Environmental Engineering Reference
In-Depth Information
12.3.1.1.
Random sampling
Here the sampling points are selected in a random manner (see Figure 12.5).
This method is used where there are no observable signs of contamination present
on the site. Random sampling is done by ensuring that at any stage of sampling the
selection of a particular sampling point is not influenced by other sampling points
already chosen. The most common way to ensure this aspect is to assign a number to
each sampling point in the ground and make a random selection of points with their
spatial coordinates, regardless of the mesh.
Figure 12.5.
Random sampling
The determination of random sampling points is generally carried out before
going to the site so as to avoid any bias related to topography or other aspects. The
number of samples can be determined based on the following expression [CRE 93]:
t
2
s
2
[12.1]
n
=
D
2
where:
−
t
is a number from a statistical table (distribution -
t
) for a given precision level
(90 or 95%); the degrees of freedom for
t
are first chosen and then modified by
reiteration;
−
s
2
is the variance, either already known from other investigations or estimated
by the expression
s
2
= (
R/4
)
2
where
R
is the estimated concentration range that can
be found in the samples [FRE 62];
−
D
is the variability in the average estimation of the potentially acceptable
contaminant concentration.
We should note that in small sites (< 0.5 ha) that have been uniformly
contaminated, as few as five to six samples can be enough based on this approach.
For larger areas, the sample number can go up to 25. There is little improvement in
the precision when the sample number exceeds 25 [WEB 90]. If a better evaluation
is necessary, the area has to be stratified.
