Design of sampling grid
methods. This approach is similar to the way ducks
Samples were collected on the basis of methods
feed and allows us to see if white phosphorus is
described by Gilbert (1987) for locating hot spots
available to dabbling ducks and swans, the recep-
with a chosen level of confidence that they will not
tors at most risk.
be missed. These methods assume the following
Objective of sampling
conditions:
We must determine if lethal quantities of white
The hot spot is circular or elliptical.
phosphorus are available to waterfowl in a given
Samples are taken on a square, rectangular, or
area.
triangular grid.
The distance between grid points is much
Assumptions
larger than the area sampled at the grid
Waterfowl are being acutely poisoned by the
points. (If a large portion of the area is sam-
ingestion of white phosphorus.
pled, the probability of hitting a hot spot will
An acutely lethal dose of white phosphorus is
be greater than indicated by the model used
in the form of at least one solid particle with a
here.)
mass greater than 1 mg.
The definition of hot spot is clear.
The bulk of the solid particles resides in local-
No errors are made in detecting a hot spot.
ized hot spots with radii on the order of 1 m.
Also implied is that the area of the hot spot is small
Approach
compared to the total area of interest, analogous to
Collect samples systematically using a two-
the proverbial needle in a haystack.
dimensional grid designed to detect a single
For the most part, these conditions are valid at
hot spot, if one exists, with probabilities of
ERF for the following reasons:
missing the hot spot of 10, 20, and 40%.
The hot spots at ERF are most likely at the
Composite the samples and isolate the frac-
point of detonation of white-phosphorus-
tion that potentially contains milligram parti-
containing projectiles, which produce circu-
cles of white phosphorus (i.e., fraction retained
lar patterns of contamination (Walsh and Col-
on 0.59-mm-mesh sieve).
lins 1993).
Determine if white phosphorus particles are
Samples were taken on square and triangular
present in the composite sample.
grids.
Equal volumes of sediment were obtained at
METHODS
each sample point by way of a graduated cor-
ing device that had a diameter of only 2.65
Field tasks
cm, whereas the grid spacings used in this
Area sampled
study varied from 0.91 to 2.44 m.
We sampled an area encompassing one of the
A hot spot is defined as a localized area con-
DWRC pens mentioned above because this area
taining solid white phosphorus particles.
was known to contain white phosphorus that was
If white phosphorus particles are abundant
available to waterfowl. These pens were located at
within each hot spot, detection errors should
the edge of a permanent pond in the salt marsh.
be minimal.
The corners of pen 5 used in the DWRC studies
were relocated by surveying (Fig. 1). An area of 7
However, we do not expect the particles to be dis-
tributed evenly throughout the hot spot. Rather,
20 m, encompassing the pen, was marked with
the hot spots at ERF are likely to have more white
survey lath. Prior to any activity, the area was
phosphorus particles in their centers than towards
checked for UXOs (unexploded ordnance). At the
their edges. Therefore, the number of particles
time of the sampling (May 1996) the water depth
found cannot be used to estimate the number of
was approximately 10 cm. We visually inspected
hot spots, if more than one hot spot exists.
the surface before walking in the immediate vicin-
To determine the grid spacing G, the length L of
ity to prevent suspending sediment and limiting
the semi-major axis (radius for a circle) of the
visibility. Additionally, a hand-held magnetome-
smallest hot spot must be known. Previous studies
ter, a Heliflux Model GA52C Magnetic Locator,
was used to scan the area to detect any buried fer-
tell us that L varies from approximately 0.5 to 1 m.
Then, the acceptable probability β of not finding a
rous metal objects. Three spots produced a signal,
and these spots were marked with orange flagging
single hot spot if one exists must be specified. A
and avoided.
nomogram (Fig. 2) is then used to compute grid
5
Previous Page
