United States
Environmental Protection
Agency
Environmental Monitoring
Systems Laboratory
Las Vegas NV 89193-3478
Research and Development
EPA/600/S4-89/029 Nov. 1989
4»EH\ Project Summary
Eastern Lake Survey - Phase II
Quality Assurance Report
T. E. Mitchell-Hall, A. C. Neale, S. G. Paulsen, J. E. Pollard, and D. W. Sutton
The Eastern Lake Survey - Phase II
(ELS-II) was designed primarily to
assess seasonal variability In region-
al surface water chemistry. This
report describes and evaluates the
quality assurance program employed
during the survey. The operations
component Included quality assur-
ance and quality control procedures
to ensure that all samples were
collected and analyzed consistently
and to estimate the accuracy and
precision of the reported values with
a known degree of confidence. The
data management component estab-
lished a means to store and track
data; to Identify and correct entry,
reporting, and analytical errors; and
to keep a record of such changes.
The survey designed Identified 24
physical and chemical character-
istics of lake water for measurement
Data quality objectives for detect-
ablllty, accuracy, precision, repre-
sentativeness, completeness, and
comparability for ELS-II were based
on previous related surveys. During
data verification and validation
activities, several Issues (concen-
trated primarily on the data from the
chloride, nitrate, sulfate, and
alkalinity analyses) prompted a
Special Data Assessment. This
process produced a list of recom-
mendations and justifications for
changes to be made to the official
verified data base.
Overall, the quality assurance pro-
gram was successful In identifying
and resolving a number of data
quality issues and assuring that the
data were of known and documented
quality. For ELS-II as a whole, the
data are of acceptable quality and
every effort was made to correct any
deficiencies. The accuracy and pre-
cision of data for four analytes of
primary interest In acidic deposition
research (acid neutralizing capacity,
pH, nitrate, and sulfate) were close to
or better than the goals set for
Intralaboratory performance. The
representativeness, completeness,
and comparability of the data meet
the project objectives. Special atten-
tion should be given to the data
quality objectives for surveys with
multiple components, Including con-
sideration of specific objectives for
each component
This report is submitted in partial
fulfillment of contract number 68-03-
3249 by Lockheed Engineering and
Sciences Company under the spon-
sorship of the U.S. Environmental
Protection Agency. This report
covers a field work period from March
25 to May 3, 1986, for the Spring
Seasonal subsurvey; from July 23 to
August 11, 1986, for the Summer
Seasonal subsurvey; and from
October 8 to November 14, 1986, for
the Fall Seasonal subsurvey. Data
verification was completed In
September 1987. The Special Data
Assessment began In February 1988
and ended In March 1989.
This Protect Summary was devel-
oped by EPA's Environmental Monitor-
Ing Systems Laboratory, Las Vegas NV,
to announce key findings of the
research project that Is fully docu-
-------
merited In a separate report of trie
same title (see Project Report order-
ing Information at back).
Introduction
This report describes and evaluates the
quality assurance (QA) program em-
ployed during the Eastern Lake Survey -
Phase II (ELS-II). This survey was
designed primarily to assess seasonal
variability in regional surface water
chemistry. The ELS-II is one of a series
of surveys conducted as part of the
National Surface Water Survey (NSWS),
a component of the National Acid
Precipitation Assessment Program. The
ELS-II QA program was designed to
ensure that all samples were collected
and analyzed consistently, to verify the
report results, and to inform data users of
the quality and potential limitations of the
data base. In addition to summarizing the
results of the QA program for ELS-II, this
report includes an assessment of ana-
lytical data quality.
In ELS-II a subset of the lakes samples
in Phase I of the Eastern Lake Survey
was resampled to assess chemical
variability and biological status. Lakes
included in the ELS-II were restricted to
those lakes considered susceptible to
acidification. To assess this chemical
variability, a subset of 150 lakes that had
been sampled in the northeast region
during Phase I was sampled during each
of the ELS-II seasonal subsurveys. To
produce comparable data, procedures for
sample collection, sample analysis, and
data reporting were based on the
protocols established during Phase I.
The ELS-II, which took place in 1986,
consisted of three major components—
the Spring Seasonal, Summer Seasonal,
and Fall Seasonal subsurveys. The
assessment of analytical data quality
discussed in this report applies to these
three subsurveys as well as the data
generated by the Fall Variability study
which was conducted during the Fall
Seasonal subsurvey.
Design and Operations of the
Quality Assurance Program
The ELS-II quality assurance and qual-
ity control program was a multifaceted
design to reduce uncertainty in the data
quality. Monitoring techniques were
incorporated at each stage of collecting,
processing, and analyzing the lake
samples. On-site evaluations also were
performed to monitor the system. Data
quality objectives were established to
determine the relative quality of the data
in terms of representativeness, complete-
ness, comparability, detectability. accu-
racy, and precision. A data base manage-
ment system established a means to
store and track data; to identify and
correct entry, reporting, and analytical
errors; and to keep a record of such
changes. Table 1 lists the analytes
selected for measurement and the loca-
tion of the measurements.
Table f. Chemical and Physical Variables
Measured During the Eastern
Lake Survey • Phase II
Field Sits
Conductivity (pS/cm)
Depth (m)
Dissolved oxygen (tng/L)
pH, field (pH units)
Temperature (°C)
Processing Laboratory
Aluminum (rng/L)
Total monomeric
Nonexchangeable monomeric
pH. closed system (pH units)
Dissolved inorganic carbon, closed
system (mg/L)
True color (PCU)
Turbidity (NTU)
Analytical Laboratory
Acid-neutralizing capacity (veq/L)
Aluminum, (mg/L)
Total extractable
Total
Ammonium (mg/L)
Base-neutralizing capacity (peq/mL)
Calcium (mg/L)
Chloride
Conductivity ftiS/cm)
Dissolved inorganic carbon (mg/L)
Initial
Equilibrated
Dissolved organic carbon (mg/L)
Fluoride, total dissolved (mg/L)
Iron (mg/L)
Magnesium (mg/L)
Manganese (mg/L)
Nitrate (mg/L)
pH (pH units)
Equilibrated
Initial (acid titration for ANC)
Initial (base titration for BNC)
Phosphorus, total dissolved (mg/L)
Potassium (mg/L)
Silica (mg/L)
Sodium (mg/L
Sulfate (mg/L)
ELS-II sampling activities included fl
operations that were conducted from f
base sites by helicopter and groi
crews. They collected lake samples i
associated data on the physical <
chemical characteristics of the lafc
After collection, the samples were sen
a centrally located processing laborat
in Las Vegas, NV. Processing laborat
activities included organizing the samp
into batches; analyses for selected ph
ical and chemical characteristics; splttl
the samples into aliquots; and preservi
packing, and shipping the samples to
analytical laboratories.
Two analytical laboratories participa
in ELS-II. The QA program provided
laboratories with explicit instructions
sample analysis protocol for ELJ
samples. Each laboratory was to use
same protocols to measure the 24 a
lytes of interest.
The QA program used a variety of
and quality control (QC) samples
monitor activities during the survey,
assist in verification, and finally to assi
data quality.The data quality objectt
for ELS-II were based on those est
lished for previous NSWS surveys. <
site system audits evaluated the fie
processing laboratory, and analyti
laboratory facilities, equipment, a
operations such as record keeping, d
reporting, sample analysis, and (
procedures.
The data management system v
designed to assemble, modify, and st
data collected during the NSWS surve
An independent data management cc
pany provided these services for ELS
A number of data bases were created
ELS-II. The raw data base consists
information derived from the field forr
the processing laboratory form, and '
analytical laboratory data package. 1
survey data were entered into the r
data base twice, the two sets of d
were compared for inconsistencies, t
any errors were corrected.
The QA staff at the U.S. Environmer
Protection Agency (EPA) laboratory
Las Vegas verified the data by check
the internal consistency of sample rest
and by evaluating the QA and QC sam
results. Much of the verification procc
was computerized. Changes to the r
data base that were necessary as a re:
of verification activities were sent to '
data base administrator where the offk
verified data base was created. That d
base was compared to the QA staff d
base to ensure accuracy. Other d
bases were created by the data b{
management administrator at the din
-------
*ion of the EPA laboratory in Corvallis,
DR, which was responsible for validating
the data.
The modified verified data base is the
product of a Special Data Assessment
which occurred after verification and
validation were completed. This assess-
ment was conducted to address a
number of issues identified during data
verification and validation. The modified
verified data base was created by
copying the official verified data base and
making any changes resulting from the
Special Data Assessment.
The objective of data evaluation and
verification was to identify and they
correct or qualify suspect data as well as
to maintain continuous control over the
data acquisition process. The steps in
this process included:
• Communicating with the field base
coordinators and the laboratories.
• Reviewing field and processing
laboratory data forms.
• Evaluating the preliminary QA data
from each analytical laboratory.
• Checking on the completeness of
each data package and on the
consistency of sample data.
• Confirming or correcting suspect data.
• Requesting reanalysis of samples for
which data remained suspect after
confirmation or correction.
• Assigning data qualifier tags and flags
to data when necessary.
• Entering the corrected values and data
qualifiers into a copy of a copy of the
raw data base to create the official
verified data base.
Many concepts developed during ELS-I
were implemented in ELS-II. In addition,
further modifications based on the
experience gained from previous NSWS
surveys were made to improve the ELS-
II. These changes include: eliminating the
requirement to analyze a matrix spike
sample; implementing daily direct data
transfers via electronic media between
the analytical laboratories and the QA
staff; conducting an Interlaboratory Bias
Study; developing an improved system
for use by the QA staff to edit the data;
using boats for lake access; measuring
dissolved oxygen at the site; establishing
the processing laboratory in a central
location; measuring dissolved and non-
exchangeable monomeric aluminum at
the processing laboratory; using two pH
meters for one batch; submitting final
data packages from the analytical labora-
tories electronically; and changing the
Decision requirement for conductivity
from 1% to 2% relative standard devi-
ation.
Assessment of Operations
Field operations for ELS-II successfully
completed collection and shipping of
samples according to protocol. On-site
evaluations monitored all field operations.
The valuations concluded that a checklist
for field equipment should be used by
field crews prior to departure from the
base site and identified a need for more
care in completing forms correctly and
concisely. The on-site evaluations con-
cluded that the field sampling personnel
were adhering to QA and QC protocols
and none of the findings had an adverse
effect on data quality.
The processing laboratory successfully
processed and analyzed the ELS-II
samples. Samples were prepared for
shipment to the analytical laboratories
within the specified holding times in all
cases. Two on-site evaluations of the pro-
cessing laboratory indicated that pro-
cessing laboratory operations were satis-
factory and that laboratory personnel
performed their duties well.
Standard EPA contract laboratory
program procedures were used to secure
the services of two analytical laboratories
to perform sample analyses for the ELS-
II. Laboratory 1 performed the sample
analyses for the Spring Seasonal sub-
survey. Laboratory 2 performed the
sample analyses for the Fall Seasonal
subsurvey. Both laboratories analyzed
samples for the Summer Seasonal sub-
survey. During the Summer Seasonal
subsurvey, both laboratories analyzed
samples designed to evaluate inter-
laboratory bias.
An on-site evaluation of Laboratory 1
during the Spring Seasonal subsurvey.
resulted in several findings: sample
receipt, storage, and labelling procedures
were adequate; the ion chromatograph
system was not fully automated with
respect to controlling the pump and
detector, requiring two analyses per
sample i.e., one for nitrate and one for
chloride and sulfate; the instrument
detection limit for chloride exceeded the
contract-required detection limit; the
laboratory was well equipped to analyze
metals, but experienced aluminum con-
tamination in the total aluminum
digestion; and laboratory personnel were
preparing control charts for percent
relative standard deviation (%RSO), not
QCCS control charts as required by the
contract. In summary, Laboratory 1 had
some deficiencies at the time of the on-
site evaluation but was performing ade-
quately.
The on-site evaluation of Laboratory 2
during the Summer Seasonal subsurvey
resulted in several findings: sample re-
ceipt, storage,and labelling procedures
were adequate; the laboratory was report-
ing a calculated pH value instead of the
measured pH value; and analysts were
not warming samples in a temperature-
controlled water bath to 25 "C before
performing conductivity measurements
as required by the contract. The evalu-
ation team concluded that the overall
performance of Laboratory 2 was accept-
able and that the laboratory was oper-
ating within the contractual framework.
During data review and verification,
sample reanalysis was only requested as
a final corrective action when there was
no other alternative for correcting data
problems identified by the QA staff.
Reanalyses were requested at two
different times. A nominal number of
reanalyses were requested in 1986 as a
result of analytical problems detected
during actual sample analyses. These
analyses were performed either within or
just outside of the maximum required
holding time. The QA staff requested the
majority of the reanalyses during data
verification in 1987. The number of these
reanalyses was kept to a minimum.
The review and verification process
identified several significant problems at
the analytical laboratories. Appropriate
changes or notations were made in the
verified data base. Some of these prob-
lems are: four batches of samples
required reanalyses for dissolved organic
carbon during the Spring Seasonal sub-
survey; one Spring Seasonal batch of
samples and one Fall Seasonal batch of
samples required reanalyses for total
aluminum; chloride, sulfate, and nitrate
values for two batches were reported
incorrectly by one laboratory because of
an error in preparing standard solutions;
one laboratory experienced difficulty in
many instances in meeting the contract-
required instrument detection limit for
chloride; ammonium values were re-
ported as nitrogen values by one labora-
tory during the Summer Seasonal sub-
survey; one laboratory did not calculate
anion and cation balances and conduc-
tivity balances as required by the con-
tract; inconsistencies in the calculation of
ANC and BNC by both analytical labora-
tories were resolved by the development
of an improved calculation procedure;
and one laboratory made errors in the
sample log-in procedure.
-------
Special Data Assessment
During data verification and validation
activities, several issues (concentrated
primarily on the data from the chloride,
nitrate, sulfate, and alkalinity analyses)
prompted a Special Data Assessment.
This special assessment took place after
the completion of the official verified data
base and during the final phase of data
validation. This assessment included an
extensive examination of the raw data
from both analytical laboratories for many
parameters. Samples were targeted as
outliers by various QA and QC sample
programs, the verification programs, and
the validation process. This method of
identifying problem samples provided a
mechanism for the QA auditor to
determine if an analyte had multiple
problems.
All outliers were prioritized by analyte.
Chloride, sulfate, nitrate, and alkalinity
were of the highest priority. If a sample
was considered an outlier, then the entire
batch was reviewed using the analytical
raw data. This process produced a list of
recommendations and justifications for
changes to be made to the official
verified data base. The data base created
by these changes is referred to as the
modified verified data base.
Assessment of Data Quality
The quality assurance program was
successful in reducing to acceptable
levels errors associated with the acquisi-
tion and reporting of data, as well as in
identifying and correcting potential prob-
lems associated with data quality.
Completeness, Comparability,
and Representativeness
The ELS-II data base was at least 90
percent complete based on the ratio of
lake samples collected to the lake
samples targeted for collection and on
the percentage of acceptable data gener-
ated from those samples. The use and
documentation of standardized sampling
and analytical procedures allow for a
quantitative evaluation of the data from
ELS-I and ELS-II and other past and
future studies. Standardized protocols
helped to ensure that each of the
samples collected was representative of
the chemical condition existing in the lake
at the time of sampling. From the quality
assurance perspective, representative-
ness can be defined by how well the
audit samples reflected the matrices and
the concentration ranges of the routine
samples. Natural audit samples were
composed of characterized, stabilized
lake water from a dilute lake with
moderate ANC and from a lake
representative of an acidic system.
Experimental synthetic audit samples
were designed to represent the expected
concentration ranges of the ELS-II routine
samples and were prepared at five
concentrations.
Detectability
Detectability can be addressed at two
levels. The first is the detectability
associated with a particular instrument or
analytical method. These limits are gen-
erally determined by calibration or
reagent blanks that are not blind to the
analyst. The second level evaluates the
system detectability, i.e., the lowest value
of an analyte that can be detected when
the entire process, from sample collec-
tion through laboratory analysis, is taken
into consideration. Field blank samples
consist of reagent grade distilled water
that is passed through the sampling
device and the entire processing and
analytical procedure. Within each of
these two levels, there are two specific
limits which can be distinguished—the
decision limit and the detection limit. The
decision limit represents the lowest
measured sample value that can be
distinguished from a blank sample or
background noise. The detection limit
represents the lowest true or theoretical
concentration above the decision limit
that can be measured with a specified
level of reliability. For most data users,
the system decision limit will be of
primary interest. This assessment con-
centrates on system-level detectability
because it is of most interest for the pur-
poses of routine data interpretation.
Evaluation of detectability for the ELS-II
data is complicated by the apparent
difference in the results of the two
analytical laboratories. Laboratory 1 con-
sistently had a larger system decision
limit, as a result of either larger values,
poorer precision, or both, for non-
exchangeable monomeric aluminum,
chloride, conductivity, initial and equili-
brated dissolved inorganic carbon, potas-
sium, magnesium, and sulfate. The
importance of this apparent problem
depends entirely on the relative levels of
these analytes in the routine samples.
Several of the analytes were present at
relatively low levels in the lakes sur-
veyed. In some cases this is not sur-
prising because the lakes under con-
sideration are, in general, dilute, oligo-
trophic systems that usually have low
buffering capacity corresponding to low
DIG values. Thus, it is not unexpected to
find a high percentage of samples be
the decision limit. These results do
necessarily indicate poor data quality,
rather reflect the systems being studi
These results do mean that analysis i
interpretation of trends in these data
require more attention to the issue
detectability, and differing levels
performance from participating labo
tories can be a problem.
The analysis of concentrations found
audit samples with respect to systi
detection limits indicates that grea
care must be taken in creating
selecting audit samples. The quantity
DIG, chloride, nitrate, and total dissolv
phosphorus present in the natural ai
samples was low relative to the systi
detection limits. The lowest concc
{rations in these audit samples should i
be less than the expected system det<
tion limits.
Accuracy and Precision
The ELS-II QA report presents
different approach to assessing accura
and precision than that used in previc
QA reports for the NSWS. In this rep
summary statistics from QA data for ea
analyte are presented by season
appendices. The procedures and calcu
tions that can be used to interpret the
statistics are presented within the te
The data user is then provided with t
tools for evaluation of the level of er
associated with survey data without a
reference to pre-set data quali
objectives.
Based on the data from both audit a
routine-duplicate pair samples, the ELS
data base reflects high quality data
most analytes of interest. For examp
ANC, pH, nitrate, and sulfate values z
of high quality with accuracy ai
precision well below or near the DQC
Nitrate values were very close to tl
upper bounds of the DQOs for Sprii
Seasonal subsurvey data and should I
evaluated carefully by the data user
determine if the quality is adequate
answer the question being aske
Analytes that did not meet pr
established DQOs included Laboratory
chloride data and all analytical laborato
measurements of extractable aluminui
DIG. and DOC. Processing laborato
measurements for DIG should be usi
because those values meet the DQC
Extractable aluminum data are cor
parable in quality to that produced
other NSWS projects. High levels of to
aluminum imprecision are a function
one or two outlying data points for
seasonal subsurveys. These data shot
-------
be closely examined by the data user to
determine if the associated variability and
bias would limit the interpretability of the
data.
Interlaboratory Bias
Because the ELS-II survey design used
two laboratories over the course of the
three seasonal subsurveys, the issue of
interlaboratory bias is an especially
critical one. Three sets of samples can
be used to analyze the QA and QC data
for interlaboratory bias. The first set
consists of the triplicate-routine samples
taken during the summer subsurvey.
Equal portions of these samples were
sent to each analytical laboratory. The
two kinds of natural audit samples make
up the second and third sets. The results
of the analyses indicate that the most
clear instances of interlaboratory bias
exist for measurements of conductivity,
DIC-eq, DIC-initial, and iron. There is
somewhat weaker evidence to indicate
potential problems with interlaboratory
bias for ANC, calcium, chloride, total
fluoride, potassium, pH-eq, and turbidity.
There are other instances of inter-
laboratory bias which appear to be
supported by either the split analyses or
the natural audit sample data, but not
both. The scientific implications of the
biases which are indicated depend
almost entirely on the magnitude of dif-
ferences which one is trying to detect
between the seasonal data.
Other Methods of Assessing
Data Quality
Three checks of overall sample data
quality are provided by comparison of
measured ANC values and calculated
carbonate alkalinity, comparison of the
total ionic charge of anions and cations,
and comparison of measured and calcu-
lated specific conductance values. Com-
parison of measured ANC values to
calculated carbonate alkalinity values
provides an indication of the reliability of
pH, DIG, and ANC measurements and an
indication of the presence of unmeasured
noncarbonate protolytes. The anion and
cation balance provides an estimate of
the internal consistency of the sample
composition. Comparison of the meas-
ured and calculated values for specific
conductivity values provides an additional
check on analytical errors in the meas-
urements or on the presence of un-
measured ionic species.
Figures in the report show plots of
measured ANC versus calculated alka-
linity. These plots show that ANC results
measured by Laboratory 1 (Spring Sea-
sonal Subsurvey) may be slightly lower
than those measured by Laboratory 2
(Fall Seasonal subsurvey). The plots also
show that the DIC-initial measurements
performed by Laboratory 1 have a high
bias. The DIG data from the processing
laboratory would provide a better esti-
mate of the DIC content of the sample.
Both the plots of the sum of anions
versus the sum of cations and of the
measured versus calculated conductiv-
ities illustrate the problems related to
chloride analyses made at Laboratory 1.
When only chloride values of less than
7.0 mg/L are included in the plots, almost
all the data points fall in close proximity
to the 1:1 line.
Based on internal consistency checks,
the results of ELS-I and the ELS-II Fall
Seasonal subsurvey appear to be com-
parable. ELS-I and the Fall Seasonal
subsurvey plots for measured ANC
versus calculated carbonate alkalinity,
sum of anions versus sum of cations, and
measured versus calculated conductivity
display similar patterns.
Conclusions and
Recommendations
Overall, the QA program was success-
ful in identifying and resolving a number
of data quality issues. The program also
was effective in assuring that the data
were of known and documented quality.
The majority of the data are of accept-
able quality and every effort was made to
correct any deficiencies. The data for
nineteen of the analytes indicate no
interlaboratory bias, and the data for six
analytes indicate only slight inter-
laboratory bias. The issue of detectability
should be considered in the context of
the routine sample values. Analyte values
in low concentration ranges cannot be
evaluated using the pre-established data
quality objectives. The accuracy and
precision of data for four analytes of
primary interest in acidic deposition
research (ANC, pH, nitrate, and sulfate)
were close to or better than the goals set
for intralaboratory performance. In a few
cases, data interpretation may be limited
by considerations of data quality in terms
of precision, accuracy and detectability.
The representativeness, completeness,
and comparability of the data are ade-
quate to meet the project objectives.
In future studies of this type, on-site
evaluations at the analytical laboratories
should be scheduled early in the process
of sample analysis and at least two
should be required. The addition of
certain requirements to laboratory con-
tracts would improve the QA program;
and special attention should be given to
the requirements for surveys with mul-
tiple components. Data quality objectives
should be developed for both the total
and analytical systems. Natural audit
samples should be well characterized for
use in a QA program.
-------
T. E. Mitchell-Hall, A. C. Neate. J. £ Pollard, and D. W. Sutton are with Lockheed
Engineering and Sciences Company, Las Vegas, NV 89119; S. G. Paulson is with
the University of Nevada, Las Vegas, NV 89119.
O. r. Heggem is the EPA Project Officer (see below).
The complete report, entitled "Eastern Lake Survey - Phase II Quality Assurance
Report," (Order No. PB 89-224 919/AS; Cost: $28.95, 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 89193-3478
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
EPA/600/S4-89/029
CHICAGO
------- |