latter area was centered at the grid node 5-S-3 and was an anomaly throughout the cleanup with
respect to ratio computation and elevated levels of TRU activity for both subsurface and surface.

The initial TRU surface characterization used only the 50 m grid data and the ratios previously
mentioned. The calculated TRU values were used to fit a variogram model necessary to make the
kriging estimates and the 0.5 s upper bounds, where s is the standard deviation of the kriging error

(see Section 5.1.1). The estimated model for Pearl did not follow the usual mathematical form of

linearity seen on other islands but was a power function. The model was tested and found to fit the
data quite well. Using this model and the 50 m grid TRU data, 0.25 hectare estimates were

calculated based on original data. (See Tech Note 23 for discussion on original versus final data.)
These results indicated that Pearl was one of the more highly contaminated islands with the highest
0.25 hectare TRU estimate being 281.6 pCi/g and the lowest being 10.1 pCi/g. Approximately 3/4 of
the island was estimated to have TRU activity greater than 40 pCi/g based on the 0.5 s upper bound

numbers, and approximately 2/3 of the island had TRU greater than 80 pCi/g based on the same
upper bounds. Figure 7-85 shows the area with TRU estimated to be above 40 pCi/g for theinitial
data.
These estimates were based on data collected prior to any debris pickup.

Because this island had a

large quantitiy of debris and was also very sandy and heavily vegetated, the radiological condition of
the island changed during debris removal. Remeasurement by the IMP and collection of soil samples
were done to determine how much this heavy soil disturbance had altered the island's
characterization.

The area of the island affected by the debris removal is shown in Figure 7-84. Only this area was
remeasured by the IMP in July 1978 and four surface soil samples were collected concurrently at

locations also shown in Figure 7-84. One ratio was calculated from these soil sample results whereas
before two ratios were included in this area. It appeared that the disturbance homogenized the soil

and one ratio of 6.91 + 0.41 was appropriate. Five additional soil samples were collected to verify
this ratio but were ballmilled with contaminated balls during sample preparation so more samples

were collected. These additional results verified the ratio calculated after debris removal. For the
area of no soil disturbance, the original ratios were used to calculate the TRU values.

Using this second set of data, a new variogram model was estimated. For these data, the modelfit
was linear with a smaller constant term than was estimated before. This model was tested and fit
the raw data well. New kriged estimates were computed using this model and the new TRU values.
These 0.25 hectare averages showed lower TRU concentrations as compared to the first estimates
calculated. The highest 0.25 hectare TRU estimate based on original data was 167.1 pCi/g compared

to 281.6 pCi/g prior to debris removal. However, the areas with TRU estimated to be greater than
40 pCi/g and 80 pCi/g were basically the same for both sets of data with the exception being one
small area on the southwestern part of the island that was significantly lower after debris removal.
Because no actual soil removal occurred prior to the seconditeration, it appeared that either the soil

was mixed or the dirt and brush piles left on the island contained much of the original top soil. If
substantial churning had occurred as a result of debris removal, it could mean the TRU activity
would be distributed deeper and several soil lifts would be necessary to remove the contamination.
Based on the surface soil samples that were collected at 0-, 10- and 20-em intervals, it seemed some

mixing did occur but did not go very deep. This conclusion was also based on subsurface sidewall
samples (see Section 6.9) collected after debris removal to a depth of 120 em. The results from

these samples showed no 24lam activity greater than 2 pCi/g below a depth of 20 cm. More will be

said about the subsurface sampling later in this section.

The next sampling involved collecting soil from the dirt and brush piles remaining on Pearl following
debris removal. The piles were first surveyed with a handheld instrument and areas with higher

307

Select target paragraph3