possibility that sites containing
hazardous chemicals whose tox±city Is not widely known will be omitted
from the NPL listing, thus possibly creating false negatives. Measures
to counter this problem include continuing to develop toxicity profiles
on more chemicals and keeping good records so that sites with currently
unstudied chemicals can be re-evaluated as toxicity information on these
chemicals becomes available,
The Subcommittee recognizes the need for some flexibility in the HIS
to allow for the wide range of physical and chemical characteristics of
sites. Being flexible is compatible with reducing the possibility of
error in the HRS. EPA must reduce this possibility of error in each
component (or "factor") of the system and, even oore importantly, In the
algorithm used to relate these components into a score that is Intended
to reflect relative risk.
-------
- 22 -
To adequately evaluate Che current or modified HRS, in-depth studies
should be conducted to determine the effectiveness of the HRS to assess
risk and to identify areas for improvement. One goal of these studies
would be to improve the algorithm so that it can be defended as risk-related,
Another goal is to to determine at least a crude method for estimating
Che quantity of specific chemicals present. Any HRS evaluation studies
should be thoroughly peer reviewed before their initiation.
Future evaluation of the HRS should consider the possibility of
incorporating detailed numerical geochemical transport/fate models which
are currently under development* The advantage of using such models in
a future HRS is the explicit inclusion of quantitative representation of
specific processes and mechanisms reponsible for the release, transport,
transfomation, and retention of hazardous compounds. In comparison to
structured value (or "scoring") approaches, the models increase data
requirements, but can yield less ambiguous, more detailed simulations
appropriate for pathway calculations and, ultimately, exposure estimates*
As mentioned in the Other Comments portion -of Appendix 3, EPA should '
develop studies both to review large-volume waste sites ranked under the
model and to examine basic parameters of the model.
-------
APPENDIX I
REPORT OF
THE TOXICITY FACTOR WORK GROUP
TO THE
HAZARD RANKING SYSTEM SUBCOMMITTEE
OF THE
SCIENCE ADVISORY BQAiS
-------
U. S. ENVIRONMENTAL PROTECTION AGENCY
SCIENCE ADVISORY BOARD
HAZARD RANKING SYSTEM REVIEW SUBCOMMITTEE
TOXICITY FACTOR WORK GROUP
ROSTER
CHAIRMAN
Dr. Warner North, Principal, Decision Focus Inc. Los Altos Office Center
Suite 200, 4984 El Catnlno Real, Los Altos, California 94022
MEMBERS/CONSULTANTS
Dr. Paul Deisler, 11215 Wilding Lane, Houston, Texas 77025
Dr. John Doull, Professor of Pharmacology and Toxicology,
University of Kansas Medical Center, Kansas City, Kansas 66103
Dr. Joseph ?. Rodricks, Environ Corporation, 1000 Potomac Street, N.W.,
Washington, DC 20007
Dr. Ellen Sllbergeld, Chief Toxics Scientist, Toxic Chemicals Program,
Environmental Defense Fund, 1616 P Street, N.W., Room 150,
Washington, D*C. 20036
Dr, Rebecca T. Zagranlskl, Division of Birth Defects and Developmental
Disabilities, Center for Environmental Health and Injury Control,
Centers for Disease Control, Koger Center, Room 2008 (F-37),
1600 Clifton Road, N*E., Atlanta, Georgia 30333
EXECUTIVE SECRETARY (Until August 31, 1987)
Mr. Eric Males, Executive Secretary, Science Advisory Board (A-101F),
U.S. Environmental Protection Agency, 401 M Street, S.W.,
Washington, DC 20460 (Until August 31, 1987)
EXECUTIVE SECRETARY (After August 31, 1987)
Mrs. Kathleen W. Conway, Deputy Director, Science Advisory Board, (A-1QIF),
U.S* Environmental Protection Agency, 401 M St,, S.W.,
-Washington, D.C. 20460
STAFF SECRETARY
Mrs, Dorothy M. Clark, Staff Secretary, Science Advisory Board, (A-1Q1F)
U.S. .Environmental Protection Agency, 401 M Street, S,W,,
Washington, DC 20460
-------
TABLE OF CONTENTS
page Number
GENERAL COMMENTS Al-1
RESPONSES TO SPECIFIC TOXICITY ISSUES POSED BY DIM Al-3
How Many Chemicals Should Be Evaluated? Al-3
Route-speciftc Toxicity Scores Al-4
Noncancer Effects of Carcinogens and Acute Toxicity Al-4
Advantages and Disadvantages of Certain Toxicity
Ranking Options Al-6
Default Values Al-6
Combining Toxicity Scores Al-6
OTHER COMMENTS Al-6
Flexibility and Common Sense Al-6
General Support for a Structured ?alue Approach to the HRS Al-6
Two Additional Steps in the NPL Process Al-6
Migration Al-7
Dose Al-7
Future Studies > Al-7
ATTACHMENT: Discussion of Options for Revising the Hazard latticing
System (HRS) Toxicity Factor, May 4, 1987 pages 40-41
-------
Al-1
GENERAL COMMENTS
The Toxicity Factor Work Group reviewed Discussions of Options for
Revisionof_the Hazard Ranking System (HRS) Toxicity Factor (12).
The Toxicity Factor Work Group believes Chat toxicity Is an important
consideration In assessing sites because, if the objective of the HIS Is to
generate estimates of relative risk, the HRS must deal with toxicity and
exposure. The toxicity factor in the current HRS makes distinctions
among only a few of the 951 current and proposed site on the NFL. By
pathway, fully 85% (ground water), 87% (surface water), and 97% (air) of
chese sites received the maximum toxieity/perslstence score. Because
the Agency did not present information on sites that are not on the NFL,
the Work Group was unable to determine whether the current toxicity
factor Is adequate to distinguish between "very toxic" and "not so toxic"
sites. Not only is it unlikely that the most toxic substances at different
sites are equally toxic, it is Inconceivable that they occur at equal
concentrations even if total quantity of waste is identical. Therefore,
in revising the HRS, the EPA should consider approaches that would
better discriminate between sites of differing toxicities and subsequently
examine how well the new toxicity factor actually performs.
Some toxicity decisions are made in the Preliminary Assessment/Site
Investigation (PA/SI) stage* For example, specific chemicals are selected
for scoring in the HRS, while other chemicals are dropped from the process.
The bases for these selections do not appear to be standardized at this time.
As a result, while the toxicity data that Is handled by the HRS is already
a subset of the full information, additional (perhaps important) toxicologic
data are not part of the quantitative process. The exposure factors are
addressed in various ways in the HRS and most are handled quantitatively
in the HRS. Consequently, the HRS score actually reflects a disproportionate
amount of Information on exposure and less on toxicity, thus deemphasizirig
coxlcity in the final score* Therefore, EPA should consider the development
of some type of consistent methodology or guidelines for evaluating the
toxicity of chemicals at a site during the PA/SI stages,
The Toxicity Factor Work Group reviewed the following methods (12) for
evaluating the toxicity of individual chemicals:
o Chronic Toxicity (Noncancer)
- Reference Dose Method
- Reportable Quantity (or Composite Score) Method
- Hodified Acceptable Daily Intake (ADI) Method
o Carclnogenicity
- Cancer Potency Factor Method
- Reportable Quantltltes (RQ) or Effective Dose^Q Method
- EPA Welght-of-Evidence Method
o Acute Toxicity
- Reportable Quantities (RQ) Method
- Registry of Toxic Effects of Chemical Substances
(RTECS) Method
-------
Ai-2
The Sax chemical toxleity ratings have provided a very crude basis
for prioritizing sices because they are largely based only on the acute
toxicity of one constituent of the waste. The Sax system is insufficient
because it focuses on acute adverse outcomes instead of the range of
toxicologlcal etidpoints that need to be addressed in the HRS and because
it does not provide citations so that toxicologic ratings can be reevaluated
and verified for the purpose of the HIS.
OSRR presented three options for evaluating and scoring toxicity in the
revised HRS (12). These are a reportable-quantities-based option
-------
Al-3
o The Agency has presented the toxicity issues in a somewhat
theoretical manner. Few real world examples and few test cases
such AS example rankings of chemicals or sites, were presented
to the Work Group.
o Lastly, and probably most important, the Work Group believes
that toxicity Issues are integrally linked to dose (l»e., migration
and exposure) issues* The existing HRS system deals with exposure
by using surrogates to address the question of whether people
(receptors) will be exposed, and if so, how many will be exposed,
and to what concentrations, via ground water, surface, and air routes.
RESPONSES TO SPE_CI_FI_C TOX_IC_I_TY_ _ISSUES POSED BY OERR
The full text of the specific questions can be1 found in the
attachment to this report. They are repeated in abbreviated form
here for the convenience of the reader.
How Many Chemicals Should Be Evaluated?
A review of past Superfund risk assessments were undertaken by OERR
(10) to determine whether the health risk associated with a site is
likely to be dominated by one or a few chemicals (for each pathway), Or
by a larger number of chemicals. The study, which was limited In the
nunber and type of sites considered and because sites were selected for
convenience rather than for representativeness, generally indicated that
only a few chemicals would dominate at any given site. However, the
Work Group did not review this study in detail and cannot recommend a
fixed number,of chemicals for OERR to examine at each site. Instead,
the Work Group offers the following general guidance.
EPA appropriately considers toxicity early in the site evaluation
process. Identification of known toxic substances occurs in the PA/SI
stages prior to the HRS evaluation. Indeed, only sites containing toxic'
substances currently identified by the Superfund Laboratory Program are
likely to be candidates for HRS evaluation and listing. In addition,
the chemical receiving the highest score in the toxicity-perslstence
matrix (8) Is selected by field Staff for scoring purposes in the HRS.
The Work Group has concerns about the procedure now used to identify
chemicals In the PA/SI stages and to select a single "most toxic"
chemical for the HRS evaluation,
Because so few chemicals have been toxicologlcally well characterized,
the Agency should consider a factor for sheer numbers of chemicals known
to- be present (in addition to developing scores based on a smaller number
of chemicals of known toxicity). All things being equal, a site with
mote chemicals should receive a higher priority for further evaluation,
most probably In the ftl/FS, (See discussion of discretionary points in
Appendix 5 of the HRS Subcommittee's report.)
-------
Al-4
Should Separate Toxicity Ratings for a Chemical be Developed for
Various Exposure Routes?
Because this question is inextricably related to how the HIS deals
with migration and exposure issues, the Work Group has chosen to discuss
the advantages and disadvantages of three approaches rather than to make
a single recommendation.
One approach is to aggregate route-specific exposure scores into a
net (total) exposure score for a site* This total exposure score could
be multiplied by a net eoxlcity score. The toxicity score may not need
to be route-specific because the cruder the scheme, the less the need
for route-specific factors.
A more accurate analysis would independently evaluate route-specific
risks before aggregating. Because the relationship between exposure level
add the biologically effective dose is different for different pathways,
route-specific toxicity factors would improve the accuracy of Such an
approach,
A third approach, which recognizes that route-specific toxicity
factors are not yet available for a large number of compounds, is to use
route-specific information when it is available and relevant—such as in
the case of asbestos where the inhalation route has far greater biological
effect than the ingestlon route—and to aggregate exposure scores where
such information Is lacking or the route-specific differences are small.
Whether or not route-specific toxicity Is addressed in the HRS» it
should be considered in the Remedial Investigation/Feasibility Study (RI/FS)
stage because the more detailed information supports more realistic
analysis of risks.
How should potential carcinogens, noncancer effects and acute COjciclty •"
be evaluated as part of & scoring system used to determine listing^Qn
the NFL?
The Work Group concludes that multiple measures of toxicity should
be used In the process of evaluating waste sites because a variety of toxic
effects may be experienced by the exposed human and natural populations*
Which endpolnts are of concern to EPA is primarily a science policy
decision* However, scientific techniques exist which would permit the
HRS to consider several endpoints which are relevant to human health and
the environment.
The Work Group believes the Importance of acute effects in the HRS
should be decreased relative to chronic effects because, given the likelihood
for long-term low-level exposures around uncontrolled waste sites, chronic
effects appear more relevant. Futhermore the Work Group understands that
EPA has a separate emergency response program to respond quickly to those
uncontrolled waste sites where acutely toxic exposure levels are thought
-------
Al-5
possible. As a result, the Work Group recommends replacing the Sax method
with any of several GERR-developed approaches which include measures of
chronic eoxlcity.
Acute effects are important when they occur, and sone Subcommittee
members conclude that, since the emergency program suffices to address then,
acute effects need not be considered in the revised HRS. However, because
of reports of acute effects such as odors, nausea, headache, and respiratory
irritation around unremediated and uncontrolled hazardous waste sites and
because of the potential for catastrophic releases, or releases during
remediation which may result in exposures sufficient to cause acute
effects, the Work Group recommends that the revised HRS include acute
effects among the multiple endpoints considered.
Besides carcinogenleity, the health endpoints should include
those used to establish reference doses (including at a minimum, fetotoxic,
teratogenic, and neurotoxic effects); and acute toxlclty (Including not
only lethality, but endpolnts such as irritation, allergic sensltization,
and neurotoxiclty), The oajor concerns for most sites will be for chronic '
health effects (cancer and non-cancer). While acute toxlclty should be
examined, it will generally be less important than chronic taxiclty In
the site ranking process. Any of the variety of health and environmental
endpoints considered could motivate-a more detailed evaluation of a site.
For acute toxics, the Work Group prefers the Registry of Toxic
Effects of Chemical Substances (RTECS) method to the Reportable Quantity
(RQ) method because RTSCS facilitates route-specific toxlcity ratings
and closer attention to the raw toxiclty data. If a more sophisticated
use of the toxicologlcal data is envisioned, use of the RQ evaluations
of RTECS may be helpful to identify the preferred animal studies for
toxiclty ratings. Use of Threshold Limit Values (TLVs)» which are
contained in RTECS, may be one means of expanding the acute toxieity,
including not only lethality but also other endpoints. f
For non-cancer chronic effects the Work Group prefers the Reference
Dose (RfD) method because the- RQ method involves subjectively weighting
the severity of various endpoints and the modified ADI method has not
received extensive review and may differ from the RfD method in rating
individual chemicals.
For carcinogenic potency the Work Group does not perceive much
gross difference between the two options presented (qj* vs. ED10)* Other
measures (e.g., ED [I] and ED [0.1]) should be investigated because they
refer to a more realistically meaningful risk range for human exposures and,
where the dose-response is known to be concave upwards, they would offer an
automatic correction to potency for this fact. In evaluating relative risk
for carcinogens, both weight-of-evldence and potency should be considered
in determining relative risk for carcinogens,
The Work Group judges that the Agency's procedure in the RQ calculation
-------
Al-6
for combining weight-o£-evldence and potency information Is reasonable for
purposes of ranking different carcinogens.
Are There Scientific Advantages/Disadvantages toBasing a Toxiclty
RanId._ng_Method_on a No~Adverse-Effect Laval?
The Work Group chose to address the advantages and disadvantages of
the options presented father than enter into an abstract discussion on
NGASLs and alternatives. For non-cancer chronic effects the Work Group
prefers the Reference Dose (RfD) method to the Reportable Quantities
(RQ) method (which involves weighting the severity of various endpoints),
and to the modified Acceptable Daily Intake (ADI) method (which has not
received extensive review and may differ from the RfD method in rating
individual chemicals),
Is it Necessary to Attempt to Develop Default Values for Rating a
Substance^s Toxicity When its Toxieity Data Base is Limited?
The Work Group has no recommendation on this issue. Having little
patience with "invented toxicology" the Work Group wishes to encourage
the Agency to develop data rathtr than relying on default values. In
cases where toxicological data become available after a site has been
scored, the Work Group encourages the Agency to re-evaluate the site in
the light of the new findings.
what Quantitative Procedures are Most Reasonable for Combining
Individual Toxicity Ratings for_Multiple Substances at j^jSlte?
The Work Group does not encourage a system for weighting the severity
of different health endpoints, as is currently done for calculating
Reportable Quantities (RQs), The Work Group believes that a variety of
endpoints should be used to rank sites for further study. In the context
of an on/off switch, any health effect may be an adequate criterion for f
further study of a site.
' OTHER COMMENTS
Flexibility and Common Sense
The HIS is designed to be a simple, workable, mechanistic procedure
that generates reliable estimates of relative risk. However, considering
the state of information available at the PA/SI stage, the complexity of
risk, and the variability among sites, this may be an impossible goal.
In light of these uncertainties, there are many opportunities for
errors in both overestimating and underestimating relative risk. Thus,
the Work Group believes that any system must offer flexibility and the
opportunity for common sense.
-------
Al-7
General Support for a Structured Value Approach to theHRS.
Even though, in the Work Group's view, it may not be possible to
develop an HRS that provides reliable estimates of relative risk, the
Work Group is supportive of a HRS concept and role which attempts to
rank sites by relative risk. That is, the Work Group supports the use
of a structured value analysis that includes (even if crudely) factors
related to risk* The estimates generated by such a scheme may be simply
one mechanism for evaluating sites.
Two Additional Steps_in the_NPL Process
The Work Group has discussed the need for two additional steps in
the NPL process; (a) some type of consistent methodology or guidelines
to select chemicals during the preliminary assessment and site investigation
should be developed for use in the site evaluation process, and (b) the
option of using more detailed risk information, if available, should be
considered as a supplement to the HRS process., One approach, for example,•
is to allow adjustment of point scores to take into account special
knowledge, as discussed in Appendix 5 to the Subcommittee's report.
Migration
Migration of pollutants from a site to a receptor (considering both
time and kinetics) is an important part of the exposure assessment.
Where there is no contact between pollutants and receptors, there ia no
risk. The current HRS does not evaluate mobility at all and does not
sufficiently Devaluate migration (which the current HRS addresses in the
ground water, surface water'and air pathways) and, thus, can be expected
to misclassify some sites with respect to relative risks.
Dose
Dose information is included in the HRS evaluation only through the
very crude surrogate of total waste quantity. Concentration information
for specific substances, if available, is not used in the HRS. The Work
Group is concerned about dose as it relates to the types and severity of
effect. It is an important element for predicting risk, but it is not
apparent that any effective surrogates exist to address it in the current
HRS.
Future Studies
Some members of the Work Group are concerned that sites containing
hazardous chemicals.whose toxleity is not widely known will be omitted
from the NPL listing, thus creating false negatives* Measures to counter
this problem include continuing to develop toxicity profiles on more
chemicals and record keeping so that sites with "unknown chemicals" can
be re-evaluated as toxicity information on these chemicals becomes available,
-------
Al-8
The lack of toxicological information on many chemicals is hardly a
problem unique to the Superfund program and the Work Group does not
wish to imply that this difficulty can be solved by Superfund, or indeed
the Agencjj alone.
-------
SPECIFIC QUESTIONS FOR THE SCIENCE ADVISORY BOARD
This section lists a few specific unresolved questions about
scientific Issues involved la the development of a revised toxiclty
factor for the HIS. EPA believes that these questions are significant
need to be addressed before a revised toxicity factor can be developed,
(I) How many'chemicals per site (or exposure medium)
should be evaluated to serve as the basis for an
overall toxlcity factor score? If less than all
chemicals detected, how should they be selected?
Should selection be based on the most toxic
substances, the most frequently detected, or those
that appear to be present in the greatest quantities?
(2) Should separate toxlcity ratings for a chemical be
developed for various exposure routes (l.e«» should
chemicals have three or four different ratings,
depending on exposure route?) , or Is a single
route-independent toxiclty rating adequate for HRS
ranking purpose? [Note; It is assumed that separate
media-specific toxieity factor scores would be
developed for a site under any HRS revision option
based on the chemicals evaluated for that medium,
regardless of whether route specificity Is accounted
for In Individual chemical toxiclty ratings.]
(3) Should potential carcinogens be evaluated for
noncancer effects also, or can it be assumed that the
carcinogenlcity potential should dominate the overall
toxlcity rating?
(4) Is it advisable to include acute toxiclty evaluation
as part of a scoring system used to determine listing
on the National Priorities List (NPL)? If so, how
should it be weighted?
(5) Should more specific noncancer toxlcity types, such as
teratogeniclty and mutagenicity, be broken out and
rated separately? If so, how should they be weighted
relative to other effect types so that single
overall rating could be derived for each chemical?
(6) Are there scientific advantages/disadvantages to
basir\g a toxlcity ranking method on a no-effect level
NOAEL, such as in Options E2 and E3 versus an effect
level (MED, such as in Option El)?
-------
(7) It is necessary to attempt to develop default values
for rating a substance's toxicity when Its toxieity
data base Is limited, or should site scores be
developed based only on chemicals having "better"
toxicity data?
(8) What quantitative procedures are most reasonable for
combining individual toxicity ratings for multiple
substances at a site?
-------
APPENDIX 2
REPORT OF
THE AIR TARGET DISTANCE WORK GROUP
TO THE
HAZARD RANKING SYSTEM REVIEW SUBCOMMITTEE
OF THE
SCIENCE ADVISORY BOARD
-------
U. S. ENVIRONMENTAL PROTECTION AGENCY
SCIENCE ADVISORY BOARD
HAZARD RANKING SYSTEM REVIEW SUBCOMMITTEE
AIR TARGET DISTANCE WORK GROUP
ROSTER
CHAIBMAN
Dr. Thomas Burke, Deputy Commissioner, New Jersey Department of Health
Room 805, State Health CN360, Trenton, New Jersey 08625
MEMBERS/CONSULTANTS
Dr. Stephen L» Brown, Project Manager, Environ Corporation
1000 Potomac Street, N.W., Washington, DC 20007
Dr. Shepherd Burton, Systems Applications, 101 Lucas Valley Road
San Rafael, California 94903
Mr. George P. Carpenter, Michigan Department of Natural Resources,
Environmental Response Division, Post Office Box 30028
Lansing, Michigan 48909
Dr. Naihua Duan, Statistician, Rand Corporation, 1700 Main Street,
Santa Monica, California 90406
Dr. Donald Mackay» Department of Chemical Engineering and Applied
Chemistry and Institute for Environmental Studies,
University of Toronto, Toronto, Canada M5S 1A4
Dr. Mitchell Small, Department of Civil Engineering, Carnegie Mellon
University, Sehenley Park, Pittsburgh, Pennsylvania 15213
EXECUTIVE jECggTARY (Until August 31, 1987)
Mr. Eric Males, Executive Secretary, Science Advisory Board, (A-101F),
U.S. Environmental Protection Agency, 401 M Street, S.W.,
Washington, DC 20460 (Until August 31, 1987)
EXECUTIVE SECRETARY (After August 31, 1987)
Mrs. Kathleen W» Conway, Deputy Director, Science Advisory Board, (A-1Q1F),
U.S. Environmental Protection Agency, 401 M St., S«W.,
Washington, D.C, 20460
STAFF_TSSCftEt,ARY
Mrs. Dorothy M. Clark, Staff Secretary, Science Advisory Board, (A-101F),
U.S. Environmental Protection Agency, 401 M Street, S.W.,
Washington, 0C 20460
-------
-------
ATTACHMENT A
POIHTS FOR SAB CONSIDERATION
Use of a Single Target Distance Limit at all Sices
Applicability of Air Emissions Infornatlon from
TSDF to Superfund Sites
Approach to assessing Health Risks
(Is Cancer most Conservative?)
Range of Emissions Estimates used
-------
TABLE OF CONTENTS
PAGE
GENERAL COMMENTS A2-1
COMMENTS ON THE ISSUES RAISED BY QERR A2-2
An Alternative to the Use of a Single Target A2-2
Distance Limit at All Sites
Approach to Assessing Health Risks A2-4
Applicability or Air Emissions Information from TSDFs to A2-5
Super!und Sites and the Range of Eiiissions
Estimates Calculated
Other Factors Affecting the Potential for Release A2-8
OTHER COMMENTS A2-8
Relationship of the Air Pathway to the Rest of the HRS A2-8
Determination of Actual Release A2-9
ATTACHMENT A: Points for SAB Consideration
ATTACHMENT B: Derivation of the Air Target Score
-------
A2-1
GENERAL COMMENTS
The Air Targed Distance Work Group .reviewed QERR's Analysis of the Air
TargetPi$_t_ance_Limit in the Hazard Ranking System (9).
The air pathway in the current Hazard Ranking System (HIS) uses an
evaluation of exposed population at risk to derive a score for "air targets,"
The scheme uses a matrix of population versus distance for four concentric
circles (with radii of 1/4, i/2, 1, and 4 miles). Public comments have
suggested that the four mile distance nay be too large. EPA studies, by
contrast, have suggested that, for sites with large emission rates of
carcinogens, individual lifetime risks may remain above 10~^ for even
greater distances. The Work Group considered this and other questions
related to the air pathway (see Attachment A). "
In general, there may not be significant air risks at many potential
Superfund sites,, but the HRS needs to address them when they are present.
The Work Group is very supportive of OERR's efforts to modify the air
pathway so that it considers potential releases, instead of relying
solely on a score for observed releases* Because air emissions are
often episodic or narrowly focused along a particular wind direction,
they are difficult to observe. In -addition, there can be a declining or
changing risk depending on the nature or Integrity of containment or the
site type. There may have occurred a significant release early in a site's
history which has resulted in a loss of contaminants and decreasing
emissions at the time when sites are evaluated and scored. Lastly,
future activities at a site (such as excavation) may create the potential
for new air ^missions. For these reasons, inclusion of a potential for
release Is an important modification of the HRS.
Ideally, if comprehensive information was available, air emission
models developed for other purposes could be used to estimate potential •-
releases. These estimations, however, may require information which Is not
available to the Agency during the preliminary assessment/site investigation
(PA/SI) stage. With information on the identities of chemicals present, an
emission rate ca*n be calculated for volatile chemicals or the presence
of volatile emissions established on a crude yes/no basis. When such
Information Is lacking, volatile air emissions cannot be predicted at
all, and the subsequent consideration of air eKposure routes is very weak.
If the identities of chemicals are known, it should be possible to
modify and sinplify the procedures for modeling air emissions from hazardous
waste transport, storage and disposal facilities (TSDPs) to calculate an
approximate, order-of-nagnitude emission rate for impoundments and landfills.
Other investigative techniques could be used to confirm emission rates,
the most obvious of which la air concentration measurement in the immediate
vicinity of the source. Because of the difficulties in measuring variable
low-level ambient air concentrations, more direct sampling, such as soil
gas analysis or "inverted dish pan" surface sampling, should be considered.
-------
A2-2
COMMENTS ON THE ISSUES RAISED §Y OERR
The full text of the issues raised by OERR can be found In Attachment A.
Abbreviated versions are used here for the convenience of the reader.
An Alternative to the Use of a Single Target: Distance Limit at All Sites.
The Wotk Group recomsaends that EPA consider an alternative to the
use of a single air target distance limit for all sites,
To evaluate the exposed population at risk more accurately, the Work;
Group recommends that the Agency derive a scoring system that will properly
weight the number of exposed people according to the distance from the
site at which they live. The weighting factor should be proportional to
individual risk, as determined by concentrations that decrease with
distance from the site. In other words, the system should use population
rings that are based on distances from a site at which air concentrations •
decrease by a constant factor. These distances may vary by size and type
of site. For illustrative purposes a method of developing these rings
is presented in Attachment B. Although this method does not evaluate
maximally exposed individual (MEI) risks per se, it does explicitly
account for the higher exposures that will occur within the ring closest
to the site. Therefore, the Work Group concludes that a separate or
additional evaluation of MEI risks as part of the HRS is unnecessary.
These rings could be incorporated into the scoring system as a
simple refinement of the current matrix, as illustrated in Table I,
Table I
Suggested Matrix for Air Pathway Population Scores
Generalized Target Distance Score
Distance from Center of Site
Population
106«-107
105-106
104-105
1Q2-103
•=*
10
9
8
7
6
5
ba-b2a
9
8
7
6
5
4
J-2-*3'
8
7
6
5
4
3
b3a-b4a
7
6
5
4
3
2
-------
A2-3
The matrix Is Interpreted as follows. The distance "a" Is the
radius of the site itself. (In practice, sites will not be circular and
"a" must be taken as a. typical dimension.—derived, for example, by
caking "a" as equal to the square root of the area of the site divided
by pi,} The dimensloftless constant "b" is a multiplier of the order of 2
to 4, determined as shown In Attachment B. For example, if the site
was 200 yards in diameter (about 1/9 mile), and the multiplier was 3,
the rings would be 1/9 to 1/3 mile, 1/3 to 1 mile, I to 3 miles, and 3
to 9 miles. The population used is that in each annular ring (between
"a" and "ba", etc.), determined to one significant figure. Once scores
are assigned for each ring, the largest score is taken as adequately
representing the site. The scores are all logarithmically related to
population risk, so that they should be added to other scores representing
logarithms of variables that enter risk linearly, such as the carcinogenic
potency or the quantity of waste or emission rate.
As shown, the scores for a large site ("a" is large) will be greater -
than for a small site ("a" is small) for the same population distribution
because each of the four rings will be larger and Incorporate more population.
The population in each ring will increase as a?t so if a increases by a
factor of 3, the population will expand by a factor of 9, and the score will
go up by one point. This behavior implies incorporation of a factor related
to emission strength (in turn, proportional to a') In the score. This may
aot be appropriate. Perhaps it would be better to define a standard value
for a, taken from a typical site.
After deriving such a ring-weighted system, the Agency may want to
evaluate the effects of different diameters to see if the use of a single
diameter for all sites is reasonable because a ring width that varies
between sites, based on their diameters, may be too complex for the HIS,
,••
To place the scores from the ring-weighting process on the same scale
as other scores in the HIS with which they might be combined, the scores
could be increased In direct proportion by multiplying by a constant.
For example, if one wishes to have a score of 30 in the top left corner,
one can multiply all the scores by 3 thus obtaining, for example, 3 In
the lower right corner and 12 in the lower left corner. But if this is
done, an order of magnitude change in carcinogenic potency should also
be represented by 3 points change in score, and so on.
The above scoring system is reasonably precise for cancer for
which individual risk, given the usual assumption that risk is directly
related to average lifetime concentration in air, is directly related to
average lifetime concentration in the air. For noncancer effects,
EPA ordinarily assumes a threshold of concentration for any damage to
occur. Thus, outside some critical distance (different for sites with
different emission rates or different threshold toxiclties), no effects
will occur, and inside that distance one could assume conservatively
-------
A2-4
thac 100% Incidence of the effect occurs. There seems to be no simple
scoring scheme to take this property into account in m population score,
but the practical effect is that population beyond perhaps one mile (for
example) is simply irrelevant to the ait target score for noncaneer
endpoints. If, in the coxlelty score, a noncancer effect is responsible
for the greater part at the score, then the choice of points from Table 1
should be restricted to the two left columns (for example). As before,
the degree of danger for an individual in these two rings will have
already been captured in the toxleity score and whatever score is used
as a surrogate for exposure,
Approach to Assessing Health_ Risks
As implied above, toxicity considerations may also be different for
the air pathway than for surface or ground waters. An exclusive evaluation
of a contaminant from the standpoint of carclnogeniclty is inadequate,
Many compounds detected at contamination sites are not currently considered
carcinogens or suspected carcinogens, Acute toxicity Is generally meant
to be considered In emergency or interim response procedures prior to
HRS scoring. In reality, sites nay not pass some threshold of acute
toxicity considered in this way and» yet, still exhibit some subacute
or chronic health effect. This Is particularly true for the late-history
site, where average emissions may be relatively low, but—with breaching
of a few drums, deterioration of a building, or removal of cover—a
short-term, acute emission la again possible. If acute toxicity conditions
were discovered at a site during site inspection or HRS scoring, the site
should be automatically referred to emergency response. This seemingly
would eliminate the need for considering acute toxicity in the HRS. It
does not appear that this complete separation is appropriate, however,
given the threshold nature of emergency response or changeable site
conditions,
.»•
In addition, other subacute or chronic effects of noncarclnogens must
be considered, Neurotoxiclty and chemosenaory effects of air pollutants,
while occurring at differing levels of concern, are real responses to
airborne contaminants, and environmental toxicity is occasionally important.
The public is also often concerned about nuisance effects, such as odor
or nausea.
The Work Group favors separate consideration of acute toxicity in a
direct contact pathway. However, acute toxicity could instead be included
among the multiple measures of toxicity in a revised toxicity factor. In
addition, the exposure of the near-site population exposure to air contaminants
could be separately assessed, or discretionary points could be added to the
HRS score for smells or symptoms reasonably related to exposures to
chemicals known to be at the site (providing that these are confirmed in
some suitable manner). The direct contact pathway may be especially
useful if EPA wishes to eliminate or reduce the importance of scoring
acute toxicity for the surface and ground water pathways.
-------
A2-5
In assessing exposure, consideration of each media pathway Is
recommended including air, water, food, and direct contact. Measurement
of exposure should also consider the route of exposure through which
substances exert their effects. For example, inhalation is more important
than iagestion in the case of asbestos.
Applicability or Air Emissions Information from TSDFs to Supgrfund Sites
andthe Range of Emissions Estimates Calculated
Because air emissions are often episodic or narrowly focused along
a particular wind direction, they are difficult to detect* For this
reason, inclusion of a potential for release is an important priority
for modification of the HES. The procedures given in the Analysisof_
the AirTarget Distance Limit in the Hazard tanking System (6) for
estimating air emissions from surface impoundments and landfills at
Superfund sites are based on model equations developed for RCM Facilities
(TSDFs), These procedures are appropriate for the purpose to which they '
are applied; to determine the range of possible (long-term) emission rates
at Superfund sites. The model equations represent the state-of-the-art
in this relatively new area of environmental science.
There are, however, some limitations Identified below and suggestions
for additional analysis that can result in a clearer and more defensible
analysis. These Include the need to Illustrate previous validation
studies of the models, particularly those presented in peer-reviewed
scientific journals, the need to better illustrate the model comparisons
at existing landfills, and better identification of the potential
applicability of the models to particular sites.
Despite the difficulties in data collection, validation studies
with the surface impoundment and landfill models have been attempted.
Using these studies in the documentation for proposed changes to the HRS
would provide the user with a sense of the reliability and accuracy
of the procedures, which we suspect is about one-half to one order of
magnitude. These studies have been presented in peer-reviewed articles
in journals such as Environmental Progress and the Journal of theAir
Pollution Control Association*
In addition, the comparison of predicted and measured emission
rates (in Section 4.4.U4 of Analysis of the Air Target Distance Liplt
in theHazard RankingSystem, July 6, 198?) would be strengthed by inclusion
of a plot of predicted vs. observed rates. This comparison may indicate
significant variability on a site-by-site basis. This variability is,
in'part, due to the simplifications inherent in the models, but is also
largely a result of the difficulties in identifying nsodel parameters at
a given site, even for the most basic of Inputs, such as the identification
of chemicals In the waste. These difficulties are pertinent to the use
of these, or related model equations, as part of an HIS scheme that
considers' potential air emissions.
-------
A2-6
Ideally» if comprehensive information was available, the emission
models foe TSDFs could be uied. A brief review of these procedures is
presented below in order to highlight important variables.
Emissions from surface impoundments depend on the concentration of
chemicals in the water. Methods to estimate such emissions apply the
traditional "two phase resistance" transport equations which include an
estimate of the chemical's air-water partition coefficient or Henry's Law
Constant, The resistances depend on factors such as wind speed, fetch,
diffusivities, and the presence of oily films but, to a first approximation,
it would probably be acceptable to adopt mean values of the resistances
and, apply them to all chemicals. The key items of information are chemical
identity, chemical Henry's Law Constant, and cheaiical concentrations.
In the long-term the chemical will be depleted from inactive sites unless
there is migration from sediments to replace it.
Air emissions from landfills depend on the resistance to transfer
from the soil cover depth and porosity, coupled to the chemical's vapor
pressure as the driving force for diffusion. The key iteiss are chemical
Identity, chemical vapor pressure, and soil cover depth and porosity.
Emission rates can be estimated readily for tanks and drums given a
knowledge of the geometry and contents. Waste piles and other accumulations
of waste will require separate treatment. In all cases, the key variables
to estimate volatile emissions are chemical identity and vapor pressure.
Information on the chemical identities is valuable because it allows
the presence of emissions to be established and the calculation of emission
rates for volatile chemicals. When such information is lacking, air
emissions can not be predicted and the subsequent consideration of air
exposure routes is impossible These may be evidence of emissions as a
result of odors, for example*
If the chemical identities are known, It should be possible to
modify and simplify the TSBF procedures to calculate an approximate,
order-of-magnitude emission rate for impoundments and landfills. A
score could then be assigned based on the logarithm of the emission
rate. The population distance matrix discussed earlier includes the
effects of dispersion.
Other investigative techniques could be used to confirm the emission
rates, the most obvious of which is air concentration measurement*
Ambient monitoring, however, is subject Co the problem of dilution, as
the three examples on the next page show.
-------
A2-7
If we consider a site 100 m by 100 m emitting 100 kg/year Into a
wind of average speed 5 m/s, and assume that the air Is well mixed at
Che downstream sampling site to a height of 5 m, then the concentration
at that site will be C g/«3 given by;
LOO kg/yr x 1000 g/kg * 5 m/s x 100 m x 5 m x 31.5 x 1Q6 s/yr x C
or C =• 1.3 x 10~6 g/m3 or 1.3 ug/m3.
This low concentration may be difficult to detect quantitatively.
For surface impoundments, It is preferable to measure the water
concentration and infer an emission rate. For example, in the above
case an emission of 100 kg/yr over 10* m^ would require a concentration
of 0.023 g/m3, assuming an overall water mass transfer coefficient of
0.05 m/h,
i.e., 105 g/y - 104 m2 x 0.05 m/h x 8760 h/yr x 0.023 g/m3.
*
This concentration is a factor of 18,000 greater than the air concentration.
Likewise, if the emission Is from soil and we assume an effective
dlffuslvity (corrected for porosity) of 0.01 era^/s or Q*QQ36 m^/h and
a path length of 1 m and area 10* m^, then the effective diffusion
volume exchanged Is 0.0036 m2/h*x 10* n^/l m or 36 m3/h. This is
essentially the volume of soil pore vapor released per hour. If the
emission rate was 100 kg/yr the concentrations would be 0.32 g/m3,
i.e., LQ3 g/y - 36 «3/h x 8760 h/yr x 0.32 g/m3.
Again, this concentration is a factor of 250,000 greater than in the
ambient air and is much easier to measure*
Based on these examples and the Work Group's experience, to detect
the presence and amount of emissions it is desirable to analyze at the
source, or as close Co the source as possible. Once the chemical mixes
with the ambient' air, It suffers a concentration drop by a factor on the
order of 10* to 10^. This may produce ambient concentrations below
detection Halts,
Thus, there is thus an incentive to use methods of "probing"
sources by using, for example, soil gas analysis or "inverted dish pan"
surface emission samples. We believe that valuable confirmatory data
would be obtained at quite modest cost. These investigative techniques
could give direct measurements of emissions rates.
-------
A2-8
Site Investigations of soil, ground water and surface water have
been aided by the prior existence of sound analytical protocols for
sample collection, storage, and analysis. It appears that insufficient
effort has been devoted to developing protocols which are appropriate
for air sampling at waste sites. The traditional high volume sampler Is
often effective, but it is, to an extent, misapplied. More effort should
be devoted to developing more subtle, economic and useful air sampling
procedures which are specific to the conditions prevailing around waste
sites,
Other Factors Affecting the PotentialforRelease
The following additional factors should be considered in developing
the air pathway component of the HRSt
I. The type of containment, t*e,, impoundment vs. tanks, aad
drums vs. encapsulation, should be considered. Also included
should be an assessment of the extent to which the eontalnnent
is likely to be disturbed as part of the remediation process
at the site* Sites where more excavation and handling of
wastes are required may present a higher potential for air
releases,
2. Catastrophic releases nay create air emissions. This category
could include the susceptibility of the site to flooding,
the seismic activity of the area, and the frequency of
severe weather, e.g., extremely high winds which can lead to
fugitive particulate emissions.
3. Reported incidents by surrounding residents, such as complaints
concerning odors, eye irritation, or nausea, may result from a|r
emissions. While, in the absence of air monitoring data, these
reports may not be sufficient evidence of a "confirmed" release,
such sites should be scored for their potential-to-release*
OTHER COMMENTS
Relationship of^AirPathway to theRestof the HRS
As part of its revision of the HRS air pathway, the Work Group
recommends that OERR assure that the scores assigned for population and
distance are consistent with one another and with other scores with
which they must be combined, for example, scores for toxicity, waste
quantity, or release potential. If the HRS is to represent the relative
risk of sites, it oust be consistent with the underlying, quantitative
relationships of the factors evaluated. Weighting the risk among pathways
on an _a_ priori basis where, for example, the air pathway might be worth
one-half the ground water pathwayt is not scientifically justified.
-------
A2-9
Two aspects of risk need to be addressed* First, risk may be
declining or changing depending on Che nature or integrity of containment
or the site type. There may have been a significant release early in a
site's history which has resulted in a loss of contaminants and decreasing
emissions in later history (which often coincides with site evaluation
and scoring). This earlier release may have been of an acute nature
which no longer is detectable or representative of the site.
The second aspect is that the near-site population may be considered
to be at greater risk from che air pathway than from the water pathways.
Because of the nature of migration of air contaminants (speed of movement
and plume entrainment) the papulaeion at highest risk may change as
atmospheric conditions change,
There may be a need to consider the near-site population as exposed
to an independent direct-contact pathway rather than as a. subset of the
air pathway because acute rather than chronic toxicity may be more relevant
and because of the known presence at some sites of chronic toxicants ia
the form of particulates (such as dusts containing lead). Exposure to
the larger respirable particulates cau be s very near-site phenomenon*
which may be de-emphasized by inclusion la the air pathway with its
focus on greater distances, On-site workers (not cleanup staff, but
employees at active manufacturing or landfill sites) are also at risk in
manners different from long distance populations. Exposures may last for
fractions of days but be intense when they occur. Near-site populations
are continually exposed to a variety of routes in comparison to single
route exposure of remote populations.
Exposure also changes with the nature of contaminant migration.
Near a site, the contaminants are more varied as well as at greater
concentrations, and there are more routes of exposure. Volatile organic"
compounds are present and are subject to Inhalation. In addition, parti-
culates containing either metals or adsorbed organica tend to deposit
near sice boundaries and are subject to ingestlon and dermal absorption
as well as inhMlation* Remote from the site, concentration and variety
of contaminants are leas, and exposure is more limited to the volatile
compounds and the Inhalation route.
Determination of Actual Release
One of the important scoring factors in the current HRS is an oa-off
switch for the presence of actual release of hazardous material into the
ambient air. The determination of actual release Is usually based on a
limited amount of -air monitoring. The data provided by the program
office Indicate that 14.2% of the 951 NFL sites have shown observed
release, compared to 41,6f for surface water and 73,01 for ground water,
These statistics may be misleading and could result from differences in
detection efficiency rather than true dominance of the water routes.
-------
A2-10
The differences in detection efficiency are also thought to result in a
reluctance to collect data for this route, a problem which could be
addressed by a minimum data requirement for the HRS.
Air releases ate often Intermittent, with a substantial temporal
variation. Let "r" denote the fraction of days during which a release
occurs ac a site. If r>Q, the site should be claaslfied (In the current
HRS) as having an actual release. Strictly speaking, both the release
fraction and the presence of actual release are relative to the time
frame under consideration. If longitudinal air monitoring is available
for a long time period, and measurement error is negligible, the release
would be observed. In the absence of longitudinal monitoring^ there can
be a substantial probability that a site with intermittent release would
not show an observed release. If one assumes that air monitoring Is
conducted for one day, such as the day of site Inspection (chosen randomly).
For a site with release fraction.'V, (say 40%) the probability of not
observing the release Is 1-r, or 60%* Therefore, the prevalence of air ,
release for the NPL sites can be substantially higher than the value of
14.2%, which is based on the available data on observed air releases,
For example, if the release fraction for all NPL sites with actual release
is 40%, the prevalence of air releases would be about 35% (14,2%/40%),
among which only a small fraction Is Identified by the available data as
showing an observed release. Lt is likely that air releases can be
substantially more prevalent than what the available data indicate.
One possible remedy for the underestimation of actual releases is
to increase the frequency of air monitoring. For a site with release
fraction r » 40%, if air monitoring Is conducted for three Instead of
one days, chosen randomly (because of possible intertemporal correlation,
three consecutive days cannot be regarded as being random), the probability
of nqt^ observing the release would be 21.6t instead of 60%. However,
the amount of resources required to increase the air monitoring frequency
would be substantial, and might not be cost effective.
As dlscus,sed above* alternative sampling methods such as soil gas
monitoring may be easier than ambient air monitoring.
-------
ATTACHMENT B
Derivation of the AirTarget Score
Suppose population N is distributed uniformly
in a ring between distance x and bx. The
population density is given by
Suppose further that concentration, averaged over all
directions, declines as an inverse power of distance;
C - k/ra
If excess risk is proportional to concentration (as usually assumed for
carcinogens), the proportionality constant is P, and excess risk-PC=«P(k/r11),
then the population risk in the ring is given by
-bx
Pk
(2TTr dr)
-------
Both atmospheric diffusion theory, available data, and current
diffusion modeling practiced by EPA support a value for n of less than
or equal to 2 for ground-level sources. At close-in distances, less
than 10 km for example, the decay In the ting-averaged concentration is
inversely proportional to the square of the radial distance from the
source for average meteorological conditions. At distances greater than
10 km, the decay for average meteorological conditions is slower than
quadratic. EPA models used in performing risk assessments, such as the
Human Exposure Model (HEM), use formulas for diffusion that have the ring-
averaged concentration decaying approximately as r~l*?, for distances
out to approximately 10 ka for average metetologieal conditions.
As an example, one can refer to the table when a - 0.09 mile and
b =• 3,33 (n Is about 1.9). Figure B-l shows the rings for these assumptions
and Table B-l shows the matrix of scores.
TABLE B-l
Air Target Score for b » 3.33
Ring Boundaries (miles)
Population
106 - 107
lO5 - 106
lO4 - 105
1Q3 - 104
tO2 - LO3
id ~
100 -
(The scores above the line imply population densities greater than
30,000/ol2, which are rare.)
If one assumes that population is uniformly distributed at density
D - l,000/mi2, then the populations in the rings are about 260, 2900,
32,000, and 390,000 people. Each ring would score 6 points*, the overall
score would, therefore, also be 6.
The decline la Individual risk per ring is by definition one order-
of-magnitude. Therefore, if one arbitrarily defines risk to equal 10~^
for the innermost ring, then the corresponding population risk
is 260 x 1Q~3 * 0.26 (less than one case in a lifetime).. In ring 2 it
is 2900 x 10~4 or 0.29, in ring 3 It is 0.32 and in ring 4 it is 0.39.
Thus, total population risk continues to Increase as long as the population
density remains constant* Practically, individual risk has declined to
10~6 In ring 4 and would be negligible (10~7) in a fifth ring.
0
.09 - 0.3
10
9
Q I
7
6
5
4
0.3 - 1.0
. 9
8 r
6
5
4
3
1.0 - 3.3
8 {
7
6
5
4
3
2
3.3 -
7
6
5
4
3
2
1
U.I
-------
FIGURE B-l
Air Target Distance Score (not to scale)
Site Area
(radius about 0.09 miles)
-------
-------
APPENDIX 3
REPORT OF
THE LARGE VOLUME WASTE WORK GIOUP
TO THE
HAZARD RANKING SYSTEM REVIEW SUBCOMMITTEE
OF THE
SCIENCE ADVISORY BOARD
-------
U. S. ENVIRONMENTAL PROTECTION AGENCY
SCIENCE ADVISORY BOARD
HAZARD RANKING SYSTEM REVIEW SUBCOMMITTEE
LARGE VOLUME WASTE WORK GROUP
ROSTER
CHAIRMAN
Mr. Richard Conway, Corporate Development Fellow, Union Carbide Corporation,
Pose Office Box 8361 (770/342), South Charleston, West Virginia 25303
MEMBERS/CONSULTANTS
Or. Keith Ferguson, Environmental Protection, Environment Canada,
Rapilano 100, Park Royal, West Vancouver, British Columbia,
CANADA V7T1A2
Dr. C. Daniel Kealy, U. S. Bureau of Mines, Spokane Research Center,
East 315 Montgomery Avenue, Spokane, Washington 99207
Dr. James 0. Leckle, Department of Civil Engineering, Stanford University,
Stanford, California 94305-4020
Mr. Bruce Napier, Battelle NW, Post Office Box 999, Richland,
Washington 99352
Dr. David Pollock, U.S. Geological Survey, National Center, Matl Stop 411,
Reston, Virginia 22092
EXECUTIVE SECRETARY
Mrs. Kathleen tf, Conway, Deputy Director, Science Advisory Board, (A-101F),
U.S. Environmental Protection Agency, 401 M Street, S.W.,
Washington, D.C, 20460
' . STAFF SECRETARY
Mrs. Dorothy M. Clark, Secretary, Science Advisory Board, (A-101P),
U.S. Environmental Protection Agency, 401 M Street, S.W»,
Washington, D.C. 20460
-------
TABLE OF CONTENTS
Page Number
BACKGROUND A3_j
RESPONSES TO ISSUES/QUESTIONS RAISED BY OERR A3-3
Applicability of MRS Co Mining Waste Sites A3-3
Bias Against Mining Sites Not Demonstrated A3~3
Inadequacy of the HRS for Mining Site A3-4
Evaluations Not Proved
Inappropriateness of the Existing HIS for A3-5
Mining Sites Not Deaonstraced. But Has
that Potential
Factors to Modify the HRS for Large Volume A3-5
Wastes
Feasibility of Using Concentration A3-6
Value of Incorporating a Concentration A3-&
Factor
Desirability of Including Hazardous A3-7
Constituent Concentration Data in the HRS
Options for Incorporating Concentration Data A3-8
In the HRS
Other Coonenta on Concentration A3-9
OTHER COMMENTS A3-11
Loag^Tem Cons ide rations A3-11
Need for Additional Studies A3-12
Search for False Negatives and False Positives A3-12
SUMMARY ' _ A3-12
ATTACHMENT A—Issues from The Superfund HaaardRanking System (HRS);
Applicability to Mining Waste Sites» July 22, 1987
ATTACHMENT B—Issues from The Superfund HazardRanking System (HRS);
Feasibility of Using Concentration Pa'ta in a Revised HRS,
July 27, 1987[~
-------
A3-1
BACKGROUND
Mr. Henry L. Longest, Director of Che Office of Emergency and Remedial
Response (OERR), referred the large volume waste issue, "whether large
volume wastes should be considered differently from other wastes," to the
Science Advisory Board for review In his June 25, 1987 memorandum (22), The
Issues as stated more specifically in the July 28, 1987 memorandum (20) of
Mr. Stephen A, Lingle, Director of the Hazardous Site Evaluation Division
(OERR) are;
"1) Applicability of the HRS in scoring mining waste sites; and
2} Feasibility of using waste concentration, data in a revised HRS."
This memorandum also states, "we (QER&) believe that the Subcommittee
findings will be equally applicable to both mining and fly ash sites."
The Work Group concentrated on mining sites and assume OBRS will translate
its findings to utility waste sites.
The primary EPA document addressing the first Issue, The Superfund
Hazard Ranking _Systea (HRS); Applicability to Mining waate Sites (13) was
reviewed at the Work Group's August 20, 1987 meeting. Besides providing
background on the HRS, it presents an overview of mining waste site
characteristics, summarizes five 'previous studies on the application of
the HRS to mining waste sites, and provides an assessment of "potential
dangers" at six such sites rated by the HRS.
In the -overview of mining waste site characteristics, four major
categories of wastes from the extraction of ores and minerals were
addressed: mine waste, mill tailings, dump and heap leach waste, and
mine water, (The Work Group had the Office of Solid Wastes'* Management
of Mining Wastes (17) and chapters 1-5 of the Draft Report to Congress; '
Wastes from the Combustion of Goalby ElectricJutility PQwejr Plants (16)
as additional sources of information on the characteristics of large
volume wastes). Metals, radiomielides, asbestos, and cyanides (but not
organics) were Identified as the contaminants of concern at large volume
waste sites. The acidity (pH) was Identified as a key factor in the
nobility of metals, and the ground water and surface water pathways were
chose usually of concern at these sites,
Three of the five papers summarized and evaluated in the issue
paper were prepared by TRC Environmental Consultants, Inc. for the American
Mining Congress (27, 28, 29). The other two were Agency contractor reports
developed by MITRE Corporation (24, 25) In response to issues raised by
TRC. The TRC papers assert that the HRS does not adequately discriminate
risks posed by various mining waste sites as well as it does for non-mining
sites; the MITRE papers respond that TRC's conclusions are not supported
by their analysis, or by further MITRE analysis.
-------
A3-2
QERR briefed the Work Group on its comparison of scores for six
actual mining waste sites with the results of an assessment of the
qualitative overall risk to human health and the environment (potential
danger) at those sites (,U) * Ratings for four pathways were derived:
ground water, surface water, air and direct contact. For these six
study sites, higher HRS scores were associated with higher potential
danger ratings.
In its July briefing of the Subcommittee, the Office of Policy,
Planning, and Evaluation reported that its Site Ranking Panel (an
in-house group of EPA personnel with varying degrees of expertised
on Superfund Issues), using a different group of sites and experts,
apparently found virtually no correlation between the experts' ranking of
sites and the HRS scores (18). Since the unpublished QPPE study (23)
was not reviewed by the Subcommittee nor otherwise peer-reviewed, the
Work Group is unable to judge what credence may be placed upon its findings,
Taking the study at face value, it would appear that because most of the
sites chosen were already on the NPL, this may indicate that the HRS
discriminates poorly between listed sites, but does not necessarily mean
the HRS does not discriminate between those that should be and should
not be listed*
There, are two ways to evaluate the applicability of the HRS to
mining sites. The July 22, 1987 issue paper, The Superfund HazardSacking
System (HRS); Applicability to MiningWaste Sites, uses the approach of
evaluating the experience in ranking mining sites with the HRS system in
terns of false positives and false negatives and differentiating according
to the potential danger as determined by other means. In its evaluation,
the Work Group coupled this approach with an examination of the scientific
issues independent of the HRS experience.
The second issue paper is The Superfund Hazard Ranking System (HRS)i''
Feasibility of UsingConcentration Data in a Revised HRS, July 27, 1987.
Besides providing general background on the HRS, the paper discusses the
(non)use of waste concentration in the HRS, describes the use of waste
concentration data in other site ranking models, and presents three
options for using waste concentration in a revised HRS.
Basically, the inherent difficulties of sampling and analyzing
hazardous wastes, the heterogeneity of waste sites, cost, and the need
to quickly develop a NFL dissuaded the Agency from including concentration
in the original HRS. Pour systems which address concentration in various
ways were described: the Site Ranking System (SRS), HARM II, Site Assessment
System, and Remedial Action Priority System (RAPS). OERR'viewed the SRS
approach, a linear s'coring system which falls just short of being a
quantitative risk model, as having components of potential utility, but
cautioned that any of the systems rely heavily on the professional judgement
of field personnel.
-------
A3-3
The three options presented by the Agency for using waste concentration
data In she revised HIS are:
Wl Retain the current HRS approach where waste concentration data
are used only to identify the most toxic waste constituent.
W2 Mandate the use of waste concentration at all sites to determine
a total mass of constituent to be used in the revised HRS,
much as the waste quantity score Is currently applied.
W3 Adopt a tiered approach to encourage the use of constituent
concentration data, where available, and reliance on default
values of concentration to estimate the total mass of a
constituent at other sites.
The detailed questions posed to the Work Group on large volume
waste issue are presented in Attachments A, and B.
RESPONSES TO ISSUES/QUESTIONS RAISED BY OEM
Applicabilityof HRS to Mining Waste Sites
The full text of the questions posed by QERR can be found in
Attachment A* An abbreviated form* la used here to assist the reader.
Bias Against Mining Sites Not Peaonstrated. The TRC and Mitre
reports showed that mining waste sites scored higher in waste quantity
compared to non-mining sites. This is expected given the large volune
of wastes generated by the mining industry. TRC did not show this bias
caused false-positives or false-negatives for mining sites. Both TIC
and Mitre found waste quantity contributed relatively little to the
final HRS score,
TRC addressed whether the degree of site information available was
a major factor in HPL listing. Since some mining sites have more extensive
data bases than do others, TIC concluded that these may be more likely
to have higher scores and, therefore, to be placed on the NPL, compared
with sites of equal risk but with less available information. Most
major industrial facilities also will have more extensive data bases
than smaller sites (especially "orphan chemical dumps"). Extensive data
bases and a regulatory history may highlight major industrial sites
before others. Moreover, more frequent and "denser" monitoring has a
better chance to find environmental problems and observed releases; also,
sites with serious problems may be more thoroughly studied. In any
event, no data were 'presented by TEC showing that the relatively larger
data bases for mining sites resulted in unfair scoring. On the contrary,
smaller data bases could theoretically be biased. Use of default values
for scores where data are missing would be a better solution to the
alleged underscoring of sites with lesser data bases.
-------
A3-4
Analysis of the relative weights of components in the Hazard Ranking
System may suggest bias If one, kind of waste, such as mining wastes, always
scores high in an important component irrespective of environmental hazard,
TIC stated the final HRS score for mining sites could be predicted by
population alone. Since the TIC studied only mining sites, it did not
prove a bias against mining waste sites. Also, the more conplete Mitre
analysis indicated that population is the second-most important factor
(after observed release) for scoring of all types of waste sites and,
therefore, is a critical factor for all sites, not just mining sites.
In two areas, the Work Group concurs with TRC. First, Inorganic
metals have certain characteristics not addressed in a system that does
not take into account differences in mobility between specific Ionic
ferns. The Work Group assumes that OERS will consider these differences
in the revisions to the toxlcity/persistence score. Second, the large
volume of mining wastes often means large masses of the contaminant are
available for release, but certain conditions must exist for these releases
to occur.
in summary, while the TRC analysis identified some important ranking
priorities of the HRS system, its findings do not present a strong argument
Indicating a bias against mining sites.
Inadequacy of the HRS for MiningSiteEvaluations Hot Proven. After
reviewing the TRC and MITRE studies as addressed in the July 22, 1987
Issue paper The SuperfuadHazard Ranking System; Applicability to Mining
Waste Sites, the Work Group finds that the studies conducted regarding the
HRS listing process are preliminary in that they have examined only small
subsets of the data base, and are inconclusive. Therefore, sufficient
experiential evidence has not accumulated and been presented to show an
inadequacy of the HRS in regard to mining sites. The QPPE study (18)
presentation at the July 16th Subcommittee meeting, "Preliminary Analysis-*
of Alternative Models to Support Revisions of the C1RGLA, Hazard Ranking
System," also provided some data. That study involved mining sites
and 16 noraiining sites and used a panel of In-house EPA personnel (with
varying degrees of expertise in Superfund issues) to evaluate risk
for comparison with the various existing ranking models. Within its
limited study, QPPE found no consistency between its panel's ranking of
risk and ratings fron any of the models, including the HRS model. This
data set for raining sites seems United and cannot support a conclusion
about the adequacy of the HRS scoring system*
In summary, sufficient evidence was not presented to the Work Group
to conclude that the current HRS has been adequate or Inadequate for
mining sites (or even for non-mining sites).
-------
A3-5
Inappgopriateness _g£the Existing HRS for Mining_Sites_Not Demonscrated,
But Has ThatPotential. As stated above, the Work Group was unable to
find sufficient evidence that the HRS has treated mining sites with
systematic error. Further, although only a small subset of mining sites
was available for analysis, the Work Group was unable to identify any
sites currently listed that should not have been on the WL» The present
system has possibly rated mining sites no worse than other waste sites.
However, the present scoring system has a potential to treat mining
wastes (and others) with systematic error. The vast majority of listed
mining sites and, to a slightly lesser degree non-mining sites» have
observed releases. Since TRC reported (25, page 9) that 95% of listed
mining sites had an observed release and since observed releases require
less judgement in scoring, there was a lower risk of false positives,
The Work Group believes the present HRS Is not well suited for
scoring potential releases from mining waste sites because mobility is
not included, as discussed below.
factors to Modify_to _Make the HRS_L_for Large yoluffleWastes, In regard to
Improving the HIS, the following points should be considered:
o Mine wastes (one of four types of mining wastes) tend to be solid
rather than liquid, and contaminants may be less mobile.
o Where mining wastes contain low concentrations of hazardous substances
in large waste volumes only partially isolated from the environment,
environmental conditions instead of the extent of isolation may
control contaminant release,
o Natural geochemical processes are often required to mobilize
hazardous substances in mining wastes, such as acid generation
by the biochemical oxidation of metal sulfldes,
o The concentration of hazardous substances and the potential for
release can often be estimated for mining wastes, I.e., aeld sine
drainage (AMD) potential and leaching tests* More generally, the
HRS does not adequately consider the nature of the source, release,
transport, and transformation.
Improved ways of considering the concentration, toxicity
(inorganics have special characteristics), release (mobility in various
matrices), and trans port/trans formation factors would make the HRS more
accurate,
The concentration issue is addressed in the latter part of this
report and Appendices 1 and 2 address the toxiclty issue. QERR"s briefing
of the Work Group August 19-20, 1987 in Denver indicated that it is
considering ways to incorporate mobility factors within the HRS that may
have considerable promise (26).
-------
A3-6
Revisions of the HRS then under consideration by OIRR included
substituting a mobility factor for the persistence factor la the ground
water and surface water pathways. In the ground water pathway parameters
representative of the tendency for contaminant mobility and the sorptive
capacity of the geologic aedla would replace the persistence parameter.
The Work Group supports such changes because mobility is a more
discriminating concept for both inorganics and orgaiiics than is persistence
in the subsurface, especially with respect to Inorganic compounds.
The Work Group recommends that, during the development of a
structured value representation of the mobility concept, GERS should
explore a means of incorporating Important matrix characteristics. These
include extreme pH, any crystalline phase modifications of mining wastes
that differ from native geologic materials, and the sorptive capacity of
Surrounding geologic materials (influencing migration tendency).
Extreme pH is a particularly attractive characteristic because it is
meaningful and because the pH of mining wastes can be determined relatively
simply and accurately. Mining wastes tend to be highly buffered, either acid
or alkaline* All fluids oa a mining sites could be tested for pH and solids.
could be tested for paste pH, particularly near the surface (30). The pH
of pathway media and receiving waters are more difficult to determine
accurately. However, even here, samples should be obtained and measured
for pH in both the field and the laboratory,
In regard to the surface water route, QESR favors retaining
persistence as the parameter to couple with toxleity In the waste
characteristics portion of the HRS, but would consider four additional
transformation processes along with the current biodegradation one. The
Work Group concludes that inclusion of these additional factors would
improve the existing HRS because it more closely corresponds with what
happens In the real world.
Mobility would be Included In the air route; a gas mobility factor
value and a particulate mobility factor value would be combined. Details
were unavailable", but the basic approach of Including nobility should be
pursued due to its importance along with waste quantity in the exposure
potential scenario. (See also Appendix>2 on exposures through the air.)
Feasibility of Using Concentration Data In a Revised HIS
The full text of the questions posed to the Subcommittee by OEM
can be found in Attachment B. Abbreviated versions are used below.
Value ofIncorporating a ConcentrationFactor. The following discussion
assumes that the structured value model construct of HRS will be maintained
in the short-term*
-------
A3-7
Modifying the current HRS to incorporate factors which capture some
measure of both the physical-chemical characteristics of the hazardous
constituents and the nature of the waste matrix, a,s well as those site
characteristics responsible for risk, is clearly an improvement, Modifica-
tions relating Co the form and concentration of the hazardous constituents
(HC) would add a degree of sensitivity to the HRS and possibly capture
some measure of nobility and potential for exposure. Possible factors
include HC concentration in the waste material, total mass of the HC,
potential for release of the HC, mobility of the HC in the ground water,
and concentration of the HC in the ground water.
It appears that the most useful way to implement a modification of
the HRS to facilitate inclusion of additional factors, as well as create
flexibility to accommodate a range of site data quality and quantity, is
to augment the HRS with a multi-tiered default approach for use of HC
concentration or quantity data. Such a multi-tiered structure would
provide alternative routes for use of data differing in quality and
quantity, as well as provide the opportunity for inclusion of a decision
point on the question of acquiring new, additional data for the HRS
scoring process,
Perhaps the concept of using the most appropriate tier for a specific
situation Is more important than assuming that the top tier is best, and
only defaulting to a lower tier. That is, for solid and semi-solid
wastes at mining sites total quantity of hazardous waste aakes little
sense in the absence of mobility factors; rather, the concentration of
the contaminants is important. For drum and other liquid wastes, the
total quantity of hazardous material is probably Important.
Desirablity of Including _Hazardous, Constituent, Concent rat loo Data
in theHRS. Because both the impact of observed releases and the risk
of potential releases are related to source concentration, particularly
for the direct contact and air pathway, including hazardous constituent
concentration in the HRS would be desirable.
At preaentj the HRS is only a screening tool used by QERR, in
combination with other science policy considerations for deciding if
detailed (RI/PS) studies are required. SARA encourages the development
of a more accurate and comprehensive ranking system. Addition of
concentration data could improve the accuracy of the HRS.
EPA has proposed two options (WZ and W3) for modifying the HIS to
account for the concentration of hazardous constituents. Both approaches
translate waste constituent concentration data into an estimate of the
total mass of a hazardous constituent at the site, which Is then used to
compute the waste quantity score for the HRS. Both approaches represent
improvements over the current HRS, However, if mobility is not addressed,
even the approach which uses "total oass" of hazardous constituent could
contribute to produce false positives for some large volume wastes and
some pathways*
-------
A3-8
For the ground water pathway, the concentration of a constituent
IB a solid waste is of much less importance than the concentration
that constituent may attain In -the leaehate migrating away fronj a site*
Consequently, more emphasis should be placed on the partitioning of
hazardous constituents between the solid waste and the leachate*
Leachate concentration (at least estimated) should then be factored
into the HRS waste quantity term. For wastes that are not in solid
fora, total contaminant mass quantity should continue to be used.
Options for Incorporating Concentration Datala the HRS. Option W2
would require that waste constituent concentration data be collected at
a site before it could be scored. In contrast, Option W3 (the Work
Group's preferred approach) is a tiered system that would encourage the
use of concentration data, but would also provide the flexibility to use
indirect estimates, or default concentrations, to estimate a constituent's
total mass when direct measurements of concentrations are not available.
The direct measurement approach of option W2» modified for solid wastes
to account for partitioning between water and solid, would provide the
soundest scientific basis for the revised HRS. In moat cases, however,
it will not be practical due to safety and cost considerations to require
waste constituent concentration data for every site at the site inspection
stage. Furthermore, the Work Group" recommends the direct use of
concentration rather than conversion to a mass value for wastes in solid
or semi-solid form.
One of the most difficult tasks will be to decide which tier to
use. In the case of option W3, there is a need to decide whether
to use measured concentrations or indirect estimates of mass. It Is
likely that economic and safety considerations will determine the
decision as to which tier will be used. If a site is relatively
uniform and sufficient waste constituent data can be obtained at
reasonable cost to characterize the waste, tier one (concentration)
could be used. On the other hand> if a site Is non-uniform and costs
of obtaining the necessary concentration data (or time and safety
requirements) are prohibitive, it would be reasonable to use a higher
tier. In effect'* this approach may effectively distinguish between
inorganic (mining and utility) and organic waste sites because of the
greater heterogeneity of organic waste sites and the large expense of
organic analyses. Inorganic analyses are relatively inexpensive (assuming
speclacion ia calculated, not measured); therefore, this approach would
encourage the collection of waste constituent concentration data at
mining sites, where concentration data is often a critically important
factor* as discussed above.
In conjunction with using concentration data in the HRS, a method of
determining a "representative concentration" for a site based on a stratified
sampling strategy would be useful. In this approach, a complex site
could be subdivided into a set of more homogeneous regions or strata.
-------
A3-9
and a representative value for each stratum determined through more
limited sampling. Statistical techniques exist for manipulating stratified
data, which could then be recombined with appropriate weighting into a
final siagle HRS score for the site*
To define a factor for converting quantity of hazardous constituents
to quantity of hazardous waste, from a scientific point of view, the
median concentration obtained in the September 1986 MITRE report by
Arlene R. Wusterbarth, Hazard Ranking Systea_ Issue AnalysisjRelationship
Between Waste Quantity andHazardous Constituent Quantity (32) should be
used. QERR reported orally chat this median concentration of all HCs
at alt sites is 450 mg/kg, but there was wide variation. This value
relates to aggregates of constituents and only materials on certain
lists of hazardous constituents were sought. Hence, risk management
or policy considerations might cause EPA to select another value for the
default concentration. One alternative is to convert HC to toxicity units
(concentration times a relative toxlcity factor) rather than concentration,
A specific rule for moving between levels (tiers) in option W3
might be considered. The first level (total amount or concentration of
hazardous constituent) should be used where data are sufficient to support
an estimate of hazardous waste constituent quantity. This would occur
where there are data to determine the hazardous waste quantity accurately,
What data are available clearly show the quantity is significantly
above the amount necessary to receive the maximum score, or significantly
below the amount necessary for the next-to-miniraum score.
Where the quantity of hazardous constltuent(s) is not known, scores
could be calculated using the procedures of each of the first three
levels. Then, scores for level 1 and level 2 would be compared. If the
scores were within 251 of one another, the score for level 1 would be
used. If the scores were greater than 25% apart, the score from level 2'"
would be compared to that froo level 3. If level 2 and level 3 scores
were within 25% of one another, then the level 2 score would be used; if
not, the level 3 score would be used in the HRS. The final level 4
score would be used where data on waste quantity were virtually absent
(especially for well fields).
Other Ccgments oti Concentration, The Work Group commented on five
additional issues related to the use of concentration data in the HRS.
1. SOBtgTechnical and PracticalConsiderations. QERR's issue
paper on the feasibility of using concentration (14) addresses
only waste concentration data. The Work Group recommends
that concentration in the receiving environment, and perhaps
population densities, also be considered as discussed in the
Wolfinger report (3D* Data on receiving water concentration
data would aid in scoring observed releases. Fish bioassay
data may be useful at times ici ranking toxicity of releases in
the absence of, or in addition to, chemical concentration data.
-------
A3-10
Sites will have data bases of differing completeness and
accuracy, as occurs now in site evaluation. Modified scoring
systems must use large data bases effectively to produce more
accurate scores while handling with equal fairness sites with
fewer data.
Some toxic contaminants, such as metals, can he measured
in both solids and liquids at reasonable cost. All site
inspections at mining sites could gather valuable data on the
quantity of metals in the liquid and solid phases of the waste,
Evaluation Techniques. The revised HRS should be tested
for both high strength concentration/low volume wastes
(drums) and low strength concentration/high volutae wastes
(mine tailings), with both observed releases and potential
releases of contaminants.
The sensitivity of the new ranking system to concentration _
should be tested to determine the importance of concentration
in listing on the NFL and, thus, the need to obtain more couplete
and accurate hazardous waste constituent data. The analysis
would also be a first step in examining possible false positives
and false negatives.
W2/W3 Modification. A non-linear scale such as that currently used
for waste quantity may be appropriate. Such a scale would give low
scores to small, less significant sites, while not unduly penalizing
large sites. In a properly constructed algorithm, sites which are
less significant in terms of risk will score lower than those sites
which present greater risks.
For unknown waste quantities, the default scheme suggested by
Kushner In Hazard Ranking System Issue Analysis; Sites with "'
Unknown Waste Quantity (19) is prefered over a minimum default
value,
Quality of Waste Concentration Data. Sampling guidelines could
be developed for each waste type (i*e» drums, tanks, tailings,
impoundments, etc.). These could be modified by site inspection
staff based upon sice specific factors, but a minimum sampling
scheme should be identified.
Perhaps a relatively large number of samples could be
collected, but only a limited number analyzed initially. The
initial results and calculated HIS scores could be used to
decide on- the need for additional analyses. This approach
would apply only to contaminants that could be preserved.
-------
A3-U
More data may be required foe sites with no observed release
because they are more likely to score below the 28.5 boundary,
and waste concentration data are critical in assessing the
risk of potential releases.
Composite samples should be obtained to reduce the number
of analyses, while providing estimates of average waste
concentration.
5. Samp11ng Risk. Waste sampling schemes should be developed
based on a thorough review of existing data on hazards to
sampling personnel. Description of the waste» anecdotal
evidence, and data on the receiving environment should Indicate
the hazards posed to site Investigators, In suspected high-risk
sites, the sampling scheme should begin around the perimeter
where less contamination Is expected, and progress towards the
areas of expected higher concentrations (this Is also consistent
with good sampling technique). Small test holes and gas sampling'
equipment could be used on the actual site to test for hazardous
conditions.
Data requirements could be relaxed where severe onsite
health and safety hazards exist. In any case, these sites
should probably be addressed by the Superfund Retaoval Program
for their direct contact, fire and explosion capabilities.
OTHER COMMENTS
Long-Tern Considerations
Although SARA mandates a re-evaluation of the MRS, and presumably a
revision or alteration in a compressed time frame, it la nevertheless
useful to consider both the long-term and short-term requirements of the
HRS and subsequent site evaluation process. In that context, these
remarks focus on the Issues involved in development of quantitative
representation of .concepts involving contaminant concentration in the
waste materials, contaminant concentration in ground water and mobility
(a notion that combines release and transport).
Clear qualitative differences exist between many large volume waste
(LVW) sites and the more numerous hazardous waste sites containing admixed
synthetic organic/Inorganic hazardous constituents In both liquid and
solid form. LW may be segregated Into monofllla of a single waste type,
Some sites have multiple types of wastes (e*g», mill tailings, smelting
slag's), Sphere these wastes are spatially separate, the sites may be
relatively easy to characterize. Commonly, LVW sites contain low
concentrations of hazardous trace elements (HTE) in the waste matrix.
-------
A3-12
The release of these HTE is governed by very specific geochemical processes
and is subject to migration constraints. Scientists and regulators increasingly
understand the effects of these two factors on release*
Long-term evolution of the MRS should include consideration of detailed
numerical geoehemical trans port/fate models currently under developraent,
The advantage of eventually using transport/fate models In the HRS is the
explicit inclusion of quantitative representation of specific processes
and mechanisms responsible for the release, transport, transformation,
and retention of hazardous compounds. The use of these models entails
with enlarged data requlrewents, but can yield less ambiguous, more
detailed simulations appropriate for pathway calculations and, ultimately,
exposure estimates. Considerable time and effort will be required to
integrate these more complex transport/fate models into the overall
framework of the HIS and to test their performance,
Heed for Additional Studies
The studies of large volume waste presented to the Work Group contained "
little in the way of original investigation. Looking to future improvements
in the HRS, OERR should plan thoroughly peer reviewed studies, both to
review large volume waste sites ranked under the HRS and to examine
basic parameters in the model. A stud-y of the first type was suggested
by Mr. R. Walline of EPA Region VIII which could address the actual
reasons why release occurred at problem mining sites (reasons reportedly
are different than those initially assumed). A study of the second type
might address the characteristic(s) of mining waste that control release
(mobility) to ground water.
Search for False Negatives and False Positives
Because it is difficult to design a system that can in all cases
make the best use of all the site-specific scientific information available^
the Agency should consider the advantages of utilizing additional scientific
judgment outside of (or adjunct to) the HIS model in making decisions on
whether to list sites on the NFL.
Screening models like the HRS must be simple. They do not have much
resolving power and therefore, some false positives and false negatives
are inevitable. Because of this limitation, HRS scores should not be
overemphasized. A process should be established either to review sites
subject to scoring or to review HRS scores in an attempt to spot false
positives and negatives. Such a scientific review process could involve
the use of additional models; in many cases, however, obvious false
negatives or positives might be best handled by a "manual" review. The
important point is that the system be flexible enough to allow for a
variety of approaches to be used, as needed, during the scientific review
process.
-------
A3-L3
The extent of waste characterization data to be collected at the SI
level is part of 9 cost/benefit calculation. The economic Impact of
being falsely included or not included on the NPL should be considered.
SUMMARY
Studies of the experience of applying the HRS to mining sites have
not proved systematic error unique to such sites,
However, the HRS has a potential to produce abnormally high scores
for mining sites due to the manner in which toxicity, mass, release, and
transport/transformation processes are addressed.
-------
ATTACHMENT A
The Superftmd Hazard RankingSystem (HRS); Applicability
to _Mining Waste Sites, July 22, 1987, defines the issues for SAB
consideration as follows:
1. "Does the series of TIC analyses indicate that the HES is
unfairly biased against mining waste sites (as compared to
other types of potential Superfund sites)?
2. "Is there sufficient evidence that the existing HRS scortng/UPL
listing process has been inadequate with regard to mining
sites evaluated to date (t.e.» produced numerous false
positives or false negatives)?
3* "Has it been demonstrated that the existing HRS is an inappropriate
scoring system for determining whether mining waste sites
should be listed on the NPL (i.e., subjected to further
Investigations and analysis)? If so, what factors should
EPA consider modifying to make the current HRS more appropriate
for mining sites?"
-------
ATTACHMENT B
The Superfund Hazard Ranking System _(HRS): Feasibility of _ Us ing
Coneeatration Data ina RevisedHRS, July 27, 1987 defines the issues
for SAB consideration as follows;
1, "Based on the information presented and other daca available,
does it appear to be deslreable to include waste concentration
daca in Che HRS?
2. "What additional technical and practical considerations
ought to be factored into the decision as to whether or
not waste concentration data should be Included in the HRS?
3. "Can the SAB suggest any systematic quantitative or semi-
quantitative methodology for deciding whether to include
waste concentration data and for evaluating systems that
do so in the HRS that would consider effectiveness of the
system at generating risk-based scotes, the costs of data
development, and potential improvements in decision making?
4. "Which of the suggested new approach* (W2 or W3) appears to
be more desireable?
5. "Can the SAB suggest any" changes in the structure of options
W2 or W3 that would make them more desireable (e.g., changes
ia decision rules, parameter values)?
6. "What quality level of data on waste concentrations data
quality (what types, amounts, and level of precision)
should be required if such information is to be used in
developing HRS scores? Is there a simple, robust way to
specify data quality that could be easily applied by a
wide range of field personnel at a wide variety of sites?
7. "With regard to alternative W3: —What general advice can
be given regarding how to define rules for moving between
tiers? How 'good' does waste concentration data need to
be to allow its use?
—Does the approach taken to defining conversion factors
(particularly the assumption of 200 ppm typical waste
concentration) appear to be reasonable? What alternatives
could be suggested?
8. "What approaches can be taken to mitigating risks to field
personnel during waste sampling? Should data requirements
be relaxed at sites where severe on-site health and safety
hazards exist?"
-------
APPENDIX 4
COMMENTS ON EXPOSURE
BY THE
HAZARD RANKING SYSTEM REVIEW SUBCOMMITTEE
OF THE
SCIENCE ADVISORY BOARD
-------
TABLE OF CONTENTS
PAGE
EXPQStTRE
Why is it Importane Co Address Actual and &4-1
Potential Exposure
Options foe Assessing Exposure A.4-1
Relationship of Source Characteristics to Emissions A4-2
How Comprehensive Should the MRS Be in Addressing A4-3
Environmental pathways and Routes of Exposure
Environmental Pathways A.4-3
Routes of Exposure ' A4-3
ME I and Population Risks A4-3
How Can the Revised HRS Better Account for Exposure, A4-3
Both Actual and Potential
-------
A4-1
EjCPQSURE
Why is in Important to Address Actual and _PotentlaI Exposure
Exposure occurs when there is contact between pollutants and receptors.
Addressing exposure is Important when there are toxic chemicals present
because the determination of whether there is any actual or potential
risk to health or the environment depends on both exposure and toxicity
Sphere there are no toxic substances or where there is no exposure to
them, there Is no risk. To the extent that the HIS Is intended to assess
relative risk, it must address both toxicity and exposure sufficiently
to determine risk. Where information on either toxicity or exposure is
incomplete, risk cannot be fully assessed, Eisk can be evaluated
by assessing four factors: (a) the presence of chemicals, (b) their
toxicity, (c) their potential for release and mobility, and (d) the
probability of human and environmental contact.
Options forAssessing Exposure
Ideally, a quantitative risk ranking system would rely on data for
all variables of the risk algorithm. Because it is not possible to
obtain quantitative data for all the necessary variables, semi-quantitative
approaches must be considered. Surrogates for quantitative values may
be developed through several mean?, they include identifying the appropriate
measure of central tendency for the known data values, modeling values
from the few known data points for the site of, concern or for comparable
sites, using indirect measures (such as total waste volume as a substitute
for concentration), or simply applying points in the algorithm based on
a subjective evaluation of the available descriptive information for the
site. An example of a tiered approach for the use of waste quantity
information Is discussed in Appendix 3* A similar strategy could be
developed which uses measured values in preference to indirect estimates .-
for a variable of exposure. When the ideal data are not available,
however, options to estimate the parameter can be used in a tiered process
that moves away from actual measures toward increasingly indirect measures.
In this manner, -the maximum value of the data Is retained while the goal
of evaluating a variable is met*
Exposure should be assessed in a manner that Is both chemical-specific
and pathway-specific as noted in Appendix 2, The potential for release
and mobility Is related not only to the chemical of'concern but also to
the environmental media through which it mist travel before exposure can
occur. Without this specificity, any attempt to address exposure Is
likely to result in inaccurate estimates of risk.
The most attractive option is to determine the presence of individual
chemicals of known toxicity and then assess their potential for release,
mobility, and contact with humans and environmental receptors. Much
could be gained by adding a chemical-specific exposure score that could
-------
A4-2
be combined with chemical-specific toxicity scores to determine which
chemicals dominate the risk at 9. single site and how the risk might
compare across sites that are otherwise similar (e.g. in population
distribution, existence of valued environmental resources, hydrogeology,
meteorology, and containment). While little is known about the amounts
of chemicals present and their concentrations in environmental media
after only a site inspection, the insights to be gained from a ehenical-
and site-specific approach are so great that it should not be lightly
dismissed,
Relationship of_ Source Characteri_s_tlc_s to Emissions
The HRS has been limited by not taking into account the potential
for mobility of specific chemicals through specific pathways. Clearly,
detailed estimates of pollutant emissions by pathway are not feasible
with the information and resources available for scoring a typical
candidate site with the HRS, (Here emissions should be taken to mean
pollutant discharges generally, not just those to the air.) Given some
minimal information about a site and its chemical inventory, however, it
may be possible to make crude estimates of partitioning among environmental
media and, therefore, of emissions by route which could be used to generate
part of a chemical-specific exposure score.
Information needed to derive-such estimates includes;
1* An estimate of the total mass or source concentration of important
chemicals at the site.
2* A description of the site-specific factors influencing release,
e*g,; impoundment vs, landfill.
3. Partition factors applicable to the chemical, e.g., Henry's law "'
constant and organic carbon partition coefficient.
Some characterization of the potential for migration also allows
better consideration of time as a factor in site management. A chemical
that quickly moves off-site through air or surface water may pose high
risks to surrounding populations in the short-term, especially if it
acts as a threshold toxicant, whereas an immobile (but persistent)
chemical will remain a potential hazard for long periods, especially If
it Is a non-threshold toxicant, e.g., a carcinogen. Estimates of the
total mass deposited and the emission rates would permit an estimate of
the site's "lifetime" if not remediated. Such estimates might provide
a better appreciation of the urgency for remedial actions'.
How Comprehens iy e Should the HRS Be in Add res s ing Env_iroOB»ntal.
Pathways and _Routes of _Exp_osure?
Soth toxlcity and potential for exposure depend on environmental and
biologic processes. An evaluation of environmental pathways addresses
-------
A4-3
the environmental parameters that affect risk and an assessment of routes
of entry provides the information needed to estimate risk to humans and
and the environment.
Environmental Pathways, Environmental pathways include air, water
(both ground and surface waters), soil and intermediate compartments
such as plants and animals that become part of the human food chain.
The potential for release and mobility of chemicals through these
pathways affects risk estimates. Factors to consider include vapor
pressure, time of transport from point of origin to populations at risk,
chemical transformations that may occur during that time, and the toxicity
of the chemicals that are expected at the point of exposure. Chemicals
behave differently in different environmental pathways, so all pathways
need to be evaluated fully to estimate risk.
Routes of Exposure. Routes of exposure (entry) occur by ingeatlon
(through water, food and soil), inhalation (through air and water vapors),
and direct contact (through water and soil). Biological processes such
as metabolism and pharmacokinetics affect the dose delivered to tissues
that may experience adverse effects. Biological uptake, delivery to
tissues and transformation of chemicals to more toxic chemicals within
the body are aspects of risk that can not be directly assessed in the
HRS* As -a result, where, route-specific toxicity information exists, the
toxicity and biological impact of chemicals need to be estimated fay
route of entry and incorporated into the toxicity factor of the algorithm.
Without a syntheses of these route^speciflc evaluations, risk nay be
either incorrectly estimated or under-estimated.
MEI and Population Risks. SAM Section l05(a)(8)(A) states that
the HRS should take into account "the population at risk." The Maximally
Exposed Individual (MEI) lg not considered in CERCtA or SAM. While
population risks always come Into play due to the impact of collective
individual risks, some "minimal" individual risks (such as birth defects)
associated with the MSI approach have major lifelong impact* Furthermore,
the Mil is a very widely used measure of risk.
The current HRS addresses HEX in considering distance to nearest
well and population risks in considering air target distance.
How can the revised HRSbetter account for exposure, both actual and potential?
Exposure requires contact between person and pollutant with risk focused
on this point of contact. Sources; transport and fate; and the presence,
nature,and activities of receptors are relevant to exposure*
Measurement of exposure may fae person-based or eonnaonity-baaed.
-------
A4-4
The current HRS finesses the assessment of exposure and instead
uses the probability of release (and other factors such as persistence,
waste quantity, distance to nearest well and population affected) as
proxies for exposure*
The main difficulty of accounting for direct exposure seems from
the need to consider potential as well as actual exposure* There are
two aspects of exposure to be considered here—the concentration of
contaminants in the environment around the site and the number of people
around the site. Actual release may be demonstrated by environmental
sampling. Potential release may be addressed by considering properties
of the site and chemicals present such as distance to ground water and
volatility. Similarly, the number of people (receptors) present and
actually exposed today can be counted, while the number of people
potentially present at the location in the future can only be estimated
from studies of population dynamics.
While estimates of persons-based exposures can be developed from
biological samples, from personal household samples, from self-reported
exposures, and from self reported symptoms, each approach has its
limitations. As a practical matter, biological samples are rarely
available. The cost of personal or household samples are prohibitive
and would.allow only limited opportunities to evaluate potential ambient
exposures. Because subjectivity can easily enter into self-reported
exposures, objective measurements need to be used to limit the subjective
variability. Subjectivity also enters into reports of symptoms (whether
self-reported or based on medical record studies) which serve as a
proxy for exposure. Nevertheless, the reporting (by nearby residents)
of physical symptoms consistent with acute effects of contamination from
the sice could be considered as evidence of possible, but not confirmed
releases. This is one of the situations which could be addressed by the
assignment of discretionary points to the HRS score, (see Appendix 5)
Communicy-ba3ed_ exposures can be developed from ambient pathway
measurements, from site measurements, or from modeled exposures. Numerous
assumptions about targets (such as animals or humans) and their behavior
are needed to impute exposure from ambient pathway measurements. However,
even more assumptions are needed to impute exposure from site measurements
(see also source characterization discussion above) because the monitors
are placed further from the potentially exposed population requiring
additional assumptions about the time and mobility factors that operate
between the source and target population. Even more assuaptions are
required to model exposure from similar sites because no two sites are
exactly the same., Modeled exposures nay be based on similar sites and
can be used to predict rare events. The latter two approaches (site
measurements and modeling) are discussed further below.
-------
44-5
•The way the current HSS addresses source ch_aracterizatj.oo-~includlng
both the evaluation of the mobility and j_at_e_ of chemicals and the consideration
of concentration or quantity of chemicals—-is insufficient to model
exposure.
incorporating concentration and mobility could produce a more
comprehensive ranking .scheme. One way to do this is by using
concentration as a weighting factor for the waste quantity. The
resulting effective waste quantities for low quantities of high-
concentration waste might be similar to those for large quantities
of low concentration wastes. Similarly, the scores for toxiclty/
persistence and for effective waste quantity could be adjusted by
considerations of mobility.
From a scientific standpoint, exposure may be modeled by imputing
an exposure for a site based on what is known of similar but more fully
studied sites. If there are sufficient data on the actual exposures
from sites with similar characteristics a potential exposure for the
specific site could be imputed, but it is njore difficult to define sites
which are comparable In this context than.to define sites which are
comparable in terms of quantity and concentration. For exposure, the
sites have to be comparable In geology (especially in mineralogy) as
well.
-------
-------
APPENDIX 5
COMMENTS ON THE ALGORITHM
BY THE
HAZARD RANKING SYSTEM REVIEW SUBCOMMITTEE
OF THE
SCIENCE ADVISORY BOARD
-------
TABLE OF CONTENTS
PAGE
ALGORITHM 45-i
Background A5-}
Why is it Important to Pay Attention to the A5-1
Overall Alogrlthm?
Can the Revised HRS Represeat Relative Risk? Or A5-2
Can it Only Serve as an "Qn-Gff" Switch?
What do These Choices Mean?
What Methods are Available to Develop a Ranked List? A5-4
What Rules Should be Used to Develop Scores Based on A5-5
oo Risk?
Aggregation of Pathway Scores A5-7
How Can the HRS Make Better Use of Temporal Analysis? A5-8
What can Be Done About the Quality of the Data Used in AS-8
the HRS?
What Can Be Done When There Isn't "Enough Data" For A5-9
The HRS?
What Should Be Done When Data Are Available That Exceed A3-10
What Is Needed For The HRS?
Caveat AS-11
-------
A5-1
ALGORITHM
Background
The Hazard Ranking System operates through the following three steps:
1, To Che extent,possible, specified information regarding the
properties of the site and of the chemicals deposited there
is gathered and described quantitatively or qualitatively
according to the needs of the HRS.
2. The information is manipulated, usually through comparison
to categories in tables provided in the HIS manual, to
yield a series of factor scores. The specifications of the
conversion from raw information to factor scores may be
called scoring rules.
3. The individual factor scores are combined within categories
and then the category scores are further combined to yield
route scores and eventually an overall migration score for
the site. The specifications of how the scores are to be
combined may be called combining rules.
Together the scoring rules and the combining rules form the overall
logic and procedure, or algorithm, for the operation of the HIS. To a
substantial extent, the methods for gathering the raw data (step 1 above)
and deciding how much effort should be expended on chat step are left to
the individuals doing the scoring. The algorithm, however, is a fixed
procedure that undergoes OER1 sponsored quality control. As such> it is
the key to how well the HSS will perform ia assessing relative risk.
Both the relationship of the raw data to the factor scores and the ways
in which the factor scores are combined must be treated carefully and
consistently to optimize the performance of the HSS.
Why_is It important to pay attention tothe overall algorithm?
Improving the algorithm could potentially do more to improve the
HRS than fine-tuning individual components.
The algorithm should be approached by thinking about the best way
to assess risks for a hazardous waste site without regard to the availability
of site-specific information and then simplifying the rules (but not the
logic) as needed to take into account real-world constraints —that is,
to.identify the most important variables and then develop" data (or surrogates)
and combining rules-.
The overall algorithm needs to meet the intended purposes of the
HRS as stated by SARA. SA&A requires EPA to modify the HRS so that, "to
-------
AS-2
the maximun extent feasible, it accurately assess the relative degree of
zrisk to human health and the environment posed by sites and facilities
under review," In Mr, Longest's June 15» 1987 memorandum (see Appendix
7 for full text) he states, "QERR believes that the purpose of the HHS
is primarily a screening tool used to determine which sites will be
candidates for Fund-financed remedial response. To the degree possible,
the model also needs to provide a stratification of sites based on the
relative risks posed to human health and the environment." It is not
clear that the current system stratifies by risk, not that it can serve
as a successful on-*off screen without stratifying by risk at least in
the region of the cut-off score.
At a tainimum the HRS oust discriminate sites into two risk groups—on
or off the NPL. Ideally, the HRS would individually rank all sites
based on accurate ascertainments of risk—resulting in as many relative
risk ranks as there are sites to be ranked. While the binary ranking is
minimally sufficient, it is apparent that programmatic concerns demand a
ranking that is more than binary. The decision about the level of ranking
was not within the scope of the Subcommittee's review, instead the
Subcommittee focused on developing conclusions and recommendations for
the scientific bases of an effective ranking system. The Subcommittee
proceeded to evaluate the various components and structure of the algorithm
to work toward a scientifically valid approach to estimating risk.
One such component relating to risk is the toxiclty score. The
additive structure of the matrix weights the lower end of the two
components (toxicity and persistence) and thus implies that any toxic
effect or persistence is more important than the probability or severity
of the effect or the degree of persistence. These baseline assumptions
have the effect of treating toxicity and persistence in a binary manner,
but this method of assessment may not be sufficient to meet the goals of
the HRS and the Subcommittee did not find it to be an acceptable approacrh
if the HRS outcome is expected to be more than binary. In addition—and
as evidenced by the very high percentages of sites receiving the maximum
score for toxicity and persistences-Che Inability of the current toxicity
module to rank substances effectively on relative risk puts a severe
limitation on the entire HRS to discriminate between sites on the OTL
and thereby to prioritize appropriately on the bases of public health
and environmental protection concerns. The scoring methodology needs to
be re-evaluated in terms of the overall purposes of the HRS.
Can the reviped HRS represent relative __ri9k? _r__0r_can it only serve __as an
"on-ofj" switch?What do these choices mean?
The Subcommittee has suggested changes that will allow the HRS to
provide a more accurate and scientifically based estimate of the relative
risk of candidate sites. To the extent possible, the HRS scores should
-------
A5-3
correspond Co an objective evaluation of relative risk at sices. However,
this is not always feasible due to both scientific and data limitations,
as well as the value and policy decisions implicit when considering and
balancing human health and environmental Impacts.
Uhile recognizing that the Agency must continue to base NPX, listing
decisions on many factors in addition to the HIS, the Subcommittee believes
that a revised HRS, better designed to evaluate sites by relative risk,
will provide an improved mechanism for determining which sites should be
included on the NPL, and can potentially provide meaningful input to the
subsequent assessment of NPL .sites. Most of the changes needed to revise
the HRS are changes in the overall algorithm and not changes with vast
new data requirements. (Pages A5-5 through A5-8 discuss desirable changes
in the algorithm.)
If the HRS is to be used to set priorities based on relative degree
of risk, aany factors need to be considered systematically in setting
priorities.
Although it is possible to talk about an on~off switch and evaluate
the HRS by whether it rates most sites on when they should be "on" and
vice-versa, in fact there is a continuous range of scores with an arbitrary
cut-off level designed simply to.get 400+ sites on the original list*
Unless the list is la reasonably good order both above and below the
cutoff, its fidelity in distinguishing high and lesser risk sites will
depend markedly on the choice of the cutoff, i.e., on how many sites (or
what percent of sites) are wanted on the list,
There are costs associated with miselasslfication. If the HRS will
be used to decide whether or not to spend $850,000 on an RI/FS, the
economic cost of a false positive Is approximately $850,000 to the fund ..
(although less to society because even the low-risk site might need
remediation and would benefit from the RI/PS), The social cost of a
false positive aay be reflected In decreased property values, Increased
physical stress.and community.concern, and in extensive time and effort
expended by public and private officials attempting to address these
concerns. The social cost of a false negative is the difference in
average risk between the sites that should be on and those that should
not, less any benefit of attention independent of Superfund. Unfortunately,
the average risk of Superfund sites Is not known, let alone the risk of
those that were scored but which did not wake the NPL.
The difficulty remains that there has been no analysis to determine
whether the current HRS score is a good predictor of the risk ultimately
generated by the formal risk assessment process conducted during the
RI/FS phase. This evaluation needs to be done as a guide to future
modifications of the HRS.
-------
A5-4
The question remains, "risk of what and to whom?" To create a
risk-based iRS chat evaluates sites, EPA would first need Co reach an
acceptable definition of risk which combines the health, ecological, and
economic damages expected over the life of the facility. The Subcommittee
favor multiple measures Including a human population risk measure. The
latter is especially Important because quantitative techniques are available
to assess it. Some subjective increases or decreases in score could be
used to account for severity of human health effects and likelihood of
effects on non-human biota,
What methods are availableto develop a ranked list?
There are three common methods for developing ranked lists.
One way is to devise a highly simplified mathematical model that
produces a number that Is approximately proportional to the true ranking
variable—-in this case, some agreed-upon definition of risk. (Actually,
all that la required is that the number increase nonotonically with
risk.) The system referred to as "RAJPS" is such a model. Although
simplified modeling has the advantage of being related to risk,, It has
the disadvantage of being a very substantial departure from the current
MRS. RAPS is probably too complex' to work well with the limited data
available for most HES candidate sites, even though it has worked well
for DOE sites.
Another method is to devise a aeries of screens that successively
divide the list of all sites into categories, which are later ordered in
some fashion that Is approximately the same order as risk.
The third major technique, often simply referred to aa "scoring",
is structured value analysis, of which the current HRS is an example.
Although the other two approaches have their virtues, it is reasonable ""
to concentrate on variations .of the structured value approach because
che current HRS uses it. Even here, two different approaches are possible.
The current HSS appears to take a so-called "empirical** approach.
That is, a superficially plausible set of rules was developed and then
modified "empirically" to match a subjectively ranked test set of sites.
If the subjective ranking was truly what the HIS was supposed to achieve,
and if th* test set was adequately representative of all sites to be
scored, then this method is perfectly satisfactory.
The other approach, a risk assessment approach, is to begin with a
clear understanding of how to rank the list quantitatively if all the
needed information and resources to process it were available. Then the
risk assessment is simplified until it Is obvious how to transform it
into a scoring system. The scoring system is further simplified until
It can operate at a reasonable cost on the often very sparse information
-------
A3-5
available. For example, che population risk at a sice might be represented
by Che sum over all routes of risk by route, and each route-specific
tisk might be represented by the sum of all chemical-specific risks by
route. lach chemical-specific risk might be represented by an integral
over population at risk of the product of toxic potency and time Integrated
exposure.
What rules should be used to develop scores based on risk?
First of all, internal consistency is important. Making rules that
are consistent requires little up-front investment and no additional
costs in the scoring of specific sites. These changes may be inch wore
cost-effective than additional data collection,
The way that the various pieces/components of the score compile
should reflect ho* their real world counterparts interrelate. The approach
for accomplishing this is to begin with a physically-based exposure
assessment model for each exposure pathway, structured properly to translate
expected or potential releases into environmental concentrations and
subsequent exposures and effects. These models would, of necessity! be
highly simplified for a screening assessment* Even a simplified exposure
assessment model may not be feasible given the time, resource and data
limitations associated with the HRS process. Still, the manner in which
the pathway scores are estimated- and combined should be consistent with
the fundamental material balance and exposure principles of such an
underlying model. The following points discuss how this can be accomplished
in the context of the structured value approach used in the current HRS
systea*
In a scoring system designed to reflect relative risk, the scores
could be added across routes and chemicals, with each score being a
product of scores representing population, toxic potency and exposure
potential. To work properly, the scores for these three factors would
have to be approximately proportional to the estimated values of the
factors. For population and potency the answer would be relatively
easy, but the exposure score would probably need to be constructed from
scores representing various factors related to exposure, such as mass of
chemical deposited, containment efficacy, nobility, and persistence*
Exactly how this is done is not as important as that the rules for
scoring factors are consistent with the rules for combining scores, and
that both are reasonably consistent with how the factor enters the risk
equation in a risk assessment.
In contrast,.the rules for combining scores in the current HRS do
not always reflect relative risk. Scores often rise as the logarithm of
the factor value being scored. For example, each order-of-magnitude
Increase in potency may receive an additional point, and each factor erf
two or two-and-a-half increase in population at risk may receive an
-------
A5-6
additional point. Using logarithmic scores is an appropriate way to
account for a wide variation in magnitude of the possible values for
factor scores; furthermore, logarithmic scores can be added to simulate
multiplication of the underlying factors. For example, when the HRS
assesses a site with carcinogens, the population at risk can be estimated
by multiplying the size of the population at risk by the risk per individual
which can be estimated as the product of the exposure of each individual
by the potency of the carcinogen. The combining rule should add the
logarithmic scores corresponding to the three factors. In the case
cited, however, the logarithms are to different bases; potency is scored
by a logarithm to the base 10, but population is scored approximately
to the base 2.3* Because (2.3)3 is apptoximately equal to 10, it takes
3 population points to be equivalent to I potency point, and the potency
scores must be multiplied by 3 (or population scores divided by 3) before
they are added. In this case, the HSS "weighting factor" of 3 Is exactly
what is called for by the existing science. It is. a natural outcoae of
the simplification process and not a subjective "weight". In the MRS, the
weights do not always conform to the science, and the combining rules are
not always consistent with the scoring rules or with one another.
The use of a scoring system to place sites on the NPL is certainly
appropriate and, as scoring systems for priority setting gof the HRS is
already better than most. Almost all of the factors scored are plausibly
related to risk and generally one point means the same in different
parts of each scale, usually a change of an order-of-magnitude or a
constant factor.
However, the current HIS is not always Internally consistent in the
number of points which relate to a change in risk. For example;
1) A single point does not have the same mathematical relationship
to the final score in one part of the algorithm as a single
point has in another. This variation in scoring implies
different assumptions about how the factors relate to risk,
2) Subjective weighting factors ate included. For example,
the route scores are combined by a route mean square rule
so that when one route dominates, it alone is enough to place
a site above the cut-off, (However, at many sites, the
risk through one route will usually dominate the other two.
Whether or not such weights should be used is debatable,
However, if weighting is used. It should be explicit.)
3) The rules for combining SOUK scores In the current HRS do
not relate properly to the observed behavior of those factors,
For example, consider the way In which the total waste
characteristics score is derived from its components. It
is the sum of the component scores for quantity and toxicity/
persistence (toxictty and reactivity for the air route).
-------
AS-7
Technically speaking, a sice with much harmless material could
be eligible for the NPL if it receives the maximum scores for
observed release/route characteristics containment, quantity,
and targets.
This is an extreme case. However, the combination of summation, and
multiplication in the algorithm has problems beyond this rather artificial
example. A useful interpretation of the current HRS would be that since
both toxieity/persisteriee and quantity scores appear to be logarithmic,
summing them would be correct for assessing the total waste characteristic
score on thelogarithmic scale. However, the same interpretation does not
hold for other parts of the algorithm. For example, the component scores
for observed release/route characteristics-containment, total waste
quantity characteristics, and targets — are multiplied. If the same
logarithmic interpretation is to be applied to these components, then
all those multiplications should be changed to summations. Otherwise,
the summation in the waste characteristics score should be changed to
multiplication. The underlying question is whether the entire score
should be interpreted on the logarithmic scale.
For another important component, observed release/route characteristics
containment, the scale used for scoring appears to be the probability of
release instead of its logarithm.
There are two important concepts here* First, the way the scores
are combined should reflect the way those components relate to one another
in the real world. Secondly, the final HRS score should be on some kind
of easily understood scale (arithmetic, probability, logarithmic»
etc.) so that the relative risks presented by various sites can be easily
understood. "Normalizing" scores among routes only makes sense if one
believes that 100 points Implies the same risk through each route. On
the other hand, there is no easily implenented way of interpreting
any of the scores and the equality assumption is not disprovable.
Aggregation of Pathway, Scores
A variety of methods can be used to aggregate pathway-specific
scores* If sufficient structural information on the pathways scores is
available, such information should be used in choosing the aggregation
method. For example, if the pathway scores are constructed on the
logarithmic scale, the exponential averaging method should be used. If
the pathway scores are constructed on the arithmetic scale, the linear
averaging method should be used,
' The quadratic averaging method used in the current HIS is not
only arbitrary but,'from a scientific viewpoint, arguably incorrect*
OPPE reported that its unpublished Site Ranking Panel study compared
three different aggregation, rules? linear averaging, quadratic averaging,
and cubic averaging. The results (within the limitations of the study)
indicated- that the panel's final ranking was closest to cubic averaging.
-------
A5-8
In the absence of sufficient structural information on the pathway
scores, Che choice of the aggregation method reflects the subjective
judgment on the relative Importance of single- and multiple-pathway
sites. An important class of aggregation rules Is the k-th power
averaging method In which the scote for each pathway Is raised to the
power kt calculating the average across pathways, and then taking the
k-th root. The linear averaging method is k =• 1 and k»2 Is the quadratic
averaging method. Large values of k give less prominence to the pathways,
with lower scores, waking it easier for single-pathway sites to get on
the NPL. Small values of k give more prominence to pathways with lower
scores, making It easier for multi-pathway sites to get on the NPL.
There ace two other important aggregation rules. The worst case
method takes the worst of the pathway scores. The exponential averaging
method exponentiates the pathway scores, takes the average across the
pathways, then takes the log* The two methods should yield very similar
results, giving little prominence to pathways with lower scores and
making it easier for single-pathway sites to get on the NPL. Both methods*
can be viewed as generalizations of the k-th power averaging method with
k approaching Infinity.
How can theHRS makebetter_u3e oftemporal analysis?
Analyzing how long it takes to analyze a site through the various
steps between Identification and remediation yields information which
has implications for parts of the HRS-especlally the toxicity/persistence
score and In considering population dynamics. If, for exaajple (and this
currently seems to be the rule rather than the exception), the site
remains unremediated for a long period of time, then chronic effects
become more important than acute and the HRS should address them.
(See Appendix 2 for a discussion of acute and chronic effects at
unremediated sites.)
By analyzing Information about the source and migration (and also
anything else that would remove wastes from the site, such as decomposition),
it Is possible to estimate the length of time a site nay present a problem.
Such an estimate has inplicatlons for remediation.
Similarly demographic and other relevant changes which may be predicted
over cine could alter the nature and degree of risk posed by a site.
Because such factors alter risk, the revised HRS should not ignore then*
Whatcanbe done about the jualltyof the data used in theHRS?
A. major effort-should be made to improve the overall quality of
analytical data collected at sites. Standardized collection and laboratory
methods currently exist for only a small fraction of substances potentially
-------
• • A5-9
present. Expanded chemical characterization of all media, coupled with a
strong laboratory certification program, will Improve not only the HRS but
all aspects of the Superfund process.
Similarly, acceptable data collection procedures should be established
(or expanded upon) for non-analytical data (such as population estimates,
volume of wastes, etc.) so that the best quality data will be preferred,
Mechanisms which cowld be used to encourage the collection and use of
quality data include: a tiered data quality approach such as the one
OSRE presented to the Large Volume Waste Work Group, data collection "trees",
and the weighting of final HIS scores to account for the quality of data
used In generating It.
What can bedone when there isn't "enough data" for the HRS?
It is necessary to have a method of default (also termed missing
value replacement values) when data are limited or nonexistent for
chemicals. Lacking complete information on chemical inventory and
concentration at a site nay well be the rule rather than the exception.
Therefore, a standardized approach to the circumstance will reduce the
inconsistencies that are inevitable if such decisions are left to field
staff or if only the "better" data are somehow selected and used.
*!
Several approaches can be used for assigning a value to a scoring
component when data are not available. One way is to assign a zero. This
would be a nistake. However, because it underestimates risk. (However,
on a logarithmic scale, assigning a zero may, In fact, overestimate
risk. If the*risk should be very small, the logarithmic score should be
appropriately negative—even minus infinity, If the risk is zero
arithmetically.)
From a scientific perspective, a better way is to use auxiliary
information on the site to help impute the missing value.
One concern with this approach is whether the fact that the data
are missing might be an indication that the site might be
different from otherwise comparable sites,
The definition of comparable sites is crucial and may depend on the
scoring component. For quantity and concentration, gross categories of
sites could be used (as a minimum)* Stratification by region Is also
possible, leading to a definition of comparable sites such as those of
the same category in the same geographic region. When possible, comparable
site's should be defined to allow the use of the greatest amount of data*
There are two general statistical approaches for defining comparability
in this situation. First, one can develop a metric to measure how comparable
two sites are in terms of the auxllliary charaeteristics, then choose a
-------
A5-IO
prespecifled number of sices closest to the specific site in terms of the
given metric. Second, one can use an empirical model to describe how the
scoring component depends on the auxilllary characteristics, then use the
prediction from the model to impute the missing value for the specific
site.
Incorporation of a large number of auxilliary characteristics is
not a trivial task, it might not be worth the effort to go beyond the
use of broad site categories and regions. The decision depends on whether
other characteristics could be powerful predictors of the scoring component
being considered.
In some situations, limited data about the site will exist. Averaging
those data Is an easy measure of central tendency likely to represent
the site better than any of the individual numbers. For example, if a
few concentration measurements are available, even if variable and of
uncertain quality, their average can be taken to show relative quantity
in the absence of better information* Alternatively, some sense of
proportion can be gleaned from the types of wastes known to have been
deposited at a site. For example, relatively low concentrations (compared
with average waste materials) of all hazardous chemicals are expected in
fly ash, but metals and products of 'incomplete combustion (dioxins and
furans) would be more likely present than volatile organics.
Another approach may be used when some etaea is available. It Is
known to statisticians as the shrinkage formula and to the actuarial
literature as the credibility method. This approach uses data both from
the site and from sites thought to be comparable. The Subcommittee is
not in a position to evaluate how difficult the field staff would find
Implementation of this approach,
What should be^dqne _when _data_ are available that_e_xceed_ what Is needed^
for the HRS ?
One approach is the assignment of discretionary points (which the
Subcommittee haa sometimes called the "Bump Factor"). The Subcommittee
discussed several scenarios where such discretionary points could be
usefully assigned such as in the case of a site where;
1. Many chemicals were identified (only some of which were data
available for evaluating toxictty but where the presence of
other implied additional risk).
2. The presence of the toxic materials in the surrounding target
population had been demonstrated.
3, Acute exposures to air emissions were possible from catastrophic
release or other changes in site conditions.
-------
A5-U
4. One or more contaminants were known to be extremely toxic to
indigenous natural populations.
5. Human effects had been confirmed and those affects could be
reasonably linked to exposure to particular chemicals from
the site,
6. Nuisance levels of chemicals in the air around the site were
identified of and their presence confirmed (best analytically,
but possibly by smell).
This approach could also be used to take better account of non-human
receptors. The idea is to assign extra points which are proportional
to the extra risk the situation presents.
Caveat
If the algorithm is changed, then the 28.5 cut-off score should be
re-evaluated. The threshold Sm» which determines whether a site is
eligible for the NPL, needs to be adjusted when either the algorithm is
changed or when more routes are incorporated because the saae score
will not necessarily have the saae meaning. For routes already used in
the HRS, OSRR should investigate from existing data what alternative
threshold should be used in order to yield a pre-speeified proportion of
sites on the NPL, if the algorithm is changed.
-------
-------
APPENDIX 6
RECOMMENDATIONS TO EVALUATE AND IMPROVE THE HRS
BY THE
HAZARD RANKING SYSTEM REVIEW SUBCOMMITTEE
OF THE
SCIENCE ADflSORY BOARD
-------
A6-1
RECOMMENDATIONSTO EVALUATE AND JMPROVE THE HRS
How Good I8the HRS?
The HRS is appropriate for Che purpose for which it Is Intended.
As a scoring systems for priority setting, it is better than most. The
factors scored (with possibly one or two exceptions) are related to risk,
and higher scores for them reflect higher risks, Thus» the current HRS
is plausible. The recommendations concerning the algorithm discussed in
Appendix 5 should greatly improve it.
HOW CAN EPA LEARN FROM EXPERIENCE TO BETTER PREPARE FOR THE NEXT REVISIONS?
OPPE Study. In considering the HRS as a predictor for a comprehensive
risk assessment, It is natural to clarify how well it performed in the
past. The unpublished OPPE Site Ranking Panel study (18, 23) did not
address this question but focused entirely on subjective evaluation.
For parts of the HSS (route characteristics and containment), additional
data are available. It should be possible to conduct an empirical
evaluation using RI/FS and/or other data to evaluate the HRS score both
for sites on the NPL and for those that did not score high enough for
the NPL.
EmpiricalEvaluation. The HRS performance should be judged by an
empirical retrospective evaluation of how successfully the HRS predicts
risk or by ho« successfully its components predict phenomena (such as
release) which contribute to risk. This evaluation should be based on
an in-depth technical review. Whatever the definition of risk, the HRS
should be judged on how well it approximates that definition, not on how
well it matches some subjective notion of the relative Importance of the
sites in a test set,
With respect to empirical evaluations» the route characteristics
and containment scores are estimates for the probablity of release when
actual release is not detected. Two distinct probabilities need to be
considered: (a) the probability that current releases are not detected,
and (b) the probability that such releases will occur in the future*
For the latter, the time horizon also needs to be considered. For the
sites with sufficient RI/FS and/or data from studies by the states or
other pattiaa, it should be possible to search for the best way to
predict release from the components of route characteristics and
containment scores*
-------
A6-2
Coppariaoa of Dataand Scores for the Same Site, Given the screening
nature ofto* HRS, as a screening tool, it does not appear cost-effective
to attempt to establish confidence Intervals rigorously. However, some
seflnleicm of uncertainty would be useful for Interpreting the scores*
If re-evaluations are available for some available for some sites, it
would be useful to compare the data obtained from the separate
inspections of the site* State data, frequently used in HRS scoring,
might also be used for this purpose.
what Kinds of Data Shouldthe AgencyCollect and What Kindsof Studies
Should, It Conduct to Support Future HRS Revisions?
There are two goals for such studies. First, the Agency should
improve the algorithm to be defended as risk-related and, second, the
studies should develop at least a crude way to estimate the quantity
of specific chemicals present. The planning of any evaluative studies
should be carefully peer-reviewed before being implemented.
As mentioned in the Other Comments portion of Appendix 3» EPA also
should develop studies to review large volume waste sites ranked under
the model and to examine basic parameters of the model, A study in the
first category suggested by Mr. R« Walline of EPA Region VItt could
address the actual reasons why release occurred at problem mining sites
(reasons reportedly are different than those initially assumed) and a
study in the second category might address the characteristtc(s) of
mining waste that control release (mobility) to' ground water*
-------
-------
APPENDIX 7
REQUESTS
FROM THE
OFFICE OF EMERGENCY AND REMEDIAL RESPONSE (SUPERFUND)
TO THE
SCIENCE ADVISORY BOARD
-------
A7-l
r
'* UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
r WASHINGTON, D,C. 204SQ
DEC
Of
SQi-lO V"»AST€ A^O
MEMORANDUM
SUBJECT; Science Advisory Board Review
of Hazard Ranking System
FROM; Henry L. Longest II, Director^
Office of Emergency and Remedijf|f*RerSiIonse
TO: Dr. Terry Yosie, Director
Science Advisory Board
The Office of Emergency and Remedial Response (OERR) is
conducting a review and possible revision of the Hazard Ranking J
System (HRS). OERR is requesting that the Science Advisory Boars
review several issues in conjunction with this effort.
BACKGROUND
The HRS is the principal mechanism used by EPA to determine
whether to place sites on the National Priorities List (NPL),
promulgated under Section 105 of the Comprehensive Environmental
Response, Compensation and Liability Act of 1980 (CERCLA). The HF
was promulgated on July 16, 1982 (47 FR 31219), as Appendix A of
the National Contingency Plan (NCP). The current HRS evaluates
the relative potential of uncontrolled hazardous substances to
cause human health or safety problems, or ecological or environ-
mental damage, by taking into account "pathways* to human or
environmental exposure in terms of numerical scores. Those sites
that score 28.50 or greater on the HRS, and which are otherwise
eligible, have been placed on the NPL. ^ site must be on the NPL
for it to be eligible for remedial action financed by the CERCLA
trust fund*
On October It, 1986, CSRCLA was amended. The Superfund
Amendments- and Reauthorization Act of 1986 (SARA) requires EPA to
promulgate changes to the HRS 18 months after enactment. The
amendments require EPA to modify the HRS so that, "to the maximum
extent feasible,_ it accurately assesses the relative degree of
risk to human health and the environment posed by sites and
facilities subject to review.* Specifically, the amendments
require:
-------
-2-
* An 'assessment of the human health risks associated with
contamination or potential contamination of surface waters,
either directly or as a result of the run-off of any hazardous
substance, pollutant or contaminant. This assessment
should take into account the use of these waters for
recreation and the potential migration of any hazardous
substance, pollutant or contaminant through surface water
to downstream sources of drinking water
* An evaluation of the damage to natural resources which may
affect the human food chain and which is associated with
any release or threatened release
9 An assessment of the contamination or potential contamination
of the ambient air which is associated with a release or
threatened release
Section 125 of. SARA requires EPA in its revision of th« HIS,
to specifically assess those wastes described in section
3001(b)(3)(A)(1) of the Solid waste Disposal Act. These wastes k
include fly ash waste, bottom ash waste, slag waste and flue gaa .f
emission control waste qefiecated primarily from the combustion of*
coal or other fossil fuels. The amendments require EPA to consider:
(1) .The quantity, toxicitv, and concentrations of
hazardous constituents which are present in such
waste and a comparison with other wastes?
(25 The extent of, and potential for, release of such
hazardous constituents into the environment; and
(3) The degree oC risk to human health and the environment
posed by such constituents.
Additionally, section 118 of the SARA states that SPA shall
give a high priority to facilities where the release of hazardous
substances or pollutants or contaminants has resulted in the
closing of drinking water wells, or has contaminated a principal
drinking water supply.
in th* amendments to CERCLA, Congress has stated its intention
that the HRS remain a screening tool to enable SPA to list sites
on the NPL as expeditiously as possible, using data from the
Preliminary Assessment (PA) and Site Inspection (SI). The
legislative history of SARA makes clear that Congress did not
intend that the .revised HftS become a mechanism for making detailed
risk assessments? 'rather, it was intended to t>e consistent with
the limited purpose o£ the NPL—screening sites that might, after
further study, warrant Fund-financed remedial action.
In order to i^or-ov* f.he accuracy of the HRS, the Agency is
considering expanding hhe data collection performed before a site
is proposed Cor the *''*'., to provide data to implement a revised
HRS. To maximise !:'<.-* use of these limited resources, EPA must
-------
A7-3
-3-
target its data collection activities to those specific areas
that would moat increase the accuracy of the HRS,
ISSUES FO1 REVIEW
Based on the public comments received during the rulemakings
for the initial NCP (47 PR 31180, July 16, 1982), the original
NPL (48 FR 40653, September B, 1983}» and the subsequent NPL
updates, as well as the intent of Congress and program experience,
QERR has begun a review and revision of the HRS. In conjunction
with this review, QERR has analyzed the HRS and developed technical
issue papers on selected aspects of the ranking scheme.
Specific issues we would like the Science Advisory Board to
review at this time include?
" A review of the scientific factors in the
existing HRS to recommend how these might be
modified to enhance its effectiveness.
* An evaluation of the way the HRS evaluates waste ;
characteristics, particularly as it relates to 1
mining waste sites* a critique of several reports
on mining waste issues, and a review of a technical
paper exploring the potential use of concentration
data in the HRS*
0 An examination of the toxicity ranking scheme
employed in the existing HRS and OERR's suggestions
for modifying this scheme.
* An analysis of the distance used to determine
the target population potentially affected by
the release of hazardous substances to the air.
QERR will outline each specific issue more fully to better
focus the Science Advisory Board's review in a memo accompanyina
each report.
Rev Jew of JTh« Scientific^Factors •in...the'HRS
We would like the Science Advisory Board to review the
current HRS and those specific factors that make up each scoring
pathway. As part of this review, we are requesting that the
Science Advisory Board rank the existing HRS factors in order Q£
their importance in accurately assessing risk from a release of
hazardous substances. We would also like the Science Advisory
Board to evaluate the appropriateness of the weighting factors
currently used in the HRS and recommend how these might be
modified to enhance the effectiveness of the HRS. The intent o£
this is'to help OOP focus our HRS revisions effort, as well as
our expanded data -
-------
A" -4
In tfi« existing HRS, the score for a facility is based on
the potential for harm to humans or the environment €rom migra-
tion of a hazardous substance from a facility by routes involving
ground water, surface water or air. It is a composite of separate
scores for each of the three routes. The score for each route
is obtained by considering a set of factors that characterize the
potential of the facility to cause harm. Each factor is assigned
a numerical value (on a scale of 0 to 3, 5 or 8) according to pre-
scribed, guidelines. This value is then multiplied by a weighting
factor yielding the factor score,
Waste Character 1st ics/Hinin_q Waste __I agues
In scoring a site using the current HRSf EPA considers the
quantity of hazardous waste deposited, rather than the quantity
of hazardous constituents within these wastes. EPA also does not
consider the quantity of hazardous constituents released into the
ground water , surface water, or airf but only whether that release
is significantly above background* When EPA developed tii« HRS,
the Agency believed that determining the quantity of hazardous
constituents would require a significant amount of sampling and
analysi-s that would result in substantial delays in the ranking
of sites.
In its various rulemakings on the NPL, EPA has received
public comments on this aspect of the HHS and how it is used to
evaluate mining waste sites* The commenters have stated that the
HRS is biased against high-volume, low-toxicity wastes, such as
mining wastes, because it does not take into account quantity,
toxicity an'd concentration of the hazardous constituents, and
that EPA is unable to provide evidence that the HIS is a rational
basis by which to rank mining sites for inclusion on the NPL.
f
QERR would like the Science Advisory loard to evaluate the
HRS in light of these criticisms, and to review both a technical
paper on the HRS developed by TRC, Inc. for the mining industry/
and a critique. o€ the TRC minim report conducted by QERR's con-
tractors. As part of this review, we would like the Science
Advisory Board to examine the validity of the mining industry's
concerns and to recommend areas for revision of the HRS, if
warranted, to address these concerns, keeping in mind the use
of the HRS aft a screening tool for further studies.
In addition, w« would like the Science Advisory Board to
review a technical n-tiier developed by OfiRR that explores the
use' of concentration ;1ata in evaluating hazardous waste sites.
This review should focus on the feasibility ?f including waste
concentrations in the HRS, the data requirements and resource
implications of implement '.n-3 such a change, as well as the
reliability of sampling >Ut* to accurately determine the
quantity -of hazardon^ ;:onstituents contained in wastes.
-------
-5-
Air_Target DIst a n ce_ Limit
Currentlyr the air pathway of the HRS relies on analytical
data that shows levels of a contaminant at or in the vicinity of
a facility that significantly exceed the background levels. The
air pathway provides a score based on the toxicity of the substance
released via the air route, and the population within a four-mile
radius of the site. This population is used as an indicator of
the population which may be harmed should a hazardous substance
be released to the air.
QERR has analysed the distance used to determine the target
population potentially affected by a release of hazardous sub-
stances to the air and prepared a technical report detailing its
findings. The report presents conclusions on the general level
of cancer risk arising from uncontrolled waste sites and examines
the implications of the analysis for the distance factor of the
air pathway. The study used an SPA model called Human Exposure
Model (HEM) to make these estimates.
We would like the Science Advisory Board to evaluate th* i
target distance limit used in the air pathway of the existing |
HRS and review the technical paper developed on this subject. As
part of this review, we would like an assessment of the general
methodology used in the technical paper and the implications of
these findings on the air pathway factor of a revised HRS. In
addition, we would welcome any suggestions concerning general
approaches for revising the air targets distance limit.
Tj3xJ.cJ.ty
Many public comments have been received by EPA on the method
used in the HRS to rank the toxicity of hazardous substances. .>
The current method is based on a rating scheme developed by N.
Irving Sax (1975, 1979, 1984} and rates the toxicity of hazardous
substances on a scale of 0 to 3. The rating is primarily based
on acute toxicity. Several technical issues have been raised by
commenters that suggest the possible need for modification of the
HRS is in order to improve its ability to discriminate among
sites whose wastes have different toxicity characteristics.
As a result of these comments and the intent of Congress to
•take the HHS as accurate as possible, OERR has evaluated the method
currently used in the HRS for estimating the relative toxicity of
substances at hazardous waste sites and the methods employed by
other ranking systems to evaluate toxicity* Based on the review,
OERR has suggested improvements to the HSS toxicity factor that
incorporate measures of acute, sub-chronic and chronic toxicity.
We would like th»» Science Advisory Board to evaluate the
toxicity- ranking scheme employed in the existing HRS, to review
OERR's technical nvu.'r ->n this subject, *nd to suggest improvements
to the HRS that *•->;,' : ^nt* ^c-^urately assess the toxicity of a-
hazardous substaa,/ -.
-------
AT-6
-6-
SCHEDOLg •
The specific technical reports Cor Science Advisory Board
review will b* made available to you 30 days in advance of the
Science Advisory Board meeting. We would be more than happy to
meet with the Science Advisory Board to brief them on the existing
HRS» if appropriate.
Thank you foe your help in this project, 1C you have any
questions concerning OERR's requests, please contact Jane Metealfe
at 382-7393. We look forward to working with you on the HRS
revisions effort.
-------
A7
5 i. < v irt »jiSMo> ,',- _ -*-, -j i i
WASHINGTON, O,C, 20460
JUN 25198T
SOLID WASTE AND EMi rtGEfiG Y
MEMORANDUM
SUBJECT: Science Advisory Review of the HRS
PROM: Henry L. Longest II, Director jf
Office of Emergency and Remedial/^
TO; Dr. Terry Yosie, Director
Science Advisory Board
In December 1986, I requested the Science Advisory Board-
(SAB)'s assistance in reviewing several issues related to
OERR's efforts to review and revise the Hazard Ranking System
(HRS) as directed by Congress in the Superfund Amendments and
Reauthorization Act of 1986 (SARA). In response to this
request, an Ad Hoc HRS Subcommittee was formed by the SAB.
This Subcommittee met for the first time on May If i 20,
1987. By this memorandum I'wish to express OERR's appreciation
to the Subcommittee and SAB staff for their responsiveness to
our request and the expressed willingness to work within the
narrow time constraints for the revisions imposed by SARA. I
would also like to take this opportunity to confirm several
key points which should facilitate the Subcommittee review.
Purposeof the HRS
During the initial meeting on May 19 i 20, many members
of the Subcommittee posed questions concerning the purpose of
the HRS. As pointed out, the purpose of the HRS is a major
determinate of th» structure and function of a model. Such a
model is required to prioritize actual or potential hazards
to public health and the environment posed by hazardous
waste sit**. OERR believes that the purpose of the HRS is
primarily^*- screening tool used to determine which sites
will b« oifMidates for fund-financed remedial response. To
the degr«* possible, the model also needs to provide a
stratification of sites based upon the relative risks posed
to human health and the environment. OERR does not believe
that the HRS is intended to assess absolute risk, since it
is to be applied early in the remedial site evaluation process.
OERR intends to revise the HRS to the maximum extent feasiblei
-------
A7-8
-2-
to estimate more accurately the risks associated with a site,
In conducting these revisions, however, GERR will balance
the benefits of increased accuracy against the associated
costs, avoiding revisions that would require far more extensive
data and, in so doing, impede the ability of the HRS to
function expeditiously.
QERR'sObjective on Seeking SAB Review
Congress, through SARA, requested that in our review of
the HIS, we assess several specific areas. In reviewing these
areas, QERR is particularly concerned that our decisions
concerning the approaches used in revisions be supported by
sound technical and scientific analyses. Peer review of the
technical or scientific issues under consideration will provide
essential support for any proposed revisions.
Specific Issues For SAB Review
OERR requests that the Subcommittee examine three specific
HRS revision issues: 1} the method(s) used to assess toxicity,
2) the air target distance limit and 3} whether large volume
wastes should be considered differently from other wastes*
These issues are described briefly-in the December- 9, 1986
memorandum and will be developed more fully in issue papers
to be presented to the Subcommittee according to the schedule
agreed to at the May 19 & 20, 1987 Subcomittee meeting.
As I understand, the following schedule has been agreed to by
all parties!
1, toxicology Panel — June 29 & 30
2. Meeting 2 of Subcommittee — July IS & 17
3. Air Panel — July 27 i 28 (issue paper to
be mailed July 1}
4. Large Volume Waste Panel — Aug. 19 & 20
(issue paper available July 17}
5. Final Subcommittee Meeting — Sept. 14 § 15 (tentative)
In my December 9, 1986 memorandum, I requested SAB review
of the fourth issue involving the review of scientific factors
in the !1S» The scope of that review wad to prioritize the
various •sjticing factors in terms of their importance to
assessirt*Mri«k and evaluate the appropriateness of the
weightings assigned to those factors.
As explained by Craig Wolfe during the May 19 i 20, 1987
SAB meeting, QPPE has assembled a group of Agency officials
to discuss the importance and weighting of various risk
related factors and rank order 20 hazardous waste sites. I
believe that the work of this expert panel will be most
satisfactory in answering the questions related to the factors
and weightings in the current HRS* Consequently, I believe
it would be duplicative to continue my request to the SAB for
such a review and therefore, I withdraw this part of my
request for review to the SA1*
-------
_ 3 ~
OERR believes that the issues presented to the Subcommittee
meet the 'criteria identified in the Administrator's guidance
concerning use -of the SAB, and can be addressed as discrete
issues. While we intend to provide updates to the Subcommittee
on overall progress with respect to revisions of fi * -*ntire
model, we do not believe that review of the overall model is
either feasible or appropriate Eor the Subcommittee due to
the need Eor policy as well as scientific considerations.
OERR looks forward to continuing the open and focused
review effort which was initiated at our first meeting. If
you have any questions please contact Ms. Penny Hansen, Deputy
Director of the Hazardous Site Evaluation Division at 475-8600.
-------
A7-10
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D,C, 20460
JUL 2 8
SOLiQ WASTE AND
MEMORANDUM
SUBJECT: Science Advisory Board (SAB5 Review
of Mining Waste/Concentration Issue
FROM: Stephen A. Lingle, Director /o
-------
A7-11
-2-
gites such as mining sites* Although the papers discussed
in the first issue pertained to mining waste* OERR believes
that the issue is equally relevant to sites containing other
high volume- wastes such as fly ash.
The 'second issue is related to the first as it addresses
the feasibility of using the concentration of the hazardous
constituents of wastes. In response to public comments that
concentration data are not used in the current HIS to determine
the quantity of hazardous waste, and the requirements of SARA
Section 125(a), EPA developed an option that specifies three-
tiered methodology that allows the use of waste concentration
data. This methodology is described in the concentration
issue paper presented to the SAB.
In revising the HRS, Section I25(a) of SARA requires
EPA to address quantity, toxicity and concentration of hazardous
constituents for RCRA Section 3001{b)(3}(A)(i) special study
wastes (fly ash, bottom ash wastes, and flue gas emission
control waste generated primarily from the combustion of coal
or other fossil fuels). Although, SARA Section 10S(g) does
not require revisions to the HRS, it does require that iPA
consider the same factors (quantity, toxicity, and concentration)
in the decision to propose new mining waste sites to the NPC«
using the current HRS until the HRS is revised. EPA believes .
that the methodology described in the third option of the
concentration issue paper could be employed to satisfy section
125(a) and would also address concerns about scoring other
high volume waste sites such as mining waste sites as well.
Consequently, we believe that the subcommittee findings will
be equally applicable to both mining and fly ash sites.
I hope that this discussion provides the desired
clarification of this issue. In addition, we are in the
process of gathering available information on costs of waste
sampling and analysis at large volume waste sites. This
information,will be provided to the SAB members aa soon as
possible.
OERR appreciates the time and effort the subcommittee
members have expended and we look forward to our next meeting
in Oenvtr. If you have any additional questions, please
contact Agnes Ortiz of my staff, at (202) 475-9700.
Attachments
ccs Kathleen 'W. Conway
SAB Subcommittee Members
-------
APPENDIX 8
REF1RRENCIS CITED
-------
REFERENCES CITED
1. Electric Power Research Institute. Inorganic and Organic Constituents
in Fossil Fuel Wastes. EPRI EA-5176, VI. August 1987,
2. Electric Power Research Institute, Lgaehing_Studiesof Utility Solid
Wastes. EP&I EA-4215. August 1985
3* Electric Power Research Institute, Physleal-Chemical Charget eris tics
of Solid Wastes. EPRI EA-3236. September 1983.
4, Electric Power Research Institute. Round-Robin Evaluation pf Regulacory
Extr_actioix Methods. EPRI EA-4740. December 1986.
5. Electric Power Research Institute. Mobilization and Attenuation
of Trace Elements in Fly Ash. EPRI EA-4747. August 1986.
6, Electric Power Research Institute. Evaluation of Instrument s_in
^saturated Ply Ash. EPRI EA-5011, April 1987.
7. Electric Power Research Institute. gWlS: Deaerigtipn,Status and
Available Results. EPRI EA-5322-SR. August 1987.
8. Environmental Protection Agency. Uncontrolled Hazardous Waste Site
Ranking Syatea; A Userg___Manual (HW-10) , originally published
in the July 16, 1982. Federal Register, 1984.
9, Environmental Protectioa Agency, Office of Emergency and Remedial
Response. Analysis of the Air Target Distance Liait inthe
Hazard Ranking Sya tea. Prepared by IGF Incorporated, July 1987,
10. Environmental Protection Agency, Office of Emergency and Reisedial
Response. Chemicals Coatributiagto EstimatedRisks at Superfund
Si t ea: A Limited Study, July 16, 1987.
11. Environmental Protection Agency. Office of Emergency and Remedial
Response. Comparison, of Available Data on th_e_Concetttratlpn_
o£ Haaardoua Constituentain MiningWagte and Non-Mining Waste.
pages 5-10. Received August 1987.
12* Environnental Protection Agency, Office of Emergency and Remedial
Response. Discussion of Options for Revising the Hazard Ranking
System(HRS) ToxicIty Factor. Prepared by IGF Incorporated,
May 4, 1987.
13. Environmental Protection Agency* Office of Emergency and Remedial
Response. The S u pe_r_f und Haza r d Ranking Sy e t em _( HRS) ;
Applieabillty to Mining Wast_e Sites. Prepared by ICF Incorporated.
' July 22, 1987,
-------
_ 2 -
14. Environmental Protection Agency, Office of Emergency and Remedial
R«spome. Report on the Feasibility of Using Concentration
Pitain a Revised Hazard Ranking Systea. Prepared by ICF
Incorporated, July 27, 1987,
15. Environmental Protection Agency. Office of Policy Planning and
Evaluation, Views/Approach toHazard Ranking System Revisions,
July 1987. ~ ~~^ '
16, Environmental Protection Agency. Office of Solid Waste. Report to
Congresst Wastes from the Combustion, of Coal by Electfie
Utility Power Plants, Chapters l-S. June 1987,
17, Environmental Protection Agency. Office of Solid Waste, Management
of Mining Wastes, June 22, 1987.
18. Industrial Economics, Incorporated. Preliminary Analysis of Alternative
ModelstoSupportRevision to the CIRCLA Hazard Hanking System* -
Prepared for the Environmental Protection Agency's Office of Policy,
Planning and Evaluation, Presented to the Science Advisory Board*
July 16, 1987.
19. Kushner, L. M, Hazard RankingSystag Issue Analysis: Sites with
Unknown Waste Quantity, August 1986,
20. Lingle, S, A*, Director, Hazardous Site Evaluation Division.
Memorandumi Science Advisory Board (SAB) Review of Mining
Waste/Concentration Issues to Terry Yoaie, Director, Science
Advisory Board. July 28, 1987,
21. Longest, H, L*» Director, Office of Emergency and Remedial Response.
Memorandums Science Advisory Board Review of Hazard Ranking ,
System to Terry Yosie, Director, Science Advisory Board.
December 9, 1986.
22* Longest, H. L., Director, Office of Emergency and Remedial Response.
Memorandum! Science Advisory Board Review of the HRS to Terry
tosie, Director, Science Advisory Board, June 25, 198?,
23. Merkhof«r, L* H., et al, A Site-Ranking PaaelrEvaluation of the
Relative Risk Posed by Twenty Superfund Sites* Keystone,
Colorado: The Keystone Center. Prepared for the U.S.
Environmental Protection Agency's Office of Policy, Planning
and evaluation. Draft Received September 1987, ,pp, 1-47.
-------
* 3 -
24. MITRE Corporation. Analysis of EPA Haz_atd_ Ranking System Scoring
of Mining^ Sii-rte_s_ on the National Priorities List _. Prepared for
U.S. EPA, 1985.
25, MITRE Corporation. Estimates of Che Quantity of Hazardous
Subs cances _Present in_Mining_Waste_ at_ NFL Sites * Prepared
for the U.S. Environmental Protection Agency's Office of
Policy, Planning and Evaluation. January 9, 1986.
26. Ortiz, A« , Incorporating Mobility Factors into the Hazard Ranking
.System, Office of Solid Waste amd Emergency and Remedial
Response. August 20, 1987.
27. TRC Environmental Consultants, Analysis of Mining Sites on the
Nat ipnal Priorities List . Prepared for the Aaerlcan Mining
Congress, 1983.
28. TRC Environmental Consultants, Anajlysjta oj Miniog Si tes^ on the
Na_ti_ona_l _Pr i o r i t i es ........ Lia t_t _ Score Changes and Me«ly Propofled
Si^tes. Prepared for American Mining Congress, i9S4a*
29. TRC Environmental Consultants. Review of Mining Sites on the
Nat ional Priori ties List : _^ Cottparigon yith Non-Mlning_ Site8_.
.Prepared for the American Mining Congress, I984b,
30. Sobek, A. A« , et si . Fie 1
------- |