Supplement - Page 3.
Note:
These computations assume the data are homogeneous, i.e. there are
no trends in the data.
Since there are trends present on Janet
(increasing concentrations near GZ areas) these kinds of computations should be done separately for GZ and low level areas.
Ill.
One-Sided Confidence Limit on a Proportion
Using "Attribute Sampling" by Herman Burstein, Mc-Graw-Hill, 1971,
(Table 1) we can obtain the following probability statement:
The probability is 100(1-a) that the proportion of soil samples
with Pu concentrations greater than or equal to the cleanup
Level L is less than or equal to P.
Estimates of P for various values of a for cleanup level 40pCi/g
(using
the 139 soil samples (0-15 cm) from Janet) are:
a
0}
. 167
.10
. 133
.05
Interpretation:
P
145
Note:
Proportion of samples with Pu
concentrations 2 40 pCi/g is
13/139 = .0935.
For a = .01;
We are 99% sure that 16.7% of the soil samples on Janet
have concentrations 2 40pCi/g.
Discussion:
A possible approach to deciding whether an island needs to be
cleaned up is as follows:
The island (or parts of the island)
will be cleaned up unless P is Tess than, say, 5% for some
specified a level, say .01.
If it had happened that only 1
of the 139 samples had a Pu concentration 2 40pCi/g then we find
that P = .047 (4.7%) for a = .01.
Hence, in that hypothetical
case we would decide not to cleanup the island if the above
rule (P <.05 when a = .01) had been used.
An alternative and
perhaps preferable method of deciding whether cleanup is necessary is discussed under Question 3, part B.