0% found this document useful (0 votes)
10 views1 page

My geoENV14 - Poster

Uploaded by

Kina Bouhafs
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
10 views1 page

My geoENV14 - Poster

Uploaded by

Kina Bouhafs
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 1

Geostatistical modeling of heavy-metal

stream-sediment samples

Ahcène BENAMGHAR *, J. Jaime Gómez-Hernández**


* ENSTP, 1 Rue Garidi Kouba Algiers, Algeria E-mail : [email protected]
** Research Institute of Water and Environmental Engineering, Universitat Politècnica de València, Spain E-mail : [email protected]

INTRODUCTION Figures
Improving prediction results is commonly a principal concern of geostatistical studies.
Semi-variogram modeling is the first step of any geostatistical project. However, the
standard way to compute semi-variograms may be questionable when spatial variability
presents some particular pattern, beyond anisotropy. A good example is given by Dagbert et
al. (1984), dealing with samples taken in a folded geological layer.
In such case, the classical Euclidean distance between sample points is no longer suitable to
compute the semi-variogram.

This paper investigates this problem using two different metrics to analyze the spatial
correlation of heavy-metal sediment concentrations measured on streams in a specific
catchment.

The first approach adopts non-Euclidean metrics and computes heavy-metal semi-variograms
using distances measured along the water courses. The second approach uses the Euclidean
distance for semi-variogram calculations.

CASE STUDY and METHODOLOGY Fig. 1: Left side: western part of Amblève watershed, studied area with samples (dots)
and natural streams (curved lines). Right side: Amblève area south-eastern Belgium.
Case study
The studied area is the Amblève watershed in the eastern part of Wallonia (Belgium) sampled
for heavy metals pollution investigation (Fig. 1). The campaign consisted in geochemical
sampling of stream network. Samples were collected within the recent alluvial, by means of soil
auger at the intersection of the lower edge of the bank and the active bed, with a density of 1
sample/km2 (nearly 1 sample every 200 m along stream).

In the late 90’s, a new soil pollution law was approved in Belgium. This has led the governmental
agency in charge of pollution control in water resources to draw detailed charts of pollution risk
using geostatistical techniques. These charts should then be compared with the underlying
geology to try to understand whether the origin of the pollution was anthropogenic or geological.

The consequences for decision makers are paramount since, unless the heavy metal spreading
obeys the first alternative, any remedial operation concentrated on the streams becomes
useless and unnecessary. In order to compare these two hypotheses, we investigated two
different approaches for semi-variogram computing. A one-dimensional (1D) approach, in which
the continuity is determined by the stream, and a two-dimensional (2D) one, in which continuity
is determined by the geological background. Six heavy metals were used for this purpose (Zn,
Co, As, Ni, Cu, Pb).

Methodology
Usually, semi-variograms are computed using the Euclidean distance as the separation
distance between samples; however, when analyzing variables which may be linked to
one-dimensional, non-rectilinear features such as stream paths, it maybe appropriate to
consider a different metric to measure distance. In our case, we have heavy metal concentration
along streams and we wish to consider the length as measured along the stream path as the
separation distance between samples, and compare the resulting variograms with those
obtained using the Euclidean distance. In the first approach, semi-variograms are one-
dimensional, whereas in the second approach semi-variograms are two-dimensional and may
Fig. 2: left column: 1D semi-variograms, Fig. 3: left column: 1D semi-variograms,
exhibit spatial anisotropy. We could interpret for the first approach that the “anisotropy” is so
right column: 2D omnidirectional semi- right column: 2D omnidirectional semi-
strong that continuity only happens along stream branches
variograms; experimental (dots) and fitted variograms; experimental (dots) and fitted
model (solid line). model (solid line).

RESULTS AND DISCUSSION


On one hand, experimental semi-variograms were computed using stream distance as the
separation distance (1D); they are displayed on the left columns of Fig. 2 and Fig. 3. The CONCLUSION
semi-variograms for As, Ni, Cu and Pb show an important nugget effect, some structure
for the shorter lags, and a chaotic behavior for longer lags.
The Amblève data set, which is part of the Wallonia geochemical data, has been used
On the other hand, and for the same elements, experimental omnidirectional semi-
in this paper as a testing area. Six heavy elements were analyzed in this study: As,
variograms were computed using the Euclidean distance as the separation distance (2D);
Ni, Cu, Pb, Zn and Co. The aim of this work was to examine whether the encountered
they are displayed on the right columns of Fig. 2 and Fig. 3. The semi-variograms show
heavy metal concentration in the stream sediments was related to the flow of water
a well structured pattern for all the elements, with some elements displaying two nested
along the streams (and therefore it could be of anthropogenic origin) or merely
structures, with two distinct ranges.
reflecting the natural background. By analyzing two approaches for computing the
semi-variograms we could conclude that much of the heavy metal concentrations is
In the light of these results, one can already observe that the 2D semi-variograms
controlled by the geological background.
reveal a regional spatial phenomenon as well as a local dependency when it exists. For
this case study, the conventional approach to semi-variogram computation is sufficient to
characterize the spatial variability of samples collected along the river network, and drives
us to conclude that the origin of the contamination is due to the background geology.

You might also like