United States Environmental Protection Agency	Office of Research and Development

National Exposure Research Laboratory
Research Abstract

Government Performance Results Act (GPRA) Goal 5
Annual Performance Measure 64

Significant Research Findings:

Guidance to Identify, Quantify, and Reduce Laboratory Subsampling

Errors in Soils and Solid Wastes

Scientific	Nearly all environmental research programs require the collection of samples in

Problem and	the field. However, the overwhelming majority of efforts (in terms of time and

Policy Issues	cost) to control and quantify error components in the data are concentrated on

laboratory analyses. It has been repeatedly stated that 80% of the total error
occurs in the field for the more stable contaminants (e.g., metals, PCBs, and
pesticides) and up to 99.99% of the total error occurs in the field for non-stable
contaminants (e.g., volatile organic compounds; (VOCs)). This research effort is
aimed at examining and reducing the biases (intentional or unintentional error)
that occur due to the use of "improper' sampling techniques when collecting the
analytical subsample. An improperly collected analytical subsample will, in turn,
yield contaminant concentrations that are not representative of the site and can
lead to improper decisions being made by regulators on whether or not to
remediate a site.

A field of sampling theory and practices has been developed by Dr. Pierre Gy in
which logical steps have been identified that can reduce the error associated with
sample collection. The original sampling theory and practices were developed for
the mining industry. However, the same errors can occur when sampling
contaminated soils or other solid matrices. While the sampling theory and
practices appear to be completely logical and the examples given indicate the
sampling errors do occur, little testing as been done on environmental samples to
determine if by following the guidelines and "rules"' established by Dr. Gy, better
(i.e., more accurate, precise, and representative) samples will be obtained in the
environmental field.

If the Gy sampling theory and practices are valid in the environmental field,
resultant data being given to decision makers and regulators will be more
accurate, precise, representative of the site being sampled, and will be legally-
defensible. With this improved data, decision makers can better define the extent
and degree of contamination at any site. Further, with the more accurate and
precise data, the need for remedial actions can be better determined and the health
of the public can be better protected.

National Exposure Research Laboratory — October 2003


-------
Research	The focus of this research effort was to examine sample reduction techniques that

Approach	are typically used to take the large mass/volume field samples (typically ranging

from hundreds of grams to kilograms) and to decrease their size down to a
method defined analytical sample size (typically ranging from milligrams to
hundreds of grams) for their effects on sampling error (i.e., either reducing or
increasing sampling error).

To test the effectiveness of the selected sample reduction techniques, artificially-
created, known contaminant distributions were created in the laboratory.
Contaminants were represented by using magnetite and salt (i.e., sodium chloride)
and the "soil" was represented by a pure sand. Artificially-created contaminant
distributions included: even layering of contaminant and non-contaminant layers,
using pockets of large particle sizes, using pockets of fine particle sizes, and using
coated particles (i.e., salt-coated sand grains). Analytical results yielded either
weights of collected magnetite or total salt concentrations.

Sample reduction techniques tested were sectorial splitters, riffle splitters, manual
incremental sampling, grab sampling, and cone and quartering techniques.

Sample grinding was also examined to determine if particle size reduction was
successful in reducing error associated with sample collection. The sample
reduction technique(s) that yielded results that had the least error were those that
were most precise and the least biased (i.e., most accurate) based on the known
contaminant distribution and concentrations. To ensure the integrity of the
contaminant distributions, an exhaustive analysis (analysis of the entire sample
distribution) was performed and favorably compared against the known amounts
of contaminants placed in the sample.

Results and	The laboratory sample reduction techniques, in general, reflected agreement of Gy

Impact	sample theory and practices, indicating that these techniques and practices can be

used successfully in the laboratory to reduce the error associated with the
collection of environmental samples.

The best sample reduction technique was the sectorial splitter. This technique
was fast, easy to use, and essentially eliminated all error that is typically associated
with poor sample collection (i.e., these samples gave the minimum error
expected). The use of riffle splitters to reduce sample size gave comparable
results to the sectorial splitter. Manual incremental sampling, a process in which
about 30 small samples are collected and combined together to represent the final
analytical samples, was cumbersome and tedious to perform and yielded greater
variability in sample results (i.e., greater error that can lead to incorrect
assessments of the contaminant concentrations at a site) than either the sectorial or
riffle splitters. Cone and quartering techniques and grab sampling failed to
produce representative results as indicated by higher variability among the
replicated samples. For example, the error associated with grab sampling was
over an order of magnitude greater than the error measured by the use of the
sectorial splitter. These two methods also tended to take longer to produce the
final analytical sample. Sample grinding is highly recommended for all samples,

National Exposure Research Laboratory — October 2003


-------
either before or after employing the sample reduction techniques and where
practical, since this technique yielded the most precise and accurate contaminant
concentrations when compared to the known values.

This research project helps satisfy the FY02 Annual Performance Goal (APG) 9
entitled, "Provide at least 2 new soil sampling and on-site screening methods."
This APG is part of the larger Government Performance Results Act (GPRA)
subobjective that is aimed at improving site characterization, site monitoring, and
modeling of contaminant fate and transport in the environment. In brief,
following Gy's sampling theory and practices, a significant reduction in sampling
error can be achieved with little additional effort on the part of the sample
collector or the analytical laboratory. And finally, by knowing with greater
accuracy and precision what contaminants are present, where they are located,
and what will happen to them if they remain untreated at a site, regulators and the
general public can more appropriately prescribe actions to protect human health
and the environment.

This research project was conducted primarily by a team of National Exposure
Research Laboratory (NERL) staff scientists and staff scientists at the National
Enforcement Investigation Center (NEIC) in Golden, Colorado. Contractor
support was obtained from EnviroStat of Fort Collins, Colorado, and Lockheed-
Martin Environmental Services of Las Vegas, Nevada.

One part of this research has been published in the following manuscript:

Gerlach, R.W., D.E. Dobb, G.A. Raab, and J.M. Nocerino. Gy sampling theory in environmental
studies. 1. Assessing soil splitting protocols. J. Chemometrics. 16:321-328,2002.

Future Research Research investigating the sampling error associated with commonly used

sampling tools (e.g., scoops, shovels, coring devices) is planned in an effort to
further reduce sampling error. Additionally, this research will be extended to test
the principles and results obtained in the laboratory setting to the field setting
where greater volumes of samples are collected and processed.

Research
Collaboration and
Research
Products

National Exposure Research Laboratory — October 2003


-------
Contacts for	Questions and inquiries can be directed to:

Additional	John M. Nocerino

Information	U.S. EPA, Office of Research and Development

National Exposure Research Laboratory
Environmental Sciences Division
Characterization & Monitoring Branch
P.O. Box 93478
Las Vegas, NV 89193-3478
Phone: 702/798-2110
E-mail: nocerino.john@epa.gov

National Exposure Research Laboratory — October 2003


-------