International Journal of Coal Geology 140 (2015) 84–96

Contents lists available at ScienceDirect

International Journal of Coal Geology

journal homepage: www.elsevier.com/locate/ijcoalgeo

Lignite resource estimations and seam modeling of Thar Field, Pakistan

Fahad Irfan Siddiqui a,⁎, Abdul Ghani Pathan a, Bahtiyar Ünver b, Abdullah Erhan Tercan b,
Mehmet Ali Hindistan b, Güneş Ertunç b, Fırat Atalay b, Suphi Ünal b, Yasin Kıllıoğlu b
a
Department of Mining Engineering, Mehran University of Engineering and Technology, Jamshoro, 76062, Pakistan
b
Department of Mining Engineering, Hacettepe University, Beytepe, 06800 Ankara, Turkey

a r t i c l e i n f o a b s t r a c t

Article history: Thar lignite field in Pakistan contains more than 175 billion tons thus far not exploited. To date, the coalfield has
Received 26 November 2014 been divided into 12 exploration blocks and each block is studied and investigated separately by different agen-
Received in revised form 13 February 2015 cies. As a result of exploration programs, a large exploration database containing geological and coal quality in-
Accepted 15 February 2015
formation is constructed, but geostatistical studies regarding to quantify entire Thar lignite resources is not
Available online 23 February 2015
available yet. This paper aims to generate 3D solid model of main Thar seam and to produce spatial distribution
Keywords:
maps for various coal quality attributes by geostatistical method, ordinary kriging. 3D seam model has a volume
Block model of 17.71 billion m3. The resulting spatial map of ash content shows structured distribution and LCV map shows
Geostatistics fair agreement with ash map except in block VI where high and low values of ash and LCV occur in close proxim-
Ordinary kriging ity. The moisture distribution reveals higher values in northern and southern parts whereas the central portion
Thar field possesses lower moisture values. The sulfur content shows homogenous distribution with some higher sulfur
Solid model patches at places. The kriging variance maps are generated to delineate areas of higher uncertainty. These
Spatial distribution maps could be helpful to devise further exploration programs. Blocks were checked on the basis of global
Variograms
averages and swath plot comparison.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction
At present, Pakistan is facing serious energy crisis. Severe shortage of
electricity has badly affected the industrial and social activities. Pakistan
is spending over US$ 14.5 billion on imports of crude oil, petroleum
products, coal, LPG etc. to fulfill energy requirements (Hydrocarbon
Development Institute of Pakistan, (Hydrocarbon Development Insti-
tute of Pakistan (HDIP, 2012)). Coal is a cheap source of energy and
widely used throughout the world for power generation, cement and
other process industry. The world electricity generation by fuel reveals
that coal is the highest contributor in electricity generation and demand
of coal will remain dominant in the worldwide electricity generation
(Energy Information Administration, USA (EIA, 2011)). The installed
electricity generation capacity of Pakistan is about 22.263 GW with
29.9% hydroelectric, 35.2% oil, 29% natural gas, 5.8% nuclear/imported
and only 0.1% coal, as shown in Fig. 1 (HDIP, 2012). Share of coal in
electricity generation in Pakistan is very negligible as compared to
world average.
Pakistan possesses the 7th largest lignite resource in the world with
nearly 200 billion tons of coal, mainly concentrated in Thar region
having more than 175 billion tons of lignite resources (Singh et al.,
2011). Thar lignite field, the largest coal resource of Pakistan, is located
⁎ Corresponding author.
in the eastern part of Sindh Province (shown in Fig. 2). Thar coalfield
covers an area of approximately 9000 km2 and number of lignite
seams lie at depths between 130 and 250 m. Cumulative seam thickness
varies between 1.45 m and 42.6 m and the maximum thickness of an
individual seam is 28.6 m (Pathan et al., 2013).
The Thar lignite was fortuitously discovered by British Overseas
Development Agency (BODA) and Sindh Arid Zone Development
Authority (SAZDA) during drilling for water wells near the village of
Khario Ghulam Shah, Tharparkar, in 1988 (Fassett and Durrani, 1994).
Further exploration work was conducted by Geological Survey of
Pakistan (GSP) and United States Geological Survey (USGS) under
Coal Resources Exploration & Assessment Program (COALREAP) from
1989 to 1994. In this exploration program, total 21 boreholes were
sunk at an average spacing of 20 km to define the extent of coal
occurrence (Thomas et al., 1994). Total hypothetical lignite resources
were estimated at 175.506 billion tons. Since then 12 exploration blocks
have been investigated separately by various agencies viz: GSP,
Shenhua (China), Rheinbraun(Germany), Sindh-Engro(SECMC), Deep
Rock Drilling (DRD) from 1994 to 2012. As a result of these exploration
programs, a large exploration data including geological and coal quality
information, is available; however no resource modeling study has been
conducted to quantify entire Thar lignite resources using geostatistical
modeling.
Several research studies on coal resource estimation and seam
modeling have been carried out. Olea et al., 2011 combined different
geostatistical methods for quantitative characterization for evaluating

http://dx.doi.org/10.1016/j.coal.2015.02.003
0166-5162/© 2015 Elsevier B.V. All rights reserved.
 F.I. Siddiqui et al. / International Journal of Coal Geology 140 (2015) 84–96
Fig. 1. Pakistan's electricity generation by fuel, 2011–12 (HDIP, 2012).
the uncertainty and indicated that distance to the neighboring drill hole
is more or less completely dissimilar to the uncertainty, which
demonstrate the deficiency of distance-based resource estimation ap-
proach. Hohn and Britton (2012) also demonstrated significance of
geostatistical estimation over distance-based classification method.
Tercan and Karayigit (2001) also estimated lignite reserves in
Kalburcayiri field, Kangal basin, Sivas, Turkey using global estimation
variance (GEV) approach. Pardo-Iguzquiza et al. (2013) assessed the
risk associated with lignite reserves estimations at North-western
Spain using semi-variograms and conditional simulation. Geostatistical
modeling was conducted for quantification of error associated with re-
serves estimation at Jharia coalfield, India. The uncertainty maps were
generated using kriging variances and areas of high uncertainty were
Pakistan
Sindh
Karachi
Tharparkar district
Boundary
Exploration Blocks
Fig. 2. Location map of Thar Lignite Field, Pakistan.
85
identified for additional sampling for precise estimation (Saikia and
Sarkar, 2013).
Hindistan et al. (2010) conducted case study on dilution problem at
an underground coal mine in Turkey. The mean calorific values of the
blocks inside the production panels were estimated using ordinary
kriging to devise short term mine plan. Tercan et al. (2013) conducted
thorough study at Eynez–Soma and Ömerler–Tunçbilek coal fields,
Turkey for seam modeling and estimation of lower calorific value. The
study sites were subjected to heavy tectonic movements and erratic
coal quality. Falivene et al. (2014) carried out three-dimensional coal fa-
cies interpolations and simulations in a heterogeneous coal zone in the
As Pontes Basin (North Western Spain) to forecast coal resources and
reserves. This study ascertained that due to interpolation smoothing,
the three-dimensional facies interpolation methods tend to overesti-
mate coal resources and reserves whereas facies simulation methods
yield similar resource predictions than conventional thickness map ap-
proximations. Vasquez and Nieto (2004) estimated calorific value and
ash by weighting kriging estimation variances.
Ertunc et al. (2013) estimated the lower calorific value, ash content,
and moisture content of lignite deposit subjected to severe tectonic ac-
tivity, by using covariance matching constrained kriging and ordinary
kriging. It is found that covariance matching constrained kriging repli-
cates the spatial variability better than ordinary kriging. Heriawan and
Koike (2008a) and Heriawan and Koike (2008b) estimate thickness,
ash, sulfur and calorific value in a multiple seam environment at East
Kalimantan (Borneo, Indonesia) by means of ordinary kriging, cokriging
and factorial Kriging.
This paper aims to estimate various coal quality parameters using
ordinary kriging technique that provides improved estimates and asso-
ciated errors by means of kriging variances.
N
Coal Boundary
Indian Territory
Cross-section AB
shown in Fig. 4
Rann-of-Kutch
fault zone
86 F.I. Siddiqui et al. / International Journal of Coal Geology 140 (2015) 84–96

Thickness
Era Period Epoch Formation Lithology
(m)

Sand, silt and

Recent Dune Sand 14-93
clay
Unconformity

Quaternary
Sand, siltstone,
CENOZOIC Sub-Recent Alluvial Deposits 11-209 claystone and
Sandstone

Unconformity

Claystone, shale,
sand, Lignite
Terary

Eocene/
Bara Formaon 52-95 seams,
Palaeocene
carbonaceous
Claystone

Unconformity
- --
PRE-CAMBRIAN

- --
-- -- Basement complex -- -- - Granite and
-- - quartz diorite
-
- -

Fig. 3. Generalized stratigraphy of Thar Lignite Basin.

2. Case study
2.1. Geographical location & geological setting
Thar coalfield is located at Tharparkar district of Sindh Province,
Pakistan between latitudes 24° 15′N–25° 45′N and longitudes 69° 45′E–
70° 45′E. The Thar coal field is covered by thick dune sands with an
average depth of 80 m resting upon the structural platform in the eastern
part of the desert, underlain by relatively shallow granitic basement rocks
of Precambrian age (Fassett and Durrani, 1994). The Thar coal field is
structurally simple. Strata dips gently at around 2° to west-northwest.
The Rann-of-Kutch fault zone is the only fault in the area, which occurs
away from exploration blocks, in the most southern part of the field
(Fig. 2).
The Lignite bearing lower Indus basin contain sedimentary rock fill
comprises Mesozoic and Cenozoic sections. The present relationship of
the strata indicates the existence of a regional high in the Thar Coal
Field area during Mesozoic, resulting in erosion of Palaeozoic rocks
and exposure of basement granites. The erosion cycle was followed by
deposition during Palaeocene and Early Eocene. During this time the
environment was suitable for the formation of the Thar coals. Subse-
quently the Thar Coal Field remained stable, resulting neither in the de-
position of younger rocks nor in the erosion of the relatively thin
sedimentary layer of Palaeocene–Early Eocene. During recent times
the area subsided and was traversed by the Indus river system, which
deposited the sub-recent alluvial sediments in the area. The dunes of
the Recent-Formation formed about 20,000 years ago. Fig. 3 shows the
generalized stratigraphic section of Thar basin.
Lignite seams in the Thar area are found in Bara formation of
Palaeocene/Eocene age. The Bara formation is some 95 m thick
consisting of sandy/silty claystone and sandstone formation overlying
the basement granite lying at a depth of 100 m to 220 m (Singh et al.,
2010b). The basement rock is slightly to moderately weathered granite
comprising fine to coarse grains of quartz. The overlying Bara formation
consists of layers of carbonaceous clay stone, sandy clay stone and silty
clay stone. The carbonaceous clay stone is medium light grey to brown

Fig. 4. General lithology of Thar lignite field, Pakistan (GSP, 2002).

 F.I. Siddiqui et al. / International Journal of Coal Geology 140 (2015) 84–96 87

Table 1
Summary of drilling activities.

No. of drillholes Average drillhole spacing (m) Total drilled meters (m) Minimum depth (m) Maximum depth (m) Average drillhole depth (m)

All 12 blocks 693 1400 166,675 109.78 319.68 240.51

in color containing carboniferous petrified roots, carbonaceous mate-
rials and rare sandy resin globules. The olive grey to dark-grey claystone
containing petrified coal roots and pyretic resin globules overlies this
sediment (Singh et al., 2010a).
The sub-recent formation overlies the Bara formation. It comprises
of siltstone, sandstone, and claystone, with an average thickness of
70 m and lies between the depth of 52 and 125 m. The recent formation
overlying the sub-recent formation consists of sand dunes. This sand is
fine to medium grained, yellowish grey in color containing sub-
rounded and moderately sorted grains of ferromagnesian minerals.
Fig. 4 shows the stratigraphy and lithology of the Thar coalfield.
2.2. Drillhole Database & Exploratory Data Analysis (EDA)
The data used in this study were obtained from 693 boreholes sunk
by various governmental and non-governmental agencies from 1994 to
2012. The obtained information includes, collar coordinates, geological
intercepts, and coal quality parameters. The details of drilling activities
are given in Table 1 and the location of drillholes in all blocks is
shown in Fig. 5.
The quality data of a total 6095 drill intersected lignite samples were
collected. The quality data includes, moisture content, ash, fixed carbon,
volatile matter, sulfur and lower calorific values (LCV) on as received

5 km

Fig. 5. Drillhole location in all 12 blocks of Thar lignite field, Pakistan.

88 F.I. Siddiqui et al. / International Journal of Coal Geology 140 (2015) 84–96

basis. In the present study, only ash content, moisture, sulfur and LCV
are considered. A complete electronic database including all geologic
and quality information, were prepared for solid modeling, variogram
analysis, block modeling and ordinary kriging. The frequency distribu-
tions are presented in Fig. 6. The raw core intervals present a right
skewed distribution with a mean of 1.52 m. The ash and sulfur show
right skewed distribution as most values are concentrated on left of
the mean value. LCV histogram represents normal distribution while
moisture content shows left-skewed distribution.
2.3. Solid modeling
A solid model is a three-dimensional triangulation of data. In general
there are two approaches used in solid modeling i.e. section method and
top–bottom surface method. In the section method coal seam is
outlined in vertical sections and these sections are then combined to
construct a 3D solid model. In the top–bottom surface method the
roof and floor surfaces of coal seam are triangulated or interpolated
and then combined. In this study, Gemcom SURPAC 6.2 was used to cre-
ate solid model by outlining the coal seam in numerous vertical sections
drawn at perpendicular, parallel and oblique directions. Section method
was selected due to structural simplicity of Thar coal field. The Bara for-
mation (coal bearing formation) at Thar field contains number of coal
seams. Due to larger drillhole spacing (average 1400 m), it is difficult
to model all the seams; therefore this study is limited to model the
main Thar seam which is the thickest seam and present in all explora-
tion blocks. Fig. 7 shows complete process of generating 3D seam
model in block II from geological information. Number of vertical sec-
tions (Fig. 7a) were taken in parallel (Fig. 7b) and perpendicular
(Fig. 7c) directions. Two different solid models were generated using
parallel and perpendicular sections, as shown in Fig. 7d and e. The two
solids were then combined (shown in Fig. 7f) and further section was
drawn considering two solids together (Fig. 7g). Final 3D model of the
seam is generated by triangulating the final sections, as shown in

No. of Samples :6095

No. of Samples :6095
Mean :7.34
Mean :1.52
Std. Deviation :4.8
Std. Deviation :1.09
Minimum :1.06
Minimum :0.04
Maximum :49.51
Maximum :22.33

No. of Samples :6095 No. of Samples :6095

Mean :2799 Mean :47.24
Std. Deviation :461 Std. Deviation :5.98
Minimum :417 Minimum :5.53
Maximum :5277 Maximum :68.58

No. of Samples :6095

Mean :1.28
Std. Deviation :1.22
Minimum :0.01
Maximum :15.64

Fig. 6. Raw data histograms.

 F.I. Siddiqui et al. / International Journal of Coal Geology 140 (2015) 84–96 89

b N c N

N
d e N

f N g N

h N

Fig. 7. Process of generating seam model in Block II, a) example of single verticals section, b) parallel sections, c) perpendicular sections, d) solid model generated from parallel sections,
e) solid model generated from perpendicular sections, f) combining two solids, g) further sectioning of combined solid, and h) 3D solid seam model.

Fig. 7h. Similarly, the 3D seam models were generated individually in all
remaining blocks and finally gap between the seam models was filled by
further sectioning and triangulations. Fig. 8 presents individual seam
models in different blocks and final seam model. The total volume of
final 3D seam model is 17,716,072,160 m3 (17.716 billion m3). The
seam thickness is derived by summing up individual block thickness
in block model (Tercan et al., 2013). Fig. 9 shows the spatial distribution
of main seam thickness. The thickness map indicates that seam is
thicker in block I, II, IV & VI and this portion possibly could be the center
of the lignite depositional basin. The eastern part of block VIII also shows
thickest seam zone whereas the boreholes in the central part of block
VIII showed no coal intercept in the entire borehole column.
90 F.I. Siddiqui et al. / International Journal of Coal Geology 140 (2015) 84–96
Fig. 8. Individual solid seam models (above) and combine master solid model (below).
2.4. Compositing
Compositing is a length-weighted average sampling. The down hole
composites of equal length are used as input for geostatistical estima-
tions. The raw samples had variable intervals with a mean of 1.52 m
hence composited to 2 m down-hole lengths. The minimum 75% of sam-
ples was considered for compositing. The mean ash value of composited
data is decreased as compared to raw data. Similarly mean LCV value is
increased. The frequency histograms of composite data are shown in
Fig. 10. Frequency distribution of composite data is approximately
similar to the raw data.
2.5. Semi-variography
Experimental variograms were calculated from the composite data
in lateral and vertical directions. Vertical/downhole experimental
variograms were constructed to establish the nugget and sill values.
No serious trend has been observed from the vertical variograms for
all the quality variable. The horizontal experimental variograms were
200 m
20Xvertical
0m
exaggeration
-250 m
200 m
0m
20Xvertical
exaggeration
-250 m
constructed in four different directions (i.e. 0°, 45°, 90°, and 135°) to de-
tect anisotropy. Directional horizontal variograms reveal no anisotropy
in the all variables; hence omni-directional variograms are selected. The
experimental and model variograms of all variables in vertical and hor-
izontal directions are shown in Fig. 11. Table 2 summarizes the model
variogram parameters for all quality variables. The variogram ranges
in vertical direction are considerably lower than the ranges in horizontal
direction as it is quite expected in case of bedded deposit. The experi-
mental variograms for all the variables are fitted with spherical models.
Ash parameter exhibits lowest range in horizontal direction and highest
nugget to sill ratio when compared to other variables. The horizontal
ranges of LCV, moisture and sulfur are approximately same.
2.6. Block kriging
The generated 3D solid seam model is divided into 288,423 blocks
with size of 200 m × 200 m × 2 m and sub-blocks with size of
100 m × 100 m × 2 m. The size of the blocks is selected according to av-
erage drillhole spacing and composite length. Block mean values are
 F.I. Siddiqui et al. / International Journal of Coal Geology 140 (2015) 84–96 91

Main Seam
Thickness (m)

Fig. 9. Spatial distribution of main seam thickness.

No. of Samples :4900 No. of Samples :4900

Mean :6.39 Mean :2872
Std. Deviation :3.75 Std. Deviation :418
Minimum :1.94 Minimum :588
Maximum :35.45 Maximum :5277

No. of Samples :4900 No. of Samples :4900

Mean :47.75 Mean :1.13
Std. Deviation :5.15 Std. Deviation :0.96
Minimum :9.40 Minimum :0.05
Maximum :67.03 Maximum :15.64

Fig. 10. Composite data histograms.

92 F.I. Siddiqui et al. / International Journal of Coal Geology 140 (2015) 84–96

30 30
Ash Ash
25 25

Semi-variance (%)2

Semi-variance (%)2
20 20

15 15

10 10

5 Downhole-Expermental 5 Omni-Directional Experimental

Downhole-Model Omni-Directional Model

0 0
0 5 10 15 20 25 30 0 1000 2000 3000 4000 5000 6000
Distance, h (m) Distance, h (m)

250000
250000
LCV LCV
Semi-variance (kCal/kg)2

Semi-variance (kCal/kg)2
200000
200000

150000
150000

100000 100000

50000 50000
Downhole-Expermental Omni-Directional Experimental
Downhole-Model Omni-Directional Model

0 0
0 10 20 30 40 50 0 1000 2000 3000 4000 5000 6000
Distance, h (m) Distance, h (m)

30 30
Moisture Moisture
25 25
Semi-variance (%)2

Semi-variance (%)2

20 20

15 15

10 10

5 Downhole-Expermental 5 Omni-Directional Experimental

Downhole-Model Omni-Directional Model

0 0
0 10 20 30 40 50 60 0 1000 2000 3000 4000 5000 6000
Distance, h (m) Distance, h (m)

2 2
Sulphur Sulphur
1.8 1.8

1.6 1.6
Semi-variance (%)2

Semi-variance (%)2

1.4 1.4

1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4
Downhole-Expermental Omni-Directional Experimental
0.2 0.2
Omni-Directional Model
Downhole-Model
0 0
0 5 10 15 20 25 0 1000 2000 3000 4000 5000 6000
Distance, h (m) Distance, h (m)

Fig. 11. Experimental variograms fitted with model variograms in vertical (left) and horizontal (right) directions for all quality variables.

estimated by employing ordinary kriging technique. Model variogram Table 2

parameters were used to define maximum horizontal and vertical Model variogram parameters for all quality variables.
search distance, anisotropic ratios, bearing and dip during ordinary
Ash % Lower calorific Moisture % Sulfur %
kriging. The minimum and maximum numbers of samples used in esti-
value (kCal/kg)
mation are 3 and 16, respectively.
Nugget, C0 7 50,000 2 0
The spatial distribution of ash, LCV, moisture and sulfur is shown in
Structure C1 13.75 20,000 5 0.5
Fig. 12a, b, c and d respectively. The spatial map of ash content (Fig. 12a) Structure C2 – 80,000 12 1
shows structured distribution and LCV map (Fig. 12b) shows fair agree- Range, a1 (m) 2600 1400 500 520
ment with ash map except in block VI (refer Fig. 2 for block identifica- 21.5 (vert.) 14.5 (vert.) 6 (vert.) 4 (vert.)
tion) where high and low values of ash and LCV are separated. The Range, a2 (m) – 3500 3400 3350
– 48 (vert.) 46 (vert.) 23 (vert.)
moisture map (Fig. 12c) reveals higher values in northern and southern
 F.I. Siddiqui et al. / International Journal of Coal Geology 140 (2015) 84–96 93

a b

Ash %
LCV (kCal/kg)

Section AB Section AB

c d

Moisture %
Sulphur %

Section AB Section AB

Fig. 12. Spatial distribution of (a) ash content, (b) lower calorific value (LCV), (c) moisture content and (d) sulfur (cross-sections are 20× vertically exaggerated).
94 F.I. Siddiqui et al. / International Journal of Coal Geology 140 (2015) 84–96

Ash %2 LCV (kCal/kg)2

a b

Moisture % 2 Sulphur % 2

c d

Fig. 13. Kriging variance maps. (a) Ash content, (b) lower calorific value (LCV), (c) moisture content and (d) sulfur.
 F.I. Siddiqui et al. / International Journal of Coal Geology 140 (2015) 84–96 95

Table 3 and block estimates for all the quality variables. Averages of block model
Descriptive statistics of composite and block estimates. and composites are notably closer to each other. Block mean ash value is
Minimum Mean Maximum Std. deviation lower as compared to composite mean; correspondingly block mean
Ash % Composites 1.94 6.39 20.51 3.75
LCV value is higher than composite mean value. The swath plots of
Block model 2.45 6.32 35.45 2.18 block centroids and composites were constructed in northing, easting
LCV (Kcal/kg) Composites 588 2872 5277 418 and downward directions. Trend comparison of block models and com-
Block model 883 2909 4305 345.25 posites is shown in Fig. 14. The trend comparison shows that block av-
Moisture % Composites 9.40 47.75 67.03 5.15
erages estimated by ordinary kriging technique have almost same
Block model 17.93 47.25 59.57 4.61
Sulfur % Composites 0.05 1.13 15.64 0.96 global pattern as the composite averages. The block averages are
Block model 0.32 1.12 8.28 0.48 smoother than composite which is quite expected in case of ordinary
kriging.

parts whereas the central portion possesses lower moisture values. The 4. Conclusions
spatial map of sulfur content (Fig. 12d) shows homogenous distribution
with some higher sulfur patches at places (i.e. north-west of block I, Being an energy-deficient country, Pakistan must develop and ex-
western part of block II and north-eastern part of block III). Fig. 13a, b ploit the indigenous and sustainable energy sources to fulfill its expo-
c, and d shows kriging variance for ash content, LCV, moisture and sul- nentially increasing energy needs. Thar coal has a potential to fullfil
fur. The kriging variance maps are generated to delineate areas of higher country's energy needs for years. The exploration blocks at Thar were
uncertainty. These maps could be helpful to devise further exploration studied separately but it is important to assess the Thar coalfield as a
programs. whole; otherwise, modeling and designing of mines separately in each
block will certainly lead to serious problems in future. Therefore, this
3. Kriging validation paper aims to generate 3D seam model and to produce spatial distribu-
tion maps for different quality variables using combine exploration data
Block values for ash, LCV, moisture and sulfur estimated by ordinary from all the blocks. 3D seam model has a volume of 17.71 billion m3. Ash
kriging are compared with composite values on the basis of global aver- content in whole field shows structured distribution and LCV distribu-
ages and trend plots. Table 3 presents descriptive statistics of composite tion shows fair agreement with ash map except in block VI where

20 5000 70 4
Composite Block Model Composite Block Model Composite Block Model Composite Block Model
18 4500 65 3.5
16 60
4000 3
14 55
LCV (Kcal/kg)

2.5
Moisture %

Sulphur %
12 3500 50
Ash %

10 3000 45 2
8 40 1.5
2500
6 35 1
4 2000 30
2 0.5
1500 25
0 20 0
2340000 2350000 2360000 2370000 2380000 1000 2340000 2350000 2360000 2370000 2380000
2340000 2350000 2360000 2370000 2380000 2340000 2350000 2360000 2370000 2380000
Easting (m) Easting (m) Easting (m) Easting (m)

820000 820000 820000 820000

Composite Composite Composite Composite
810000 Block Model 810000 Block Model 810000 Block Model 810000
Block Model
800000 800000 800000 800000
Northing (m)

Northing (m)
Northing (m)
Northing (m)

790000 790000 790000 790000

780000 780000 780000
780000
770000 770000 770000
770000
760000 760000 760000
760000
750000 750000 750000
750000 0 1000 2000 3000 4000 5000 20 30 40 50 60 70 0 0.5 1 1.5 2 2.5 3 3.5 4
0 5 10 15
Ash % LCV (Kcal/kg) Moisture % Sulphur %

Ash % Lower Calorific Value (Kcal/kg) Moisture % Sulphur %

0 10 20 30 0 1 2 3 4
1000 2000 3000 4000 5000 6000 30 40 50 60 70 80
-40 -40
-40 -40

-60 -60
-60 -60

-80 -80
-80 -80

-100 -100 -100 -100

Depth

Depth

Depth

Depth

-120 -120 -120 -120

-140 -140 -140 -140

-160 -160 -160

-160

-180 -180 -180

-180 Composite Composite Composite
Composite Block Model Block Model Block Model
Block Model -200 -200 -200
-200

Fig. 14. Trend comparison of various quality attributes in easting (upper), northing (middle) and elevation (lower).
96 F.I. Siddiqui et al. / International Journal of Coal Geology 140 (2015) 84–96
high and low values of ash and LCV occur in close proximity. The mois-
ture distribution reveals higher values in northern and southern parts
whereas the central portion possesses lower moisture values. The spa-
tial map of sulfur content shows homogenous distribution with some
higher sulfur patches at places. The kriging variance maps are generated
to delineate areas of higher uncertainty. Accuracy of geostatistical
estimations can be improved by additional sampling in the areas of
higher uncertainty.
Acknowledgment
Authors are thankful to British Council, Pakistan, Higher Education
Commission, Pakistan and Mehran University of Engineering & Technol-
ogy, Jamshoro, Pakistan (SP-247) for providing financial support under
INSPIRE program for this research. Thanks are also due to Thar Coal En-
ergy Board (TCEB), Energy Department, Government of Sindh, Pakistan
for providing exploration data, vital for the conduct of this academic
research. Special acknowledgements to Hacettepe University, Ankara,
Turkey for providing necessary support for smooth conduct of the
research.
Appendix A. Supplementary material
Supplementary data associated with this article can be found in the
online version, at http://dx.doi.org/10.1016/j.coal.2015.02.003. These
include Google map of the most important areas described in this
article.
References
EIA, 2011. International Energy Outlook 2011 vol DOE/EIA-0484(2011). Energy Informa-
tion Administration, U.S.
Ertunc, G., Tercan, A.E., Hindistan, M.A., Unver, B., Unal, S., Atalay, F., Kıllıoglu, S.Y., 2013.
Geostatistical estimation of coal quality variables by using covariance matching
constrained kriging. Int. J. Coal Geol. 112, 14–25. http://dx.doi.org/10.1016/j.coal.
2012.11.014.
Falivene, O., Cabrera, L., Sáez, A., 2014. Forecasting coal resources and reserves in hetero-
geneous coal zones using 3D facies models (As Pontes Basin, NW Spain). Int. J. Coal
Geol. 130, 8–26.
Fassett, J.E., Durrani, N.A., 1994. Geology and Coal Resources of the Thar Coal Field, Sindh
Province, Pakistan. U.S. Geological Survey Open File Report.
GSP, 2002. Coal Resources of Four Blocks in Thar Coalfield, Sindh, Pakistan. vol. 115.
Director General, Geological Survey of Pakistan.
HDIP, 2012. Pakistan Energy Yearbook, 2012. 2012 edn. Ministry of Petroleum and
Natural Resourses, Islamabad.
Heriawan, M.N., Koike, K., 2008a. Identifying spatial heterogeneity of coal resource quality
in a multilayer coal deposit by multivariate geostatistics. Int. J. Coal Geol. 73, 307–330.
Heriawan, M.N., Koike, K., 2008b. Uncertainty assessment of coal tonnage by spatial
modeling of seam distribution and coal quality. Int. J. Coal Geol. 76, 217–226.
Hindistan, M.A., Tercan, A.E., Ünver, B., 2010. Geostatistical coal quality control in longwall
mining. Int. J. Coal Geol. 81, 139–150. http://dx.doi.org/10.1016/j.coal.2009.12.014.
Hohn, M.E., Britton, J.Q., 2012. A geostatistical case study in West-Virginia: all coals are
not the same. Int. J. Coal Geol. http://dx.doi.org/10.1016/j.coal.2012.12.004.
Olea, R.A., Luppens, J.A., Tewalt, S.J., 2011. Methodology for quantifying uncertainty in coal
assessments with an application to a Texas lignite deposit. Int. J. Coal Geol. 85, 78–90.
Pardo-Iguzquiza, E., Dowd, P.A., Baltuille, J.M., Chica-Olmo, M., 2013. Geostatistical model-
ling of a coal seam for resource risk assessment. Int. J. Coal Geol. 112, 134–140. http://
dx.doi.org/10.1016/j.coal.2012.11.004.
Pathan, A.G., Singh, R.N., Stace, R., 2013. Geotechnical Assessment of Block VIII at Thar
Coalfield, Pakistan. Paper presented at the 23rd World Mining Congress, Montreal,
Canada, 11–15 August, 2013.
Saikia, K., Sarkar, B.C., 2013. Coal exploration modelling using geostatistics in Jharia Coal-
field, India. Int. J. Coal Geol. 112, 36–52. http://dx.doi.org/10.1016/j.coal.2012.11.012.
Singh, R.N., Atkins, A.S., Pathan, A.G., 2010a. Determination of ground water quality
associated with lignite mining in arid climate. Int. J. Min. Environ. Issues 01, 65–78.
Singh, R.N., Atkins, A.S., Pathan, A.G., 2010b. Water resources assessment associated with
lignite operations in Thar, Sindh, Pakistan. Arch. Min. Sci. 55, 425–440.
Singh, R.N., Pathan, A.G., Reddish, D.J., Atkins, A.S., 2011. Geotechnical appraisal of the Thar
open cut mining project. Paper presented at the 11th Underground Coal Operators'
Conference, University of Wolongong, Wollongong, New South Wales (NSW),
Australia, 10 – 11 February 2011.
Tercan, A.E., Karayigit, A.I., 2001. Estimation of lignite reserve in the Kalburcayiri field,
Kangal basin, Sivas, Turkey. Int. J. Coal Geol. 47, 91–100.
Tercan, A.E., Ünver, B., Hindistan, M.A., Ertunç, G., Atalay, F., Ünal, S., Kıllıoğlu, Y., 2013.
Seam modeling and resource estimation in the coalfields of western Anatolia. Int.
J. Coal Geol. 112, 94–106. http://dx.doi.org/10.1016/j.coal.2012.10.006.
Thomas, R.E., Shah, A.A., Khan, S.A., Tgar, M.A., 1994. Analytical results on coal samples
collected from drilling activities in the Thar Desert, Sindh, Pakistan. U.S. Geological
Survey Open File Report.
Vasquez, C., Nieto, A., 2004. A statistical means of modeling lignite quality from e-log
and lab data. Paper presented at the SME Annual Meeting, Denver, Colorado, USA,
Feb. 23–25.