903R91006
CBP/TRS 58/91
May 1991
Chesapeake Bay Coordinated
Split Sample Program
Implementation Guidelines
Revision 3
v^^^
Chesapeake
Bay
Program
CB 00824
-------
Chesapeake Bay Coordinated
Split Sample Program
Implementation Guidelines
Revision 3
Prepared By:
Analytical Methods and Quality Assurance Workgroup
of the Chesapeake Bay Program Monitoring Subcommittee
c/EPA
Environmental Protection Agency
Chesapeake Bay Program Office
410 Severn Avenue-Suite 110
Annapolis, Maryland 21403
(301) 267-0061
Printed by U.S. Environmental Protection Agency for the Chesapeake Bay Program
-------
TABLE OF CONTENTS
LIST OF FIGURES ill
LIST OF TABLES iii
BACKGROUND 1
PROGRAM OBJECTIVES 1
PROGRAM DESIGN CONSIDERATIONS 2
SPLIT SAMPLE PROGRAM RESPONSIBILITIES 3
Component Program Responsibilities 3
Data Management and Reporting Responsibilities 3
Coordinated Split Sample Program Oversight Responsibilities ... 4
SPLIT SAMPLE COLLECTION AND PROCESSING PROTOCOLS 4
LABORATORY SAMPLE HANDLING AND ANALYSIS PROTOCOLS 5
DATA MANAGEMENT REQUIREMENTS AND PROTOCOLS 10
Chain-of-custody form 10
Diskette submission 10
Data format and parameter names 10
Diskette formats 13
Diskette labeling . 13
Hardcopy submission . 14
Accompanying narrative 14
Data verification 14
STATISTICAL DATA ANALYSIS AND REPORTING 14
COORDINATED SPLIT SAMPLE PROGRAM IMPLEMENTATION 18
SPLIT SAMPLE COMPONENT PROGRAMS 19
Chesapeake Bay Coordinated Split Sample Program 19
Mainstern/Tidal Tributaries Component 19
Virginia Mainstem/Tributaries Component 19
Tidal Potomac River Component 23
Non-tidal Tributaries/Fall-line Component 23
REFERENCES 26
11
-------
LIST OF FIGURES
Figure 1. Sequential Field Split Sample Dispensing Order 6
Figure 2. Schematic of the Operational Flow of Analyses 8
Figure 3. Sample Chain of Custody Form 11
Figure 4. Data Submission Form 15
Figure 5. Coordinated Split Sample Program Components 20
Figure 6. Mains tern/Tidal Tributaries Component 21
Figure 7. Virginia Mains tern/Tributaries Component 22
Figure 8. Tidal Potomac River Component 24
Figure 9. Non-tidal Tributaries/Fall Line Component 25
LIST OF TABLES
Table 1. Parameters and their CBP Abbreviations, Holding Times
and Temperatures 9
Table 2. Sample data set submission format for Coordinated Split
Sample data 12
Table 3. Analysis of variance design for testing inter-
organization differences 16
Table 4. Analysis of variance design for testing splitting
effectiveness 17
111
-------
Split Sample Plan
Revision 3, 5/6/91
Page 1 of 26
BACKGROUND
In 1988, the Chesapeake Bay Program's- Monitoring Subcommittee identified
the need to assess the comparability of the water quality data produced by the
many agencies participating in the basinwide data collection programs. The
Monitoring Subcommittee's Analytical Methods and Quality Assurance Workgroup
recommended the implementation of a basinwide coordinated split sample program
to address this programmatic need. Although individual laboratories can
evaluate the performance of their own analytical operations against standard
reference materials, the most complete mechanism for the evaluation of total
sampling and analysis system variability is through the use of field split
samples. These include both field and laboratory sources of variability. The
Coordinated Split Sample Program (CSSP) was started in June 1989, following the
earlier revision of these guidelines (CBP 1989). This revision incorporates
changes and refinements based on the first two years of CSSP operation.
PROGRAM OBJECTIVES
The major objective of the Coordinated Split Sample Program is to
establish a measure of comparability between sampling and analytical
operations for water quality monitoring basinwide. A secondary objective is
to evaluate the in-matrix dilution of standard EPA reference materials. These
standard reference materials are analyzed in appropriate matrix, fresh to
saline, and concentration level to match the sample.
Continued implementation of the Coordinated Split Sample Program provides
the institutional structure to address three important program coordination
needs:
o implemention of a valid statistical approach for the evaluation of
split sample results to assure the use of these data in data quality
assessment;
o facilitation of efforts to identify problems and achieve solutions
in individual programs as revealed through the comparability
evaluation; and
o improvement of communication of split sample analytical findings
among organizations •through a central computerized data base for
split sample results.
The CSSP provides the forum and information necessary to promote an on-going
refinement of the field and laboratory techniques rather than assuming that the
system is static and never changing. The statistical assessment of the data
allows the field and laboratory personnel to improve their respective
techniques. In addition, the description of the data quality provides the
necessary information for assessment and application of the data by the
intended user. :
-------
Split Sample Plan
Revision 3, 5/6/91
Page 2 of 26
PROGRAM DESIGN CONSIDERATIONS
The Analytical Methods and Quality Assurance Workgroup determined through
a series of surveys that ,a limited number of split sampling programs were in
place. There was considerable diversity in the objectives of these split
sampling operations and, therefore, their designs varied accordingly. Prior
to the implementation of the Coordinated Split Sample Program, few programs
were able to work out a well organized method of data submission, compilation,
analysis and timely distribution to participating laboratories and agencies.
Most of the other programs use only two-way split samples and therefore do not
provide the multiple comparisons that are part of the CSSP results.
Because of the difficulties associated with measurements of waters of
varying salinities and the Monitoring Subcommittee's desire to link tidal and
nontidal monitoring programs, linkages were created between laboratories
routinely analyzing samples of. comparable salinities. Through common "third
party" laboratories, those laboratories analyzing only estuarine samples are
linked with laboratories whose analytical responsibilities are limited to
freshwater tidal/nontidal samples.
The Coordinated Split Sample Program's field split samples and
laboratory duplicate and spike samples provide an estimate of overall sampling
and analytical precision and accuracy. Since field split samples are defined
as samples divided into portions following sampling, they provide precision and
accuracy information about all steps after sample acquisition including effects
of sample splitting procedures, sample processing, storage, shipment, analysis,
and data processing. CSSP field split samples are divided from one large
sample rather than consecutive co-located samples. Therefore, they provide an
estimate of overall sampling and analytical precision and accuracy assuming
analytical comparability. Combined with routine analysis of standard reference
materials such as EPA certified materials, results from the Coordinated Split
Sample Program can be used to verify analytical comparability and provide an
independent measure of accuracy.
The actual structural design of the Coordinated Split Sample Program is
based on a series of interconnected and interrelated split sample component
programs organized around common geographical areas and similar sample
salinitiy ranges. The four Component programs are the Mainstern/Tidal
Tributaries Component, the Virginia Mainstem/Tributaries Component, the Tidal
Potomac River Component, and the Non-tidal Tributaries/Fall-line Component.
The Mains tern/Tidal Tributaries Component and Virginia Mainstem/Tributaries
Component form the central core of the Coordinated Split Sample Program,
interrelating laboratory and field operations working the Bay mainstem and
tidal tributaries. The other components build upon this network of
laboratories, linking directly with a laboratory or group of laboratories
associated with monitoring the mainstem of the Chesapeake Bay.
-------
Split Sample Plan
Revision 3, 5/6/91
Page 3 of 26
SPLIT SAMPLE PROGRAM RESPONSIBILITIES
Component Program Responsibilities
For each component of the Coordinated Split Sampling Program, one agency
has been assigned the coordination responsibility for that component's
operation. The responsibilities of that lead agency are as follows:
A. Maintain contact with all the participating field and lab organizations
to coordinate all logistics.
B. Provide for the necessary sampling equipment, sample containers, labels,
chain of custody paperwork, etc. at the time of the quarterly split sample
collection.
C. Collect the sample and prepare the splits according to the protocol
described within this document.
D. Arrange for the direct exchange, transfer or shipment of the individual
split samples to each participating laboratory within the component
program. This may be accomplished by meeting the other organizations'
crew at some mutually satisfactory point, personal delivery or by common
courier. The goal should be the most rapid, feasible means of sample
delivery. Every attempt should be made to adhere to normal holding times,
temperatures, preservatives and filtration arrangements followed routinely
by each organization. Records should be maintained of all handling
conditions and practices so that data evaluations may be facilitated.
Data Management and Reporting Responsibilities
The routine submission of split sample data is the responsibility of each
individual laboratory/agency and its in-house data management organization.
The Chesapeake Bay Liaison Office contractor for the management and operation
of the Chesapeake Bay Program Computer Center, Computer Sciences Corporation
(CSC), is the designated recipient of all data generated through the
Coordinated Split Sample Program. CSC is responsible for the processing,
routine statistical analysis, report development, and timely distribution of
results back to the participating laboratories and agencies within their
respective component programs when all the data are received from all agencies
within the individual component program.
The Chesapeake Bay Program's Monitoring Coordinator and Quality Assurance
Officer both review and evaluate the results of each split sampling component
program and consult with the appropriate individuals in each organization to
determine the appropriate response to any significant findings.
-------
Split Sample Plan
Revision 3, 5/6/91
Page 4 of 26
Coordinated Split Sample Program Oversight Responsibilities
The Chesapeake Bay Liaison Office is responsible for overall coordination
of the Coordinated Split Sample Program. The Chesapeake Bay Program's
Monitoring Coordinator assists organizations in working out any logistical
problems which cannot be resolved by the participating organizations. The
Chesapeake Bay Quality Assurance Officer helps resolve any technical concerns
which arise concerning the sampling and analysis of the split samples. The
Analytical Methods and Quality Assurance Workgroup reports regularly to the
Monitoring Subcommittee on the Program's results and any blockages to the full
implementation of the Program which are based on resource constraints.
SPLIT SAMPLE COLLECTION AND PROCESSING PROTOCOLS
For each sample collected, the sampling crew fills a single large vessel
such as a large carboy according to normal sample handling protocol. For
instance, if a submersible pump is normally used this would be employed; if a
cubitainer is normally dipped manually, this would be done. If multiple grabs
are required to collect adequate volume for the splitting, these multiple grabs
should be composited in a larger vessel prior to splitting. Field triplicate
sub-samples obtained from a "single sample" potentially reduces additional
sources of variability caused by sampling sequentially.
It is imperative that a strong mixing be applied to the large vessel to
ensure that the sample is uniformly mixed and the sub-samples will be
representative. This may be accomplished with a stir bar arrangement or other
form of mechanical mixing as long as a vortex is created. This mixing must
continue for the duration of the splitting operation. When more particulates
are present in the sample, the splitting operation will be more difficult to
accomplish a representative sample therefore, more effort must be applied to
provide a good mix. As of March 1991, the sample collection, stirring and
splitting methods used by each sampling organization (defined in the Split
Sample Component Program section) were:
MDE (Mainstem Component, at CB4.4)—Samples are collected by submersible pump
into a 15 gallon carboy; sub-samples are split on the boat from the carboy
agitated with paint stirrer turned by electric drill.
VWCB (Virginia Component, at TF5.5)—Samples are collected by submersible pump
into three 9-liter churn splitters, filled in sequential sections (a few
liters at a time in sequence 1-2-3-1-2-3, etc.) until all are filled.
Sub-samples are split on the boat from the churn splitters, one used for
sub-sample 1, one for 2, and one for 3. This deviates from the
recommended single large vessel.
DCRA (Potomac Component, at PMS-10)—Samples are collected in three 5-gallon
carboys by dipping. In the lab, these are composited into one 15-gallon
carboy, and the sub-samples are split from the large carboy'via a spigot,
while the carboy is agitated manually with a paddle.
-------
Split Sample Plan
Revision 3, 5/6/91
Page 5 of 26
USGS Towson (Fall Line Component, at CBl.O)—They have used a large churn
splitter, but did not have enough water to split three sub-samples for
each laboratory. They may get a cone splitter.
From the single large vessel of continuou5~sly well mixed sample, all four
organizations draw three sub-samples for each laboratory in rotational
succession (see Figure 1). These sub-samples were previously called "aliquots"
(CBP 1989), and the name was changed to indicate that they are split in the
field, not in the laboratory.
For example, if there are three laboratories receiving the split samples,
the sample in the large vessel would be divided into nine sub-samples in nine
bottles, three bottles for each laboratory. The first bottle for each
laboratory would be filled first, then the second and the third, resulting in
the first laboratory receiving bottles #1, #4 and #7. Data from this type of
sample collection would indicate if there was any type of bias caused by non-
uniform mixing resulting in samples that are not representative. Splitting
effectiveness is checked statistically in CSSP reports (see Data Analysis).
Each laboratory evaluates these three sub-samples as discrete samples.
This approach provides an estimate of the variability of the material in the
composite container which must be established to provide assurance that any
inter-organization variability observed is not a function of a poorly mixed
initial sample. Three sub-samples are statistically the minimum number
required to permit an effective analysis of variance. For similar reasons, the
minimum number of laboratories involved in any split sampling operation should
be three wherever possible.
Once the splitting is completed, the sampling organization processes their
three sub-samples through their normal handling and preservation procedures.
The remaining sub-samples for the other organizations are iced down until they
are delivered to the appropriate organization. Samples are delivered to the
participating laboratories as rapidly as possible. The chain of custody form
is sent with the samples (see Data Submission). Normal sample handling and
preservation methods and holding times for CBP samples are adhered to as
closely as possible. Any deviations from normal procedures should be noted on
the chain of custody form and sent to CSC.
LABORATORY SAMPLE HANDLING AND ANALYSIS PROTOCOLS
Samples should be analyzed as soon as possible after arrival at the
laboratory to minimize holding time effects. In some components (e.g.,
Virginia Component), participants have agreed to filter the samples the morning
after collection, so that all are analyzed at approximately the same time.
This would not be possible when some samples are delivered by mail (in the Fall
Line Component). Participants in each component should discuss and agree on
when samples will be delivered and analyzed.
-------
£
'8
2L
is
Q
_a>
a.
i
a>
iZ
15
1
a>
J>
"
w
a. a>
I 2
W Q.
a> o
•E8
IS
If
o- E
I8
« a>
£-"i
«= 1
.0 Q.
"O a>
§m
g'-s
2
cB
00
-------
Split Sample Plan
Revision 3, 5/6/91
Page 7 of 26
All samples are analyzed for the parameters listed in Table 1 within the
recommended Chesapeake Bay Monitoring Program holding'times. If there are
parameters in Table 1 which are not routinely measured in CBP samples by a
participating laboratory, these parameters do not need to be analyzed and
reported solely for the purposes of the CSSP, unless specific parameters are
requested for comparison to results from other laboratories in the component.
CSC computes any calculated parameters used in the reports, using the methods
outlined in D'Elia et al. (1987). Currently the calculated parameters include
Total Nitrogen (TN) for all laboratories, and Total Phosphorus (TP),
Particulate Phosphorus (PHOSP), Total Dissolved Nitrogen (TON), and Particulate
Nitrogen (PN) for those laboratories that do not calculate them directly.
A complete schematic of the operational flow of analyses is outlined in
Figure 2. Within the laboratory, at least one of the three sub-samples is
subjected to the normal quality control (QC) routine. If a duplicate and spike
sample are analyzed every tenth analysis, the sub-sample identified for QC
should be analyzed as a routine quality control sample. This lab QC sample is
divided into a duplicate and spiked with the appropriate standard. Additional
QC samples from any source can be analyzed to fit the lab QC sample frequency.
If the laboratory elects to run more than one sub-sample for additional quality
control, these data are to be submitted for evaluation as well.
To perform adequate diagnostics in the event that significant inter-
organization differences are found, it is essential that quality control data
are available for each laboratory on their system's performance with the
specific matrix under consideration in this program. It will be useful to
compare the precision and accuracy during analysis of the sub-samples (and
their matrix) against the routine precision and accuracy limits of the
laboratory over all matrices.
To supplement the analyses of the three sub-samples and the respective QC
sample, EPA standard reference material for each parameter are analyzed where
available. The analysis of standard reference materials provides a strong
measure of comparability between all laboratories and within one laboratory's
analytical system over time. Quarterly analysis of standard reference
materials is the most independent evaluation of laboratory performance
available at this time. It is a critical element of any diagnostic efforts
associated with the Coordinated Sample Split Program.
The standard reference material (SRM) should be diluted with
deionized/distilled water in all components. In the Mainstem Component, an
additional standard diluted in the appropriate concentration saline matrix
should be analyzed. The concentration of the estuarine dilution water should
be subtracted as a blank value.
-------
LARGE
VESSEL
(see Figure 1
for dispensing
order)
Figure 2. Schematic of the Operational Flow of Analyses,
Coordinated Split Sample Program
Split Sample Plan
Revision 3, 5/6/91
Page 8 of 26
Triplicate Subsamples
(Field precision estimate)
Normal Laboratory
Quality Control
Procedures
Analyze for
Routine Parameters
Analyze for
Routine Parameters
Spike Sample
Laboratory QC
Analyze for
Routine Parameters
Analyze for
Routine Parameters
Analyze for
Percent Recovery
(Lab accuracy estimate)
(Lab precision estimate)
Deionized/distilled
water dilution
EPA Standard
Reference Material
(Accuracy estimate)
Matrix water
dilution
Analyze for
SRM Parameter
Analyze for
SRM Parameter
-------
Split Sample Plan
Revision 3, 5/6/91
Page 9 of 26
Table 1. Parameters and their CBP code, holding times and temperatures
Parameter (me/1 except CHLA)
Total phosphorus as P
Total dissolved phosphorus as P
Particulate phosphorus as P
Dissolved orthophosphate as P
Total nitrogen as N
Total dissolved nitrogen as N
Particulate nitrogen as N
Total Kjeldahl nitrogen as N
Dissolved Kjeldahl nitrogen as N
Ammonium as N (filtered)
N02 + N03 as N (filtered)
Nitrite as N (filtered)
Nitrate as N (filtered)
Total organic carbon
Dissolved organic carbon
Particulate organic carbon
Particulate carbon
Silica as Si (filtered)
Total suspended solids
Chlorophyll a (ug/1)
Pheaophytin
Biological Oxygen Demand 5 day
HOLDING
CODE TIME (davs)
TP
TOP
PHOSP
P04F
TN
TON
PN
TKNV
TKNF
NH4
N023
N02
N03
TOC
DOC
POC
PC
SI
TSS
CHLA
PHEA
BODS
28
28
28
28
N/A
28
28
28
28
28
28
28
28
28
28
28
28
28
7
30
30
N/A
TEMPERATURE (°C)
-20
-20
-20
-20
N/A
-20
-20
-20
-20
-20
-20
-20
-20
-20
-20
-20
-20
4
4
-20
-20
N/A
Notes:
1. Report all parameters that are measured directly. Calculated parameters
will be calculated by CSC using the outline in D'Elia et al. 1987.
2. If there are parameters noted above which are not routinely measured by a
participating laboratory, these parameters do not need to be analyzed and
reported solely for the purposes of the Coordinated Split Sample Program. In
some cases, laboratories may be requested to analyze parameters performed by
other laboratories in the component. Please report all of the parameters
listed that you routinely analyze.
3. Please report any deviations from these maximum holding times in a narrative
accompanying the submitted data.
4. Parameters without holding times and temperature were determined by
calculating the concentration from parameters that were measured directly (per
D'Elia et al. 1987).
-------
Split Sample Plan
Revision 3, 5/6/91
Page 10 of 26
EPA standard reference materials are available through the Cooperative
Research and Development Agreement (CRADA) of the U.S. EPA Environmental
Monitoring and Support Laboratory, Cincinnati, Ohio. • These EPA certified
materials are obtained from the company awarded the specific contract by EPA.
Standards are or will be available for organic compounds, toxic and hazardous
compounds, pesticides and inorganic compounds. Any questions in regards to the
EPA standard reference materials should be directed to Claudia Walters, CBP
Quality Assurance Officer, CBLO, at (301) 267-0061 or (800) 523-2281.
DATA MANAGEMENT REQUIREMENTS AND PROTOCOLS
Chain-of-custody form
Upon sampling, the chain of custody form is filled out completely by the
sampling agency to convey the necessary information to the other participants
regarding the sampling site, time of collection, handling and any special
observations which might affect the results. A sample chain of custody form
for the Mainstem Component (Fig. 3) should be adapted for other components if
not already in use. This form should be sent with the split samples to the
laboratories, and sent to CSC with the data submission. Each laboratory should
perform the requisite analyses and report concentration values for each
determination on each sub-sample of the split sample, with the associated QC
and SRM data.
Diskette submission
Data sets are submitted to CSC as an ASCII text file on an IBM diskette,
unless other arrangements have been made with CSC. CSC has no keypunching
staff, since virtually all of the data they receive are on diskettes or tapes.
The provided Data Submission form (Fig. 4) is optional (see below), and may be
used to prepare the data for keypunching. If questions arise concerning data
submission, please contact Peter Bergstrom, CSC, at (301) 267-0061 or (800)
523-2281.
A. Data format and parameter names
The preferred data formal: is standard columnar text (ASCII), with each
variable in a separate column, and columns separated by spaces. This can be
created and edited with a word processor if it is saved as an ASCII or text
file. If you use LOTUS or dBASE please send an ASCII file (.PRN from Lotus).
Include all the variables listed above, in the same order; values for PARAM are
in Table 1. Please report values that are below detection limits as you
normally report them, and be sure they get the '<' flag. Report missing values
for numeric variables as a period. The data received to date will fit in a 132
column width. If you cannot fit your data into a 132 column width, please
submit them as two files, each less than 132 columns wide, on the same
diskette. A sample format with hypothetical data is in Table 2/. This format
is available on diskette from CSC to facilitate data entry.
-------
Figure 3.
Split Sample Plan
Revision 3, 5/7/91
page 11 of 26
COLLECTED FOR: Q P 1C
BOTTLE NUMBER: Q i . ^
MAIN BAY SPLIT SAMPLE CUSTODY LOG
COLLECTION DETAILS: DATE:
TIME: / V / fa DEPTH: Q, £
LOCATION: ftl £ vHtH SALINITY: I \ , ^ T
COMMENTS: (unusual conditions, problems, floating algae, high solids, etc.)
SPLITTING DETAILS:
COMPOSITE CONTAINER
FILLED BY:
multiple grabs
pump ^-"
other
COMPOSITE SUBSPLIT BY:^-
sequential bottles *-^"^
cone splitter
other
TRANSFER SEQUENCE:
SPLITTING SEQUENCE BOTTLE LABELLED
bottle
bottle
bottle
bottle
bottle
bottle
bottle
bottle
bottle
bottle
bottle
bottle
bottle
bottle
bottle
DATE TIME
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
MDE
VIMS
CBL
VWCB
MDE
VIMS
CBL
CODU
VWCB
MDE
VIMS
CBL
^OP.U
VWCB
BY WHOM?
- Al
- Bl
- --C1
- El
- A2
- B2
- C2
'-~E2
- A3
- B3
- C3
""- E3
TEMP. OF SAMP
composite collected & split ,'^L-lQ-^/J
Subsamples picked up / -2-~l&~7&
(circle one)
0°C 4° C ambie
Subsamples delivered to lab 17-fQ-QQ 2Q,
0°C
ambie
CV. ambie
FIELD PROCESSING INFORMATION
BOTTLE
FIELD PROCESSING DONE ON SAMPLE
DATE/TIME
BY
D I
NOTE: PLEASE SEND A COPY OF THIS COMPLETED FORM TO: PETER BKRGSTROM, CSC.
SEVERN AVENUE, SUITE 110, ANNAPpLIS, MD, 21403, (800) 523-2281 -
-------
Split Sample Plan
Revision 3, 5/6/91
Page 12 of 26
Table 2. Sample data set submission format, for Coordinated Split Sample data
STATION DATE LAB PARAM APC MDL SUEISAMP REP MUM RESULTS PCRECOV SPIKE SRM EPA SRM DE
CB5.3 890515 DCLS PHOSP
CBS.3 890515 DCLS PHOSP
CBS.3 890515 DCLS PHOSP QQ
CBS.3 890515 DCLS PHOSP
CBS.3 890515 DCLS TKNW
CBS.3 890515 DCLS TKNW
CBS.3 890515 DCLS TKNW
CBS.3 890515 DCLS TKNW AA
CBS.3 890515 DCLS SI
CBS.3 890515 DCLS SI
CBS.3 890515 DCLS SI
CBS.3 890515 DCLS SI
1
2
3
1
1
2
3
1
1
2
3
1
2
1
1
1
2
1
1
1
2
1
1
0 .010
0 .011
0.009
0 .013
0 .754
0.712
0 .657
0 .774
2.42
2.44
2.17
2.39
106
.
t
92
105
.
.
Ill
0 .023
.
t
1.11
1.56
.
.
3.45
,
0.016
0.645
1.65
0.024
SRM_MA
0.019
0.623 0.554
1 .54
1.62
The parameter names to use and their meanings are:
STATION is the sampling station number (CBS.3, PMS-10, etc.)
DATE is the sampling date, as YTMMDD
LAB is the abbreviation for the analysis laboratory (DCLS, OWML,
etc.), not the collecting agency
PARAM is the CBP abbreviation for the parameter (see revised Table
1). Please do not use other abbreviations.
APC (Analysis problem code): One or two letters, from Table 20 of
the CBP Data Management Plan for Water Quality Data (latest
version is Revision 2, July 1990).
MDL (Method Detection Limit): For values below the Method Detection
limit record '<'. For samples requiring a dilution record '>'
(list the dilution factor in the comments section or narrative).
Otherwise leave this blank.
SUBSAMP is the sub-sample number (1, 2, or 3). Previously this variable
was called 'ALQ' and included the lab replicate designation, but
the codes used by different submitters have been so diverse that
identification of the replicates was uncertain. In the future,
indicate ONLY the field replicate or sub-sample number here, and
show the lab replicate number with the following new variable;
REP_NUM is being requested for the first time in CSSP submissions. It
indicates the LAB replicate number (usually 1 or 2, sometimes
3, A, or 5) just as it does in other CBP submissions. This
needs to be a separate number for data analysis.-
-------
Split Sample Plan
Revision 3, 5/6/91
Page 13 of 26
RESULTS is the concentration for that sub-sample. Report the results
the way you normally repor-t them to thfe CBP (corrected for
dilution if necessary).
PCRECOV is the percent recovery from the spiked sample. List it in the
row of the sub-sample that was spiked. This can be done for
more than one sub-sample if desired; please report all spikes
run.
SPIKE is the amount of spike added. This value will be added to the
RESULTS in that row to get the theoretical total concentration
used for-calculation of percent recovery.
SRM_EPA is the EPA value for the Standard Reference Material. The SRM
results can be put in any row for that parameter. Please attach
a copy of the "Answer Sheet" that came from the EPA with each
standard.
SRM_DE is the lab value for that SRM, diluted in deionized water.
Please list the dilution used in the Comments section, or in
your narrative.
SRM_MA is the lab value (if done) for the SRM diluted in lowest
concentration saline matrix.
B. Diskette formats
CSC/CBLO staff are currently able to read the following diskette formats:
IBM: 5.25", 360KB (PC/XT) and 1.2MB (AT); OR 3.5", 720KB and 1.44MB
Macintosh; 3.5", 800KB
C. Diskette labeling
All diskettes submitted must be labeled in the following manner (based on
page 4-4 of the Chesapeake Bay Program's Data Management Plan for Water Quality
Data):
a. Data format used and the software and version number used for data
storage
b. Creation date
c. Submitting individual, organization, and telephone number
d. Names and a brief description of all files on the diskette
Please contact Peter Bergstrom of CSC at (301) 267-0061 (or toll free (800)
523-2281 or FTS 691-6873) if you have any questions about data submission.
-------
Split Sample Plan
Revision 3, 5/6/91
Page 14 of 26
Hardcopy submission
A hardcopy submission is optional. Use the data submission form provided
(Fig. 4). Data entry errors will be minimized if each laboratory puts their
own data on the standard form. Please double check all decimal points and all
numbers for clarity, and put decimal points in a separate box.
Accompanying narrative
Please attach to the data sheets a narrative detailing the methods and
procedures followed in each step of the analysis process. Describe how the
sample was handled before and after you received it and any sample preparations
including any digestions, preservatives added or filters used. Give details
of each analytical method used (include method code if known), any dilutions
used, and any unusual conditions or problems. This information will be very
useful if differences are found in the results. This narrative can be included
on the diskette (if used) as an ASCII text file if desired.
Once the narrative has been sent, future submissions only need to include
any changes in procedures. Please indicate whether the changes are temporary
or permanent.
Please attach copies of the original lab sheets if there are any ambiguous
parameters or other features of the data that may need clarification. These
are currently received from MDHMH, and have been requested from DCLS.
Data verification
CSC/CBLO staff will upload the submitted data to the VAX computer and
combine the data for each component into a single data set. Since errors can
be introduced in this process, each laboratory should verify that their data
are correct in the combined data set. Normally a printout with a Data Set
Checklist will be sent with each Interim Report, requesting verification of the
new data in the report. The graphs and analyses in the report will help
identify any outliers in the data. The data should be checked against the
original lab sheets, and any changes sent to CSC before the Annual Report is
produced. Any data submitted between the Interim Report and the Annual Report
will have to be .verified before the Annual Report is finalized. Although data
verification is tedious work, the use of the data requires that the numbers be
as correct as possible.
STATISTICAL DATA ANALYSIS AND REPORTING
The data analysis and reporting scheme was developed by Peter Bergstrom
in consultation with members of the Analytical Methods and Quality Assurance
Workgroup (AMQAW) during 1990. The current approach uses two main tools:
graphs of the data with "precision bars," and two-way analysis of variance.
-------
page 15 of 2b
^_
o —
•o
CM
• 0>
O) CNJ
o TT
0. 5
°- o
LLJ
>
_
^*
^
Q£
_ 0
^™ i*^
a 2l tt
J*f LLJ C^
O — ' LU
Z£ Q_
QL ~ f
Q. 3> *"
" •*. ^^ y -J
-d" x~ ^ i^
^^
•
CO
o.
JT
^J
P
~^~
O
O
o
?i
CO
..J
Ou
^j
^H
CO
^
2
£*
a
Ul
LU
D.
CO
LU
h™
• -
•™
CO
^
M)
z
in ^.^
Z.
X
..
g
E
o
*-
>•
D
fid
"c" ^
<5 o
5 « p
s'x-S
Ge £ C
?i
v*'*-'
-
Q-
°|T3
.U
B
f
< 5
5 -^
» ^
x
&
UJ
^
&5
>
5
a
P
O) w
£•0
S *
X' «
8
t£
csu,^r35
<-0
So--
0EI
oT^O
^ TfT
<^-S
^ ^
oi *^^
^>
^
0
r-
CM
CM
CM
"O
CM
0
-
-.
j
j
J
I
1
. .
t?
2?
LU
^.
^
^£
O
U
1
1
1
1-
1"
CN
1—
^—
CM
CM
CM
-
CM
1
t?
^_
LU
^>
^
«^
O
u -
•~
•—
CN
'-
CM
CM
r>
CM
.
c/
^
LI
^
c
-------
Split Sample Plan
Revision 3, 5/6/91
Page 16 of 26
Graphs of the split sample results show which differences were larger than
the within-laboratory precision. Based on a discussion with AMQAW members on
A/24/90, within-laboratory precision for th.e graphs is estimated by the larger
of: 1) the Method Detection Limit (MDL); or, 2) the standard deviation of the
three sub-samples for each sample, which estimates analytical and field
precision. In most cases, the MDL is larger than the field precision so the
MDL usually sets the precision limit. Graphs of the cruise means for each
laboratory show this estimate as "precision bars." Any laboratory means with
non-overlapping precision bars have differences that are larger than within-
laboratory precision.
The Friedman two-way non-parametric repeated measures analysis of variance
(ANOVA) with replication within blocks (Marascuilo and McSweeney 1977) is used
to test statistically for differences among results from different
organizations. In this design, the organizations are treatments, the sampling
dates are blocks, and the three sub-samples are replicates within blocks. The
null hypothesis is that there are no consistent differences among the
concentrations measured by the different organizations. The ANOVA design is
shown in Table 3.
Table 3. Analysis of variance design for testing inter-organization
differences, Coordinated Split Sample Program.
TREATMENTS (Organizations)
(Potomac Component shown)
CRL/DCRA DCLS MDHMH
BLOCKS
(Sampling
dates)
March |
1
June |
1
September |
1
December |
1
1
1
1
1
2
2
2
2
3* |
3 1
1
3 1
3 1
1
1
1
1
1
2
2
2
2
3 1
3 1
1
3 1
1
3 1
|
1
1
1
1
2
2
2
2
3
3
3
3
*Replicates within
blocks (sub-samples)
The Friedman program used before (Bergstrom 1990) did not allow for
replicates, so cruise means had to be used in previous reports. This change
to using replicate data uses more information in the data and increases the
power of the test, or its ability to detect real differences. The test is done
with a computer program written by Peter Bergstrom in SAS using .the formula in
Marascuilo and McSweeney (1977), including their formula for post hoc pairwise
-------
Split Sample Plan
Revision 3, 5/6/91
Page 17 of 26
comparisons. The program was tested with the example in Marascuilo and
McSweeney (1977).
The Friedman test is also used for a preliminary test of splitting
effectiveness. This is done to check whether the results from one sub-sample
were consistently higher or lower than the results from other sub-samples.
This would most likely occur if the later sub-samples, drawn from lower in the
carboy, had more sediment in them or if some of the sample bottles are
contaminated. This test is done separately for each parameter and sampling
date, using the three sub-samples as treatments and the organizations as
blocks. The null hypothesis is that there are no consistent differences among
the concentrations of the three sub-samples. The ANOVA design is shown in
Table 4. If the results do not show any differences caused by splitting, the
analysis for inter-organization differences is done. If there are significant
effects of splitting, the data must be examined to see if these effects will
bias the tests of inter-organization differences.
Table 4. Analysis of variance design for testing splitting effectiveness,
Coordinated Split Sample Program.
TREATMENTS (Sub-samples)
1 23
Blocks
(Organi-
zations)
DCLS
HRSD
ODU
VIMS
1
1
1
1
1 2
2
2
2
3 1
3 1
3 1
3 1
(Note: No replicates within blocks.)
Statistical significance is assumed when P < 0.01 (Bergstrom 1990).
Standard quality control procedures use the P = 0.01 level as the "control" or
action level for precision and accuracy charts (e.g., Montgomery 1985). The
P < 0.01 standard is now attainable for most of the parameters most components,
due to larger sample sizes and the use of replicates in the Friedman test.
Accuracy data, from percent recoveries and Standard Reference Material
(SRM) analysis, are included in tables for each report, but are not the subject
of statistical tests. These data are used to supplement the split sample
results, and for diagnostic purposes if the split sample results show inter-
organization differences that should be investigated. Confidence limits for
SRMs (provided by EPA) are included only when the standards were analyzed
-------
Split Sample Plan
Revision 3, 5/6/91
Page 18 of 26
without dilution, since the confidence limits do not apply to diluted samples.
SRMs analyzed after dilution are reported as the diluted value, with the
expected value based on the dilution used.-
Variability data are reported for within-organization and inter-
organization variability, as the standard deviations and coefficients of
variation (CV). Within-organisiation variability comes from the variability of
the triplicate sub-samples (field replicates) for each sampling date, using
REP_NUM=1 if there were laboratory replicates done, using the MEANS procedure
in SAS. Variability of laboratory replicates is used for diagnostic purposes
only, since it includes fewer sources of variability than the field replicates
(sub-samples). Inter-organization variability is calculated among the mean
concentrations of the three sub-samples reported by each laboratory for each
sampling date, using the MEAN and CV functions is SAS. All the variability
estimates are reported in one table to facilitate comparisons.
Each quarterly meeting of AMQAW includes an update on the latest CSSP
results by CSC/CBLO staff. In addition, CSC/CBLO staff contact data submitters
as soon as possible after receipt of the CSSP data if there appear to be any
errors or outliers in the data or other problems with the data.
Written results of the statistical analysis of the split sample data are
routinely distributed to agencies and laboratories within the individual
program components. These are called Interim Reports. The original quarterly
reporting schedule has proved impractical due to slow submission of data and
the time needed for report preparation. The current schedule is one Interim
Report per year per component, including data from two or three sampling dates
from that year. It may be possible to prepare two Interim Reports a year per
component if data submission and reporting are streamlined. An Annual Report
is prepared summarizing the results of all components and any responses to the
Interim Reports (Bergstrom 1990). The Annual Report is approved by AMQAW,
distributed to all participating agencies and laboratories, and formally
presented to the Monitoring Subcommittee.
COORDINATED SPLIT SAMPLE PROGRAM IMPLEMENTATION
Implementation of the Coordinated Split Sample Program is an ongoing
process, and the program requirements change slightly over time. Since the
program's inception in June 1989, most organizations have made changes in their
protocols to achieve full implementation. The few remaining gaps in full
implementation are the result of problems with obtaining splitting equipment
or Standard Reference Materials. Hopefully these will be remedied in the near
future.
Samples are split quarterly. Discussions of CSSP implementation and
statistical analysis of the data occur during the quarterly AMQAW meetings
attended by the coordinating agency, field members and lab representatives for
each component and the CBP Quality Assurance Officer and the OBP Monitoring
Coordinator.
-------
Split Sample Plan
Revision 3, 5/6/91
Page 19 of 26
SPLIT SAMPLE COMPONENT PROGRAMS
Chesapeake Bay Coordinated Split Sample Program (See Figure 5)
Coordinating Agency: U.S. EPA Chesapeake Bay Liaison Office
Coordinated Split Sample Program Coordinator: Joe Macknis (800) 523-2281
Technical Program Coordinator: Claudia Walters (800) 523-2281
Data Management Coordinator: Peter Bergstrom, CSC (800) 523-2281
Statistical Analysis Coordinator: Peter Bergstrom, CSC (800) 523-2281
The overall Coordinated Split Sample Program will be coordinated through the
Chesapeake Bay Liaison Office (CBLO). Computer Sciences Corporation (CSC)
staff at CBLO will have the major responsibility for centralized data
management, statistical analysis and reporting on behalf of all the
participating agencies and laboratories.
Mainstem/Tidal Tributaries Component (See Figure 6)
Coordinating and Field Sampling Agency: Maryland Department of the Environment
Component Program Coordinator: Bruce Michael
Contact Persons:
Robert Magnien, MDE (301) 631-3680 (MD MONITORING PROGRAMS)
Bruce Michael, MDE (301) 631-3680 (MD MAINSTEM FIELD)
Carolyn Keefe, CBL (301) 326-4281 (MD MAINSTEM LAB)
Sally Bowen, MDE (301) 974-3238 (MD TRIBUTARY FIELD)
Alvin Sober, MD DHMH (301) 225-6200 (MD TRIBUTARY LAB)
Frederick Hoffman, WCB (804) 367-6683 (VA MONITORING PROGRAMS/VA TRIE FIELD)
Betty Salley, VIMS (804) 642-7213 (VA UPPER MAINSTEM FIELD AND LAB)
Steve Sokolowski, ODU (804) 683-4524 (VA LOWER MAINSTEM FIELD AND LAB)
Norma Roadcap, DCLS (804) 786-4853 (VA TRIBUTARY NUTRIENT LAB)
Robert Potts, DCLS (804) 786-4826 (VA TRIBUTARY CARBON/TSS/BOD LAB)
The Mainstem/Tidal Tributaries Component forms the central core of the
Coordinated Split Sample Program, interrelating laboratory and field operations
working the Bay tidal mainstem and tributaries. It is the only component that
analyzes saline samples. Sampling is performed by a field crew from MDE,
currently at Station CB4.4, and before June 1990 at Station CBS.3.
Virginia Hainstem/Tributaries Component (Figure 7)
Coordinating and Field Sampling Agency: Virginia Water Control Board
Component program Coordinator: Frederick Hoffman
Frederick Hoffman, WCB (804) 367-6683 (VA MONITORING PROGRAMS/VA TRIE FIELD)
Betty Salley, VIMS (804) 642-7213 (VA UPPER MAINSTEM FIELD AND LAB)
Steve Sokolowski, ODU (804) 683-4524 (VA LOWER MAINSTEM FIELD AND LAB)
-------
JB
ID in
CO
CM
to §>
W DC 0.
41
I
O
U
2
2
a,
V
"a
ea
a
•a
"eS
•3
s
- - "S
|||
o is 03
CL .9 o
ssi-
Q) co £
:— c: O
m
o
ID
-------
S2 9?
°-!x«>
& IO CM
9-co o
W-2 w
w W m
S > O)
Q. O) TO
w tr Q.
NC
a
6JD
1
UJ
O
o^n
o I—I
-------
J5
a> m CM
-co'B
•= > 0>
CO CC D!
C/)
w c
8.
O 0
W CO
CC CC
I X
I
D
-------
Split Sample Plan
Revision 3, 5/6/91
Page 23 of 26
Norma Roadcap, DCLS (804) 786-4853 (VA TRIBUTARY NUTRIENT LAB)
Robert Potts, DCLS (804) 786-4826 (VA TRIBUTARY CARBON/TSS/BOD LAB)
Drew Francis, HRSD (804) 460-2261 (HRSD FIELD AND LAB) '
The Virginia Mainstem/Tributaries Component links the field agencies and
laboratories involved in sampling the Virginia Chesapeake Bay mainstem and
tributaries, building on the existing, routine split sampling program between
the Virginia Institute of Marine Science and the Old Dominion University.
Sampling is done by a field crew from VWCB at Station TF5.5, in the James River
at Hopewell, VA.
Tidal Potomac River Component (Figure 8)
Coordinating Agency: Metropolitan Washington Council of Governments
Component Program Coordinator: Tom An
Field Sampling Agency: DC Department of Consumer and Regulatory Affairs
Contact Persons:
Tom An, MVCOG (202) 962-3366 (POTOMAC COORDINATED MONITORING PROGRAM)
Morris Hennesy, MDE (301) 974-3677 (MD TRIBUTARY PROGRAM)
Sally Bowen, MDE (301) 974-3677 (MD TRIBUTARY FIELD)
Alvin Bober, MD DHMH (301) 225-6200 (MD TRIBUTARY LAB)
Hamid Karimi, DC DCRA (202) 404-1120 (DC MONITORING PROGRAMS)
Sheila Besse, DC DCRA (202) 404-1120 (DC FIELD)
Al Robertson, DC DCRA/CRL (301) 266-9180 (DC LAB)
Frederick Hoffman, VWCB (804) 367-6683 (VA MONITORING PROGRAMS)
Jeff Talbott, VWCB/NRO (703) 490-7352 (VA TRIBUTARY FIELD)
Norma Roadcap, DCLS (804) 786-4853 (VA TRIBUTARY NUTRIENT LAB)
Robert Potts, DCLS (804) 786-4826 (VA TRIBUTARY CARBON/TSS/BOD LAB)
Building upon the existing Potomac Regional Monitoring Program's co-located
split sampling program, the laboratories and field agencies involved in
sampling the Potomac River are incorporated into the Baywide split sample
program through the Tidal Potomac River Component. The samples are collected
by a field crew from DCRA at Station PMS-10, at Key Bridge.
Non-tidal Tributaries/Fall-line Component (Figure 9)
Coordinating and Field Sampling Agency: USGS Mid-Atlantic Regional Office
Component Program Coordinator: Joel Blomquist
Contact Persons:
Dwayne Womer, PA DER (717) 787-9637 (PA PROGRAM)
Ken Walizer, PA DER (717) 787-8184 (PA FIELD)
Lynn Schaffer, PA DER (717) 783-1998 (PA LAB)
Bruce Michael, MDE (301) 631-3680 (MD FALL-LINE PROGRAM)
Linda Zynjuk, USGS Towson (301) 828-1535 (MD FALL-LINE FIELD) -
Joel Blomquist, USGS Towson (301) 828-1535 (MD FALL-LINE LAB)
-------
in
to
g-to o
E c Tt
~ > 0)
Q. 03 CD
CO DC Q.
2
S
a.
a
a.
OJ
•o
4»
•S *•*
2 B
C « 03
£*&
W° TJ
£2 S
e|i
£53 »
o>r g
p re O
£i*
its
Lu < •
w
00
Ss
> Q
-------
_0>
CM
.
.52 o>
= > o>
Q. 0> TO
W DC D.
§
a
OJD
-------
Split Sample Plan
Revision 3, 5/6/91
Page 26 of 26
Alvin Bober, MD DHMH (301) 225-6200 (MD TRIBUTARY LAB)
Tom Grizzard, OWML (703) 361-5606 (MD POTOMAC FALL-LINE PROGRAM)
Harold Post, OWML (703) 361-5606 (MD POTOMAC FALL-LINE FIELD)
David Sirois, OWML (703) 361-5606 (MD POTOMAC FALL-LINE LAB)
Frederick Hoffman, VWCB (804) 367-6683 (VA MONITORING PROGRAMS/TRIB FIELD)
Norma Roadcap, DCLS (804) 786-4853 (VA TRIBUTARY NUTRIENT LAB)
Robert Potts, DCLS (804) 786-4826 (VA TRIBUTARY CARBON/TSS/BOD LAB)
Donna Belval, USGS Richmond (804) 771-2427 (VA FALL-LINE FIELD/LAB)
The Non-tidal Tributaries/Fall-line Component links those field agencies and
laboratories involved in sampling the fall line stations of all the Chesapeake
Bay tributaries. The Susquehanna River Basin Commission's monitoring programs
are also linked to this component via the Pennsylvania Department of
Environmental Resources laboratory. All samples are collected by a field crew
from USGS Towson at the Susquehanna River fall line station (CB1.0) at
Conowingo, MD.
REFERENCES
Bergstrom, P. 1990. Chesapeake Bay Coordinated Split Sample Program Annual
Report, 1989. CBP/TRS 51/90, EPA Chesapeake Bay Program, Annapolis, MD.
Chesapeake Bay Program. 1989. Chesapeake Bay Coordinated Split Sample Program
Implementation Guidelines, Revision 2. EPA Chesapeake Bay Program,
Annapolis, MD.
Chesapeake Bay Program. 1990. Data Management Plan for Water quality Data.
EPA Chesapeake Bay Program, Annapolis, MD.
D'Elia, C., R. Magnien, C. Zimmermann, P. Vaas, N. Kaumeyer, C. Keefe, D. Shaw,
and K. Wood. 1987. Nitrogen and phosphorus determinations in estuarine
waters: A comparison of methods used in Chesapeake Bay Monitoring.
CBP/TRS 7/87, EPA Chesapeake Bay Program, Annapolis, MD.
Marascuilo, L. A., and M. McSweeney. 1977. Nonparametric and distribution-
free methods for the social sciences. Brooks/Cole Publishing Co.,
Monterey, CA.
Montgomery, D. C. 1985. Introduction to statistical quality control. Wiley
& Sons, NY.
-------
OS
(U ••=;
(ft c
>.~ i>
ro S F
CD Q. £
ro *>
v -S
Q. c
CO -=:
O
(N
03
O
O
m
o
Q
Q
------- |