An Approach to Industrial Stormwater Benchmarks: Establishing and Using Site-Specific Threshold Criteria at Lawrence Livermore National Laboratory Page: 4 of 12
This article is part of the collection entitled: Office of Scientific & Technical Information Technical Reports and was provided to Digital Library by the UNT Libraries Government Documents Department.
The following text was automatically extracted from the image on this page using optical character recognition software:
In the first step, data is plotted over time to identify any major trends or large
discontinuities that may indicate a change in analytical technique, laboratory, or
discharge process. Control charts are developed by plotting data that have been
standardized by subtracting the mean and dividing by the standard deviation. Control
charts have a mean of zero and the variability around the mean is standardized by the
standard deviation. Therefore, the value 2.0 on the Y-axis corresponds to two standard
deviations. Values that are many standard deviations above or below zero need to be
examined because they may be outliers and a decision should be made whether to
keep these values in the analysis. The trend plots and control charts are a fast way to
quality check large amounts of data.
Step two involves calculating the confidence intervals for a given probability level for a
select group of water quality parameters. As environmental data are often not normally
distributed, this may require a correction to normalize the data (if possible). If not
possible, alternative calculations developed for other distributions (e.g., Chi-squared
distribution) need to be used. After normalization, the standard deviation is multiplied
by the test statistic (for a normal or t-distribution) for the desired confidence interval
(e.g., 95, 97, or 99 percent) and added to the mean value. For LLNL's site-specific
thresholds, a value exceeds the threshold if it is greater than the 95 percent confidence
interval (usually about two standard deviations) computed for the historical mean value
for a specific parameter.
The purpose of steps three and four is to determine if the value for the UCL calculated
is acceptable to use at your site. This is determined by comparing the UCL value to
published regulatory reference values, such as ambient water quality goals, basin plan
objectives, or drinking water maximum contaminant levels. If the calculated UCL values
exceed regulatory reference values, the reasons and background information should be
documented. Another validation of the UCL value is to see how often it would trigger
further investigation. If this value is used as an "action level", but is exceeded once a
year (or more), the exceedence probability may be set too low. It is important not to set
the action level so low that it triggers a "chicken little" response. These two approaches
basically determine if the UCL value calculated is too high or too low to be an effective
action level. If it is not a useful value, then return to step two and increase the
probability to 97 percent to see if a better value is returned.
After this fine-tuning of the UCL value, it can be used as a threshold criteria to identify
out-of-the-ordinary data that should be investigated further to determine if
concentrations of that parameter are increasing in the stormwater runoff.
As an example, presented here is the process of data analysis and decision points
using more than 10 years of copper data from the LLNL Livermore site. Quality
assurance procedures are well established and in place at LLNL and quality control
samples, including sample duplicates, were examined. In plotting the data we noticed
that the detection limits provided by the analytical laboratory for copper analyses
changed from 0.050 mg/L to 0.01 mg/L in the mid 1990s and changed again to
0.002 mg/L in 2001 (Figure 1). The data presented in Figure 1 are for the effluent
stormwater sampling location along the Arroyo Las Positas called WPDC. All values
listed as less than the limits of detection were removed from the calculation of
Here’s what’s next.
This article can be searched. Note: Results may vary based on the legibility of text within the document.
Tools / Downloads
Get a copy of this page or view the extracted text.
Citing and Sharing
Basic information for referencing this web page. We also provide extended guidance on usage rights, references, copying or embedding.
Reference the current page of this Article.
Campbell, C G & Mathews, S. An Approach to Industrial Stormwater Benchmarks: Establishing and Using Site-Specific Threshold Criteria at Lawrence Livermore National Laboratory, article, September 7, 2006; Livermore, California. (digital.library.unt.edu/ark:/67531/metadc886697/m1/4/: accessed October 17, 2018), University of North Texas Libraries, Digital Library, digital.library.unt.edu; crediting UNT Libraries Government Documents Department.