A Geostatistically Based Probabilistic Risk Assessment Approach - Evolving Application Domains of Data Warehousing and Mining - page 296

Approach 1: Classic Risk Assessment Concentration value has been obtained by means

of Chebyshev's inequality. The tests carried out
by means of software ProUCL have recognized
as representative a value of the Source Represen-
tative Concentration equal to 53,0779 µg/l, that
corresponds to the 97.5% of the mean. For this
approach, the surficial extension of the potentially
contaminated area is individuated from the area
delimited by the most external sampling meshes
containing a point in which at least one contami-
nant has been detected with a concentration higher
than CSC (APAT, 2008).

The base map of the location of piezometers has
been drawn up (Figure 11a) in order to evaluate
the uniformity in their spatial distribution and
therefore the possible presence of clusters and/or
of not sampled areas. The statistical parameters
reported in Table 2 have been determined: mean,
standard deviation, coefficient of variation, skew-
ness and kurtosis, and the histogram of the relative
frequencies has been calculated (Figure 11b).
From table 2, concerning the basic descriptive
statistics, it can be pointed out how the distribu-
tion of the pollutant shows high positive values
of skewness and kurtosis; the histogram of the
relative frequencies is not symmetric because of
the presence of high outliers, that means that the
variable has considerable displacements from the
normal distribution. The Source Representative

Approach 2: Geostatistical
Approach: UCL Method

The area in which the concentration values exceed
the CSC value for arsenic in groundwater (10µg/l)
has been determined by means of geostatistic

Figure 11. a) Base map of georeferenced arsenic values b) histogram of relative frequencies of arse-
nic

Table 2. Significant statistical parameters

Var Mean µg/l Min µg/l Max µg/l StD. µg/l Skewness Kurtosis Coeff of variation (%)

As 22.25 1 184 36.93 2.54 9.22 1.66