Assessment of Sampling Error Associated with
Collection and Analysis of Soil Samples at a
Firing Range Contaminated with HMX
THOMAS F. JENKINS, MARIANNE E. WALSH,
PHILIP G. THORNE, SONIA THIBOUTOT, GUY AMPLEMAN,
THOMAS A. RANNEY, AND CLARENCE L. GRANT
about 150 mg/kg for a second discrete sample
INTRODUCTION
collected only 61 cm distant. Either of these were
legitimate discrete samples that could be used to
Background
Recently, we reported large short-range spatial
represent this grid location, according to typical
heterogeneity of explosives contaminant concen-
sampling plans now being used at explosives-
trations in surface soils at explosives-contaminated
contaminated installations. Very different deci-
sites (Jenkins et al. 1996). Nine locations were
sions relative to the need for site remediation
sampled at three installations; the sampling loca-
might be made, though, depending on which of
tions varied in the principal contaminant present,
these results happened to be the sample collected
mode of contamination, and soil type. Seven dis-
to represent the grid location.
crete samples were collected at each sampling
A positive result from this work, however, was
location from a 1.22-m circle. The samples were
the demonstrated ability to prepare composite
analyzed on-site using colorimetric methods
samples on site that were a good physical average
(Jenkins and Walsh 1992), and in the laboratory
of the mathematical mean of the discrete samples
using SW846 Method 8330, the standard reversed-
making up the composite (Jenkins et al. 1996).
phase high performance liquid chromatography
The use of composite sampling with on-site analy-
(RP-HPLC) method for the determination of
sis provides an attractive alternative to the con-
nitroaromatics and nitramines in soil (EPA 1995).
ventional use of discrete sampling strategies and
The results indicated that regardless of whether
analysis at off-site laboratories.
the major contaminant was 2,4,6-trinitrotoluene
While a fairly large body of information is avail-
(TNT), 2,4-dinitrotoluene (DNT) or ammonium
able relative to explosives concentrations at sites
picrate (AP), or whether on-site or laboratory re-
impacted by manufacturing operations (Walsh et
sults were used, extreme heterogeneity in con-
al. 1993), very little information is available on
taminant concentrations was found among the
the levels of accumulation of explosives residues
seven samples at each location. Ratios of the high-
at active firing ranges. In addition, we have been
est concentration obtained among the seven dis-
unable to locate reports of site characterization
crete samples divided by the lowest of the seven
where HMX (octahydro-1,3,5,7-tetranitro-1,3,5,7-
ranged from about 3 to greater than 600, with a
tetrazocine) (Fig. 1) was the principal contami-
mean value of about 73 (median value was 50).
nant. We recently reported that a colorimetric on-
These results have implications for developing
site method for RDX (hexahydro-1,3,5-trinitro-
sampling designs to characterize explosives-
1,3,5-triazine) would detect HMX (Jenkins et al.
contaminated sites, if data are expected to repre-
1995), but no on-site evaluation of this method for
sent average site conditions. For example, the TNT
quantitative estimation of HMX has been reported.
concentration at one location varied from about
The firing range characterized in this study is a
40,000 mg/kg for one discrete sample to only
Canadian antitank firing range, which had been
Previous Page
