See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/279914688

Kriging of Groundwater Level -A Case Study of Dibdiba Aquifer in Area of

Karballa-Najaf

Article · April 2008

CITATION READS

1 260

1 author:

Waqed Al-Mussawi
University of Kerbala
34 PUBLICATIONS   99 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Optimum Design of Sewer Networks View project

Investigation of Dibdibba Aquifer in Kerbala,Iraq View project

All content following this page was uploaded by Waqed Al-Mussawi on 09 July 2015.

The user has requested enhancement of the downloaded file.

Journal of Kerbala University , Vol. 6 No.1 Scientific .March. 2008

Kriging of Groundwater Level - A Case Study of Dibdiba Aquifer in

Area of Karballa-Najaf
Assist. Teacher –Waqed Hameed Al-Mussawi
Department of Computers - College of Science- Karballa University
Email:[email protected]

Abstract:
The application of the spatial statistical technique (kriging) is used in this research, for the spatial
analysis of groundwater levels is shown. The data set consists of groundwater levels measured at
about 13 hand dug wells were selected in the studied area for the observation water table (vary for
ten month, from June 2002 to March 2003 ) in an area of 20x20 km2 Dibdiba hydrologic basin;
which lies between Karbala and Najaf provinces. With the use of measured elevations of the water
table, experimental semivariograms were constructed that characterizes the spatial variability of the
measured groundwater levels. The experimental semivariograms were fitted into many models as
Spherical, exponential and gaussian semivariogram. The finally selected models were used to
estimate the groundwater levels and estimation variance (which express the accuracy of the
estimated groundwater levels) at the nodes of a square grid of 2.5km x 2.5km and to develop
corresponding contour maps .Also, used the Inverse Square Distance (ISD) method in order to
interpolate the groundwater levels for the study area. It was found that ISD method resulted in higher
errors as compared to kriging method. The groundwater table maps resulted by kriging method were
compared with the groundwater table maps prepared using the ISD method.
: ‫الخالصة‬
‫( فٌ انتخًٍَ انًكبٌَ نًستوى انًَابِ انووفَةااذازم يوًو اة‬Kriging)‫تى فٌ هزا انبحث تطبَق تقَُة انتخًٍَ االحصبئٌ انكشٍكُك‬
‫) بئاشا يختابسيفٌ يُطقاة انذساسة انًشاشبةيا شٍب ةنًاذي اشي ايا شميٍ يا ش‬31 ٍ‫بَبَبم نًستوى انًَبِ انووفَة انًقبساة حقهَاب يا‬
ٌ‫كى شض)يٍ حوض انذبذبة انشيهاٌ باٍَ ياذٍُت‬20 x ‫كى طول‬20 ‫)كى‬000 ‫ )انًسبحة‬2001 ‫ اني ي ش اراس‬2002 ‫حزٍشاٌ نعبو‬
‫كشبالء ةانُوف انًقذستٍَا‬
‫(توشٍبَاة ةانتاٌ هاٌ ازء ياٍ انت ابٍش‬semivariograms) ‫ببستخذاو انًُبسَب انًقبسة نًستوٍبم انًَبِ انووفَة تى اَ بء يخططابم‬
(Spherical, ‫( يُ ب‬Semivariograms) ٍ‫( نعذي يودٍالم ي‬fitted)‫انًكبٌَ انًقبس نًستوى انًَبِ انووفَةا تى ًم تقشٍب يالئى‬
‫ا‬exponential and Gaussian)
‫ةبعذ ا شاء االذتببسام نعذي يودٍالم تى اذتَبس انًودٍم انُ بئٌ يٍ هزِ انًودٍالم نتخًٍَ يستوى انًَبِ انووفَة فٌ انًُطقة ةكزنك‬
‫كى)ةسساى ذبسطاة‬2‫ا‬2 x ‫كاى‬2‫ا‬2 ‫يقذاس انت بٍش انزً ٍعبش ٍ دشة انتخًٍَ فٌ يستوى انًَبِ انووفَة)فٌ َقبط ن ابكة يشبعاة بببعابد‬
‫كُتوسٍة نًستوى انًَبِ انووفَةا‬
‫كزنك تى تخًٍَ يستوى انًَبِ انووفَة نًُطقة انذساسة بطشٍقة يشبع انًسبفبم انًعكوسة‬
‫( حَث ة ذ اٌ َتبئج هزِ انطشٍقاة تحاوً هاي ذطابء كبَاش ارا يبشوسَائ بُتابئج‬Inverse Square Distance (ISD) method)
‫( اكزنك تى يقبسَة انخبسطاة انكُتوسٍاة نًساتوى انًَابِ انووفَاة فاٌ انًُطقاة انًشساوية بطشٍقاة‬Kriging) ‫انتخًٍَ بطشٍقة انكشٍكُك‬
‫( يع انخبسطة انكُتوسٍة انتٌ سسًئ بطشٍقة يشبع انًسبفبم انًعكوسةا‬Kriging)‫انكشٍكُك‬

Aim of the study

Using a new technique (kriging) to guess the fluctuating of groundwater levels for the study area (AL-
Dibdiba Basin). Also confirm the accuracy of this method by comparing the results with other methods,
such as Inverse Square Distance (ISD) method.

170
Journal of Kerbala University , Vol. 6 No.1 Scientific .March. 2008

Introduction
Groundwater is one of the major sources of water. Management of this resource is very important to
meet the increasing demand of water for domestic, agricultural and industrial use. Various management
measures need to know the spatial and temporal behavior of groundwater. Observed groundwater levels
serve as one of the main input data in studies related to groundwater simulation for various purposes as
required in water balance studies, estimation of groundwater recharge potential, in the design of
drainage structures etc. However, the measurement of groundwater levels are generally carried out at
spatially random locations in the field, whereas, most of the groundwater models requires these
measurement at a pre-specified grid. Some interpolation method is generally employed to get these
values at grid nodes. The accuracy with which this interpolation can be carried out affects the accuracy
of the model output.

Literature review
Basic concepts of the kriging technique and its application to natural phenomenon have been reviewed
by the ASCE Task Committee (1990a, b). Kriging of groundwater levels was carried out by Delhomme
(1978); Volpi and Gambolati (1978); Aboufirassi and Marino (1983); Virdee and Kottegoda (1984);
Kumar (1996) and Kumar and Ahmed (2003); Kumar and Remadevi (2006). In this research,
application of kriging to interpolate the groundwater levels, as observed in the part of Dibbdiba
hydraulic basin has been shown.

Methodology
Kriging is a technique of making optimal, unbiased estimates of regionalized variables at unsampled
locations using the structural properties of the semivariogram and the initial set of data values. Kriging
takes into consideration the spatial structure of the parameter and hence score over other methods like
arithmetic mean method, nearest neighbour method, distance weighted method, and polynomial
interpolation. Also, kriging provides the estimation variance at every estimated point, which is an
indicator of the accuracy of the estimated value. This is considered as the major advantage of kriging
over other estimation techniques.
Spatial variations with interdependence are commonly described with a variogram (Isaaks and
Srivastava, 1989).In geostatictics, the concept of variance from classic is extended to semi variance
.Considering a transect with equally spaced samples and measurements of ground water level z, a set of
values z(x1), z(x2) …z (xn) at location x1, x2…xn were obtained .The experimental semivariance γ*(h) is
estimated as:

N (h)

 z( x )  z( x  h) 
1 2
γ*(h) = i i (1)
2 N ( h) i 1

Where:
N (h) = the number of pairs separated by lag distance h;
z (xi) = measured variable value at point i; and
z (xi+h) = measured variable value at point i+h.

Experimental semivariograms were calculated for ten months between the years 2002-2003 using the
computer software (Geostatistics for the environmental sciences) Ver. (5.1).A lag distance of 2.5km
and a tolerance of 1.25km were used for the calculation of semivariogram.
The experimental semivariograms were fitted with various theoretical models like spherical,
exponential, gaussian, linear and power by the weighted least square method. The theoretical model
that gave minimum standard error is chosen for further analysis. The adequacy of the fitted models was

171
Journal of Kerbala University , Vol. 6 No.1 Scientific .March. 2008

checked on the basis of validation tests. In this method, known as jackknifing procedure, kriging is
performed at all the data points, ignoring, in turn, each one of them one by one. Differences between
estimated and observed values are summarized using the cross-validation statistics (de Marsily and
Ahmed 1987): mean error (ME), mean squared error (MSE), and kriged reduced mean error (KRME),
and kriged reduced mean square error (KRMSE). If the semivariogram model and kriging procedure
adequately reproduce the observed value, the error should satisfy the following criteria.
1 N *
ME   ( z ( xi )  z( xi ))  0
N i 1
(2)

1 N
MSE   ( z * ( xi )  z ( xi )) 2 Minimum (3)
N i 1

 ( z 
1
KRME  ( xi )  z ( xi ))
 ki   0
*
(4)
N
1 N
 *
( xi )  z ( xi )) 2 
KRMSE 
N
 ( z
i 1
 ki2 
1 (5)

Where, z*(xi), z (xi) and  ki2 are the estimated value, observed value and estimation variance,
respectively, at points xi . N is the sample size. As a practical rule, the MSE should be less than the
variance of the sample values and KRMSE should be in the range 1±2√2/N.
In all interpolation techniques, interpolated value of z at any point x0 is given as the weighted sum of
the measured values i.e.

N
z * ( x 0 )   i z ( x i ) i = 1, 2, 3,..........., N (6)
i 1

Where:
λi is the weight for the observation z at location xi. In kriging, the weights λi are calculated by
equation (7) so that z*(x0) is unbiased and optimal (minimum squared error of estimation).

j 1
 j  ( xi , x j )     ( xi , x 0 ) i= 1, 2, 3,..........., N

(7)
N


j 1
j 1

Where:
μ = Lagrange multiplier
γ (xi, xj) = semivariogram between two points xi and xj.
The minimum squared error estimation is also a measure for the accuracy of estimates, which is known
as estimation variance, or kriging variance, and is given by
N
 k2 ( x0 )   i  ( xi , x0 )   (8)
i 1

172
Journal of Kerbala University , Vol. 6 No.1 Scientific .March. 2008

Inverse Square Distance (ISD) method, widely used in geohydrology, was also employed to interpolate
the groundwater level data. In this method, the weights λi are inversely proportional to the square of
distance from the estimation point as:

1
(d oi ) 2
i  N
(9)
1

i 1 ( d oi )
2

Where, d oi is the distance between the sample point and the estimated point.

The case study

The Study area Al-Dibdiba basin (Fig. 1), it lies in the area between Karbala and Najaf provinces, Iraq.
It located between longitude 430 30/ - 440 20/ E and latitude 320 00/ - 320 40/ N.
Tested the area within Al-Dibdiba basin by size 400 sq.km, (20km length x 20km width). This area is
located between Khan Al-Rubu in the north to Khan Al-Nuss in the south and the Euphrates in the east
to the strategic pipe line in the west which located in the (UTM) lines follows:
Longitude: 412 – 432
Latitude: 3575 – 3595

The area is falls within semi dry weather that characterized by rain shortage except for some of the
rainy storms that covers the area from time to time (annual rainfall of about zero in August and 23.34
mm in January), with extremes of temperature (maximum upto 45C0 in July and minimum upto 6C0 in
January), and very high potential evapotranspiration (6.67mm in January and 370.88mm in July), (Al-
Ani, 2004).
The main soil types of the study area are Sedimentations of sand, gravel and gravelly sand with the
existence of clayey lenses which are generally take the form of compacted clayey balls interfered with
small amount of sand and gypsum working as agent material. The maximum depth of the formation is
about 80 km in the western south of the basin, the depth decreases towards Al Saeed Tar, Al Najaf Tar
and towards the river of Euphrates that considered as formation boundaries, (Al-Khateeb, 2001).

The water in the basin of Dibdiba is generally saline, but it still highly used in irrigation, since the soil
is sandy which is do not allow the accumulation of slats in the upper layer of soil.
For this study, groundwater level data pertaining to ten months from June to March seasons over the
years from 2002 to 2003 covering an area of 400 sq. km (Fig.1) were selected, [quoted by (Al-
Ani,2004)]. Fig. 2 shows plan of study area and the location of observation wells. The descriptive
statistics of the observed groundwater levels are shown in Table (1). Mean values of groundwater
levels indicate decrease in groundwater level in summer season (minimum in August), and then starting
rise in groundwater levels (maximum in March).There is very small change but that is due to as mean
values are provided. Also study area receives very little rainfall.

173
Journal of Kerbala University , Vol. 6 No.1 Scientific .March. 2008

Fig.(1): Location map of study area

174
Journal of Kerbala University , Vol. 6 No.1 Scientific .March. 2008

100

Eu
ph
Khan Al-Nukhaila

r at
es
95

riv
D-1

re
D-13
D-2
D-3
90
Oi
lL

D-10
ine

D-11
D-12
85

D-7

D-8

80 D-9

D-4
D-5
D-6
Observation wells Kerball
75 Al-Haydaria a-Naja

0 1 2 3 4 5
km
f

70
10 15 20 25 30 35

Fig. (2): Plan of study area and location of observation wells.

175
 Journal of Kerbala University , Vol. 6 No.1 Scientific .March. 2008

Table (1): Statistical parameters of select data .

No. Data No. of Mean Variance Coeff of Max. Min.
2
Year Month Wells (m) (m ) Variance G.W.L G.W.L
(m.a.s.l.) (m.a.s.l.)
1. 2002 Jun. 13 32.36 63.80 0.2468 42.5 16.5

2. Jul. 13 32.35 63.90 0.2470 42.52 16.5

3. Aug. 13 32.34 63.83 0.2469 42.44 16.51

4. Sep. 13 32.38 64.15 0.2473 42.66 16.52

5. Oct. 13 32.44 63.89 0.2464 42.7 16.62

6. Nov. 13 32.47 64.03 0.2464 42.84 16.66

7. Dec. 13 32.51 64.34 0.2467 42.92 16.6

8. 2003 Jan. 13 32.54 64.11 0.2460 42.99 16.72

9. Feb. 13 32.62 64.43 0.2460 43.05 16.75

10. Mar. 13 32.65 64.76 0.2464 43.15 16.73

Results and Discussion

Fig. (3) Shown the experimental semivariograms and the best-fitted theoretical model for all the all
data sets. For all the data sets, Gaussian model resulted in the minimum standard error and so
considered the best-fit model. The theoretical fitted gaussian semivariogram is of the form:

   h  2 
 (h)  C0  C 1  Exp      (10)
  a  
  

Where:
C0= nugget effect (m2),
C = intercept between sill and nugget effect (m2),
a = range of influence (The distance at which samples become independent of each other) (km).
Then from the best fit model Fig.3, The theoretical fitted gaussian semivariogram for August 2002 data
is of the form:

   h  2 
 (h)  5.7  95.981  Exp    
  8.06  
(11)
  

176
Journal of Kerbala University , Vol. 6 No.1 Scientific .March. 2008

June-2002 July-2002

Semivariance(sq.m)

Semivariance(sq.m)
98.3 98.5

73.8 73.9

49.2 49.3

24.6 24.6

0.0 0.0
0.00 4.25 8.49 12.74 16.98 0.00 4.25 8.49 12.74 16.98
Lag Distance(km) Lag Distance(km)

August-2002 September-2002
Gaussian model (Co = 5.5000; Co + C = 95.2400; Ao = 7.99; r2 = 0.752; Gaussian model (Co = 5.8000; Co + C = 95.5300; Ao = 8.01; r2 = 0.751;
RSS = 932.) RSS = 937.)

Semivariance(sq.m)
Semivariance(sq.m)

98.8 98.7

74.1 74.0

49.4 49.3

24.7 24.7

0.0 0.0
0.00 4.25 8.49 12.74 16.98 0.00 4.25 8.49 12.74 16.98
Lag Distance(km) Lag Distance(km)

October-2002
Gaussian model (Co = 5.7000; Co + C = 95.9800; Ao = 8.06; r2 = 0.753;
November-2002
Gaussian model (Co = 6.2000; Co + C = 95.5500; Ao = 8.01; r2 = 0.748;
RSS = 938.) RSS = 947.)
Semivariance(sq.m)
Semivariance(sq.m)

97.9 97.3

73.5 72.9

49.0 48.6

24.5 24.3

0.0 0.0
0.00 4.25 8.49 12.74 16.98 0.00 4.25 8.49 12.74 16.98
Lag Distance(km) Lag Distance(km)

December-2002 January-2003
Gaussian model (Co = 5.1000; Co + C = 94.3100; Ao = 7.85; r2 = 0.752; Gaussian model (Co = 3.7000; Co + C = 92.7000; Ao = 7.60; r2 = 0.753;
RSS = 923.) RSS = 911.)
Semivariance(sq.m)
Semivariance(sq.m)

97.6 97.2

73.2 72.9

48.8 48.6

24.4 24.3

0.0 0.0
0.00 4.25 8.49 12.74 16.98 0.00 4.25 8.49 12.74 16.98
Lag Distance(km) Lag Distance(km)

February-2003 March-2003
Gaussian model (Co = 3.9000; Co + C = 93.2000; Ao = 7.63; r2 = 0.755; Gaussian model (Co = 3.8000; Co + C = 92.6100; Ao = 7.59; r2 = 0.753;
RSS = 908.) RSS = 908.)
Semivariance(sq.m)

Semivariance(sq.m)

97.4 97.7

73.0 73.2

48.7 48.8

24.3 24.4

0.0 0.0
0.00 4.25 8.49 12.74 16.98 0.00 4.25 8.49 12.74 16.98
LagDistance(km) Lag Distance(km)

 Experimental
Gaussian model (Co = 4.0000; Co + C = 92.8500; Ao = 7.58; r2 = 0.752;
RSS = 913.) Fitted (Gaussian)
177andRSSfitted
Fig. (3): Experimental
Gaussian model (Co = 3.0000; Co + C = 92.8500; Ao = 7.49; r2 = 0.757;
= 905.) semivariogram for different data

sets
8
Journal of Kerbala University , Vol. 6 No.1 Scientific .March. 2008

The summary details of the best-fit gaussian model for ten months data set are given in Table 2. An
important feature which has emerged from the best fit models (Table 2) is that while the gaussian
model is the best fit for all the data set, the parameters have changed over the months. Nugget effect
(C0) shows random change between 3.0 and 6.2. The range in which the intercept C lies between 92.61
and 95.98. The range (a) exhibits constant increase through summer season (maximum in August) and
it’s a general decreasing trend in winter season (minimum in March).
Table (2): Parameters of fitted Gaussian models.
Data
No. C0 C a
Year Month
1. Jun. 5.5 95.24 7.99
2. Jul. 5.8 95.53 8.01
3. Aug. 5.7 95.98 8.06
4. 2002 Sep. 6.2 95.55 8.01
5. Oct. 5.1 94.31 7.85
6. Nov. 3.7 92.70 7.60
7. Dec. 3.9 93.20 7.63
8. Jan. 3.8 92.61 7.59
9. 2003 Feb 4.0 92.85 7.58
10. Mar. 3.0 92.85 7.49

Table (3), shown the cross validation results for the ten months. Results of Jackknifing procedure for
March 2003 data with the fitted gaussian model resulted in a mean error (ME) of 0.082, (which is very
near to zero), mean square error (MSE) of 23.67, (which is very low as compared to the variance of the
data), kriged reduced mean error (KRME) of 0.0162, (which is very near to zero) and a kriged reduced
mean square error (KRMSE) of 0.934, (which is very near to 1). The above cross validation results
show that the chosen model and its parameters are adequate.

Table (3): Cross validation results with gaussian model

Data ME*1 MSE*2
No. 2 KRME*3 KRMSE*4
Year Month (m) (m )
1. Jun. 0.137 23.55 0.0278 0.970
2. Jul. 0.148 23.79 0.0300 0.983
3. Aug. 0.137 23.54 0.0270 0.967
4. 2002 Sep. 0.166 24.00 0.0339 0.997
5. Oct. 0.137 23.40 0.0279 0.961
6. Nov. 0.102 23.29 0.0204 0.939
7. Dec. 0.108 23.65 0.0216 0.953
8. Jan. 0.108 23.25 0.0217 0.939
9. 2003 Feb. 0.117 23.60 0.0236 0.954
10. Mar. 0.082 23.67 0.0162 0.934
*1 *2 *3
= Mean error, = Mean sq error, = Kriged reduced mean error,
*4
= Kriged reduced mean sq error

Groundwater levels and estimation variances were calculated by kriging at the nodes of a square grid of
2.5km x 2.5km for August of 2002 and March of 2003 months. These estimated level values are used
to draw the contour maps of groundwater levels and estimation variance. Fig. 4 and 5 are shown the

178
 Journal of Kerbala University , Vol. 6 No.1 Scientific .March. 2008

contour maps of the groundwater levels and estimation variance obtained for August 2002 and March
2003 respectively. Fig. 5 can be interpreted as the map of the reliability of the kriged ground water
level in Fig. 4. As seen from the Fig. 5, the estimation variance is low at 3m2 in the middle of the study
area (where most of the observation points are located) and increase rapidly towards the boundaries,
where no observation well is located. It indicates that the estimated groundwater level are highly
reliable in the middle of the study area and at or near the boundary, these are to not reliable the same
extent.

94 94

92 92

90 90

88 88

86 86
KM

KM

84 84

82 82

80 80

78 78

76 76

12 14 16 18 20 22 24 26 28 30 32 12 14 16 18 20 22 24 26 28 30 32
KM KM

August 2002 March 2003

Fig. (4): Groundwater level contours maps of the study area (m) by Kriging Method.

179
 Journal of Kerbala University , Vol. 6 No.1 Scientific .March. 2008

94
` 94

92 92

90 90

88 88

86 86
KM

KM
84 84

82 82

80 80

78 78

76 76

12 14 16 18 20 22 24 26 28 30 32 12 14 16 18 20 22 24 26 28 30 32
KM KM

August 2002 March 2003

Fig. (5): Estimation variance (m2) by Kriging method.

Fig. (6), shown the ground water level contour obtained by inverse square distance (ISD) method for
August 2002 and March 2003. The contour map provided by two interpolation methods (Fig. 4 and 6)
are different as kriging takes into consideration the spatial structure of the parameter and ISD method
consider only distance between estimated and observed points. The comparison of ISD map with the
map obtained by kriging (Fig. 4) indicated that kriged map provided a more regular gradient of the
groundwater table, which seems more likely than the mound and valley combination provided by the
inverse square distance method.
For more comparison of these two techniques was obtained by comparing the ME and MSE obtained
by jackknifing procedure (Table 4). ISD resulted in a ME of 0.913m to 0.926m whereas kriging gave a
ME of 0.082m to 0.166m. Similarly, ISD gave a MSE of 35.89m2 to 36.32 m2 and kriging 23.25 m2 to
24.00 m2. It is concluded that for this study, kriging performed better than the inverse square distance
method and more importantly.

180
 Journal of Kerbala University , Vol. 6 No.1 Scientific .March. 2008

94 94

92 92

90 90

88 88

86 86

KM
KM

84 84

82 82

80 80

78 78

76 76

12 14 16 18 20 22 24 26 28 30 32 12 14 16 18 20 22 24 26 28 30 32
KM KM

August 2002 March 2003

Fig. (6): Groundwater level contours (m) by Inverse Square Distance Method.

Table (4): Comparison of errors of two interpolation methods.

Data ME*1(m) MSE*2(m2)

No. *3 *4
Year Month ISD K ISD K
1. Jun. 0.924 0.137 36.01 23.55
2. Jul. 0.923 0.148 36.17 23.79
3. Aug. 0.926 0.137 36.01 23.54
4. 2002 Sep. 0.924 0.166 36.32 24.00
5. Oct. 0.922 0.137 35.94 23.40
6. Nov. 0.918 0.102 35.91 23.29
7. Dec. 0.917 0.108 36.19 23.65
8. Jan. 0.913 0.108 35.89 23.25
9. 2003 Feb. 0.919 0.117 36.12 23.60
10. Mar. 0.915 0.082 36.22 23.67
*1
Mean error, Mean square error, Inverse square distance, *4 Kriging
*2 *3

181
 Journal of Kerbala University , Vol. 6 No.1 Scientific .March. 2008

Recommendation
1. Kriging method gave very good accurate guess fluctuations in groundwater levels (especially in
center of the study area).
2. Kriging methods gave more accurate results through the Inverse Square Distance (ISD) method.
3. The level of groundwater near the surface in the west of the study area reach (42m.a.s.l) in the
winter, and then sliding towards the east to reach a level (22m.a.s.l) in the summer.

Conclusions
In this study, kriging, a type of geostatistical techniques, is applied to the groundwater level data of
minimum recharge season (August) and maximum recharge season (March) over a period of ten
months in years (2002-2003). The gaussian model is found to the best model representing the spatial
variability of groundwater level data over the months. The modeling results indicate that the kriged
groundwater levels satisfactorily matched the observed groundwater levels. The degree of difference
between the kriged values and the estimates using ISD are significantly high. Also, kriging out
performs ISD in giving reliability indices and in the present study the reliability of the estimates is high
as indicated by low level of variance.

References :
 Aboufirassi, M., and Marino, M. A., (1983) "Kriging of water levels in the souss aquifer Morocco",
Math. Geol., vol.15, 537-551.
 Al-Ani,A.Ahmad,(2004)"Selection of best technique to evaluate groundwater recharge a case study
of Dibdiba aquifer in area of karballa–Najaf",Ph.D.Thesis,departmentofgeology,Baghdad university.
 Al-Khateeb, H.M., (2001)"Problem of shallow groundwater level in the center of kerballa city:
Evaluation and simulation", Ph.D.Thesis, department of civil engineer, university of Al-Mustansiria.
 ASCE Task Committee, (1990a) "Review of geostatistics in geohydrology, I: Basic concepts", J. of
Hydraulic Engg.(ASCE), vol.116,612-632.
 ASCE Task Committee, (1990b) "Review of geostatistics in geohydrology, II: Applications", J. of
Hydraulic Engg. (ASCE), vol.116, 633-658.
 Burgess, T.M., and Webster, R., (1980) "Optimal interpolation and isarithmic mapping of soil
properties, I: The semivariogram and punctual kriging", J. of Soil Sci., vol.31, 315-331.
 Delhomme, J.P., (1978)" Kriging in the hydroscience", Adv. in Water Resour. vol.1,251-266.
 De Marsily, G., and Ahmed, S., (1987)"Application of kriging techniques in groundwater
hydrology", J. Geol. Soc. of India, vol.29, 57-82.
 Isaaks,E.H., and Srivastava,R.M. (1989) "Applied Geostatistics",New York:Oxford University
press.
 Kumar, V.,(1996)" Space time modeling of ground water with assistance of remote sensing", Ph.D.,
Indian Institute of Technology, New Delhi, India.
 Kumar, D., and Ahmed, S., (2003)"Seasonal behaviour of spatial variability of groundwater level in
a granitic aquifer in monsoon climate", Current Science, vol. 84, 188-196.
 Kumar,V.,and Remadevi,(2006)"Kriging of groundwater level a case study",Spatial Hydology
Journal,Vol.6;No.1 pp 81-91.
 Virdee, T.S., and Kottegoda, N.T., (1984)" A brief review of kriging and its application to optimal
interpolation and observation well selection", Hydro. Sci. J., vol.29, 367-387.
 Volpi, G., and Gambolati, G., (1978)" On the use of main trend for the kriging technique in
hydrology", Adv. Water Resour, vol.1, 345-349.

182
View publication stats