with greatest precision (smallest variance).
would be different for Line 1 and Line 2.
This optimum allocation
This allocation could be
estimated using the general approach given by Cochran (1977) as discussed
above for the aliquot study.
The variogram for the soil data of Line 2 is given in Figure 16.
The
values for ¥(h) for 5-, 10~-, and 15-inch spacings are 6.4 x 10712,
8.9 x 10712, and 7.5 x 107!%, respectively.
These data are about half
the value observed for 150 ft (see Figure 16), which suggest the real
increase in y(h) between 15 inches and 150 ft.
However, no data are
available to assess the form of the curve between those points.
The
form of the {(h) curve beyond 150 ft suggests little if any correlation
between Am concentrations spaced greater than 150 ft.
The FIDLER readings corresponding to the soil samples for Line 2 are
given in Figure 17.
The same absence of pattern seen in the soil data
is seen here.
Note that a high percentage of the surface readings are
at background levels.
The estimated variograms for both sets of FIDLER readings are given in
Figure 18.
With the possible exception of a within-cluster (spacings
less than 15 inches) structure at the 1-ft height, the readings appear to
be uncorrelated,
The 100 adjacent FIDLER readings were taken 10 ft south of Line 2, for a
distance of 42 ft, which corresponds to only the first cluster of observations from the original sampling plan.
The readings are given in
Figure 19.
The 1-ft height readings are uniformly higher than the
surface reading; but unlike the Line 1 adjacent FIDLER data, the 1l-ft
height readings are slightly more variable than the surface readings.
This is evident by examining the variograms for the adjacent FIDLER
readings in Figure 20.
The variograms are essentially flat, with the
1-ft height variogram being slightly above the surface height variogram.
This latter point indicates slightly greater variability for the 1-ft
data.
The flatness of these variograms suggest there is no correlation
structure between FIDLER readings along Line 2 no matter how close the
readings are taken.
This is probably a function of the Am activity
being at near background levels.
CONCLUSION FOR VARIABILITY WITH DISTANCE STUDY
For relatively high Am concentrations (nCi/g range), there appears to be
a correlation between observations as a function of distance for FIDLER
readings and possibly for the soil Ge(Li) analyses. For lower levels
(pCi/g range), this does not appear to be the case. This suggests that
in these low-level (pCi/g) areas, kriging may not be feasible using soil
or FIDLER data as collected here since kriging depends on the existence
of a correlation structure.
However, for higher activity level areas
within the inner fence in Area 13, Delfiner and Gilbert (1978) successfully use kriging on both FIDLER and soil data to estimate spatial
distribution and inventory of Pu in soil.
433