AIL
United States
Environmental Protection
Agency
Environmental Monitoring
Systems Laboratory
Las Vegas NV 89193-3478
Research and Development
EPA/600/S4-86/011 Jan. 1988
SEPA Project Summary
National Surface Water Survey
Eastern Lake Survey
(Phase I - Synoptic Chemistry)
Quality Assurance Report
M.D. Best, S.K.Drouse, LW. Creelman, DJ. Chaloud, and D.W. Sutton
The quality assurance program for
the Eastern Lake Survey - Phase I
included quality assurance and
quality control activities for field and
laboratory operations and for
verification and validation of the data.
A combination of blank, duplicate,
and audit samples were analyzed to
provide an external check on the
quality of 32 physical and chemical
parameters and to allow early
detection of problems in sample
collection, processing, and analysis.
The statistical analysis of the
verified data set included estimates
of instrumental and system detection
limits, system decision limits, overall
and analytical within-batch
precision, and overall and analytical
among-batch precision. Quantita-
tion limits also were calculated for
use in evaluating the precision of the
survey data. The results of these
statistical analyses were compared
to the data quality objectives for
detectability and precision.
This report was submitted in
partial fulfillment of Contract
Numbers 68-03-3050 and 68-03-
3249. This report covers a planning,
implementation, and data review
period from March 1983 to January
1986, and work was completed as of
December 1986.
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
The National Surface Water Survey
(NSWS) is a three-phase project within
the National Acid Precipitation
Assessment Program (NAPAP). The U S.
Environmental Protection Agency (EPA)
initiated the NSWS in 1983. The purpose
of Phase I of the NSWS was to document
the present chemical status of lakes and
streams in areas of the United States that
are potentially susceptible to the effects
of acidic deposition. This report provides
an overview of the quality assurance (QA)
activities and results for one component
of the NSWS, the Eastern Lake Survey -
Phase I (ELS-I), which evaluated the
chemical status of 1,798 lakes in the
eastern United States. The lakes were
selected from three regions east of the
Mississippi River that are potentially
susceptible to acidification. The EPA
Environmental Monitoring Systems
Laboratory in Las Vegas, (EMSL-LV),
Nevada, had primary responsibility for
the ELS-I QA program and sampling
operations.
Thirty-two chemical and physical
parameters were selected for in situ or
laboratory measurement. Data from these
measurements of single lake samples
provided information to evaluate the
present status of lakes on a regional
basis. Data quality objectives (DQOs)
were defined in terms of the precision
and accuracy of measured values for
-------
each parameter and of the
representativeness, comparability, and
completeness of the resulting data base.
The anticipated range of values and the
required detection limits were also
specified for each measurement.
Equipment and protocols for sampling,
chemical analysis, and data processing
were based on the best available
methods and were standardized in order
to achieve the DQOs.
A draft QA plan and a draft analytical
methods manual were used during pilot
studies which were conducted to test all
aspects of the ELS-I research plan.
Activities tested include the lake
selection process, the proposed
sampling and analytical protocols, the
DQOs, the QA program, and the data
management system. The final QA plan
and the final methods manual
incorporated revisions based on results
obtained from the pilot studies and on
changes implemented during the ELS-I.
Sampling and Analytical
Methodologies
The primary QA goals of ELS-I
base site operations were to obtain
accurate physicochemical and
geographical data at each lake site, to
collect representative lake samples
without introducing contamination, to
preserve the integrity of samples until
their analysis at contract laboratories,
and to perform selected analyses at field
laboratories. Field personnel used
helicopters equipped with fixed floats to
collect samples.
Activities of field laboratory
personnel included:
1. Receiving lake and QA samples
and field data from each
sampling team and assessing
sample condition upon receipt.
2. Reviewing field data forms and
field laboratory forms for
accuracy and completeness.
3. Incorporating audit samples with
lake samples to form a batch of
samples for analysis.
4. Analyzing the batch of samples
for pH, dissolved inorganic
carbon (DIG), true color, and
turbidity.
5. Performing aluminum extrac-
tions.
6. Filtering, preserving, and
shipping samples to contract
analytical laboratories for
additional analyses.
7. Coordinating sample shipment
information with the EPA
Sample Management Office in
Alexandria, Virginia, and with the
EMSL-LV in Las Vegas.
8. Distributing copies of data and
data forms to the appropriate
offices.
Standardized forms were developed
to record measurements made at each
lake and at the field and contract
analytical laboratories. One copy of each
form was sent to Oak Ridge National
Laboratory for entry into the NSWS data
base, and a second copy was sent to QA
personnel in Las Vegas. Figure 1 shows
the flow of samples and data from the
field to the field laboratory and to the
analytical laboratory.
Data Comparability Studies
Standardized techniques for
sampling and chemical analyses of water
samples from the ELS-I ensured that
the effect (if any) of variance from
sampling and analysis procedures on the
differences between lakes could be
identified.
Studies were also undertaken to
determine whether ELS-I data could be
compared to survey data from other
countries and to data obtained by using
different analytical methods. Subsamples
from 215 ELS-I samples collected in
the southern Blue Ridge Mountains were
shipped via commercial courier to
Norway for chemical analysis of 14
parameters. Similarly, 105 subsamples
from the Adirondack Mountains were
analyzed in Canada for 18 chemical
parameters.
A second study used 2,047 split
samples from the ELS-I to compare
data from chemical analyses by flame
atomic absorption spectroscopy (AAS)
and inductively coupled plasma emission
spectroscopy (ICPES). The ICPES
analyses were performed at the EPA
Environmental Research Laboratory in
Corvallis, Oregon, and by ELS-1
contract laboratories. The flame AAS
analyses were performed by ELS-I
laboratories only.
Procedures
Operational Quality Assurance
and Quality Control Program
The QA program used a combination
of blank, duplicate, and audit samples to
provide an external check on the quality
of the data obtained from measurements
of the 32 physicochemical parameters
and to allow early detection of problems
in sample collection, processing, and
analysis. Quality control (QC) protocols
for field sampling, field and analytical
laboratory activities, and data base
management were implemented
ensure reliability of the data.
The estimated number of samples
be analyzed and the estimated rate
sample collection required contract!
with more than one analytical laborati
to meet the QA and QC requiremer
Laboratories that submitted low bids
response to the formal bidding proce
were evaluated on the basis of th
analytical laboratory performance. The
that passed were visited to verify th
qualifications and capabilities.
Data quality also depended on i
ability of the field and laboratc
personnel to properly collect, proce
and analyze the samples. Training v\
essential to ensure consistent applicati
of all operational and QA-C
procedures.
Coordination of the ELS
operations required close communicat
among all participants to ensure that 1
program objectives were met. Da
monitoring of field sampling, fie
laboratory, and analytical laboratc
activities facilitated safety and logisti
coordination. During the actual sampli
phase, the most critical lines
communication were between t
logistics personnel in Las Vegas and I
field stations and between the <
personnel in Las Vegas and the contr
analytical laboratories. Logisti
personnel also coordinated and track
the shipment of samples to the analyti
laboratories and coordinated t
shipment of supplies to field stations. 1
QA staff made daily calls to the analyti
laboratories to ensure that the <
procedures were being implement
according to survey requirements a
that the samples were being handled z
analyzed properly.
Sampling and Field Laboratc
Quality Control
Field sampling QC procedur
included daily calibration of the Hydro
units, measurements of QC che
samples for pH and conductance, a
sampling site verification. The <
procedures for field laboratory operatic
included daily instrument calibrate
measurements of QC check samples <
laboratory duplicates, and preservation
sample aliquots for additior
measurements by the analytii
laboratories.
The objectives of samp
preservation at the field laboratories w
to (1) inhibit chemical and biologi
activity, (2) prevent changes due
volatilization, and (3) prevent effects c
-------
Routine
Lake
Samples
Field Blank
Samples
1
1
— — — - 11
Daily S<
of Sam
Field Duplicate
Sample
\
Field Laboml
itch
pies __
Field/Laboratory
Audit Sample(s)
ory
\
r
Relabeling
Analysis
(DIC. pH, Turbidity, True Color)
OC Check Batch Trailer Duplicate
Samples Samples Sample
-*• Data "•*-
Aliquot Preparation
Aluminum Extraction
Preservation
Shipment to
Analytical Laboratory
Analytical Laboratory
Analysis
r
Internal
OC Samples
Laboratory Blank,
Matrix Spike,
OC Sample Check
t
Batch
Samples
Laboratory
Duplicate
-»- Data -*-
Figure 1. Flow of samples and data through field and analytical laboratories, Eastern Lake Survey - Phase I.
-------
to precipitation or adsorption. Seven
preserved aliquots were prepared from
each bulk sample (routine, field
duplicate, field blank, or field audit). Four
parameters (pH, DIG, true color, and
turbidity) required immediate analysis.
Use of a mobile field laboratory for these
analyses permitted these four
measurements to be completed within 16
hours of sample collection.
Analytical Laboratory Quality
Control Protocols
Analytical laboratory QC included
the use of QC charts, calibration of
instruments and analysis of QC check
samples (QCCS), detection limit QCCS,
and duplicate samples. A maximum
sample holding time, from the time of
sample collection to sample analysis,
was established for each parameter
measured in the analytical laboratories.
These holding times were based upon
information from the literature, the best
scientific judgment related to the defined
needs, and the logistical demands and
limitations of the ELS-I.
Data Base Quality Assurance
Oak Ridge National Laboratory
(ORNL) personnel managed the data
base for the ELS-I. The data are stored
in four major data sets: (1) a raw data set
of field and analytical laboratory data, (2)
a verified data set, (3) a validated data
set, and (4) a final data set. Oak Ridge
personnel entered the field and
laboratory data into the data base and
used a double entry and comparison
process to minimize data entry errors.
The verified data set provides a data
base in which any values that are
questionable on the basis of known
physicochemical relationships are
qualified with a flag. The validated data
set provides data that have been
evaluated, using all available information,
for internal and regional consistency. The
final data set provides a representative
summary of sample values for use in
generating population estimates.
Data Verification
The raw data were verified by: (1)
communicating daily with the field and
analytical laboratories; (2) assessing the
completeness and consistency of each
data package (one per sample batch) on
receipt, and reviewing any comments or
questions associated with the batch or
sample under evaluation; (3) evaluating
the preliminary routine and QA sample
data; (4) obtaining confirmation,
correction, or reanalysis data from the
laboratories as needed to address
atypical values; and (5) providing
correcting entries to ORNL for
establishing the verified data set.
EMSL-LV QA personnel developed a
computer software package (AQUARIUS)
to automate this procedure as much as
possible.
These five steps produced a verified
data set in which all values that did not
meet a number of criteria based on
physicochemical relationships were
flagged, replaced with either corrected or
reanalyzed data, or replaced with codes
indicating a missing value.
Data Validation
The data validation process
identified potential errors in chemical
analyses that could not be revealed by
the verification procedures. The
validation process also evaluated the
quality of nonchemical variables. Data
validation employed pre-existing
software to identify possible outliers and
to evaluate possible systematic error in
the measurement process. Both of these
aspects were exploratory (as opposed to
test-oriented). The objective of data
validation was to identify individual
values or sets of values that warrant
special attention or caution when used
for analysis of survey results or when
used for building models based upon
survey data. Final decisions regarding
data quality were based on all available
information.
Data Base Review
An independent data audit team
inspected routine and duplicate data
from each of the four data sets (raw,
verified, validated, and final) that
constitute the ELS-I data base. The
data base review included an evaluation
of the adequacy of documentation for
value changes made between the raw
and the final data sets. For some of the
parameters, the audit team evaluated the
documentation by a search and review of
all of the data. For other parameters,
they selected a random subsample of
the changed values for review. The
review consisted of a comparison of the
old and new values against the field and
laboratory records and the verification
reports.
The review also included a
determination of the correctness of data
entry. The completeness of the data
base was estimated for each of the
parameters by examining the
justifications for missing value codes.
The correctness of the calculations of
population estimates was verified I
implementing the estimation algorithr
on a test data set which was a subset
the final ELS-I data set.
Results
Operations Evaluation
The ELS-I was completed with
the time required by the research plj
and the statistical requirements f
adequate sample size were achieved
all field stations. There were no maj
interruptions in field operations due
accidents, weather, or equipment failui
The sampling and laboratory protocc
were successful for most procedure
This survey design should serve as
model for future field studies of a simi
nature.
Lake and Sample Information
Approximately 90 percent of tl
lakes initially selected for sampling we
visited by sampling crews. Of those lak
visited, 96 percent (1,612) were actua
sampled. In addition to those lakes, 1
special-interest lakes were sampled 1
a total sample of 1,798 lakes.
Less than 20 percent of the lak
visited were sampled at a depth otr
than the one originally specified. Only
percent of the lakes sampled we
thermally stratified; thus, 95 percent of
samples were acceptable in terms of t
research plan requirement that a sin(
water sample was to be collected frc
each lake during a period when the lak
were isothermal.
In total, 2,389 routine, field duplica
and field blank samples were deliver
from the field laboratories to the contr<
analytical laboratories. One matrix spi
and one laboratory duplicate we
analyzed for each batch of sampl
making a total of 2,639 sets of analyses
Field Problems and Resolution;
The field crews collected samples
planned during the ELS-I with
average yield of 20 processed sampl
per operating day from each fie
laboratory. Problems which we
identified and corrected prior to or duri
field operations included an inaccur.
pH meter, use of contaminated bla
sample water at one field laboratory, a
destruction of one set of split samples
the commercial courier.
At the conclusion of the survey,
number of recommendations were ma
for improving field operations. A syst<
for efficient and complete informati
transfer should be used between t
management team and field personnel
-------
ensure that all new developments or
modifications in operational protocols are
consistently disseminated and
understood. Field personnel should be
prepared to make special arrangements
for shipment of samples to the analytical
laboratories on weekends. All field
samplers should receive comprehensive
instruction to ensure consistency of
sample collection and processing
activities. Rotation or replacement of field
personnel should be minimized because
it may lead to inconsistencies in data
reporting and it requires additional time
and effort for training.
Analytical Laboratory Problems
and Resolutions
Several problems were identified
and corrected during the analytical
laboratory operations. Two significant
difficulties centered on the inaccurate
calibration method used with silica
measurement at one laboratory and on
aluminum contamination problems at
three laboratories.
Methods Evaluation
Methods studies addressed a
number of analytical questions before
and during the survey operations. The
measurement of free dissolved fluoride
was found to be impractical, and ion-
selective electrodes (ISE) were favored
over ion chromatography (1C) for
measurement of total dissolved fluoride.
The Gran analysis method was found to
be useful for determinations of acid-
neutralizing capacity (ANC); however,
problems were encountered with the
calculations for base-neutralizing
capacity (BNC). The procedure for
measuring total extractable aluminum
and the design of the pH sample
chamber were modified during the pilot
study. Effects of filtration on dissolved
iron and aluminum concentrations in the
audit samples were detected during the
survey operations and were attributed to
sample instability.
A nitrate contamination problem in
blank samples was also encountered. To
meet analytical laboratory quality control
specifications, nitrate concentrations in
field blank samples were required to be
less than 0.01 mg I/1. During the pilot
study, up to 18 mg L'1 nitrate were
detected in field blanks, suggesting a
serious contamination problem. This
contamination was not present in the
analytical laboratory blank samples.
During the pilot study, two sources of
contamination were found to be the
deionized water and the sample
containers. Contamination from the
filtration apparatus was a third source.
The ELS-I was about half completed
before nitrate contamination from the
filtration apparatus was isolated and
eliminated. Nitrate data from samples
that were processed on or before the
change in the filtration protocol were
replaced with data from reanalyzed
samples.
Evaluation of Split Sample
Analyses from Norway and
Canada
In a study undertaken to determine
whether data obtained using analytical
methods used in this survey are
comparable to the data obtained by
methods used in other regions where
surface water surveys take place, 110
pairs of aliquots (split samples) from 97
lakes in the southern Blue Ridge
Mountains were sent to the Norwegian
Institute for Water Research in Oslo and
to an ELS-I analytical laboratory. Each
laboratory analyzed the split samples for
11 chemical parameters.
This study also used 105 split
samples from 92 lakes in the Adirondack
Mountains. Some of these split samples
were shipped to the Water Quality
National Laboratory of the Canada
Centre for Inland Waters in Burlington,
Ontario, and to an ELS-I analytical
laboratory. Other split samples were sent
to the Water Quality Section of the
Ontario Ministry of the Environment in
Rexdale and to an ELS-I analytical
laboratory. Analytical data from the three
laboratories were compared for 17
parameters by using linear regressions
and sign tests.
The data from the pairs of
laboratories that analyzed the samples
for the same parameters are comparable
for most of the parameters as indicated
by the high values of the regression
coefficients (r2 > 0.95 for 38 of 56
comparisons and r2 > 0.90 for 48 of 56
comparisons). The linear regression
analysis also indicated that there are
significant differences (uniform bias or
concentration-dependent bias) between
the ELS-I and Norwegian
measurements and between the ELS-I
and Canadian measurements for nearly
all of the parameters.
The sign test results show that as a
set, the ELS-I values are greater in
magnitude than the Norwegian values for
5 of the 11 parameters and less than the
Norwegian values for 3 of the
parameters; there is no significant
difference between the ELS-I and
Norwegian values for the remaining 3
parameters. The magnitude of the set of
ELS-I values is greater than the
magnitude of the Canadian values for 11
of the 17 parameters and it is less than
the magnitude of the Canadian values for
4 of the parameters; there is no
significant difference between the ELS-I
and Canadian values for the remaining 2
parameters.
The components of overall variability
in these split sample data include random
error (measurement imprecision) and
systematic error (accuracy and specificity
of the method). Measurement imprecision
and differences in the sample
composition over time may have
contributed to the overall variability for all
of the parameters measured by the
ELS-I laboratories and by the
Norwegian and Canadian laboratories.
Differences in analytical methodology
contributed to the overall variability for
some parameters.
Evaluation of Data Base Quality
A review of the data base showed
that less than 3 percent of the raw data
was classified as reporting errors. These
errors were corrected in the verified data
set. Sample reanalysis was requested for
less than 4 percent of the originally
reported raw data values. Less than 1
percent of the reported data required
correction because of transcription or
data entry errors.
An independent review of the data
base indicated that nearly all of the value
changes had been documented; that
there were few if any data entry errors;
that the data base was essentially
complete; and that the calculations of
population estimates were correct.
Evaluation of Quality Assurance
Data
Blank Sample Data
Field blanks and analytical laboratory
blanks were analyzed during the ELS-I.
The 245 field blanks were analyzed for
turbidity at the field laboratory and for 21
physicochemical parameters at the
analytical laboratory. The analytical
laboratories used calibration blanks or
reagent blanks to determine background
levels and to calculate instrumental
detection limits.
Evaluation of the blank sample data
demonstrated that the goals for
instrumental detection limits were
generally achieved. However, to interpret
the data, results from the field blanks
-------
must also be taken into consideration.
Extremely low detection limits achieved
in the laboratory are of limited value in
defining usable data when they are lower
than the system detection limit (the
background from sample collection and
handling). The system decision limit
should be considered during data
interpretation when comparing population
estimates.
Duplicate Sample Data
Analysis of field duplicate samples
provided data for estimating the overall
within-batch precision. The 125 field
duplicate samples that were processed
by the field laboratories were analyzed
with the routine samples and field blanks
at the analytical laboratories. The field
and analytical laboratories also
performed duplicate chemical analyses
on one sample per batch as a QC check
on analytical within-batch precision.
Overall within-batch precision
estimates were numerically larger, as
was expected, than the analytical
within-batch precision estimates for all
but five parameters. The estimated
overall and analytical within-batch
precision was considered to be adequate
to meet the DQOs.
Audit Sample Data
Six types of audit samples (2 field
naturals, 2 field synthetics, and 2
laboratory synthetics) were analyzed for
23 parameters during the ELS-I. Field
naturals and field synthetics were
handled in the same manner as were
routine, field blank, and field duplicate
samples to estimate the overall among-
batch precision (including the effects of
sample processing, but not of sample
collection). Laboratory synthetics were
prepared as processed aliquots by a
support laboratory; these were relabeled
at the field laboratory and were
incorporated with the sample batch to
estimate the analytical among-batch
precision.
The among-batch precision
estimates from field natural audit sample
measurements were within the DQOs for
all of the parameters except total
extractable Al, total Al, and Cl- in the
FN2 (low ANC) samples, and they were
within the DQOs for all of the parameters
in the FN3 (high ANC) samples. The
among-batch precision estimates from
field and laboratory synthetic audit
sample measurements were generally
within the DQOs for parameters with
sample values above the quantitation
limit, and they were generally not within
the DQOs for parameters with values
below the quantitation limit.
The natural audit samples were also
used to estimate the amount of relative
interlaboratory bias by comparing
measured values from the contract
laboratories with each other and with the
referee laboratory measurements. The
synthetic audit samples were used to
provide information on absolute
interlaboratory bias, and thus had to be
prepared from solutions of known
composition. Examination of the
theoretical values and the measured
values for the synthetic audit samples
indicated that the actual sample
composition may have differed on
different days; the amounts of
measurement imprecision and
interlaboratory bias were thought to be
small by comparison to the differences in
sample composition.
Quantitation limits were a useful
means of classifying the data in order to
objectively evaluate the among-batch
precision estimates for the 23
parameters. High relative precision
(percent relative standard deviation) was
not expected for measurements close to
the detection limits, and it was not
achieved for any parameters with means
less than quantitation limits except
dissolved organic carbon (DOC).
Conversely, high relative precision was
expected and was generally achieved for
measurements with higher mean values.
The overall and analytical within-batch
precision estimates for DOC
measurements showed patterns which
were opposite to the patterns shown by
all other parameters; i.e., measurements
at higher concentrations exhibited greater
variability than measurements at lower
concentrations.
An independent statistical analysis
concluded that, overall, the imprecision
of a perfectly calibrated instrument under
constant conditions is small in
comparison to the bias and trend
introduced by procedural variations and
by changes in the sample composition
over time, that precision estimates from
field natural audit samples are the best
indicators of measurement uncertainty in
the routine samples; that measurements
of laboratory synthetic audit samples
provide estimates of accuracy and
relative interlaboratory bias; that daily
lots of audit samples should be prepared
and divided into aliquots and the aliquots
should be assigned to different analytical
laboratories at random; that
interlaboratory comparisons should be
made using data from the same audi
sample lot; that audit sample compositioi
should be varied among the lots in orde
to estimate the precision and accuracy 2
various concentrations and to prever
recognition of the audit samples by th<
analytical laboratories; and that a
empirical approach should be used t
calculate detection limits. An EMSL-L1
review of this independent analysi
questioned two of the statistic*
assumptions: a zero median wa
assumed for blank sampl
measurements, and the measuremer
error variance was assumed to b
independent of the analyte concentratio
for some purposes but to be proportion;
to concentration for other purpose;
Other issues raised by the EMSL-L
reviewers concern the evidence used t
support the statement describing th
blank sample data as non-normal; us
of the root mean square for estimatin
the mean of the percent relative standar
deviation; use of a single slope i
correcting the bias and trend in aud
sample data; use of standard deviatic
instead of standard error to estimal
precision in the corrected data; an
omission of the DQOs as a basis for ft
data quality evaluations.
Variability
Four types of precision estimate
(overall within-batch, overall amont
batch, analytical within-batch, an
analytical among-batch) identify th
amounts of data variability that can t
attributed to sample collectioi
processing, storage, and analysis. Fi
the ELS-I, each type of precisic
estimate was used to estimate a differe
aspect of data variability.
Overall within-batch precisic
estimates were expected to t
numerically larger than analytical withi
batch precision estimates by an amou
equal to the variability from samp
collection, processing, and storag
Similarly, overall among-batch precisii
estimates were expected to t
numerically larger than analytic
among-batch precision estimate
Analytical and overall among-bat<
precision estimates were expected to I
numerically larger than the correspond'!!
analytical and overall within-bati
precision estimates by amounts that we
equal to the temporal variability.
Exceptions to the expect!
relationships were generally associat
with the presence of one or me
extreme outliers in the verified data s
with values close to the detection limit,
with a methodological problem. Ma
-------
exceptions involved small differences in
the estimated precision. In several cases
it was necessary to retain confirmed but
questionable values in the verified data
set; these values were later deleted
during data validation. The confirmed
questionable values influenced the
statistical evaluation of the ELS-I data.
For subsequent surveys, a data qualifier
was added to ensure that such values
are retained in the verified data set but
are not included in statistical calculations.
Summary
Data quality objectives were
established for precision, accuracy,
representativeness, comparability, and
completeness. Pilot studies were
conducted prior to the ELS-I to identify
and correct problem areas in field and
laboratory operations. Field and
laboratory QA and QC protocols were
implemented during the survey to detect
and eliminate equipment and procedure
problems. QA and QC procedures in the
data entry and verification provided a
means for detecting and correcting
transcription, transposition, and
typographical errors as well as analytical
errors in the data base. The raw data
were verified using physicochemical
relationships. Values that did not meet
criteria were flagged, corrected, or
replaced with codes for missing values.
The data quality objectives that were
specified for detectability and precision
were generally achievable. On the basis
of a limited amount of information the
amount of bias in the data appeared to
be small in relation to the imprecision.
Field operations were a successful
means for obtaining samples and field
data consistent with the ELS-I research
plan. The statistical requirments for
adequate sample size were achieved at
all field stations. Comparability of data
obtained by different analytical methods
and different laboratories was addressed
by independent studies. Less than 3
percent of the raw data was classified as
reporting errors, and these were
corrected in the verified data set.
-------
M.D. Best, S.K. Drowse', LW. Creelman, D.J. Chaloud, and D. W. Sutton are
with Lockheed Engineering and Management Services Co. ,lnc..
Las Vegas. NV 89119.
Eugene P. Meier is the EPA Project Officer (see below).
The complete report, entitled "National Surface Water Survey, Eastern Lake
Survey (Phase I - Synoptic Chemistry) Quality Assurance Report," (Order No.
PB 88-1337491 AS; Cost: $19.95, subject to change) will be available only
from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Exposure Assessment Research Division
Environmental Monitoring Systems Laboratory
Las Vegas, Nevada 89193-3478
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
BULK RATE
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
Official Business
Penalty for Private Use $300
EPA/600/S4-86/011
0000329 PS
------- |