United States
Environmental Protection
Agency
Environmental Monitoring Systems
Laboratory
Las Vegas NV 89114
Research and Development
EPA/600/S4-85/048 Aug. 1985
&ERA Project Summary
Sediment Sampling Quality
Assurance User's Guide
Delbert S. Barth and Thomas H. Starks
This report is to serve as a companion
to an analogous document on soil
sampling quality assurance. Prior to the
design of an adequate quality assur-
ance/quality control (QA/QC) plan for
sediment sampling, there must be
agreement on the objectives of the
sampling program. Answers to the
following questions should be available:
How will the resulting data be used to
draw conclusions? What actions may be
taken as a result of those conclusions?
What are the allowable errors in the
results? Once answers to these ques-
tions are available, an experimental
protocol may be prepared with an
appropriate statistical design and
QA/QC plan.
An overview of selected sediment
models is presented to serve as a
foundation for stratification of study
regions and selection of locations for
sampling sites, methods of sampling.
and sample preparation and analyses.
Discussions of situations relating to
rivers, lakes, and estuaries are included.
Statistical considerations presented
include experimental statistical designs
to enable ANOVA to be accomplished,
discussion of Type I and Type II errors,
numbers and locations of sampling
sites, bias, confidence and prediction
limits, outliers, and testing hypotheses.
The importance of an exploratory study
to the cost-effective achievement of the
overall objectives of a sediment sam-
pling program is emphasized.
This Project Summary was developed
by EPA's Environmental Monitoring
Systems Laboratory, Las Vegas, NV, to
announce key findings of the research
project that is fully documented in a
separate report of the same title (see
Project Report ordering information at
back).
Introduction
U.S. Environmental Protection Agency
(USEPA) quality assurance policy requires
that every monitoring and measurement
project must have a written and approved
quality assurance (QA) project plan.
Among the sixteen elements which must
be contained in all QA project plans are
the following:
Project description.
QA objectives for measurement data
in terms of precision, accuracy, com-
pleteness, representativeness, and
comparability.
Data analysis, validation, and report-
ing.
Specific routine procedures used to
assess data precision, accuracy, and
completeness.
This report, which is a companion to an
analogous document on soil sampling
quality assurance, addresses selected
factors associated with the application of
quality assurance/quality control (QA/
QC) guidelines to sediment sampling. In
order to make this report more self-
contained, chapters from the companion
soil report covering such topics as sample
handling, analysis of QA/QC data, and
system audits, which are equally appli-
cable to sediment sampling, are contained
verbatim in the appendices.
The most important consideration for
sediment sampling is the objective for
which the sampling is being done. The
statement of objectives should contain
clear answers to the following questions:
-------
How will the resulting data be used to
draw conclusions?
What actions may be taken as a result
of those conclusions?
What are the allowable errors in the
results?
Once answers to these questions are
available an appropriate statistical design
for the sampling and analysis program, to
include an adequate and verifiable
QA/QC project plan for the study, can be
devised.
Prior to the establishment of an ade-
quate, cost-effective QA/QC plan for
sediment monitoring programs, a deci-
sion-making official, after careful analysis
of the consequences, must specify allow-
able Type I and Type II errors in the
results. A Type I error, for a situation in
which a measured population mean is
being compared to either an action level
or a control level, is committed when it is
concluded that the population mean
exceeds the action or control level when
in fact it does not. For the same situation,
a Type II error is committed when it is
concluded that the population mean does
not exceed the action or control level
when in fact it does. The desired minimum
detectable difference between a mea-
sured population mean and either an
action, or a control level must also be
specified.
The goal of this document is to provide
a flexible, but technically sound, frame-
work within which the user can devise a
QA/QC plan consistent with the specific
objectives of any sediment monitoring
program. The document has been devel-
oped to serve as a user's guide for anyone
designing, implementing, or overseeing
sediment monitoring programs.
The extent to which adequate field-
validated models exist for describing sedi-
ment transport and deposition has a direct
bearing on the design of cost-effective
sediment monitoring programs. General-
ly, when adequate models exist, fewer
monitoring measurements are required
to assess pollutant levels and their signif-
icance. Accordingly, this report presents
a brief review of some available sediment
transport models after first providing
some background definitions and discus-
sions.
The models range from simple, steady
state, dissolved oxygen relationships to
very complex models describing the inter-
relationships among pollutant additions
and removals, organic matter concentra-
tions, and life processes occurring in
aquatic environments. Many pollutants
can be transported in suspended solid
form or adsorbed on suspended particu-
lates. Unfortunately, the dynamics of the
movement of pollutants adsorbed on
sediments is not well understood.
Sediments play an important role in the
transport of pollutants as well as in the
transport of nutrients. Both the pollution
and nutrient aspects must be considered.
Sediments can overwhelm bottom fauna,
but the nutrients they carry can give rise
to new biota.
In choosing an appropriate model, a
comparison should be made of available
models. A model should be fitted to the
problem and not vice versa. If complete
validated models are not available for the
pollutants and other site-specific condi-
tions of a problem, it still may be possible
to use portions of available models, or
other empirical field experience in the
cost-effective design of sediment sam-
pling programs.
The responsibilities of National Program
Managers in the USEPA Mandatory
Quality Assurance Program include en-
suring that data quality acceptance
criteria and QA Project Plans are prepared
for all data collection projects sponsored
by their offices.
This requires the development of data
quality objectives (DQOs). DQOs are
qualitative and quantitative statements
developed by data users to specify the
quality of data needed from a particular
data collection activity.
DQOs are the basis for specifying the
quality assurance and quality control
activities associated with the data collec-
tion process. QA Project Plans clearly
describe what will be done at each stage
of data collection (i.e., sample site selec-
tion, sample collection, sample handling
and analysis, and data handling and
analysis) and include instructions or
standard operating procedures for each
field and laboratory activity.
Some possible objectives for sediment
sampling are:
Determining the extent to which sedi-
ments act as either sources or sinks for
water pollutants,
Determining presence and distribution
of selected pollutants in sediments in
both space and time,
Determining the risk to human health
and/or the environment from sedi-
ment contamination by selected pollu-
tants, and
Taking measurements for validation of
sediment transport and deposition
models.
Under most circumstances, background
data will not be available for a given
monitoring location. These data must be
acquired before, or preferably during, any
sediment monitoring program. The inten-
sity of the background sampling that is
undertaken depends upon the pollutants
being measured, the sediment character-
istics and variability, the levels of pollutant
likely to be found in the study area and the
purpose of the study. QA/QC procedures
are just as critical for the background
measurements as they are for the study
area measurements.
When sediments are contaminated,
drinking water or human foods, contami-
nated directly or indirectly through con-
tact with sediments, may be unfit for
human consumption. As the hazardous
constituents move through different
trophic levels, substantial biomagnifica-
tion of contaminants may take place.
The steps outlined below are designed
to provide a sediment monitoring effort
with minimal needed sample precision
and representativeness.
Determine the components of variance
that should be built into the statistical
design.
Choose the allowable probabilities for
Type I and Type II errors and the
difference in means considered to be
significant. (These are the DQOs and
they are needed together with an
estimate of the coefficient of variation
to determine the number of samples
required in each stratified region.)
Obtain sampling data from studies
with similar characteristics to the one
of interest. (Estimates of coefficients
of variation are of particular impor-
tance.)
Calculate the mean and note the range
of each set of duplicates (co-located
independent samples).
Using results from previous studies,
develop a table of critical difference
values for duplicate sample results for
various concentrations that span the
range of concentrations of interest.
Use this table to accept or reject sets of
duplicates.
Suggestions for additional elements of
a more complete QA/QC plan are pro-
vided in the text.
The DQO guidelines below are sug-
gested for the indicated operational situa-
tions.
-------
Confidence
Level Power Relative
(1-g) (1-ffl Increase*
Preliminary
Site
Investigation 70-80% 90-95% 10-20%
Emergency
Cleanup
Planned
Removal and
Remedial
Response
Activities
80-90% 90-95% 10-20%
90-95% 90-95% 10-20%
Relative Increase from Background or an
Action Level to be Detectable with Probability
Statistical sampling plans are based on
assumptions concerning the probability
distributions of the measurements to be
made. The properties of a normal distribu-
tion are so desirable that, if the data are
not normally distributed, a transformation
is sought to convert the existing distribu-
tion into a new distribution which is
approximately normal.
The maximum probability allowed for a
Type I error is called the significance level
of the test of hypothesis and is commonly
denoted by alpha (a). The probability of a
Type II error is usually denoted by beta (ft)
and is typically a function of a, sample
size, and the size of the deviation from the
null hypothesis. The probability that the
alternative hypothesis will be accepted
when it is true is called the power of the
test and may be denoted by (1 -ft). Typical-
ly, the experimenter will specify the
smallest deviation from the null hypothe-
sis that he considers to be scientifically,
economically, or environmentally impor-
tant to detect and then specifies the
power of the test that he wants for that
specific alternative.
The Quality Assurance Officer, sup-
ported by a qualified statistician, should
be intimately involved in the review of the
experimental or sampling design pro-
posed by the investigator. He should
insure that the information obtained
provides measures of the components of
variance that are identified in the field.
Composite samples provide only an
estimate of the mean of the population
from which the samples forming the
composite are drawn. No estimate of the
variance of the mean, and hence, the
precision with which the mean is esti-
mated can be obtained from a composite
of samples. Since the primary purpose of
QA/QC is to measu re the precision of the
samples obtained, the compositing of
samples should be avoided if at all
possible.
Split samples, spiked samples and
blanks are used to provide a measure of
the internal consistency of the samples
and to provide an estimate of the compo-
nents of variance and the bias in the
analytical process. The number of QA/QC
samples needed is suggested as one out
of every twenty samples for most cate-
gories of samples. In some instances this
guideline may not be adequate while in
others it may provide more samples than
are necessary. It is good practice to
perform an initial exploratory study in
which, among other things, QA/QC sam-
ples in excess of the guideline recom-
mendations are collected and analyzed.
Analysis of the resulting data will provide
a better estimate of the optimum required
number of QA/QC samples of different
types.
Typically, one wishes to estimate the
concentration of measured pollutants in
the sediments and to indicate the pre-
cision of these estimates. To indicate
precision of an estimate, one may provide
the standard error or a confidence interval
for the expected value of the concentra-
tion. The confidence interval is bounded
by confidence limits. Confidence limits
are bounds of uncertainty about the
average caused by the variability of the
experiment.
Prediction limits are similar to confi-
dence limits but are used to identify an
interval into which a randomly chosen
future sample value should fall. Equations
for both confidence and prediction limits
are provided along with an example
calculation.
A problem that is particularly prevalent
in data obtained from field samples is that
of outliers. The cause of the outlier may
be an error of procedure in sampling,
subsampling, chemical analysis, or the
transcribing of data; or it may be due to an
anomaly that would indicate that a change
is required in the assumed model for the
process. Guidelines are provided for
rejecting outliers, however, there are
many problems with outlier tests. If at all
possible, prior to rejecting values as
outliers, repeat measurements should be
made on the same or nearly identical
samples.
Once objectives have been defined
which involve the need for sediment
sampling, the next step is to develop a
total study protocol including an appro-
priate QA/QC project plan. The recom-
mended approach is to conduct an ex-
ploratory study first that includes both a
literature and information search along
with selected field measurements made
on the basis of some assumed transport
model.
To provide a framework for the discus-
sion, a hypothetical situation involving an
abandoned hazardous waste site is de-
scribed. The established objective for this
hypothetical situation is to conduct an
environmental assessment of the site
and its environs to determine whether a
short or long term hazard to man or the
environment exists. If a hazard exists, its
nature and extent must be defined and
appropriate recommendations made to
bring the hazard under control. A study
team is organized to address the problem
and the sediment study group's task is to
identify and make an assessment of
potential problems associated with sedi-
ments in a nearby river and estuary.
Questions which must be answered, at
least in part, by the exploratory study
include:
What wastes have been placed at the
disposal site over what time periods?
What chemicals in what amounts have
escaped from the site via what trans-
port routes and what is the present
geographical extent of these chem-
icals?
What adverse effects on human health
or the environment have been reported
in the site vicinity?
What is an appropriate background
region to use for the study?
Before taking any field measurements, a
comprehensive literature and information
search should be conducted to determine
what information may already be avail-
able. The results of the exploratory study
will provide information and field data
that will serve as the basis for the design
of a more definitive monitoring study.
Thus, any field measurements taken
should include appropriate QA/QC
measures to determine the quality of the
data.
The hypothetical case study is devel-
oped step by step. Data quality objectives
are identified, a grid system is defined,
the study area is stratified, a background
region is selected, number and locations
of sites for sampling are determined, and
an appropriate QA/QC project plan is
prepared.
In general, the simplest sampling tool
deemed to be adequate should be used.
The advantages and disadvantages of
some bottom samplers and some coring
devices are presented in tables.
-------
One of the possibilities for error during
the sampling process is discarding non-
sediment material collected with the
sediment samples prior to analysis. It is
suggested that all such discarded material
be retained. Ten percent of these samples
should be sent to the analytical laboratory
for analysis with the remainder being
archived.
If the exploratory study is conducted
well, it will provide some data for
achieving the objectives of the study; it
will provide data concerning the feasibility
and efficacy of most aspects of the study
design including the QA/QC plan; it will
serve as a training vehicle for all partici-
pants; and it will pinpoint where addi-
tional measurements need to be made.
Following analysis and interpretation
of the information and data resulting from
the exploratory study, the next step is the
design of the final definitive study. Any
problems with the QA/QC plan noted
should be solved by appropriate modifica-
tions of the plan. The procedure is illus-
trated by extending the hypothetical case
study based on assumed data obtained
from the exploratory study.
In view of conclusions reached on the
basis of the assumed data, the following
questions which should be answered in
the definitive study are identified:
How far down the stream are the
sediments significantly contaminated?
What are the relative contributions of
surface water and ground water to the
contamination of sediments?
How are the sediment levels changing
as a function to time?
What levels of contamination in human
foods are derived directly or indirectly
through contact with sediment?
What is the impact of contaminated
sediments on aquatic biota?
How should the study area be stratified
in the definitive study?
A table is provided givng the number of
samples required in a one-side, one-
sample t-test to achieve a minimum
detectable relative difference at confi-
dence level (1 -a) and power (1 -/)). In this
table the coefficient of variation varies
from 10 to 35%, the power from 80 to
95%, the confidence level from 80 to
99%, and the minimum detectable relative
difference from 5 to 40%. An equation is
provided to calculate values not included
in the table.
The required frequency of sampling
depends on the objectives of the study,
the sources and sinks of pollution, the
pollutant(s) of concern, transport rates,
and disappearance rates. Assessment of
trends in time will establish whether
sediment concentrations are increasing,
decreasing, or remaining fairly level.
Evaluations of these trends will be im-
portant to selection of appropriate re-
medial response measures.
The analysis and interpretation of
QA/QC from the more definitive study
should show how all aspects of the total
QA/QC plan combine to give an overall
level of reliability for various aspects of
the resulting data. Another goal may be to
determine whether all QA/QC procedures
used were necessary and adequate. It is
desirable to provide summarized tables of
validated QA/QC data in the final report.
From such tables it is possible to deter-
mine bias; precision; component random
errors associated with reproducibility,
extract matrix, sample matrix, and sample
homogeneity; inter laboratory precision;
and uncertainty. Presentation of QA/QC
data also contributes to the building of a
body of data in the literature which allows
comparisons to be made between and
among studies.
Data from the more definitive study
describing variations in sediment con-
centrations with depth will show how
effective dredging to different depths
might be in the removal of the contami-
nation. If dredging is even contemplated,
safe and effective methods for disposing
of the dredge spoil must be available.
Delbert S. Barth and Thomas H. Starks are with University of Nevada, Las Vegas,
NV89154.
Kenneth W. Brown is the EPA Project Officer (see below).
The complete report, entitled "Sediment Sampling Quality Assurance User's
Guide, "(Order No. PB 85-233 542/AS; Cost: $14.50. subject to change) will be
available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Environmental Monitoring Systems Laboratory
U.S. Environmental Protection Agency
Las Vegas, NV 89114
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
BULK RATE
POSTAGE & FEES PAII
EPA
PERMIT No G-35
Official Business
Penalty for Private Use $300
EPA/600/S4-85/048
------- |