tions have been offered to describe the qualifica-
we believe it is an attractive option to improve
tions of representative samples (Gilbert 1987,
representativeness of samples.
Barcelona 1988, Smith et al. 1988, Barnard, in
press). We are partial to the Gilbert definition, "A
Objectives
representative unit is one selected for measure-
The major objective of this work was to charac-
ment from the target population in such a way
terize the short-range heterogeneity of contami-
that it, in combination with other representative
nants at explosives-contaminated sites. This was
units, will give an accurate picture of the phe-
done by conducting field sampling and analysis
nomenon being studied." According to Barnard,
studies at a number of explosives-contaminated
"Representativeness is a statistical concept that is
sites that varied in explosives analytes present,
a measure of how well a data set of sample mea-
mode of contamination, soil type and geohydrol-
surements yields information concerning the
ogy. Statistical analyses of the results were con-
population."
ducted to determine the following:
Explosives are solids at ambient temperature,
1. Analytical error, which was estimated from
dissolve slowly and sparingly in aqueous solu-
the pooled variances from duplicate analyses of
tion and have low vapor pressures. These proper-
seven grab samples collected within a localized
ties limit modes of mobility compared to other
area. Short-range sampling error was estimated
contaminants such as fuels or solvents. Thus, the
from the variance computed from the differences
areas of high concentrations that serve as sources
of mean values of the seven grab (soil) samples.
for contamination of ground water remain at or
2. The degree to which some form of compos-
near the surface where deposited, unless the soils
ite sampling could be used to reduce sampling
themselves are moved. Thus, characterizing the
error.
contamination distribution for explosives will of-
3. Whether inexpensive, colorimetric on-site
ten be possible using samples of near-surface soils.
analysis methods could be used to provide an
In this study we focus on how to obtain repre-
accurate description of contaminant distribution
sentative samples from surface soils contaminated
and a reliable estimate of sampling error.
by munitions residues. Too often, local spatial
heterogeneity is bypassed in favor of grab sam-
pling on the theory that heterogeneity will be
EXPERIMENTAL
"averaged out" if sufficient samples are taken.
While there is validity in this position, it hardly
Throughout this report the following terminol-
qualifies as cost-effective, especially when analy-
ogy will be used: installation will refer to the
sis cost often outpaces sample collection cost by
government facility where sampling was con-
orders of magnitude. In addition to our experi-
ducted; sampling location will refer to any one of
ence, several authors have reported large local
the nine areas (three at each installation) where
spatial heterogeneity, often of the same magni-
sampling was conducted; and sample position
tude as present on a much larger scale (see, for
(or sample number) will refer to the specific spa-
example, Parkin 1987, Sabbe and Marx 1987, van
tial position where a discrete sample was col-
Ee et al. 1990, Starr et al. 1995). To address this
lected.
problem, others have used or recommended com-
posite sampling (Cameron et al. 1971, Schaeffer et
Sampling sites
al. 1980, Gilbert 1987, Garner et al. 1988, Paasivirta
Sampling studies were conducted at three in-
and Paukku 1989, Parrish et al. 1990, Huesemann
stallations. These are Monite, a BLM (Bureau of
1994, Fabrizio et al. 1995). We decided to investi-
Land Management) installation near Sparks, Ne-
gate the feasibility of this approach, coupled to
vada; Hawthorne Army Ammunition Plant (AAP),
both on-site analysis and conventional laboratory
Hawthorne, Nevada; and Volunteer AAP, Chatta-
analysis.
nooga, Tennessee (Fig. 1).
Compositing is sometimes discouraged because
The Monite installation is a small former in-
it eliminates information regarding the variabil-
dustrial area that has about 1.5 acres of land con-
ity of the individual samples composited. When
taminated with TNT and DNT. The company that
applied to large areas, this limitation may repre-
owned the site reportedly reclaimed explosives
sent a valid concern, especially when concentra-
from out-of-date military munitions, but since that
tions are near a regulatory limit. However, when
company declared bankruptcy and abandoned the
used on localized areas in lieu of grab sampling,
site many years ago, the history of contamination
2
Previous Page
