a
at locations where samples are not collected than did fits obtained on
untransformed data.
untransformed scale,
Fits in either scale, but particularly in the
gave questionable estimates in regions of sparse
data where Pu concentrations change rapidly within short distances.
Plots of sample data on estimated contour maps suggest regions where
more data are needed.
The two-phase grid estimation procedure used here does not give estimation
variances for the grid values.
Kriging is mentioned as a method that
does provide such estimates.
An evaluation of the applicability of
Kriging to estimating spatial pattern of radionuclides in the environment
is encouraged.*
INTRODUCTION
Estimating the spatial pattern or geographical distribution of environ-
mental contaminants is often of interest in environmental sampling
programs.
Concentrations of the contaminant are measured at various
locations and an estimate is desired of the "true" concentration "surface"
for the area from which samples are collected.
One approach to this
problem is to use the observed data to estimate the surface at regular
grid points over the study site.
This estimated grid matrix can then be
displayed as a three-dimensional concentration surface or as a contour
map showing lines of constant concentration.
In this paper, we examine whether an iterative fitting procedure for
estimating the grid matrix of concentrations would improve estimates of
the true concentration surface.
The iterative procedure is applied to
239°240by concentrations in surface (0-5 cm) soil samples collected
according to a stratified random sampling plan at the Area 13 (Project 57)
"safety-shot" site on the Nevada Test Site (NTS).
This study site is one of 10 safety-shot sites on the NTS or the adjacent
Tonopah Test Range currently being studied by the Nevada Applied Ecology
Group (NAEG).
These are sites where, during the period 1954-1963,
assemblies or devices composed of plutonium and/or uranium were blown
*This paper was prepared for presentation at the 1976 Annual Meeting of
the American Statistical Association in Boston, Massachusetts,
August 23-26, 1976. This analysis of Area 13 (Project 57) 239°240py
data was performed prior to the kriging analyses by Delfiner and Gilbert
(1978) reported elsewhere in this volume.
The plutonium soil concentration data used in this latter paper is identical to that listed in
Appendix A of the present paper except as noted in Appendix A and
footnote 2 in Delfiner and Gilbert (1978).
320