In Reply Refer To:                             May 12, 1981
EGS-Mail Stop 412


QUALITY OF WATER BRANCH TECHNICAL MEMORANDUM NO. 81.12

Subject: WATER QUALITY--Trace Metals; Questionable Values for
                        Dissolved and Total Selenium

It was recently brought to our attention that the pattern of 
occurrence of selenium concentrations in streams in the 
Southeast showed few values above 1 ug/L after 1974 whereas 
before that date, values greater than 5 ug/L were common.

A retrieval and analysis of selenium data collected at 
Hydrologic Benchmark Network stations during the past 13 
years revealed that this pattern occurred nationwide. As 
shown in Table 1, dissolved selenium concentrations commonly 
exceeded 5 ug/L prior to 1975 but rarely did so in more 
recent years. The two high values in 1978 and 1980 were 
obtained at a single station where anomalously high values of 
other trace metals have been detected also. A similar pattern 
can be seen in the total selenium concentrations as well 
(Table 2). The one value over 5 ug/L in 1978 occurred at the 
same station as noted above.

In an attempt to detect a pattern in the values > 5 ug/L, the 
data were grouped into seven regions for a frequency 
analysis. It was apparent by inspection of the grouped data 
that no regional pattern existed in either the dissolved or 
total selenium concentrations. Thus it appears that the high 
values are approximately evenly dispersed nationwide and 
occurred, primarily, before 1975.

The cause for these anomalously high values is not known. If 
it is assumed that the high concentrations are real, we can 
think of no change in sources of airborne selenium that would 
have caused a radical decrease in stream water concentrations 
nationwide. Local shifts in electrical power plant fuel from 
petroleum to coal should have caused, if anything, an 
increase in selenium after 1974 and the increase should have 
shown a regional pattern and been steady, not precipitous.

On the other hand, there is reason to believe that values
> 5 ug/L during the period studied were caused by random 
errors in chemical laboratory analysis. The decrease in 
occurrence of high values coincided with a change in 
analytical method from the colorimetric diaminobenzidine 
method described by Brown and others (TWRI, Book 5, Chapter 
Al, 1970) to the atomic absorption with hydride generation 
method described by Skougstad and others (TWRI, Book 5, 
Chapter Al, 1979). Brown and others acknowledged no 
interferences in the diaminobenzidine method but the 
coincidence of the change in methods with the decrease in 
occurrence of values > 5 ug/L is strong circumstantial 
evidence that positive interferences were encountered, 
apparently at random, in routine analysis by that method. No 
such interferences were detected in the development research 
on the method.

With this evaluation in mind, we are advising all data users 
that values of dissolved and total selenium exceeding 5 ug/L 
in samples collected before 1975 are probably incorrect and 
should be used with great caution. Moreover, values of 
dissolved selenium greater than 1 ug/L collected before 1975 
should be considered questionable, although a fair percentage 
of them may be, in fact, correct.

The Water Quality File will not be purged of the high 
selenium values because it is not certain that the logic 
presented here is correct. The Automatic Data Section has 
added to the E771 program a cautionary note, referencing this 
memorandum, to alert all users to this potential problem with 
the historical data on selenium.



                         R. J. Pickering
                         Chief, Quality of Water Branch

Enclosures

WRD Distribution: A, B, FO, PO

Key Words: water quality, data handling, trace metals, 
           selenium.

This memorandum does not supersede any previous QW Technical 
Memorandum.