International Journal of Coal Geology 112 (2013) 107–113

Contents lists available at SciVerse ScienceDirect

International Journal of Coal Geology

j ourna l homepage: www.e lsev ie r .com/ locate / i j coa lgeo

Geostatistical drillhole spacing analysis for coal resource classification in the BowenBasin, Queensland

Olivier Bertoli a,⁎, Andrew Paul b, Zach Casley c, Doug Dunn d

a Geovariances Pty Ltd, PO Box 979, Wynnum,QLD 4178, Australiab Ex BHP Billiton Mitsubishi Alliance Coal, GPO Box 1389, Brisbane QLD 4001, Australiac Geovariances, 49b Av F. Roosevelt, 77215 Avon Cédex, Franced BHP Billiton Mitsubishi Alliance Coal, GPO Box 1389, Brisbane QLD 4001, Australia

⁎ Corresponding author.E-mail addresses: bertoli@geovariances.com (O. Bert

Andrew.Paul@bmacoal.com (A. Paul), zach.casley@macqDoug.L.Dunn@bmacoal.com (D. Dunn).

0166-5162/$ – see front matter © 2013 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.coal.2012.12.010

a b s t r a c t

a r t i c l e i n f o

Article history:Received 20 July 2012Received in revised form 14 December 2012Accepted 15 December 2012Available online 10 January 2013

Keywords:Drill hole spacing analysisPolygonal krigingAustralian Coal GuidelinesBowen BasinGlobal uncertainty

Geostatistical drill hole spacing analysis (‘DHSA’) for resource classification using the global estimation variancetechnique has been used across BHP BillitonMitsubishi Alliance (‘BMA’) Coal Operation's variousmines and pro-jects since 2004. Analysis of the results points to the emergence of possible patterns in the results for projectspertaining to specific coal measures being mined by BMA. This correlation may be a useful guide to assist in de-veloping resource classifications for projects based on the coal measures in which they occur. Comparison of theresults of classification using the Coal Guidelines versus classification using the geostatistical DHSAmethod for aselection of BMA's operating mines in Queensland's Bowen Basin indicates that the non-geostatistical approachleads to level of uncertainty that does not always agree with the complexity of the geology.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

The coal industry in Australia has been actively working in recentyears towards the (re-)integration of geostatistical techniques to theprocess of coal resource estimation and overall management of thecoal resource. Whilst the actual estimation of key project variablesmay still not be routinely obtained by implementing geostatistical esti-mation techniques, the area of resource classification has been a topicwhere these integrative efforts have been more widely implementedin the industry.

BMA (BHP Mitsubishi Alliance) has been one of the key proponentsof that success, having supported for almost a decade now the explicitintegration of geostatistical techniques to the characterisation of globalresource risk with a view to offer a quantitative framework to the deri-vation of adapted resource classification schemes.

Resource classification is a multi-facetted problem that needs toencompass a detailed characterisation of a wide range of factors (in-cluding but obviously not restricted to the estimation of mineabletonnages of coal) capable of impacting the level of confidence thatcan be placed on a coal resource. Any acceptable scheme of classifica-tion needs to be devised on the basis of quantitative measures of theuncertainty attached to these factors.

oli),uarie.com (Z. Casley),

rights reserved.

Bertoli et al. (2010) presented the use of a specific geostatisticaltechnique to support classification of coal resources, namelygeostatistical drill hole spacing analysis (‘DHSA’) based on global es-timation variance which provides a quantitative measure of theglobal estimation precision with which a particular variable for agiven seam/domain combination may be estimated at a particulardrilling spacing.

From a theoretical view point, the current paperwill not dwell uponthe simple geostatistical setting that underpins the global estimationvariance calculations but rather focus on the limitations and caveatsfor its use so that the results which are presented can be qualifiedaccordingly.

Many deposits, variables, and most of all geological settings havebeen analysed in the course of this decade-long application of DHSAto the varied suites of deposits and projects being mined and exploredby BMA in Australia. A simple taxonomy is proposed by which the re-sults are broken down according to the coal measures to which the dif-ferent seams being exploited belong. The simple but rather ambitiousobjective is to try and detect the existence of potential patterns of clas-sification for the different coal measures being mined in Australia.Without presuming if such patterns (that are highly dependent on thecorporate decisions being used to convert global precisions into re-source categories) may be turned into industrially accepted guidelines,their mere existence may constitute an interesting platform to stimu-late further work aimed at guiding the competent person for the classi-fication of the resource in their derivation of sustainable andtransparent classification schemes for coal deposits.

108 O. Bertoli et al. / International Journal of Coal Geology 112 (2013) 107–113

2. Regional geology in the Bowen Basin

The Bowen Basin is part of a connected group of Permo-Triassic ba-sins in eastern Australia that includes the Sydney and Gunnedah Basins.It occupies an area of approximately 160,000 km2, the southern half ofwhich is covered by the Surat Basin. Maximum sediment thickness inBowen basin reaches about 10,000 m, concentrated in two northtrending depocentres, the Taroom Trough to the east and the DenisonTrough to thewest (Fig. 1, Sliwa et al., 2008 provides the structural con-figuration of the Bowen Basin).

Tectonically, the Bowen basin may be subdivided into NNW–SSEtrending platforms or shelves separated by sedimentary troughs. Theunits, from west to east, are Springsure Shelf, Denison Trough, Collins-ville Shelf/Comet Platform, Taroom Trough, Connors and AuburnArches, interrupted by the Gogango Overfold Zone.

Basin development started with an extensional phase during theEarly Permian whereby volcanic, fluvial and lacustrine sedimentswere deposited in a series of half-graben in the east and a thick succes-sion of coals and non-marine clastics were laid in the west.

Fig. 1. Structural configuration of Bowen Basin (Sliwa et al., 2008).

The mid-Early to Late Permian is characterised by thermal subsi-dence. During this stage, basin-wide transgression allowed depositionof deltaic and shallow marine, predominantly clastic sediments aswell as extensive coal measures (GeoscienceAustralia, 2009). West-ward foreland loading during the Late Permian resulted to a period ofaccelerated subsidence, enabling the deposition of a thick successionof marine and fluvial clastics, accompanied by coal and Early- toMiddleTriassic fluvial and lacustrine clastics. Basin sedimentation was termi-nated by a compressional event during the Middle to Late Triassic.

3. Coal geology

The economic coal seams of interest for BMA in the Bowen Basinare hosted by 3 coal-bearing units deposited during the Late Permian:a) the Moranbah Coal Measures, b) its facies equivalent, GermanCreek Formation, and c) the Rangal Coal Measures (Fig. 2). The readeris kindly referred to Fig. 3 for the regional stratigraphy of the BowenBasin.

The Moranbah Coal Measures comprise the most extensive coal mea-sures in the northern Bowen Basin. The formation was deposited inmid-late Permian and is characterised by several laterally-persistent,relatively thick coal seams ofmedium to low volatile bituminous rank, in-terspersed with several thin minor seams. Relatively uniform thicknessesof about 230–300 m are noted for the Moranbah Coal Measures on thewestern margin of the Bowen Basin, which increase eastwards towardsthe depocentre to a maximum thickness of 760 m (Mallett et al., 1995).

The German Creek coalmeasures, deposited in early Late Permian, issubdivided into a lower 160 m-thick, marine-influenced unit barren ofcoal and an upper coal-bearing interval about 110 m thick (Falkner andFielding, 1990). The coal bearing unit of the German Creek coal mea-sures is correlatable to the Moranbah Coal Measures.

The youngest coal-bearing units in the Bowen Basin Permian se-quence are the Rangal Coal Measures (Quinn, 1985). It comprises of100–300 m light grey, cross-bedded, fine to medium-grained labilesandstones, grey siltstones, mudstones and coal seams. Cementedsections are common in the sandstones, at times reaching 40 m inthickness.

Underlying the Rangal Coal Measures are the Fort Cooper CoalMeasures which are typically comprised of tuffaceous sandstones,siltstones, mudstones and coal seams. At its type section in HailCreek Syncline, the unit reaches 400 m thickness (Jensen, 1968).The transition between the Rangal Coal Measures and the Fort CooperCoal Measures is generally clearly marked by the Yarrabee Tuff — abasin-widemarker bed comprised ofweak, brown tuffaceous claystone.The presence of tuffaceous beds within the Fort Cooper Coal Measuresdistinguishes it from the Rangal Coal Measures.

The boundary between the Fort Cooper and the Moranbah CoalMeasures is taken as the basal part of the lowermost tuffaceous seamin Fort Cooper. This boundary is sometimes difficult to identify becausethe Moranbah Coal Measures contain scattered tuffaceous units.

4. Australian Coal Guidelines

The Australian Guidelines for Estimating and Reporting of InventoryCoal, Coal Resources and Coal Reserves (‘the coal guidelines’), are a setof non-prescriptive rules destined to guide the competent person intheir classification of a coal inventory (exploration potential) or coal re-source. The central element to the scheme of classification proposed inthe guidelines is the essential notion of points of observation and thespacing between points of observation used to characterise a resource.

Points of observation as defined in the guidelines are intersections ofcoal bearing strata, at known locations, which provide Informationabout the coal. A point of observation for coal quality evaluation is nor-mally obtained by testing samples obtained from surface or under-ground exposures, or from bore core samples having an acceptablelevel of recovery.

Fig. 2. Distribution of coal-bearing formations and BMA areas of interest in Bowen Basin (from BMA).

109O. Bertoli et al. / International Journal of Coal Geology 112 (2013) 107–113

The guidelines state thatMeasured resourcemay be estimated usingdata obtained from points of observation usually less than 500 m apart,Indicated resource from points of observation less than 1000 m apartand Inferred resources less than 4000 mapart. The guidelines also spec-ify that these spacings may be extended (the authors would argue thatthe term “varied”would bemore adapted) if there is sufficient technicaljustification to do so; for example if supported by geostatistical analysis.

5. Geostatistical method of characterisation of global uncertainty:global estimation precision

Global estimation variance is an operational concept characterizingthe error associated with a particular sampling pattern and a given ge-ometry to be estimated. It is the required input to drill spacing analysissince themethod attaches a level of precision (or uncertainty) to a givendrilling budget (at a fixed drill spacing a given areawill require a certainbudget to be drilled).

Several geostatistical algorithms are available to obtain global esti-mation variances, and include:

• The combination of elementary estimation variances which offers afirst pass approximation of the global estimation variances thatworks rather well for known regular geometries;

• The post processing of a series of conditional simulations which of-fers a much more involved, but more precise alternative; and

• Polygonal kriging (PK) which is particularly adapted when the ge-ometry of the global areas to be characterized is fixed.

PK is designed to provide an estimated value of a variable of interestinside a set of areas delineated by polygons. Each polygon receives a sin-gle global estimation value. As for Block Kriging, PK requires each poly-gon to be associated with an internal discretization grid to perform thecalculations.

In order to illustrate the last point (PK), suppose we want to esti-mate the average raw ash content over an area that corresponds to a

Fig. 3. Bowen Basin stratigraphy (from BMA).

Table 6-1Estimation precision associated with Coal Resource Categories.

110 O. Bertoli et al. / International Journal of Coal Geology 112 (2013) 107–113

mining period of one year and characterize the precision of estimationthrough an estimation variance. This estimation variance depends on:

• The level of spatial continuity captured by the variogram model(Srivastava, 2013–this volume);

• The ‘geometry of the data’, i.e. the particular data locations used forestimation; and,

• The geometry of the area to be estimated.

Once a variogram model γ is obtained for the specific seam/domain/area being investigated the following variance can be obtained by PK forboth accumulation and thickness (Armstrong, 1998):

σ2 ¼ 2NV

∑i∫γ xi−yð Þdy− 1V2∬γ y−y′

� �dydy′− 1

N2 ∑i∑jγ xi−xj� �

ð1Þ

where N is the number of samples used to estimate the polygonal areaand V denotes the area to be estimated.

This leads to the estimation of the standard deviation for Ash usingthe formula for standard deviation of estimation ratio established byJournel and Huijbregts (1978):

σ2ash

ash2¼ σ2

Accu

Accu2 þ σ2thickness

Thickness2−2�ρAccu=Thickness

�σAccu

Accu� σ thickness

Thickness: ð2Þ

It must be noted that the uncertainty, which is in that case orders ofmagnitude lower than the other uncertainties listed above, attached tothe geometrical definition of the estimation area (see Chilès andDelfiner (1999)) is not taken into account in Eq. (1).

DHSA at BMA has actually been based on the industrial applicationof the first option (combination of elementary estimation variances)

performed directly on the raw variables (see Bertoli et al., 2010). Thedual simplification of working directly on the raw variables and usingthe combination of elementary variances has meant that an industrialapplication of the technique to awide range of projects, seams and vari-ables was possible. However the clear departure from optimal andsound geostatistical premises has also meant that careful monitoringand constant benchmarking of the results produced through the yearswere mandatory. The conceptual benchmarking was obtained by regu-larly comparing the spacings derived from implementing DHSA to thespacings obtained by post-processing a platform of conditionalco-simulations (CCS) of accumulation and thickness variables for a se-ries of seams capturing the diversity of coal measures being mined byBMA. Invariably the conceptual benchmarking concluded that DHSAspacings based on global estimation variance were producing resultsfor seam thickness and raw ash (the two variables used by BMA toguide their classification decisions) in close agreement with CCS resultsover periods equating to 5 years of production. To bemore specific, thecomparisons showed that DHSAwas offering good to very good predic-tions for thickness and more conservative predictions for raw ash (i.e.the spacings derived for Ash by DHSA tended to be larger than theones resulting from the application of CCS).

30.0

25.0

20.0

15.0

10.0

5.0

0.0

D24

H16

H15

D14

D52

P08

H19

D14

2

P14

P07

D14

1

D22

D21

H18

D30

D53

D17

Q02

H17

D31

Q03

% contribution of seam to resources within reserves

Fig. 4. Contribution to resources by major coal seams within Saraji Mine.

111O. Bertoli et al. / International Journal of Coal Geology 112 (2013) 107–113

The practical application and benchmarking of the techniqueactually confirmed that the global precisions derived in DHSA wererobust results that:

• Apply only to the deposit, seam, domain and variable considered;• Can only be used to assign precisions to estimation of the mean ofan attribute of interest for a global area equivalent to a certain pro-duction period, assuming a fixed mining rate; and,

• Are not applicable to any other area than the one implicit in the calcu-lations and, in particular, are not suited to assigning local confidenceintervals.

• Are more reliable for longer time periods (i.e. beyond the five yeartimeframes).

Their overall validity strongly depends on the adherence for thedeposit, seam, domain and variable considered to the hypothesis ofstationarity (statistical homogeneity) (Bertoli et al., 2010) that un-derlies the geostatistical calculations involved in the process.

6. Conversion of global estimation precisions intoresource classification

In general, coal resource classification in Australia is guided by theAustralian Coal Guidelines (Coalfields Geology Council of NSW andQueensland Mining Council, 2003). The Coal Guidelines recommendspacing between points of observation of 500, 1000 and 4000 m to de-fineMeasured, Indicated and Inferred resource categories, respectively.

4500

4000

3500

3000

2500

2000

1500

1000

500

Saraji (TK: BHP01022)

spac

ing

[m

]

Measured (+10 %)

Indicated (+ 20 %)

Inferred (+ 50 %)

750

1400

2500

0

Fig. 5. Comparison of the results of DHSA to the recom

These recommended that spacings are based on historically-provencontinuity of coal seams in the Hunter Valley and in the Bowen Basin.As a measure of confidence in the estimation results, BMA assignsequivalent ranges of errors for each of the resource categories, suchthat precision increases from Inferred to Indicated and Measured. Tosubstantiate the application of the coal guidelines in coal resource esti-mation, current BMA practice also involves an analysis of the spacingsobtained by DHSA for raw ash content and seam thickness for themajor seamcontributors of each project over an area equating to a nom-inal 5 year period. So for each project/mine, the seams contributing to acumulative tonnage exceeding 75% of the yearly production are select-ed in decreasing order of importance. The total areamined over a 5-yearperiod is derived from the cumulative seam thickness, in situ bulk den-sity and projected annual production. This is then used as an input tothe combination of elementary variance method so that a precisioncan be attached to any drill spacing being considered over that areafor ash and thickness. From the resulting precisions/variances, equiva-lent resource categories are derived using the same parameters asshown in Table 6-1.

For example an area being mined over a 5 year period if classifiedas Indicated should return an average value for thickness and raw ashwithin 20% of the estimated values.

7. Main results

7.1. Application of DHSA to Saraji Mine

BMA is currently mining the Saraji Mine in Central Queensland(Fig. 2), producing coking coal products and a small amount of ther-mal coal.

Three stratigraphic sequences which lie within the Moranbah CoalMeasures are mined at Saraji: the Harrow Creek seams (H seams), theDysart seams (D seams) and the P seams. Coal is uncovered by opencut methods utilising four draglines supplemented by truck/shovelstripping operations.

The seams investigated for DHSA at Saraji consisted of the following(Fig. 4):

• Harrow Creek Seams: H15, H16,• Dysart Seams: D14, D24, D141, D142, D52, D63,• P Seams: P07, P08, P14.

Coal Guidelines

500

1000

4000

mended spacings in the Australia Coal Guidelines.

Table 7-1DHSA spacings in metres (m) for various BMA coal projects.

operation

Blackwater (TK, RA) 550 1050 2100

Caval Ridge (RA) 800 1400 2800

Caval Ridge (TK) 500 1000 2450

Crinum M Block (RA) 1100 1900 3600

Daunia (RA) 650 1250 2800

Goonyella Riverside (TK) 650 1250 3150

Gregory Crinum (RA) 1100 1900 3600

Lotus North (TK core) 350 700 1850

Norwich Park (TK) 750 1450 3550

Peak Downs (TK) 700 1300 2600

Peak Downs (TK) 850 1700 4200

Poitrel (TK) 400 750 1800

Saraji (TK) 750 1400 2500

South Walker Creek (TK) 250 500 1000

Coal guidelines 500 1000 4000

Measured

(±10 %)

Indicated

(±20 %)

Inferred

(±50 %)

112 O. Bertoli et al. / International Journal of Coal Geology 112 (2013) 107–113

The variables investigated were seam thickness, raw ash content,washed ash, phosphorous and sulphur, relative density and yield atdiffering floating settings.

The variables retained for DHSA as per BMA's corporate guide-lines were thickness and raw ash. The Annual tonnage of 12.7 Mtwas converted to an annual area of ~1.6 million m2 equating to a5 year area of approximately 8 million m2.

DHSA based on the above decisions results in the following spacingsat Saraji (the spacings are compared to the spacings recommended in

Measured (+10 %)

Indicated (+ 20 %)

Inferred (+ 50 %)

4500

4000

3500

3000

2500

2000

1500

1000

500

spac

ing

[m

]

0SouthWalker

Creek (TK)

Poitrel(TK)

CavalRidge (TK)

Blackwater (TK, RA)

GoonyellaRiverside

(TK)

Daunia(RA)

PD

1000

500

250 400

750

1800 2450

1000

500 550

1050

2100 3150

1250

650 650

1250

2800

Fig. 6. Tabulation and plotting of drill spa

the coal guidelines that are presented on the right hand side of the fig-ure (Fig. 5)).

The geostatistical analysis thus indicates that the characterizationof spatial continuity of thickness and raw ash for the major tonnagecontributors can lead to a variation of the proposed spacings forthe different categories. In that instance it is argued that at SarajiMeasured resource may be estimated using data obtained frompoints of observation 750 m apart, Indicated resource from pointsof observation less than 1400 m apart and Inferred resources lessthan 2500 m apart (instead of 500 m, 1000 m and 4000 m as beingrespectively suggested by the guidelines).

7.2. Summary of DHSA results from all coal measures studied at BMA

The following table (Table 7-1) lists all the deposits the methodolo-gy has been applied at to date. They offer an actual sampling of the threemain Australian coal measures being exploited by BMA in the BowenBasin. The variable between brackets refers to the actual driver for theclassification proposed.

The results are then summarised by ascending required spacings(in m) for the Measured category (Fig. 6).

A close analysis of the results shows that this ranking allows agrouping of the deposits into the three principal coal measures beingmined:

1. Rangal Coal Measures (deposits/mines: South Walker, Poitrel,Blackwater, Daunia) return spacings for the Measured categoryaround 500 m or below, for Indicated typically at 1000 m orbelow, and for Inferred, less than 2000 m between points of ob-servation; the seams pertaining to these coal measures usuallyhave levels of variability for secondary variables (phosphorous,sulphur) such that no control of mine variability is possible unlessadapted short term in pit quality estimates are being undertakenat an operational level;

2. Moranbah coal measures (deposits/mines: Goonyella Riverside,Caval Ridge, Peak Downs, Norwich Park, Saraj) return spacings forthe Measured category around 750 m, for Indicated typically at1250–1500 m, and for Inferred around 2500 m; and,

eakowns(TK)

NorwichPark (TK)

Saraji (TK)Caval

Ridge (RA)

PeakDowns

(TK)

GregoryCrinum

(RA)

Crinum MBlock (RA)

CoalGuidelines

2600

1300

700 750

1450

3550 2500

1400

750 800

1400

2800

850

1700

4200

1100

1900

3600 3600

1900

1100 500

1000

4000

DHSA results

cings for various resource categories.

113O. Bertoli et al. / International Journal of Coal Geology 112 (2013) 107–113

3. German Creek coal measures (deposits/mines: Gregory Crinum,M-Block) for which spacings for the Measured category sit at1000 m, for Indicated at 2000 m, and for Inferred around 3500 m.

8. Conclusions

The following specific conclusions may be drawn upon the inspec-tion of the above results:

1. First, the summary above is in complete keeping with the under-standing of the differing geological environments for the three setsof coal measures. The stationarity hypothesis lying at the core ofthe implementation of DHSA comes to the fore eventually by pro-ducing patterns of category spacings that are consistent with theoverall level of geological complexity attached to each measure;

2. Second, it seems that for deposits in the Rangal Coal Measures,using the coal guidelines probably does not provide sufficientlevels of confidence for Measured resources. For these coal mea-sures the uncertainty in global estimation is too large when esti-mation is based on points of observation that are 500 m apartand that spacing needs to be reduced;

3. In the German Creek formation, the coal guidelines are probablytoo strict for Indicated resources, in other words larger spacingsbetween points of observations may generally be utilised to classi-fy the seams from the German Creek formation;

4. And generally speaking, 4 km seems to be too large a spacing forInferred resources in all three formations.

The general conclusion from this work is that the use of a “one sizefits all” classification scheme, in this case, the Coal Guidelines, for classi-fication of resources may result in inappropriate resource classifica-tions. The use of a geostatistical method, whereby the classification ofthe resource is driven by the actual in situ variability (or conversely

spatial continuity) of the resource under consideration stronglyrecommended by the authors as best practice for the industry.

References

Armstrong, M., 1998. Basic Linear Geostatistics. Springer Verlag, Berlin (166 pp.).Bertoli, O., Casley, Z., Mawdesley, C., Dunn, D., 2010. Drill hole spacing analysis for Coal

Resource Classification. Proceedings of 6th Bowen Basin Symposium 2010, Mackay,QLD, Australia.

Chilès, J.P., Delfiner, P., 1999. Geostatistics — Modeling Spatial Uncertainty, Wileyseries in Probability and Statistics, New-York, 696 p, Second edition. Wiley,Hoboken, NJ (734 pp.).

Coalfields Geology Council of NSW and Queensland Mining Council, 2003. Australiaguidelines for estimating and reporting of inventory coal. Coal Resources andCoal Reserves. 8p. Accessed from the Internet in December 2013 from http://www.jorc.org/pdf/coalguidelines.pdf.

Falkner, A.J., Fielding, C.R., 1990. Late Permian coal-bearing depositional systemsof the Bowen Basin. The GSA (Qld Division) Field Conference: Bowen BasinSymposium Proceedings, p. 36.

GeoscienceAustralia, 2009. Bowen Basin. Accessed in May 2012 from http://www.ga.gov.au/oceans/ea_Browse.jsp.

Jensen, A.R., 1968. Upper Permian and Lower Triassic Sedimentation in Part of the BowenBasin. Bureau of Mineral Resources Geology and Geophysics, Queensland (91 pp.).

Journel, A.G., Huijbregts, Ch.J., 1978. Mining Geostatistics. Academic Press, London(600 pp.).

Mallett, C.W., Pattison, P., McLennan, C., Balfe, P., Sulivan, D., 1995. Bowen Basin. In:Ward, C.R., Harrington, H.J., Mallett, C.W., Beeston, J.W. (Eds.), Geology of Aus-tralian Coal Basins: Geological Society of Australia Incorporated Coal GeologyGroup. Special Publication, 1, pp. 299–339.

Quinn, G.W., 1985. Geology of the Rangal Coal Measures and Equivalents. The GSA CoalGeology Group, Bowen Basin Coal Symposium Proceedings, p. 93.

Sliwa, R., Hamilton, S., Hodgkinson, J., Draper, J., 2008. Bowen Basin Structural Geology2007: an interpretation based on airborne geophysics. Petroleum Exploration Societyof Australia Special Publication, Sydney.

Srivastava, R.M., 2013. Geostatistics: a toolkit for data analysis, spatial prediction andrisk management in the coal industry. Special issue on geostatistics of the Interna-tional Journal of Coal Geology 112, 2–13 (this volume).