Results for Line 1]

The soil concentrations for Line 1 are given in Figure 9.

The horizontal

axis is distance along the line in feet and the vertical axis is Am

concentrations (nCi/g dry soil).

The mean of the two aliquots at each

location is also plotted.
The means in each cluster are connected by
bold lines to give a better visual presentation of the variability with
distance within a cluster.
The ends of the vertical bars are the observed
concentration for the two aliquots.
The concentration peak that occurs at about 140 ft corresponds to Line 1
crossing a higher activity level stratum.
The clusters with the highest
concentrations have the most variability, both among the four samples in
a cluster and between the aliquots.
The clusters past 220 ft tend to be
reasonably constant.
The greatest proportion (82%) of the total variance
of these Am data is due to differences between clusters.
Variation
between adjacent samples within clusters accounts for 15%.
The remaining
variability (3% of the total) is accounted for by variability between
the 70-g aliquots within each soil sample.
The estimated variogram for Line 1 soil samples, as computed using
Equation 17, is given in Figure 10.
Notice the peak at 120 ft.
This
occurs because of the peak in the data at that distance in Figure 9.
This variogram is a biased estimate of the true underlying variogram
that should be used for kriging purposes.
This bias results from the

strong trend (called "drift" in the kriging literature) in concentrations

along the line evidenced by the peak kriging literature) in concentrations
along the line evidenced by the peak at 140 ft.
We have noted above
that Equation 17 should not be used under these circumstances.
Methods

are available (Delfiner, 1975) for obtaining variograms that are free
from the effects of drift.

Delfiner and Gilbert

(1978) discuss in

greater detail the estimation of the variogram in the presence of a

drift.
Kriging could use unbiased estimates of the variogram to estimate
average Am concentrations at unsampled locations along the line or over
unit areas where the same correleation structure applied.
The net FIDLER readings for Line 1, both surface and 1-ft height, are
given in Figure 11.
The data plotted represent the mean of two readings.
The solid line connects the means within a cluster for the 1-ft readings,
while the dashed line connects the surface readings.
The star is the
1-ft reading at the center of the cluster.
The 1~ft and center readings are generally higher than the surface
readings, which is expected since the instrument held at l]-ft height
receives input from a larger area (Tinney, 1968).
Comparing Figures 9
and 11, soil concentrations and FIDLER readings show the same pattern

of variability within clusters with a peak at 140 ft.

The estimated variograms for the FIDLER readings are given in Figure 12;
the solid-line and dashed-line curves are the 1-ft height and surface

427

Select target paragraph3