Statistical Sampling Strategies for Survey of Soil Contamination - Dealing with Contaminated Sites: From Theory Towards Practical Application - page 203

Fig. 4.15 Example of the
formation of composites from
aliquots taken on a grid.
Aliquots from locations with
the same symbol form a
composite sample
a threshold concentration, by measuring some or all individuals of some compos-
ite samples, assuming that some of the soil material of the individuals is kept for
later analysis (Gore and Patil 1994 ,Goreetal. 1996 ). In these methods, the com-
posite samples are sorted on the basis of their concentration in descending order.
For each composite sample, an upper bound of the maximum value is calculated by
assuming that the concentration of all except one of the individuals is zero. Under
this assumption, the maximum equals the total of the composite sample, i.e., the
composite sample concentration multiplied by the number of individuals in that
composite sample.
In the simple sweep-out method (Gore and Patil 1994 ), all individuals of the first
ordered composite sample are measured, and the largest value in this composite
sample is identified and recorded as the tentative maximum of the individuals of
all composite samples (global maximum). If the upper bound of the maximum of
the next ordered composite is smaller than this tentative global maximum, then it
is clear that this is the true global maximum, and no further composites need to
be measured. If the upper bound of the maximum for the next ordered composite
is larger than the tentative global maximum, then this composite may contain an
individual with a value larger than the tentative global maximum, and the individuals
of this composite have to be measured. If the maximum identified in this second
composite is larger than the maximum identified in the first composite, then the
tentative global maximum is updated. This procedure is continued until the tentative
global maximum is larger than the upper bound of the maximum in the next ordered
composite.
If the aim is to identify the two largest values, the second largest value is also
identified each time the individuals constituting a composite are measured, and the
procedure is continued until the tentative second largest value is smaller than the
upper bound of the maximum value in the next ordered composite. If the aim is to
identify all individuals with values larger than a threshold concentration, then the
procedure goes on until the individual with the largest value below the threshold
concentration has been identified.
Contrary to the simple sweep-out method, where, once a composite is selected,
all individuals except one constituting this composite are measured, the sequential