Overview
We recently
received for review the Draft Work Plan for Off-site Soil Sampling, LCP
Chemicals Superfund Site, Brunswick, Georgia, dated May 7, 2004. This plan
summarizes past sampling activities in the neighborhood next to the site,
the ARCO neighborhood, and proposes some additional testing.
Overall, the
actual testing proposed is very limited. Few new samples are collected and
tested under this plan. Only very limited conclusions could be drawn from
such a simplistic look at pollution in the neighborhood. The study appears
designed to “prove” the old sampling was correct rather than find out if
the neighborhood is safe.
Background
The LCP Chemicals
superfund site is a 550-acre site along the Turtle River in Glynn County
Georgia. The site was the location of several industrial chemical plants
producing bleach, petroleum products, or paint. These operations left the
area heavily contaminated with known chemical toxins from ground, water
and air pollution. The site is in the process of being cleaned with
Federal and State oversight. One issue for the local community is the
extent of contamination from air pollution and any surface runoff. Various
industries at the site used “stacks” and fans to vent chemical processes
directly into the air. Most pollutants travel only a short distance before
settling back to the ground, and the ARCO neighborhood is in the pathway
of any possible contamination. In an earlier evaluation the Environmental
Protection Agency noted: “It is very likely that this airborne
deposition traveled with prevailing winds, and mercury was therefore
distributed over a potentially rather large area. The prevailing wind
direction for this area is from the northwest, away from the marsh area.”
[emphasis added; Section 7.1.3, airborne redeposition (page 42,
Operable Unit One Upland Soils and Estuary March 2002)].
The Testing Plan
Plan Designs:
All testing plans
are designed to answer a question. Sampling locations, the number of
samples, and the testing methods are all chosen to provide specific
answers. As a result some testing plans are broad with numerous samples
along a grid pattern in order to provide as much detail as possible on the
way in which chemicals are found in the area. Broad testing plans are
designed to answer questions such as “how extensive is contamination,” or
“are all areas free of toxins.” However, other sampling pans are very
narrow, only designed to answer a very specific question. Narrow testing
plans are directed to very limited questions, a confirmation test for a
previously tested area is one example of a limited sampling plan. The
testing discussed in the Draft Work Plan is a narrow testing plan, only
directed to one small area near the site fences.
Single versus Composite Samples:
Single samples,
sometimes called “discrete” samples, are one test for each sample taken in
the field. Composite samples are mixed samples where several samples from
over a wide area are mixed before testing. Single discrete samples provide
more detail with less possibility of error. There are two types of error:
false-positives occur when a sample is reported as contaminated when it is
really clean, and false-negatives are reporting a sample as clean when it
is really contaminated. Both false-positives and false-negatives are rare
for discrete samples. Composite sampling has a much higher error rate for
false-negatives. When several samples are mixed any contaminated samples
are diluted with clean samples. If there are five samples in the composite
and all five are contaminated then the sample will show up as a positive.
However, if only one sample would be positive by discrete sampling and
four are clean, then the one positive sample will be a false negative in
composite sampling from dilution. Composite sampling is less expensive
than discrete sampling. All of the sampling proposed for the ARCO study is
composite sampling.
Phased Testing:
Fourteen composite
samples are planned for an initial phase 1 study. Each sample is taken
from a very broad area of 10,000 square feet (according to section 3.2 the
sampling station is a 100’ by 100’ square, with 5 samples taken and mixed
together from the square). All total the area actually being composite
sampled in phase 1 for the 14 samples is about 3 acres. However, these
sampling squares are not “contiguous” or joined together, they are
dispersed over an area of about 60 acres (on Figure 2 provided for
review, the phase 1 area is shown with a scale of 7/8” equals 400 feet,
the measured phase 1 area was about 6” by 2”).
So the 14 phase 1
samples are just a few pounds of surface soil mixed together from an area
covering nearly 60 acres. The results from these few pounds of dirt are
then used to determine if EPA will require a larger study, called phase 2,
that will actually test the neighborhood.
Discussion
In this proposed
study each mixed sample represents about 4 acres. This is far below
recommended sampling guidelines, and is not a valid scientific sampling
plan. The combination of mixing samples before testing and sampling over
such a wide area can result in false-negative reporting of actual site
conditions.
Suggestions for
changes to the plan include:
·
Five of the fourteen
proposed phase 1 samples are on LCP property and are not properly called
“ARCO neighborhood” samples, all of the proposed samples should be within
the neighborhood blocks;
·
Closer grids—100’ by 100’
is too coarse, a better grid size is on the order of 25’ by 25’;
·
The plan should clearly
state that testing will evaluate toxins at residential safety levels;
·
Sediment testing of the
pond shown on the EPA’s sampling map is needed, it is large enough to
easily have received fallout during plant operations and residents
indicate that children often fish this pond;
·
Include existing testing
data from local school grounds in the site description and add sample
locations to the neighborhood plan to fill in any data gaps;
·
For proper evaluation of
the plan the figures should be representative of the actual site area, as
provided features such as streams and railroads were omitted, a detailed
plan should be provided;
·
An air dispersion model
is needed—usually sample patterns for airborne depositions are made after
careful atmospheric modeling.
Conclusions
The sample pattern
chosen for this study seems arbitrary and capricious, not scientific. The
number of samples chosen are too few and too far apart. The small sampling
footprint is only a small percentage of the actual neighborhood area of
concern.
Earlier, EPA noted
that pollution may be present in this neighborhood. This study as
described will neither prove nor disprove that possibility. If the study
finds some pollution then it only means that at least one more study is
needed, the phase 2. However, if nothing is found it fails to prove the
site is safe because of the poor study design. The Plan as proposed will
not answer the question: “Is the ARCO neighborhood safe from LCP toxins?”
Written
by R. Kevin Pegg, Ph.D.; edited by Dr. Mary S. Saunders. Copies of the
newsletter are available from the GEC, at the Glynn County library, or at
www.Enviro-Issues.com on the Internet.
P. O. Box 2443