ARTICLE IN PRESS

Applied Radiation and Isotopes 65 (2007) 10701077

www.elsevier.com/locate/apradiso

Spatial dispersion modeling of 90Sr by point cumulative semivariogram

at Keban Dam Lake, Turkey
Fatih Kulahca,b,, Zekai S- enb
a

Science & Arts Faculty, Physics Department, Frat University, 23169 Elazg, Turkey
Civil Engineering Faculty, Hydraulics Department, Istanbul Technical University, Maslak, 34469 Istanbul, Turkey

Received 13 January 2007; received in revised form 6 March 2007; accepted 16 March 2007

Abstract
Spatial analysis of 90Sr articial radionuclide in consequence of global fallout and Chernobyl nuclear accident has been carried out by
using the point cumulative semivariogram (PCSV) technique based on 40 surface water station measurements in Keban Dam Lake
during March, April, and May 2006. This technique is a convenient tool in obtaining the regional variability features around each
sampling point, which yields the structural effects also in the vicinity of the same point. It presents the regional effect of all the other sites
within the study area on the site concerned. In order to see to change of 90Sr, the ve models are constituted. Additionally, it provides a
measure of cumulative similarity of the regional variable, 90Sr, around any measurement site and hence it is possible to draw regional
similarity maps at any desired distance around each station. In this paper, such similarity maps are also drawn for a set of distances. 90Sr
activities in lake that distance approximately 4.5 km from stations show the maximum similarity.
r 2007 Elsevier Ltd. All rights reserved.
Keywords: Strontium; Point cumulative semivariogram; Spatial analysis; Lake; Articial radionuclide

1. Introduction
The total yield of atmospheric nuclear tests carried out
until the end of 1980 has been estimated at 217.2 Mt.
Nearly 90% of the resulting releases occurred before 1963
and only about 1% originated from tests between 1976 and
1980 (UNSCEAR, 1982). Depending on the particular type
of the test, up to 50% of the radioactive debris was
deposited in the vicinity of testing sites giving rise to local
fallout. The remainders were injected into the atmospheric
strata and entered the global circulation of air masses. The
portion injected into the stratosphere (i.e., at the altitude
above 917 km) represents a major reservoir of worldwide
(global) fallout. It accounts for most of the worldwide
contamination of long-lived ssion products (UNSCEAR,
1982; BSEP, 1995).
Corresponding author at: Science & Arts Faculty, Physics Department, Frat University, 23169 Elazg, Turkey. Tel.: +90 424 2370000/3835;
fax: +90 424 2330062.
E-mail address: fatihkulahci@rat.edu.tr (F. Kulahc).

0969-8043/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apradiso.2007.03.012

The semivariogram (SV) (or variogram), one of the

fundamental tools of geostatistical analysis, is used
commonly to measure the spatial similarity of a regionalized variable. General procedures are available for
obtaining experimental SVs and tting theoretical smooth
mathematical models. Considerable emphasis is placed on
anisotropic models and their relation to directional
geologic features (Clark, 1979). A variety of important
topics related to the SV are discussed, including the
proportional effect, relative SVs, nested structures, and
the geometric meaning of many SV models (Journel and
Huijbregts, 1978). Complexities in the development of SV
models include the existence and dealing with outliers,
robustness of SVs, and SVs in curved coordinate space
(Sinclair and Blackwell, 2002).
Traditional variance techniques and classical autocorrelation methods are not helpful to account directly for the
regional dependence or for the variability in terms of
sample positions (Luetkepohl et al., 2004). The downsides
are due to either non-normal distribution of data and/or
irregularity of sampling positions. However, the SV

ARTICLE IN PRESS
F. Kulahc, Z. S- en / Applied Radiation and Isotopes 65 (2007) 10701077

technique developed by Matheron (1963) and used by

many researchers (Bunzl et al., 1994; S- en, 1997; Golosov
et al., 1999; Kumar et al., 1999; Sanchez-Vila et al., 2004)
in rather different elds, such as radon risk mapping,
mining, heavy metal analysis, and contaminated soil can be
applied to characterize regional structural variability. None
of the available regional behavior measuring techniques is
able to provide specic calculations at a single sampling
point. However, S- en (1989, 1992, 1998) developed the
concept of cumulative SV (CSV) and point cumulative
semivariogram (PCSV) techniques, which is capable of
quantifying the point properties of any regionalized
variable.
Modeling of the big ecosystems like lakes is rather
difcult (McEvoy, 1980; Brugmann, 1984; Benes and
Cern k, 1992; Choppin and Wong, 1996; Maringer et al.,
2004). It is shown in this paper that the application of
PCSV helps to overcome such a difculty and therefore the
PCSV modeling technique is applied to 90Sr data obtained
from the analysis of Keban Dam Lake surface water
samples. The PCSV technique also provides an opportunity to construct regional similarity maps at xed levels,
which guide to further interpretations such as the regional
heterogeneity.
2. Mathematical theory
The mathematical denition of the variogram is given by
Matheron (1963) as
gDx; Dy 12  fZx Dx; y Dy  Zx; yg2 ,

(1)

where Z(x, y) is the regionalized value of interest at

location point (x, y), and e[ ] is the statistical expectation
operator. Note that the variogram, g, is a function of the
separation (distance), (Dx, Dy), between two points and not
a function of the specic location (x, y). This theoretical
mathematical denition is a useful abstraction, but not
easy to apply to observed values.
On the other hand, let {(x1, y1, z1), (x2, y2, z2), y, (xn, yn,
zn)} be a sequence of n irregularly located point data, where
(xi, yi) is the location of observation point i, and zi is the
associated measured regionalized value. Hence, there are
n(n1)/2 distinctive pairs of observations between n points.
For each of these pairs, it is possible to calculate the
associated separation vector:
Dxi;j ; Dyi;j xi  xj ; yi  yj .

(2)

To infer the variogram for a particular separation vector,

(Dx, Dy), it is required to use all data pairs whose
separation vectors are almost equal to each other.
Dxi;j ; Dyi;j  Dx; Dy.

(3)

In general, even though there may be several points, say

7, there will be (7  6)/2 21 such pairs, which means that
as the number of measurement sites, n, increases the

1071

number of SV calculation pairs increases according to the

aforementioned n(n1)/2 formulation.
Let S(Dx, Dy) be the set of all such pairs, then
SDx; Dy fi; jjDxi;j ; Dyi;j  Dx; Dyg.

(4)

Furthermore, let N(Dx, Dy) represent the number of such

pairs in S(Dx, Dy), then the experimental SV is dened as
g^ Dx; Dy

X
1
zi  zj 2 .
2NDx; Dy i;j2SDx;Dy

(5)

Herein, one-half of the differenced squares of the

z-values is summed over all pairs of observations separated
by approximately that vector (Matheron, 1963; Barnes,
2006).
The CSV is equal to the successive summation of the
SV values, which yields a non-descending scatter diagram.
The CSV has all the advantages claimed for SV, and it
appears to be a better tool in the identication of
theoretical models for the spatial variability. Furthermore,
standardization of CSV provides an equal footing base for
the identication of regional stationary stochastic models
(S- en, 1992).
On the other hand, PCSV functions are presented for
the identication of the spatial behavior of any regionalized variable around any concerned sampling point. In
other words, it presents the regional effect of all of the
other sites within the investigation area on a nite area
surrounding the sampling point location. Consequently,
the number of PCSVs is equal to the number of stations.
Hence, their comparisons with each other and their
respective groupings provide valuable information about
the heterogeneity of the regionalized variable concerned.
The following steps lead to the desired sample PCSV for
the concerned site among n observation (measurement,
sampling) sites.
(i) Calculate distance between the concerned site and the
remaining sites. If there are n sites, the number of
different distances is n1, dj (j 1, 2, y, n1).
(ii) For each pair, nd the half-squared differences
between data values, in this case 90Sr values. In this
way, each distance will have its half-squared value, 1/
2(zizj)2 where zi and zj are the regionalized 90Sr
variable values at the concerned and jth sites,
respectively.
(iii) Plot distances versus corresponding successive cumulative sums of half-squared differences. This procedure
gives a non-decreasing function, which is the sample
PCSV at the concerned site. Its mathematical expression is given as

gd j

n1
1X
zi  zj 2 .
2 j1

(6)

(iv) Apply previous steps by considering different concerned sites, to give n sample PCSVs.

ARTICLE IN PRESS
1072

F. Kulahc, Z. S- en / Applied Radiation and Isotopes 65 (2007) 10701077

Fig. 1. The locations of sampling stations on Keban Dam Lake in Turkey.

3. Experimental
3.1. Research area
Uluova Region at Keban Dam Lake has almost 35 km
length and approximately 1012 km width, with a surface

area of
eld is
38150 N
altitude
1000 m.
covered

360370 km2. However, the area of the research

approximately 250 km2 and lies at the latitude
and 38140 E longitude in Eastern Turkey. The
of the lake plain changes between 845 and
One-third of this plain, i.e., about 110 km2, is
by the Keban Dam reservoir lake. This region is

ARTICLE IN PRESS
F. Kulahc, Z. S- en / Applied Radiation and Isotopes 65 (2007) 10701077

encircled by high mountains such as Mastar Mountain

(2171 m), C
- elemlik (1658 m), and Sivrice (1908 m) in the
south; Hasret Mountain (1621 m) in the north; and
Meryem Mountain (1471 m) in the west. Hasret Mountain
limits Uluova in the west. The altitude is 1621 m in the
western part of Hasret Mountain. It reaches 1724 m in the
Kartal Hill in the eastern part. The evaluation decreases
eastwards so that it reaches 1200 m in Murat Valley. The
slope values change between 15% and 45% in this
mountainous area.
The surface water samples taken from approximately
20 cm depth have been collected from 40 stations, in order
to characterize the lake during March, April, and May
2006 (Fig. 1).
3.2. Determination of

90

Sr

A known amount of stable Sr, in the form of strontium

nitrate Sr(NO3)2, is added as a carrier. The alkaline earths
and rare-earths carrier is precipitated as the carbonate to
concentrate the radio-strontium. Along with the radionuclides of Sr, it is separated from other radioactive
elements and inactive sample solids by precipitation as
Sr(NO3)2 from fuming nitric acid solution. The Sr carrier,
together with the radionuclides of Sr, is precipitated nally
as SrCO3, which is dried, weighed to determine recovery of
carrier, and immediately measured for radioactivity. The
activity in the nal precipitate is due to radioactive Sr only,
because all other radioactive elements have been removed.
Due to the impossibility to separate the 89Sr and 90Sr
isotopes by any chemical procedure, the 90Sr amount is
determined by separating the measurement of its daughter
activity, 90Y. After the equilibrium is reached, the 90Y
activity is exactly equal to the activity of 90Sr. Two
Table 1
Locations and

90

1073

alternate procedures are given for the separation of 90Y. In

the rst method, 90Y is separated by extraction into tributyl
phosphate from concentrated nitric acid solution. The
second method consists of an addition of yttrium carrier,
separation by precipitation as yttrium hydroxide, Y(OH)3,
and nally precipitation of yttrium oxalate for counting. In
order to determine 90Sr activity was used the second
method and the activity of 90Sr was calculated as follows
(Clescerl et al., 1989):
90

Sr Bq=l

net cpm  0:0166

,
abcdf elt

(7)

where a is the counting efciency for 90Y, b is the chemical

yield of extracting or precipitating 90Y, c is the ingrowth
correction factor if not in secular equilibrium, d is the
chemical yield of Sr gravimetrically, f is the volume
of sample, liters, and elt is the 90Y decay factor
(t1/2 2.671d). In order to determine 90Sr radioactivity,
beta spectroscopy is used which has been performed by a
low background plastic b-scintillation system supported by
a plastic b-scintillator (2059) and a photomultiplier tube
through a SR8 dual radiation counter from NE Technology Inc. Lead shielding was used for both systems for the
protection against external radiation. The precision (random error) at the 95% condence level was 10% and 12%
for the two sets of paired samples.
4. Results and discussion
The PCSV technique is applied to 90Sr radioactivity
concentrations measured at 40 irregularly distributed sites
in the Keban Dam Lake as shown in Table 1.
The PCSV versus distance plots obtained from 90Sr
radioactivity analyses at 40 sampling stations are drawn

Sr radioactivity measurements

Station

Latitude (North)

Longitude (East)

90

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

38.37
38.37
38.36
38.37
38.36
38.37
38.37
38.38
38.38
38.39
38.39
38.40
38.40
38.40
38.40
38.39
38.39
38.38
38.38
38.38

39.24
39.25
39.25
39.26
39.27
39.27
39.29
39.30
39.31
39.33
39.34
39.35
39.34
39.33
39.32
39.31
39.30
39.29
39.28
39.27

0.10
0.09
0.11
2.02
0.89
0.97
0.10
0.09
0.10
0.28
0.28
0.38
1.22
1.18
1.17
1.84
1.78
2.23
1.35
1.45

Sr (Bq/l) (  102)

Station

Latitude (North)

Longitude (East)

90

21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40

38.38
38.37
38.38
38.38
38.39
38.40
38.40
38.40
38.40
38.41
38.41
38.41
38.41
38.42
38.41
38.40
38.40
38.39
38.38
38.38

39.26
39.24
39.25
39.26
39.28
39.29
39.30
39.31
39.32
39.34
39.33
39.32
39.30
39.30
39.29
39.29
39.28
39.27
39.25
39.23

1.35
0.09
0.11
0.12
0.08
0.09
0.10
1.33
1.40
1.33
1.80
1.88
1.80
2.30
0.23
0.24
0.23
1.55
1.50
1.55

Sr (Bq/l) (  102)

ARTICLE IN PRESS
F. Kulahc, Z. S- en / Applied Radiation and Isotopes 65 (2007) 10701077

1074

and then their similarity groups yielded ve PCSV models

as shown in Fig. 2, where each group is labeled as A, B, C,
D, E model. For the sake of brevity, only one sample PCSV
is presented in the same gure and the ve PCSV models
are plotted in idealized form in Fig. 2 from A to E. Each

6.50x10-3
C

5.35x10-2

6.00x10-3

5.30x10-2
Point CSV

5.50x10-3
Point CSV

one can be described according to its characteristic shape.

For instance, the rst four models are bounded because
they reach maximum PCSV values, which are known as the
sill in the geostatistical literature (Matheron, 1963). For the
theoretical exponential and Gaussian model, the sill is

5.00x10-3
C1

4.50x10-3

3.50x10-3
0.00

5.20x10-2

Co {
5.10x10-2
0.00

a
4.00

8.00 12.00 16.00

Distance (km)

20.00

No.1, modelA

7.52x10-2

4.64x10-2

7.48x10-2

4.60x10-2
4.56x10-2

Co

8.00
12.00
Distance(km)

a
4.00

8.00
12.00
Distance (km)

16.00

No.22, model C

7.44x10-2
7.40x10-2

7.32x10-2
0.00

4.00
8.00
Distance(km)

No.35, model D

3.00x10-2

Point CSV

2.96x10-2

2.92x10-2

2.88x10-2

2.84x10-2
0.00

16.00

7.36x10-2

C1

4.48x10-2
0.00

a
4.00

No.25,model B

4.68x10-2

Point CSV

Point CSV

5.25x10-2

5.15x10-2
C1

4.00x10-3
Co {

4.52x10-2

4.00

8.00
12.00
Distance (km)

16.00

No.14, model E
Fig. 2. The examples of point cumulative semivariograms (PCSV) of

90

Sr concentrations.

12.00

ARTICLE IN PRESS
F. Kulahc, Z. S- en / Applied Radiation and Isotopes 65 (2007) 10701077
Table 2
The PCSV models assigned to stations
PCSV
models

Stations

Model A

1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 16, 17, 18, 19, 21, 23, 24,
26, 29, 30, 31, 32, 33, 34, 37, 38, 39
20, 25, 27, 28
22, 40
35, 36
14, 15

Model
Model
Model
Model

B
C
D
E

approached asymptotically and these models never reach

the sill (Journel and Huijbregts, 1978).
None of the PCSVs modeling passes through the origin,
which means that all the sampling stations are under the
inuence of some regional control structure (geology, deep
currents, etc.). This study is adopted as a pilot regional
study due to its interesting geological and geographic
characteristics. The lake basin is completely covered with
the clay minerals (Gunek and Tonbul, 1995).
The PCSV categories at Keban Dam Lake stations are
given in Table 2.
Models A, B, and C in Fig. 2 show a discontinuity
(Co) at the origin, followed by a gradual increase in g(h) as
the lag increases to h a (the range); for lags beyond
the range (h4a), the experimental values of g(h) are
roughly constant (C). These models have an intercept on
the y-axis except D model, which intersects the horizontal
x-axis. However, all the PCSV models have a gradual
increase in g(h) up to sample spacing a at which g(h)
becomes uniform.
Models A, B, and C in Fig. 2 are characterized by two
components: (a) a purely random component referred
to as the nugget effect, or Co; and (b) a structured
component in which the structure is characterized by the
range a of an autocorrelation function. The range is the
distance over which average variability for the structured
component rises from zero to C1 (or from Co to C).
Furthermore, the models A and B show Gaussian shape up
to spacing a. As for model C, it appears in the form of an
exponential model (model C, Fig. 2), but approaches the
sill more sharply. Herein, exponential model is
g(h) C[1eh], where g(h) is the semivariogram and C
is the scale for the structured component of the variogram
(Cressie, 1991). The range, therefore, is shorter than the
exponential range. Moreover, the slope decreases for
models A, B, C, and D with increasing distance gradually
reaching the saturation. The smaller the slope, the less is
the effect of the atmospheric environment as the global
factor, and there is more effect of the geologic formation
(the absorption to elements on clay minerals, etc.) as the
regional factor.
Model C in Fig. 2 shows an exponential model variation,
which approaches the sill asymptotically. The variance is
lower than the spherical variance for all distance values less
than the range. Besides, this model applies when the spatial

1075

dependence decreases exponentially with increasing distance, which disappears completely only at innite
distances.
The PCSV in model D in Fig. 2 intersects the horizontal
axis, which indicates that 90Sr radioactivity dispersion
between the concerned site and the remaining sites at
distances less than the distance of the intersection is almost
the same.
Model E in Fig. 2 has two parameters: the slope in
addition to the intercept on the vertical axis. The line
rises without limit and consequently has no sill. It applies
when the spatial variability observed for a variable
increases linearly with distance and never levels off.
Its shape near the origin comes close to that of the
Gaussian and exponential models. S- en (1989) has shown
that the slope of the straight line of linear model is
represented exactly by the variance of the underlying
random eld or the regionalized variable, which is 90Sr in
this case. Hence, the smaller the slope, the smaller is the
random uctuation in the regionalized variable. For this
reason, in the model E, due to the sample PCSVs
appearance as a single straight line, it implies the existence
of rather homogeneous areas of inuence around these
stations.
Finally, after the explanations of the PCSV models
earlier in the text, and in order to show the similarities of
each station to others with respect to 90Sr radioactivity
measurements, representative distances of Ro 1.5, 4.5,
and 7.5 km are considered from which corresponding
PCSV values are read for each station. The regional
distribution of these values on the sampling points, as in
Fig. 1, results in a similarity map. Fig. 3 shows the
similarity maps obtained from the PCSV graphs at
Ro 1.5, 4.5, and 7.5 km distances and Table 3 presents
the PCSV values obtained for these distances.
The similarity maps in Fig. 3ac give very detailed
information about the 90Sr radioactivity dispersion in the
lake. According to S- en (1998), if the two PCSVs at
different sites follow the same pattern within the limits of
sampling errors, then they are said to be similar. Such a
similarity implies the existence of heterogeneity between
the two sites.
As seen in Fig. 3, there is not much variation with the
distance Ro and, especially in Fig. 3a and b, the PCSV
values are almost at the same. However, the central
regions for all the Ro distances of the lake basin show
elevated radioactivity levels. In particular, the radioactivity of lake increases gradually towards southwest. It
has already been stated that there are substantial clay
layers at the bottom of the lake (Gunek and Tonbul,
1995). Clay absorbs radioactive minerals, so this area is
expected to be higher in 90Sr radioactivity than other
regions.
Similarity contours (for Ro 4.5 km) in Fig. 3b are
more intensive than the others. Essentially, levels of
similarity for Keban Dam Lake in Fig. 3 may be arranged
as b4a4c.

ARTICLE IN PRESS
F. Kulahc, Z. S- en / Applied Radiation and Isotopes 65 (2007) 10701077

1076

Ro= 1.5 km
38.41

Latitude

38.4
38.39
38.38
38.37
38.36
39.24 39.25 39.26 39.27 39.28 39.29 39.3 39.31 39.32 39.33 39.34
Longitude
Ro= 4.5 km
38.42

Latitude

38.4

38.38

38.36
39.24

39.26

39.28

39.3
39.32
Longitude

39.34

Ro= 7.5 km
38.42

Latitude

38.4

38.38

38.36
39.24

39.26

39.28
39.3
Longitude

39.32

39.34

E
S

Fig. 3. Similarity maps for Ro 1.5, 4.5, and 7.5 km on the Keban Dam Lake.

5. Conclusions
Keban Dam Lake in Turkey is chosen as a pilot area to
study the effect of the global fallout and Chernobyl nuclear
accident. The PCSV technique is employed for the
modeling purpose in order to identify possible regional
features and similarities of 90Sr. The following ad-

vantages make the PCSV model interesting in practical

applications:
(i) The PCSV model may be used for irregularly distributed sample positions within the study region.
(ii) The PCSV does not require complex mathematical
theorems and it is fairly easy to apply in practical studies.

ARTICLE IN PRESS
F. Kulahc, Z. S- en / Applied Radiation and Isotopes 65 (2007) 10701077

Acknowledgments

Table 3
Similarity levels at different distances
Station

Fixed PCSV levels

Ro 1.5 km

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40

3

3.9  10
6.2  103
8.6  103
1.0  102
1.4  102
1.5  102
1.6  102
1.8  102
2.1  102
2.3  102
2.5  102

2.8  102
2.9  102
3.1  102
3.2  102
3.4  102
3.7  102
4.2  102
4.3  102
4.5  102
4.6  102
4.9  102
5.1  102
5.3  102
5.5  102
5.8  102
6.0  102
6.1  102
6.3  102
6.4  102
6.7  102
7.0  102

7.5  102
7.6  102
7.8  102
8.0  102
7.5  102

1077

Ro 4.5 km
3

4.2  10
6.7  103
9.0  103
1.2  102
1.4  102
1.5  102
1.6  102
1.9  102
2.1  102
2.3  102
2.5  102
2.7  102
2.8  102
3.0  102
3.1  102
3.2  102
3.5  102
3.9  102
4.2  102
4.4  102
4.5  102
4.7  102
4.9  102
5.1  102
5.4  102
5.6  102
5.9  102
6.0  102
6.2  102
6.3  102
6.5  102
6.8  102
7.1  102
7.3  102
7.5  102
7.7  102
7.9  102
8.1  102
8.2  102
8.4  102

Ro 7.5 km
4.7  103
7.4  103
9.6  103
1.3  103
1.5  103
1.6  103
1.7  103
2.0  103
2.2  102
2.4  102
2.6  102
2.7  102
2.9  102
3.0  102
3.1  102
3.4  102
3.6  102
4.1  102
4.3  102
4.4  102
4.6  102
4.7  102
5.0  102
5.2  102
5.5  102
5.8  102
6.0  102
6.1  102
6.2  102
6.4  102
6.6  102
6.9  102
7.1  102
7.4  102
7.6  102
7.8  102
8.0  102
8.1  102
8.3  102
8.5  102

The PCSV modeling technique is applied to 90Sr surface

water data taken from Keban Dam Lake. After the
identication of individual PCSV at each location, they
are then grouped according to their similarity, which
led to ve distinctive groups in the study area. All together
there were 40 locations and hence 40 PCSVs, but the
representative number is deduced as ve distinctive
PCSVs. The spatial dispersion of 90Sr radionuclide is seen
to be heterogeneous and spatially uncorrelated at big
distances.
It is seen that changing in the lake of the 90Sr approves
generally the model A. In this situation, last but not the
least, changing of the 90Sr radioactivity, nowadays, is
related to geological structure of the study area rather than
atmospheric effects.

This work is a part of the post-doctorate research project

supported by The Scientic and Technical Research Council
of Turkey (TUBITAK). The authors would like to thank
TUBITAK for nancial support and encouragement.
References
Barnes, R., 2006. Variogram Tutorial. Golden Software Inc., p. 6.
Benes , P., Cern k, M., 1992. Kinetics of radionuclide interaction with
suspended solids in modeling the migration of radionuclides in rivers.
J. Radioanal. Nucl. Chem. 159, 1526.
Brugmann, L., 1984. Electrochemical speciation of trace metals in sea
water. Sci. Total Environ. 37, 4160.
BSEP (Baltic Sea Environment Proceedings), 1995. Radioactivity in the
Baltic Sea 19841991, no. 61. Baltic Marine Environment Protection
Commission, Helsinki Commission.
Bunzl, K., Kretner, R., Szeles, M., Winkler, R., 1994. Transect survey of 238U,
228
Ra, 226Ra, 210Pb, 137Cs and 40K in an agricultural soil near an exhaust
ventilating shaft of a uranium mine. Sci. Total Environ. 149, 225232.
Choppin, G.R., Wong, P.J., 1996. Current status of radioactive waste
disposal. J. Radioanal. Nucl. Chem. 203, 575590.
Clark, I., 1979. Practical Geostatistics. Applied Science Publishers,
London, p. 281.
Clescerl, L.S., Greenberg, A.E., Trussell, R.R., 1989. Standard Methods
for Examination of Water & Wastewater. American Public Health
Association, Washington, pp. 748.
Cressie, N., 1991. Statistics for Spatial Data. Wiley, New York, 900pp.
Golosov, V.N., Walling, D.E., Panin, A.V., Stukin, E.D., Kvasnikova,
E.V., Ivanova, N.N., 1999. The spatial variability of Chernobylderived 137Cs inventories in a small agricultural drainage basin. Appl.
Radiat. Isot. 51, 341352.
Gunek, H., Tonbul, S., 1995. Uluova ve c- evresinde toprak erozyonu ve
alnmas gereken onlemler. Univ. Firat Social Sci. 7, 79100.
Journel, A., Huijbregts, A., 1978. Mining Geostatistics. Academic Press,
London, p. 710.
Kumar, U.S., Navada, S.V., Rao, S.M., Nachiappan, Rm.P., Kumar, B.,
Krishnamoorthy, T.M., Jha, S.K., Shukla, V.K., 1999. Determination
of recent sedimentation rates and pattern in Lake Naini, India by 210Pb
and 137Cs dating techniques. Appl. Radiat. Isot. 51, 97105.
Luetkepohl, H., Kraetzig, M., Phillips, P.C.B., 2004. Applied Time Series
Econometrics. Cambridge University Press, West Nyack, NY.
Maringer, F.J., Tesch, R., Hrachowitz, M., Gruber, V., 2004. Long-term
environmental monitoring and application of low-level 3H, 7Be, 137Cs
and 210Pb activity concentrations in the non-biotic compartments of
the Danube in Austria. Appl. Radiat. Isot. 61, 313317.
Matheron, G., 1963. Principles of geostatistics. Econ. Geol. 58, 12461266.
McEvoy, J.E., 1980. Lead modelinga decision-making tool. Sci. Total
Environ. 16, 231237.
Sanchez-Vila, X., Carrera, J., Gomez-Hernandez, J., 2004. GeoENV IV:
Geostatics for Environmental Applications. Kluwer Academic Publishers, Dordrecht, p. 498.
S- en, Z., 1989. Cumulative semivariogram model of regionalized variables.
Math. Geol. 21, 891903.
S- en, Z., 1992. Standard cumulative semivariograms of stationary
stochastic processes and regional correlation. Math. Geol. 24, 417435.
S- en, Z., 1997. Objective analysis by cumulative semivariogram technique
and its application in Turkey. J. Appl. Meteorol. 36, 17121724.
S- en, Z., 1998. Point cumulative semivariogram for identication of
heterogeneities in regional seismicity of Turkey. Math. Geol. 30, 767787.
Sinclair, A.J., Blackwell, G.H., 2002. Applied Mineral Inventory
Estimation. Cambridge University Press, West Nyack, NY, p. 192.
UNSCEAR, 1982. Ionizing Radiation: Sources and Biological Effects.
United Nations Scientic Committee on the Effects of Atomic
Radiation, New York.