Producing Tables of Rounded Water-Quality Data from NWIS

In Reply Refer To:
Mail Stop 412

                           MEMORANDUM

                                             October 27, 2000

To:       All WRD Employees
From:     Janice R. Ward, Acting, Chief, Office of Water Quality
Subject:  Producing Tables of Rounded Water-Quality Data from NWIS

Modifications to the Parameter Code Dictionary (PCD) will be provided with 
the release of NWIS 3_2 Patch 6 in November 2000.  These modifications are 
designed to help with the production of annual-data reports.  The Office 
of Water Quality recommends that most water-quality tables produced for 
publication be prepared using default (PCD-based) rounding after 
installation of the revised PCD.

Background:

Retrieving tables of rounded water-quality data from NWIS has been 
problematic since April 1997 when the National Water Quality Laboratory 
(NWQL) began to send unrounded values with a 'precision' code for 
rounding.  The problems have ranged from inconsistent precision codes 
supplied with the lab values to incorrect precision codes inserted by the 
NWIS software with the field values.  Quite often, users have had to 
manually edit the publication tables to correct the confusing and 
erroneous output.

A complete overhaul of rounding procedures that involves changes in 
processes at NWQL and software changes in both the lab and NWIS is planned 
for NWIS 5_1.  The release of 5_1 will not be earlier than Spring 2002. 
Until the release of NWIS 5_1, an interim rounding process is recommended.

Interim Rounding Process:

The water-quality-user group (Phoenix group) has developed an interim 
process for NWIS to help with the annual report production and general 
data requests until the planned overhaul can be implemented.  The 
recommendation is to use the default option for rounding (uses the PCD for 
rounding information) for the annual report production unless the District 
is using many unusual analytical schedules for data analysis.  The output 
from this process should require less editing.  However, if default 
rounding causes inappropriate precision for the selected data, then use 
the user rounding option. The changes in NWIS to support the interim 
process are:

·       Modifications to the PCD to make a single rounding array work for 
values 
from multiple methods:  In the revised PCD, the rounding arrays will be 
updated to describe the 
precision of the most commonly used analytical method.  Pete Rogerson has 
provided most of the changes for the constituents analyzed by the NWQL. 
Jon Scott has provided changes to repair systematic errors in the rounding 
array.  A description of the logic used by both Pete and Jon and the 
changed rounding arrays will be made available on the Web when the new PCD 
is released.  The revised PCD will be provided in the NWIS 3_2 Patch 6 
release. 

·       Modifications to the rounding software to manage very small values 
that 
round to zero:  Most very small values are estimated results produced from 
methods that 
can provide data below the method reporting level.  Use of the PCD for 
rounding will often round these values to zero. Uncensored, non-zero 
values (including estimated values) that round to zero will be replaced 
with a 'M' remark code and no value in the output.  The 'M' remark code is 
defined as 'presence of material verified, but not quantified'.  More 
details about this software change and how less-than values and negative 
values that round to zero will be treated can be found at the following 
url: http://ok.water.usgs.gov/nawqa/phoenix/www/roundzero.html
This software change was provided in the NWIS 3_2 Patch 5 released during 
the summer of 2000. 

The changes to the PCD rounding arrays and the rounding software for very 
small numbers are designed to produce better publication tables of rounded 
data.  However, data tabled using these new procedures will appear 
different from data tabled with user defined rounding or previous versions 
of the PCD.  Some expected differences:

·       There will be trailing zeros on values from a less precise method than 
what was used to define the rounding array.
·       'M' remark codes will be used in place of a very low value where the 
rounding array defines a less precise method.

Although the differences may require some explanation, the Office of Water 
Quality believes the changes generally improve the presentation of the 
data and will result in more consistent publication tables that require 
less editing. 

Any questions regarding the new PCD should be directed to John Briggs 
(jbriggs@usgs.gov).

*********************************************
Janice Ward
Acting Chief, Office of Water Quality
US Geological Survey
Reston, VA
703-648-6871
jward@usgs.gov