United States
Environmental Protection
Agency
Region III
Office of Superfund
Hazardous Waste Management
Philadelphia, PA 19103
EPA/903/8-91/002
November 1991
Region III
Technical Guidance Manual
Risk Assessment
Exposure Point Concentrations
In Groundwater
EPA Contact: Dr. Debra L. Forman
EPA
Region
Hazardous Waste Management Division
November 1991
The EPA method of risk assessment uses long term or chronic exposure as a basis for determining the excess cancer risk
at a Superfund site. Oftentimes, the risk from exposure to contaminated groundwater is inappropriately calculated from the
single highest confirmed concentration found in a groundwater well. This approach is mathematically and conceptually
indefensible since a single measurement cannot represent the contamination in an entire plume at a Superfund site. Instead,
a sufficient database is required to effectively represent site risk during a lifetime of exposure. The larger database serves
to reduce the uncertainty inherent in risk analysis, and the Remedial Project Manager is provided with a more scientifically
sound risk evaluation on which to trigger a remedial decision. While this approach applies to most Superfund sites, factors
such as calculation method, well placement and use of the historical database attain particular importance at sites where
groundwater contamination is not clearly established. This guidance is intended to improve the quality and consistency of
deriving exposure point concentrations in groundwater in risk assessments performed in Region III. (EPA/903/8-91/002)
COMMUNICATION
In accordance with our longstanding policy of involving
scientists at the early stages of the RI/FS process, this
Guidance document stresses communication. Clear lines
of contact both between the technical support staff and
the risk manager as well as among the technical
personnel are essential to the process. The Guidance
outlines a sampling strategy, including both spatial and
temporal collection and handling of groundwater data. This
strategy promotes a coherent technical approach to the
RI/FS process, initiating the proper experimental design
and correct data usage. Hence, the risk manager is
provided with a justifiable risk conclusion based on sound
scientific methodology.
The risk associated with groundwater usage at a site is
generally calculated by combining the pollutants'
concentrations in the aquifer of concern along with site-
specific exposure parameters. This result is then
combined with chemical specific exposure factors to
obtain the final risk value. The approach assumes that the
pollutants' concentration is linearly related to risk, thus,
changes in concentration may have a significant influence
on the risk analysis for the site. A clear understanding of
this relationship and its potential impact on the final risk
value underscores the requirement for a conceptually
correct derivation of the exposure point concentration.
-------
WELL PLACEMENT
During the scoping meeting, the toxicologist may present
the guidelines for risk analysis from contamination in
groundwater. These may include selecting the location of
groundwater wells and proposing analytical methods of
sampling for suspected contaminants. The choice of
groundwater wells is of prime importance in determining
the appropriate concentrations of pollutants in the aquifer
of concern. Placement of wells in both the horizontal and
vertical planes should be considered. In general, both
horizontal and vertical placement of groundwater
monitoring wells should be designed so that monitoring
well data can be extrapolated to future residential well
usage. Consultation with the hydrogeologist is required to
outline any hydrological and/or geological concerns which
may impact the subsequent well selection.
S))))))))))))))))))))))))))))))))))))))))))
Both horizontal and vertical placement of groundwater
monitoring wells should be designed so that monitoring
well data can be extrapolated to future residential well
usage.
A. Horizontal Well Placement
Hydrogeologists may locate wells for a variety of
purposes, yet toxicologists primarily utilize water quality
data to assess the potential risks to human health. Since
toxicologists usually do not direct well placement, the
body of data obtained for hydrogeological objectives may
be used by the toxicologist for a different purpose.
For example, groundwater wells may be located by the
hydrogeologist purposely to identify the fringe of
contamination. On the other hand, the toxicologist
requires information concerning the reasonable maximum
concentration of pollutants in the aquifer of concern. In
this case, the ideal placement of wells for risk purposes
is near the apparent center of the plume. The choice of
wells may be different for on site and off site scenarios or
if multiple sources are present.
B. Vertical Well Placement
The aquifer of interest should provide sufficient water for
residential use. In some cases, monitoring well data from
two independent aquifers may be combined if each aquifer
cannot supply enough water individually. If the aquifer is
not currently used as a drinking water source, consider
the likelihood of its future use as a drinking water source.
For example, monitoring well data from a perched aquifer
is not appropriate for risk assessment because it usually
does not provide sufficient water for residential use. In
any case, the appropriateness of spatial placement may
depend on hydrogeological factors. Thus, consultation
with a hydrogeologist is required to outline potential
problems.
Identification of wells should be such that the toxicologist
may combine water quality data from several wells in
order to achieve a reasonable maximum estimate of
groundwater contamination. Those wells which meet the
criteria discussed above may be grouped for spatial
analysis. Temporal analysis may be achieved by multiple
sampling of the chosen wells.
It is important to recognize that the combined data from
multiple well sampling should belong to the same
statistical data population data, i.e. the apparent center of
the plume.
C. Well Construction
Once the well locations have been determined, the
hydrogeologist should be consulted to determine the
adequacy of well construction. The problems identified
with well construction may also influence the choice of
data to be used by the risk assessor.
Although both filtered and unfiltered data should be
collected (USEPA, 1990b), the data is evaluated on a well
by well basis by the risk assessor for its potential use in
extrapolating monitoring well data to a residential well
scenario. Generally, unfiltered data is preferred, however,
if there is an obvious discrepancy in the levels of
inorganics, or if secondary MCLs are exceeded, filtered
data may be selected for use in the risk assessment.
This issue is addressed more fully in a separate Region
III guidance document which is currently in draft form
(USEPA, 1991b).
S))))))))))))))))))))))))))))))))))))))))))
The appropriateness of spatial placement in both
horizontal and vertical planes may depend on
hydrogeological factors.
-------
RISK ASSESSMENT
HISTORICAL DATABASE
During the scoping phase, the complete historical
database should be thoroughly examined. If the historical
data demonstrate clear trends, the toxicologist should
incorporate relevant site-specific information into the risk
calculation. Site-specific information should also be
considered in determining the confidence assigned to the
trend direction. In addition, the historical database should
be evaluated for landmark actions, such as emergency
removal or remedial action prior to the RI/FS. Use of the
historical database should include consideration of
potential inconsistencies in analytical methods, data
validation protocols and QA/QC practices which may
have changed with time (USEPA, 1990c).
If the available information is inadequate to substantiate
the risk assessment, additional sampling events should
be performed for each well identified for risk assessment
purposes. The sampling events should be spaced such
that an independent sample population is obtained. The
selected time interval should be acceptable to all
members of the investigation team.
As data is collected, the results should be reviewed for
trends, the number of sampling rounds should be
sufficient to yield a database with clear trends. The
sampling effort may be a continual process, such that the
RI/FS process is not delayed. In this respect, information
obtained from ongoing sampling efforts may be submitted
as addendums to the Remedial Investigation report.
DATA QUALITY OBJECTIVE
A high data quality objective is recommended. Depending
on site conditions, analysis of samples using SAS
procedures may be warranted. For example, EPA method
500 series for drinking water, which have lower detection
limits for some contaminants, can provide greater
sensitivity for assessing contaminant concentrations.
Thus, a clearer evaluation of the relevance of
contaminants detected at concentrations below the
detection limit may be attained. In some cases, this
approach may eliminate the need to apply the "0.5 times
the detection limit" rule (USEPA, 1989). In addition, and
if logistics permit, provisions should be made for
reanalysis of rejected or estimated samples within their
holding times.
A. Current Scenarios
The current, on site risk should be based on the most
reliable database obtained during the entire site
investigation which may include studies other than the
RI/FS. The data to be included in the calculation consists
of useable, water quality data obtained from repeated
sampling of the wells identified for risk assessment
purposes as well as useable historical information.
Treatment of non-detects is considered in a separate
Region III guidance document (USEPA, 1991 a). The
reasonable maximum concentration of pollutants in the
aquifer can be calculated as the upper 95th percent
confidence limit of the arithmetic mean, UCL95 (See
Highlights). If the database is sufficient, a preliminary
conservative risk assessment may be performed following
the Phase I investigation. Current off site risk may be
assessed using water quality data from a set of wells
independent of those identified for on site risk (possibly
residential wells).
B. Future Scenarios
Future risk may be estimated using the results of a fate
and transport groundwater modelling effort. Consultation
with the hydrogeologist is recommended to determine the
appropriate modelling approach. If the hydrogeologist
determines that groundwater modelling is not appropriate
due to site specific conditions, current monitoring well
data may be used to assess future risk.
-------
HIGHLIGHT #1:
LOGNORMAL DISTRIBUTION
The following calculations are used to
determine the UCL-95 for the useable
groundwater dataset.
1. Identify the frequency distribution of the
sample population. A lognormal distribution
can be characterized as having no zero values
and the relative percentage of data points
greater than or less than the mean is not equal.
The W test by Shapiro and Wilk may be used
to test the distribution type (Gilbert, 1987).
According to Dean, 1981, most environmental
datasets are skewed lognormally and the data
can be assumed to be lognormally distributed.
Note that this assumption is supported only by
a large dataset and may not necessarily apply
to small datasets available at Superfund sites.
2. If the sample frequency distribution is
lognormal, transform the detected data to
logarithmic equivalents using the expression
t = ln(x)
where:
where:
• g = arithmetic mean of log transformed data
•2 = variance of log transformed data
H = Tabular H statistic, depends on geometric
•, n, and selected degree of probability
(Gilbert, 1987).
n = sample size
5. If the UCL-95 is greater than the maximum
value, use the W test to examine the sample
population for normality.
HIGHLIGHT #2:
NORMAL DISTRIBUTION
The following calculations are used to
determine the UCL-95 for the useable
groundwater dataset.
1. Identify the frequency distribution of the
sample population as outlined in highlight #1.
2. Calculate the UCL-95 using the following
expression
UCL-95 = x, +1(' /n0-5)
where
x = raw groundwater data
t = transformed data
3. Obtain an estimate of the arithmetic mean of
the transformed data, if desired, using the
expression
= exp(»
• 2
/2)
4. Obtain the UCL-95 using the expression
UCL-95 = exp(' g + • 2/2 + • H / (n-lf5)
(Land, 1971, 1975)
x,, = arithmetic mean of the raw data
• = arithmetic standard deviation of the raw
data
t = Tabular t statistic, depends on degree of
freedom (df = n-1) and selected degree of
probability (one tailed @ p<0.05).
n = sample size
3. If it is determined that the sample population
is neither lognormally distributed nor normally
distributed, omit the non-detect data and obtain
a maximum likelihood estimate of the detect
data (Gilbert, 1987).
-------
References
Dean, R.B. (1981). Use of Log-Normal Statistics in
Environmental Monitoring, in Chemistry in Water Reuse.
Vol. 1. (ed.) Cooper, W.J., Ann Arbor Science, pp. 245-
258.
Gilbert, R.O. (1987). Statistical Methods for
Environmental Pollution Monitoring. Van Nostrand
Rheinhold Co., New York, pp. 152-185.
Land, C.E. (1971). Confidence intervals for linear
functions of the normal mean and variance. Annals of
Mathematical Statistics 42: 1187-1205.
Land, C.E. (1975). Tables of confidence limits for linear
functions of the normal mean and variance, in Selected
Tables in Mathematical Statistics. Vol. 3. American
Mathematical Society, Providence, R.I., pp. 385-419.
USEPA (1989). Risk Assessment Guidance for
Superfund. Volume I, Human Health Evaluation Manual
(Part A). EPA/501/1-89/002.
USEPA (1990a). Guidance for Data Useability in Risk
Assessment. EPA/540/G-90/008.
USEPA (1990b). Field Filtration Policy for Monitoring Well
Groundwater Samples Requiring Metals Analysis, Region
III QA Directive, Bulletin #QAD009, USEPA, Region III,
Philadelphia, PA.
USEPA (1990c). Guidance for Data Useability in Risk
Assessment, EPA/540/G-90/008.
USEPA (1991 a). Chemical Concentration Near the
Detection Limit, Region III Technical Guidance Manual,
EPA-3/HWMD/11-91/001.
USEPA (1991b). Useability of Filtered vs. Unfiltered
Metals data for Risk Assessment, Region III Technical
Guidance Manual, (Draft document).
For additional information, (215) 597-6626.
Approved by:
Thomas C. Voltaggio, Division Director
------- |