QUALITY OF WATER BRANCH TECHNICAL MEMORANDUM NO. 81.18

                                                   June 29, 1981



QUALITY OF WATER BRANCH TECHNICAL MEMORANDUM NO. 81.18

Subject: PROGRAMS AND PLANS--Data Handling, corrections to the 
                             WATSTORE Water-Quality File


         File Action Will Be Taken July 31-Aug. 3, 1981

As noted in my memorandum of April 24, 1981, on the same subject 
to Regional Hydrologists and District Chiefs, the Analytical 
Services Coordinator has identified in the WATSTORE Water-Quality 
File a problem that has evolved over the years as a result of 
changes in the precision and detection limits of the analytical 
procedures used in the Central Laboratories and the manner of 
reporting some analytical results in one of the Central 
Laboratories.

The investigation that uncovered the problem was prompted by 
recent queries from the Environmental Protection Agency (EPA) and 
the President's Council on Environmental Quality (CEQ) on the 
significance of 1973-79 NASQAN trace metal data they had retrieved 
from STORET. The questions pertained to water-suspended sediment 
("whole water") samples analyzed for total cadmium, copper, lead, 
mercury, and silver. For example, an examination of the data sets 
in question revealed that for lead at a particular site, the 
analytical results for 1973-1977 contained approximately equal 
numbers of less than 100 and 100 ug/L values, while the analytical results 
for 1978-79, after detection limits had been lowered through new 
methodology, contained roughly equal numbers of O and 1 ug/L 
values.

This situation was brought about by a practice started in the Salt 
Lake Central Laboratory in which an analyst, when reporting 
analytical results at or below the detection limit by atomic-
absorption methods, would enter a 100 or less than 100 (or a 1 or O) 
depending on whether the instrument reading was above or below the 
point on the instrument scale designated as the limit of 
detection, rather than rerunning the sample using a more sensitive 
method with a lower detection level.

The investigation also revealed a problem with the existing 
quality of Water Branch policy concerning the convention to be 
used to indicate detection limits and precision of water-quality 
data. In 1972, the Branch adopted a policy of using a fixed number 
of "significant zeros" to specify both the precision and the lower 
limit of detection of each water-quality parameter in WATSTORE. 
That convention was valid as long as the detection limit for a 
given constituent was essentially the same for all types of 
samples analyzed and analytical procedures used.  Due to 
improvements in methods and instrumentation over the last few 
years a variety of analytical methods, each with different 
detection limits, can be used for analysis of the same 
constituent.  Moreover, the chemical matrix of the sample affects 
the detection limit for a number of determinations.  Because of 
these two considerations, the use of "less thans" rather than 
"significant zeros" is now considered more useful and appropriate.

The Automatic Data Section (ADS), working closely with the Quality 
of Water Branch, will correct the problems described above by a 
mass change of data stored in the Water-Quality File to be 
conducted simultaneously with the routine merge of the current 
file and back file on the weekend of July 30-Aug. 3, 1981.  The 
changes to be made are based on Central Laboratories System 
criteria and are listed in Tables 1, 2, and 3.  The changes will be
applied to all data for samples collected during the period 
October 1, 1972 through September 30, 1978, regardless of 
analyzing laboratory.  Districts will be responsible for 
reentering any cooperator laboratory data changed inappropriately 
by the ADS procedure.  Central laboratories System criteria may 
very well be appropriate for cooperator and District laboratory 
data in many cases, thus minimizing the updates to be done by the 
District, but the Districts should verify this and carefully 
document such criteria prior to updating the file.  The choice of 
different criteria than those listed in Tables 1-3 for data from a 
given District or cooperator laboratory should be based on that 
laboratory's performance on reference samples, round-robin 
samples, or other quality assurance documentation.

Action must be taken in the following sequence.  First, Districts 
will identify all data from District and cooperator labs for which 
changes as specified in the attached tables may not apply and 
retrieve those data utilizing the retrieval list option (non-blank 
character in column 44 of the "M" card) in the retrieval deck to 
make a listing of the data.  Use of the A534 Water Quality Tables 
program for this listing could be prohibitively expensive and 
should not be necessary.  Next, ADS will make the wholesale 
changes as described.  Finally, the Districts will make another 
retrieval and listing of the same data retrieved in step 1, 
compare the two listings, and make appropriate updates.

Many people have objected to the fact that the ND, "not detected", 
remark will be applied to some data in the corrected file.  For 
clarification, the ND remark is only used to indicate that an 
analysis was made, that the detection limit of the technique was 
unknown, and that the constituent was not detected.  The remark 
provides no quantitative information and is merely holding a place 
in storage until definitive information on the analytical 
technique becomes available, at which time the remark can be 
replaced with a "less-than" value.

District offices are requested to replace any ND remarks in the 
changed file with "less-than" detection limit values wherever 
sufficient information is available from reference samples or 
other inter-laboratory calibration activities of the District or 
cooperator laboratory to enable a determination of the detection 
limit of the analysis.  Advice and assistance in this work will be 
provided by the Branch as needed.

The Branch plans to pursue the detection-limit problem for the 
period prior to 1973. An attempt will be made to determine the 
detection limits from documented performance on Standard Reference 
Water Samples. The exercise will have to be performed on a lab-by-
lab, constituent-by-constituent basis and may take considerable 
time to complete. District advice regarding this effort will be 
welcomed.

                             R J. Pickering 
                             Chief, Quality of Water Branch

Enclosure

WRD Distribution: A, B, S, FO, PO

Key Words: Water quality, data handling, WATSTORE, 
           trace metal data

This memorandum describes action planned as a result of responses 
to the memorandum of April 24, 1981 on the same subject from the 
Branch Chief to the Regional Hydrologists and District Chiefs.