In Reply Refer To:                              September 8, 1981
EGS-Mail Stop 412



QUALITY OF WATER BRANCH TECHNICAL MEMORANDUM NO. 81.22

Subject: PROGRAMS AND PLANS--WATSTORE, Remark Coded Data

Quality of Water Branch Technical Memorandum 81.18 described 
modifications to be made to the water-quality data stored in 
WATSTORE. The principal effect of these changes will be a greatly 
increased number of data values being qualified by the 'less than' 
(<) remark code. The purposes of this memorandum are to define the 
'less than' and ND remarks, review the implications of remark 
codes, and outline the options available to the user relative to 
WATSTORE, SAS, and statistical procedures in general.

The ND, "not detected", remark means that an analysis was made, 
that the detection limit of the analytical technique was unknown 
and that the constituent was not detected. The data value stored 
in such a record is 0.0. This value has no quantitative validity 
and is merely a holding place in storage until the detection limit 
of the analytical technique is determined at which time the ND 
remark can be replaced with a "less than" remark and the value 0.0 
replaced with the appropriate value for the detection limit.

The <, "less than" remark means that the analysis was made, that 
the detection limit was known, and, that any readout from the 
analytical device was within two standard deviation units of 
background noise, meaning that there was not sufficient confidence 
in the readout to assign a real, numerical value to the 
concentration.

In any given data-analysis application, the decision to use or not 
use remark-coded data must be based on the statistical analysis 
method being used. Furthermore, for a number of chemical 
constituents, the analytical results stored in the Water Quality 
File were obtained using several different methods, each of which 
had a different detection limit. The presence of multiple 
detection limits for a given constituent presents additional 
challenges to interpretive analysis. For difficult data-analysis 
situations such as these, the analyst should consult a recognized 
statistical expert early in the planning stage.

The remark codes presently available for use in WATSTORE are 
described in Volume 3, Chapter II, Section A, Page 16 of the 
WATSTORE manual. Each parameter in a record of the Water-Quality 
File is described by three data fields as follows:

BYTE POSITION            DATA TYPE               IDENTIFIER

137-140                  Fixed bin (31)          Parameter code
141-145                  Float (15)              Value
140-149                  Char (4)                Remark

When a data value is qualified by a remark code, the value itself 
is not modified in any way; only the four byte field assigned to 
the remark code is filled. Therefore, unless remarked data are 
specifically excluded from a retrieval, the value in the data 
field would be passed to an application program; that is ~ <200 
becomes 200 in the application program. Analysis of a data set 
containing such data can lead to serious interpretive errors.

The WATSTORE retrieval program (E771) described in Volume 3, 
Chapter III, Section A of the WATSTORE manual has two options for 
the exclusion of remarked data from the output data set.

A. Code a nonblank character in column 24 of the "M" card to 
exclude an entire analysis from the output data set if any 
parameter value in the analysis is qualified by a remark.

B. Code a nonblank character in column 25 of the 'M' card to 
exclude a given parameter code, value, and remark from the output 
data set if that parameter value is qualified by a remark. All 
other parameters in the analysis will be written to the output 
data set.

A SAS user routine (a "macro") written by WRD called QWSAS, 
described in Volume 3, Chapter IV, Section P of the WATSTORE 
manual, converts data retrieved from the QW File in backfile 
format to a SAS data set. Each record in a QWSAS SAS data set, as 
in the QW file backfile format, contains all the parameters for a 
given analysis. In the SAS records each parameter code is 
converted to a label for the parameter in question and the remark 
code, at the option of the user, can be made into an additional 
variable in the record. The remark code variable can then be used 
in a SAS DATA statement to either exclude or otherwise qualify a 
particular data value for use in subsequent processing steps.

Another SAS macro, QWINPUT, available for use with data retrieved 
from the Water-Quality File creates a data set with a slightly 
different format than that created by QWSAS. QWINPUT produces as 
many records as there are data values in the Quality of Water file 
output data set. Each record contains a single parameter value 
(labeled VALUE), a parameter code (labeled PARMCODE), and a remark 
code (labeled REMARK). As in QWSAS the remark code variable can be 
used in a SAS DATA statement to exclude or qualify the parameter 
value for use in subsequent processing steps.

Methods for using remark-coded data have been the subject of 
widespread discussion within the Division and a number are being 
tested at the present time. We welcome suggestions, questions, and 
comments on this subject and will update current information as 
appropriate.



                       R. J. Pickering
                       Chief, Quality of Water Branch

This memo supplements Quality of Water Branch Technical Memorandum 
No. 81.18.

Key Words: WATSTORE, Remark codes, Statistics, SAS, Water Quality

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