United States Office of Water EPA-823-F-06-013
Environmental Protection Agency 4305T August 2006
yf, jj Water Quality Standards for Coastal Recreation Waters:
Using Single Sample Maximum Values in State Water Quality Standards
This fact sheet addresses common questions regarding the appropriate use of the single sample
maximum (SSM) values component of EPA 's 1986 bacteria criteria in coastal recreation waters.
States monitor their Great Lakes and marine coastal recreation waters for the presence of fecal
contamination, specifically for the indicator organisms E. coli and enterococci. States typically
collect water samples and analyze them for bacteria content at regular intervals. States then
evaluate these bacteria levels against the applicable water quality standards.
What are coastal recreation waters?
Coastal recreation waters are those Great Lakes and marine waters (including coastal estuaries)
that are designated under section 303(c) of the Clean Water Act for use for swimming, bathing,
surfing, or similar water contact activities. Inland waters or waters upstream from the mouth of a
river or stream having an unimpaired natural connection with the open sea are not considered
coastal recreation waters. (See CWA Section 502(21) and 40 CFR 131.41(b)).
What is EPA's current position on using the SSM in BEACH Act Waters?
EPA's current position is the same as it was in the November 2004 Water Quality Standards for
Coastal and Great Lakes Recreation Waters rule (69 FR 67217, November 16, 2004), commonly
referred to as the BEACH Act rule. In the preamble to the BEACH Act rule, EPA clarified its
expectations with regard to how the single sample maximum (SSM) values could be used in the
context of beach notification and closure decisions, and in the context of the implementation of
other Clean Water Act programs. EPA indicated that it expected that the single sample
maximum values would be used for making beach notification and closure decisions. EPA
recognized, however, that States and Territories also use criteria in their water quality standards
for other purposes under the Clean Water Act in order to protect and improve water quality.
Other than in the beach notification and closure decision context, the geometric mean is the more
relevant value for ensuring that appropriate actions are taken to protect and improve water
quality. The geometric mean is generally more relevant because it is usually a more reliable
measure of long term water quality, being less subject to random variation, and more directly
linked to the underlying studies upon which the 1986 bacteria criteria were based. States,
however, retain the discretion to determine whether and how to use the SSM in other Clean
Water Act programs.
Why did EPA choose to address the SSM in the 2004 BEACH Act rule?
The 1986 EPA Ambient Water Quality Criteria for Bacteria document clearly identifies the SSM
values and the geometric mean values as part of the criteria. Therefore, consistent with section
303(i)(2)(A) of the Clean Water Act, EPA included both components of the criteria in the
BEACH Act rule. However, the SSM discussion in the 1986 bacteria criteria document refers
1
-------
only to beach monitoring, and does not discuss how or whether the SSM should be implemented
for other Clean Water Act applications, such as establishing Total Maximum Daily Loads or
National Pollutant Discharge Elimination System permit limitations.
EPA continues to believe that the SSM values in the criteria are very useful for making beach
notification and closure decisions. Beach managers frequently need to make beach notification
and closure decisions based on as little as one single grab sample. Of the 2,823 beaches reporting
information to EPA in 2002, 65% reported that pathogen levels were monitored at least once per
week (EPA's Beach Watch Program: 2002 Swimming Season, EPA 823-F-03-007, May 2003,
http://www.epa.gov/waterscience/beaches/beachwatch2003-newformat.pdf). This means that at
35% of the beaches, the beach managers had fewer than four samples each month for making
decisions to open or close the beach. Indeed, in numerous cases, managers had only one sample
in any given 30-day period. Thus, having a trigger level against which just one lone sample value
can be compared is quite helpful. Furthermore, beach management programs need to be able to
respond rapidly to short-term changes in water quality.
This does not mean the single sample maximum has no value outside of the beach monitoring
and notification context. Using an SSM is especially important for beaches and other recreational
waters that are infrequently monitored or prone to short-term spikes in bacteria concentrations,
e.g., waters that may be affected by combined sewer overflow outfalls. The BEACH Act rule
preamble recognized that States and Territories, having identified that a water is prone to short-
term spikes in bacteria concentrations due to pollution episodes, have significant flexibility in
how they address those episodes in permitting and assessments consistent with the Clean Water
Act and implementing regulations. For States and Territories that are subject to the rule, they
have the flexibility to determine how they choose to apply the SSM outside the beach monitoring
and notification context. States that have already adopted their own water quality standards as
protective of human health as EPA's 1986 bacteria criteria should consider in their water quality
assessment methodology the applicability of the SSM to assessment of waters where fewer than
five samples are available during the time over which the samples would normally be averaged if
the states specify a minimum number of samples for calculating a geometric mean.
As noted above, the SSM may, but need not, also play a role in implementing other Clean Water
Act programs. Except in the beach notification and closure context, EPA expects that States will
determine whether and how to use the SSM criteria in the context of their other programs
implementing the Clean Water Act.
Was the SSM developed to serve as an acute criterion to protect swimmers against short
term exposures?
-------
No. The SSM values in the 1986 EPA Ambient Water Quality Criteria for Bacteria were not
developed as acute criteria; rather, they were developed as statistical constructs to allow decision
makers to make informed decisions to open or close beaches based on small data sets. This does
not mean the SSM values serve no purpose outside of beach notification decisions. For example,
they may give States and Territories the ability to make waterbody assessments where they have
limited data for a waterbody. However, the SSMs were not designed to provide any more
protection of health than provided by the geometric mean criterion.
In developing the 1986 EPA Ambient Water Quality Criteria for Bacteria, EPA derived the SSM
as upper percentiles of the frequency distributions around the geometric mean. The 1986
bacteria criteria document recognizes that there will be instances where the concentration of
bacteria in one or more individual samples will be higher than the acceptable geometric mean
concentration. This is to be expected when dealing with water quality criteria expressed as
average concentrations over a period of time. For example, in a waterbody with a 30-day
average concentration exactly at the water quality criterion, it can be expected that approximately
half the time the waterbody concentration will be above the criteria level. Hence, half of the
samples collected will have a concentration above the criterion concentration (e.g., 126/100 ml
for E. coif). Thus, that the value of any one individual sample is greater than the numerical value
of the geometric mean criterion, or even the numerical value of the SSM, does not necessarily
indicate that the geometric mean criterion has actually been exceeded. Likewise, the fact that one
sample out of a set of samples has a concentration lower than the criteria level does not
necessarily mean that the actual average concentration is lower than the criteria level.
Can a beach meet the geometric mean without always meeting the SSM?
Yes. Based on the derivation of the SSM as percentiles of a distribution of bacterial
concentrations around the 30-day geometric mean, using the SSM as values not to be surpassed
at any time for all Clean Water Act applications could impart a level of protection much more
stringent than intended by the 1986 bacteria criteria document. For example, in marine waters
the geometric mean criterion for enterococci is 35/100 ml, and the SSM is 104/100 ml at
designated bathing beach waters based on the 75th percentile of the distribution of individual
values around the mean. If that SSM were used as a value-not-to-be-surpassed, it would become
a maximum value and all other values in the statistical distribution of individual measurements
would have to be less than the maximum. EPA typically uses the 99th percentile of a distribution
to derive regulatory maximums. Assuming a waterbody had the same standard deviation in
concentration of bacteria employed in deriving the SSM (e.g., 0.7 for marine waters), then the
waterbody geometric mean needed to keep the waterbody concentration below 104/100 ml 99%
of the time would need to be 2/100 ml. This would be far more stringent than the level of
protection provided by the actual geometric mean criterion for enterococci of 35/100 ml.
Treating the SSM as equivalent to acute criteria (i.e., with a specified duration of exposure of
just one second) for purposes of complying with Section 303(d) of the Clean Water Act would
result in a large number of waters being listed as impaired even though the waters may not have
exceeded the applicable geometric mean criteria. Therefore, EPA intends that States and
Territories covered by the BEACH Act rule retain the discretion to use SSM values as they deem
-------
appropriate in the context of Clean Water Act implementation programs other than beach
notification and closure, consistent with the Clean Water Act and its implementing regulations.
Can states establish more stringent water quality criteria than that recommended by EPA?
Of course, states are always free to establish water quality criteria more protective of human
health and aquatic ecosystems than those required or recommended by EPA. Hence, though
EPA is pointing out the effects of using the SSMs in the ways described in the answer to the
previous question - ways the Agency had not envisioned when it published the 1986 bacteria
criteria document - states, territories, and authorized tribes retain the discretion to be more
stringent and use an SSM in such a fashion.
What additional flexibility do states have with regard to using the SSM?
Although EPA promulgated default SSM values based on the 75, 82, 90, and 95% confidence
levels, the BEACH Act rule also includes the equation that can be used to calculate site-specific
SSM values. EPA calculated the values for the SSM included in the rule using the standard
deviations observed during the EPA epidemiological studies. The Agency recognizes that the log
standard deviations observed in EPA's epidemiological studies may not coincide with that for a
particular waterbody. EPA encourages states to collect data to calculate site-specific standard
deviations. To compute a site-specific log standard deviation in a statistically meaningful way,
the rule requires that States and Territories collect at least 30 bacterial samples in a single
recreation season (see 40 CFR 131.41(c)(3)). EPA recognizes the difficulty in collecting the
required number of samples over a single recreation season, but the Agency nonetheless
concluded that collecting this much data during a single season is necessary in order to capture
the variability inherent in bacteria concentrations at a site over the period of a single season
without introducing additional variability from extreme weather conditions such as drought or El
Nino conditions. Using 30 samples over more than one recreation season could affect the
outcome of the SSM such that it may not be as protective of human health as EPA's 1986
bacteria criteria. If this requirement is met, the state may use the resulting site-specific standard
deviation to calculate a corresponding SSM or set of SSMs.
EPA considers that the calculation of site-specific SSM values, as specified in 40 CFR
131.41(c)(3), provides enough detail on the calculation that states included in the BEACH Act
rule can implement this provision of the rule without needing to adopt it as a site-specific water
quality criterion. As a result, states included in the BEACH Act rule do not need EPA review and
approval under 40 CFR Part 131 in their application of 40 CFR 131.41(c)(3).
Should states included in the BEACH Act rule use the SSM in implementing all of their
Clean Water Act programs, such as the TMDL and NPDES permitting programs?
The BEACH Act rule was not intended to constrain the states included in the rule in how they
could use single sample maximum values in the context of Clean Water Act implementation
programs such as the Total Maximum Daily Load program and the National Pollutant Discharge
Elimination System (NPDES) program.
-------
Normally, states use all applicable water quality criteria included in their water quality standards
for all purposes specified under the Clean Water Act and implementing EPA regulations,
including water quality assessments, establishment of TMDLs, and setting of water quality-based
effluent limits (WQBELs) under the NPDES program. However, the BEACH Act rule presented
a unique situation in that it promulgated both the geometric mean component - without
establishing a minimum number of samples - and an SSM component. This leads to some
special considerations in CWA section 303(d) assessments and NPDES permits.
CWA 303(d) listing
In making CWA 303(d) listing decisions, the geometric mean is generally more relevant than the
SSM because it is usually a more reliable measure of long term water quality, as discussed
above. However, because EPA did not include a minimum sample size in expressing the
geometric mean criterion when EPA promulgated criteria for coastal recreation waters, if there is
only a single measurement (sample) of a waterbody, a state could use the SSM instead of the
geometric mean to determine whether to include that waterbody on the CWA 303(d) list. This is
because the state could not calculate a statistically reliable geometric mean without more than a
single grab sample. It would not be appropriate for a state to use the SSM as a substitute for the
geometric mean if there is more than one measurement (sample) because a geometric mean could
be calculated with more than one grab sample.
When developing TMDLs, states should consider the availability of additional pathogen data
besides that used for the original listing in developing their estimates of loading capacity, load
allocations and wasteload allocations needed to meet their specific water quality standards. For a
more detailed discussion of pathogen TMDLs, see EPA's January 2001 "Protocol for
Developing Pathogen TMDLs" (US EPA 841-R-00-002).
NPDES Permitting
Although states have flexibility in deciding whether and how to apply SSM values in the NPDES
permitting program, this does not mean that maximum daily or seven-day average permit limits
for bacteria are inappropriate for National Pollutant Discharge Elimination System permits.
EPA's Technical Support Document for Water Quality-based Toxics Control describes how
maximum daily and 7-day average effluent limits can be calculated based on 30-day average
conditions and an understanding of effluent variability (EPA-505-2-90-001, March 1991). The
procedures in Section 5.4.4 of the Technical Support Document are based on statistical
methodologies similar to those employed in deriving the SSM in the 1986 water quality criteria
for bacteria. EPA's decision to promulgate criteria that include SSM values does not address the
question of how States' should establish maximum daily permit limits.
For a state now adopting water quality standards for coastal recreation waters to be as
protective of human health as EPA's 1986 bacteria criteria and seeking EPA approval,
would a state need to adopt into its water quality standards for coastal recreation waters
both the geometric mean and SSM components of the criteria?
Yes. As stated in the preamble to the BEACH Act rule, one of the considerations EPA uses in
reviewing state water quality standards submissions related to bacteria criteria for coastal
-------
recreation waters is whether or not those standards include not only a geometric mean, but also
appropriate SSMs for all coastal recreation waters.
EPA expects those states adopting their own water quality standards as protective of human
health as EPA's 1986 bacteria criteria to use the SSM to make short-term decisions in the beach
monitoring and notification context. States adopting their own water quality standards have the
flexibility to determine how to use the SSM in other Clean Water Act programs. In adopting
new standards in coastal recreation waters, States may elect to include a minimum sample set
size as part of its geometric mean criterion. If it does so, it would need to have another
component of its criteria that would apply when there are fewer samples than the minimum
sample set size. This is because the criteria have to be as protective of human health as EPA's
1986 bacteria criteria. If the geometric mean criterion does not apply for some specified
condition, such as when there are fewer than five samples, then some other aspect of the criterion
would need to apply, in this case the SSM. Alternatively, states may elect to apply the geometric
mean criterion regardless of the number of samples used to calculate the geometric mean, which
is the approach EPA envisioned when it promulgated the BEACH Act rule. In this case, the
geometric mean criterion would apply in all instances, and a state could elect not to use the
single sample maximum for Clean Water Act purposes as a substitute for the geometric mean,
other than for beach monitoring and notification. This approach ensures a situation does not
arise where there is no applicable criterion.
When states adopt the 1986 bacteria criteria into water quality standards for their coastal
recreation waters, states should specify, in the water quality standards, how the SSM will be used
in particular Clean Water Act applications. This helps make it clear to EPA, state regulators,
dischargers, and the public how the SSM will be used in NPDES permits, assessment, TMDLs,
and beach monitoring and notification. The flexibility afforded to states in deciding whether and
how to apply SSM values in the National Pollutant Discharge Elimination System permitting
program does not mean that maximum daily or seven-day average permit limits for bacteria are
inappropriate for National Pollutant Discharge Elimination System permits. EPA's Technical
Support Document for Water Quality-based Toxics Control describes how maximum daily and
7-day average effluent limits can be calculated based on 30-day average conditions and an
understanding of effluent variability (EPA-505-2-90-001, March 1991). The procedures in
Section 5.4.4 of the Technical Support Document are based on statistical methodologies similar
to those employed in deriving the SSM in the 1986 water quality criteria for bacteria. EPA's
decision to promulgate criteria that include SSM values does not address the question of how
States' should establish maximum daily permit limits.
How should states that already have water quality standards that are as protective of
human health as EPA's 1986 bacteria criteria apply the single sample maximum?
States that already have water quality standards as protective of human health as EPA's 1986
bacteria criteria should apply the single sample maximum in a manner consistent with their water
quality standards.
-------
Some states have included a provision in their water quality standards that limits the application
of the geometric mean component of the criterion to situations only where five or more samples
are available. EPA expects these states to use the single sample maximum for all Clean Water
Act purposes if the geometric mean cannot be used because, for example, the sample set is fewer
than the minimum number of samples required by the state's water quality standards. EPA has
this expectation because the criteria have to be as protective of human health as EPA's 1986
bacteria criteria. If the geometric mean criterion does not apply for some specified condition,
such as when there are fewer than five samples, then some other aspect of the criterion would
need to apply, in this case the SSM. Alternatively, states may elect to apply the geometric mean
criterion regardless of the number of samples used to calculate the geometric mean, which is the
approach EPA envisioned when it promulgated the BEACH Act rule. In this case, the geometric
mean criterion would apply in all instances, and a state could elect not to apply the single sample
maximum for Clean Water Act purposes other than beach monitoring and notification, as is
discussed in the preamble to the BEACH Act rule. This approach ensures a situation does not
arise where there is no applicable criterion.
May states adopt a single sample maximum only for their beaches, rather than for all
coastal recreation waters?
No. This is because EPA has interpreted CWA section 303(i)(l)(A) of the Clean Water Act to
require states to adopt a geometric mean and at least one SSM for all of their coastal recreation
waters, not just for their beaches. In determining whether or not a state's coastal recreation water
quality standards for bacteria are as protective of human health as the 1986 bacteria criteria
document, EPA will look to see if the water quality standards include, along with a geometric
mean, at least one SSM for each of the state's coastal recreation waters. EPA will also look to
see if criteria for designated bathing beaches include an SSM based on the 75% confidence level,
because EPA considers this approach to be as protective of human health as the 1986 bacteria
criteria. EPA included in the BEACH Act rule any state or territory that did not cover all coastal
recreation waters with a SSM and that for designated bathing beaches did not have a SSM based
on the 75% confidence level. EPA does not expect a state or territory to use all four of the use
categories identified in the criteria document for its standards to be at least as protective as the
1986 bacteria criteria. For example, a State that applied the 75% confidence based maximum to
all waters would clearly be as protective as the 1986 bacteria criteria, even though it would only
have a single use category.
To get more information
Contact Lars Wilcut at:
1200 Pennsylvania Avenue, NW
Mail Code 4305T
Washington, DC 20460
(202) 566-0447
wilcut.lars@epa.gov
------- |