The first set of preliminary IMP data, 21-23 July 1977, consisted of measurements at the nodes of a 5 x 8 grid of sampling points at about 23 m (75 foot) spacing at the Lawrence Livermore Laboratory experimental garden on Janet. These data were used to help get the IMP data base started and checked out as well as to shake down thein situ system. A second preliminary survey was done 29 July to 7 August 1977, at the nodes of an 11 x 12 grid with 25 m spacing at the north central edge of Janet. (Because the area was later remeasured on the regular 50 m Janet grid, neither the absolute coordinates nor the exact compass orientation of this grid was ever determined.) The 11x 12 grid, known as the Test Grid, provided enough data to complete checkout of the in situ system and the data base programs. There was sufficient 241 Am data from the Test Grid to begin the statistical analysis of Janet data by fitting an initial variogram model. (The variogram and its use in estimation are explained in section 5.2.1.) A plot of the raw variogram led to the conclusion that the 24. Am activity distribution pattern was anisotropic, that is, not the same in all directions. The difference could have been caused by the effect on fallout plumes of the strong prevailing northeast trade winds. The mH 6.4.2 Preliminary Statistical Analyses lam activity changed most rapidly from northwest to southeast, perpendicular to the prevailing wind, and slowest along the path of the wind. The pattern was exactly what would be expected for fallout from a wind-elongated plume. The effect was especially noticeable in these data because the Test Grid is almost due southwest, that is, directly downwind, of Item ground zero, and directly It was very desirable for practical reasons to use 50 m instead of 25 m spacing for the cleanup sampling grids. In order to check whether 50 m spacing would yield adequate data, the Test Grid was split into four disjoint 50 m subgrids, and raw variograms computed for data from each subgrid. The variogram model estimated from the complete data set fit each subgrid raw variogram fairly well. i upwind of the Easy/X-Raysites. The models estimated on the subgrid raw variograms were also very similar to the original model, except that one subgrid yielded a model which underestimated the nugget effect (see Section 5.2.1). On the basis of the good agreement between the original model and the subgrid data variogram models, the IMP measurements of Janet after the Test Grid data analysis were on a 50 m grid. A more detailed discussion of these and the following statistical analyses was published previously (Barnes, 1978). An area on the west tip of Janet had already been staked at 25 m spacing on the standard Janet grid (origin at benchmark PORKY) before the Test Grid data analysis was complete (see Figure 7-65). The IMP had taken data at most of the points in this area by the time the change to a 50 m spacing was made. Data were therefore also taken at the remainder of the 25 m grid points already staked, but the rest of Janet was staked and measured initially on a 50 m grid. When the initial IMP characterization measurements were complete, new variogram models werefit to the data, treating the 25 m data from the western area separately from the rest of Janet. The separation was based on the significant differences in TRU activity distribution between the western the reason is apparently soil recontouring activities during the testing years. For example, the TRU activity is much lower in the west, despite the presence of two ground zero sites, Easy and X-Ray, in that area. Also, the ratio of TRU to 241Am (see Section 7.5.2) is different in the west than anywhere else on Janet. Later subsurface sampling revealed the presence of asphalt below the surface (see discussion on asphalt sampling below). This may have been deposited by post-event cleanup activities during the testing years. Whatever the cause of the activity differences, the result was substantially different variogram model parameters for the west data than for Test Grid data. However, both models have the same mathematical form. 160 mom area and the rest of the island. Although the reason for the differences is not known, at least part of