Healthy HeadWaters
Coal Seam Gas Water Feasibility Study
Lithofacies transition probabilities within the
Walloon Subgroup and its coal measures
Technical report
Prepared for the Department of Natural Resources and Mines
March 2013
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
This document presents outcomes of Activity 1.3 (Scales of geological heterogeneity within the
Walloon Subgroup and its coal measures) of the Healthy HeadWaters Coal Seam Gas Water
Feasibility Study. This report is in addition to Esterle et al (2013).
The Healthy HeadWaters Coal Seam Gas Water Feasibility Study is analysing the opportunities
for, and the risks and practicability of, using coal seam gas water to address water sustainability
and adjustment issues in the Queensland section of the Murray-Darling Basin.
The study is being funded with $5 million from the Commonwealth Government, with support
from the Queensland Government, as part of the Healthy HeadWaters Program, which is
Queensland’s priority project funded through the Commonwealth Government’s Water for the
Future initiative. The study is being managed by the Queensland Department of Natural
Resources and Mines (DNRM), and is due to finish in 2012.
This report was prepared by UniQuest Pty Ltd.
Citation
This report may be cited as:
Hamilton, SK,1
Desassis, N,2
Esterle, JS,1
and Tyson, S,1, 2013. Lithofacies transition probabilities within the Walloon Subgroup and its
coal measures. Report by School of Earth Sciences, University of Queensland for Queensland
Department of Natural Resources and Mines. Brisbane.
1. University of Queensland School of Earth Sciences; 2. Geovariances Pty Ltd
18 March, 2013
UniQuest Project No: 00572 - addendum
Disclaimers
This document was prepared exclusively for the State of Queensland (Department of Natural
Resources and Mines) and is not to be relied upon by any other person. UniQuest Pty Ltd has
made every effort to ensure that the information provided is accurate but errors and omissions
can occur and circumstances can change from the time that the report or document was prepared.
Therefore, except for any liability that cannot be excluded by law, UniQuest Pty Ltd excludes
any liability for loss or damage, direct or indirect, from any person relying (directly or indirectly)
on opinions, forecasts, conclusions, recommendations or other information in this report or
document.
© State of Queensland (Department of Natural Resources and Mines) 2013.
The State gives no warranty in relation to the contents of this report (including accuracy,
reliability, completeness, currency or suitability) and accepts no liability (including without
limitation, liability in negligence) for any loss, damage or costs (including consequential
damage) relating to any use of the contents of this report.
UniQuest File Reference: 00572 addendum
Page i
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
Executive Summary
Heterogeneity can be defined by the range of sedimentary lithofacies ‐ coals, mudstones, siltstones and sandstones, their grainsize and composition, and their thickness, shape and lateral extent. Together with faults and fractures, these attributes influence reservoir properties, hydraulic connectivity and potential flow pathways for water and gas. Ongoing research aims to stochastically model the spatial geological heterogeneity within the Walloon Subgroup coal seam gas reservoir. Unlike natural outcrop and manmade exposures, where strata can be directly mapped, 3D subsurface models rely on interpretation and interpolation of drilling data. As such, lithofacies models draw on the principle of Walther’s Law (Walther, 1894), i.e., that a conformable vertical succession reflects the lateral distribution of facies. Data diagnostics are an important first step in any modelling workflow. In this context, vertical transition probability is a useful tool for quantifying geological heterogeneity, testing depositional models, and interpreting the structural characteristics of the input data to geostatistical models. We investigated the vertical stacking characteristics (vertical facies transitions) of the Walloon Subgroup as part of a plurigaussian truncated1 (PGS) modelling exercise. A regionally‐consistent stratigraphic framework model forms the foundation to the project, interpreted from open‐file digital wireline log data. Density and gamma‐ray logs were used to assign lithofacies (clean sandstone, silty sandstone, siltstone, mudstone, carbonaceous mudstone, dirty coal, clean coal) down each well. Lithofacies contacts were then analysed by formation using border effects curves. These border effects curves represent the combination of two geostatistical structural tools (the negative ratio between the indicator non‐centred covariance and the univariate variogram). A sliding scale of various lag lengths was applied to each down‐hole lithofacies log, and used to calculate transition probabilities according to the example formula below: Determination of lithofacies transition probabilities, in both the up‐ and down‐hole directions, revealed that vertical stacking patterns are non‐random and show trends that can be used to guide realistic stochastic simulations. This work identified meaningful differences between the constituent formations of the Walloon Subgroup. The majority of lithofacies transitions are symmetrical, i.e., there is an equal probability of leaving, say, clean sandstone for siltstone in both the up‐ and down‐hole directions over short distances. However, a small number of transitions are asymmetrical and likely reflect autogenic (local intrabasinal) sedimentary processes. Whilst this asymmetry cannot be reproduced by current simulation methods, its geological interpretation enhances understanding of the depositional system and assists in its conceptualisation. Overall, the results demonstrate that border effects curves can be used to quantitatively model cyclicity in the Walloon Subgroup. 1
Stochastic model that is able to reproduce lithofacies proportion variations and contact characteristics
UniQuest File Reference: 00572 addendum
Page ii
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
TABLE OF CONTENTS
EXECUTIVE SUMMARY ..................................................................................................... II 1. INTRODUCTION AND OBJECTIVES .................................................................... 6 1.1 Context: Geological heterogeneity within the Walloon Subgroup.......... 6 2. SCOPE ........................................................................................................................ 7 3. GEOLOGICAL SETTING ......................................................................................... 8 4. 5. 6. 3.1 Depositional setting ....................................................................................... 8 3.2 Stratigraphy .................................................................................................. 10 DATA ......................................................................................................................... 11 4.1 Lithofacies interpretation ............................................................................ 11 4.2 Diagnostics: lithofacies thickness distributions, proportions and
univariate variograms .................................................................................. 12 LITHOFACIES CONTACT ANALYSIS................................................................. 15 5.1 Border effects calculation ........................................................................... 15 5.2 Example output ............................................................................................ 17 RESULTS ................................................................................................................. 18 6.1 Interpreting the border effects curves: Durabilla Formation example . 18 6.2 Leaving lithofacies 1-4 ................................................................................ 21 6.3 Leaving mire lithofacies 5, 6, 7 .................................................................. 24 6.4 Close-up examples of asymmetry: interpretation ................................... 25 6.5 Overview of results ...................................................................................... 32 7. CONCLUSIONS....................................................................................................... 32 8. REFERENCES......................................................................................................... 34 UniQuest File Reference: 00572 addendum
Page iii
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
LIST OF FIGURES
Figure 1. Conceptualised flow through isotropic vs anisotropic reservoirs. The
vertical and lateral continuity and associations of sedimentary lithofacies will
influence the connectivity of coal reservoirs, aquitards and aquifers, and the way
that water and gas flow through the sequence. ........................................................................7 Figure 2. Map showing location of the Surat Basin and related Mesozoic basins,
Surat Basin structure, Walloon coal outcrop/subcrop and the area of this study.
Outlier of the Bowen Basin south of Dalby based on Day et al. (2008); structure
modified from Day et al. (2008) and Geological Survey of Queensland (2011a). ..................8 Figure 3. Block diagram of a meandering river system, showing similar
depositional environments to those proposed for the Walloon Subgroup (from
Draper in Beeston & Gray 1993). ............................................................................................9 Figure 4. Schematic illustration of various coal type associations. .......................................10 Figure 5. Simplified lithostratigraphy of the Walloon Subgroup in the study area
(modified from Hamilton et al., 2012). ..................................................................................11 Figure 6. Schematic chart of contact relationships between the different
lithofacies, based on the density and gamma ray thresholds given in Table 1. .....................12 Figure 7. Relative thickness of different lithofacies in the Walloon Subgroup as
derived from wireline logs. N=350 wells and n=190299 counts of different
lithofacies. ..............................................................................................................................13 Figure 8. Vertical indicator experimental variograms per lithofacies for the
Juandah Coal Measures..........................................................................................................14 Figure 9. Example stratigraphic correlation through the Walloon Subgroup. The
gamma ray log is colour-coded by the different lithofacies determined by the cutoff values................................................................................................................................15 Figure 10. Sequential diagram illustrating an up-hole border effects calculation
for a transition from Lithofacies A to Lithofacies B. The resultant transition
probability is a ratio of lengths, generated by sliding a lag bar (h) up each
lithofacies log and blocking out sections of core over which the transitions are
possible. .................................................................................................................................16 Figure 11. Schematic diagram illustrating an up-hole border effects calculation
for a transition from Lithofacies A to Lithofacies B. The resultant transition
probability is a ratio of lengths, generated by sliding a lag bar (h) up each
lithofacies log and blocking out sections of core over which the transitions are
UniQuest File Reference: 00572 addendum
Page iii
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
possible. It is possible to land in siltstone (brown) when leaving sandstone
(yellow) with a step h, when we are leaving intervals 1 and 4. .............................................17 Figure 12. Example output. A single border effects curve was generated for each
formation. The different formations are represented using the colour scheme
shown left. The red border effects curve was generated from all the study data
across all formations. These colours are used in subsequent figures. ....................................18 Figure 13. Vertical proportion curve and un-normalised border effects matrix for
the Durabilla Formation. Border effects matrix - The red lines represent the
probability of being in that lithofacies, knowing we are not in the diagonal
lithofacies, irrespective of lag distance. The pink rectangle is expanded in Figure
14. For a given column, the border effects curves represent the probability of
landing in that lithofacies when leaving the lithofacies with an index displayed on
the diagonal. The X axis is the transition probability and the Y axis is the lag
distance (range -8 to +8 m) in the up and down-hole directions. ..........................................19 Figure 14. Example Durabilla Formation interpretation key when leaving
sandstone (Lithofacies 1). Border effects curves are un-normalised. Inverted
border effects curve = separation. The initial probabilities are a function of the
thresholds (wireline log cut-off values). The red lines represent the probability of
being in that lithofacies, knowing we are not in sandstone, irrespective of lag
distance. .................................................................................................................................20 Figure 15. Example of a Durabilla Formation border effects matrix where the
curves have been normalised for proportions. The X axis is the normalised
transition probability ratio and the Y axis is the lag distance (range -5 to 5 m) in
the up and down-hole directions. ...........................................................................................21 Figure 16. Normalised border effects matrix – leaving lithofacies 1, 2, 3 and 4 for
other channel and overbank lithofacies. Line colours as per Figure 12.................................22 Figure 17. Normalised border effects matrix – leaving lithofacies 1, 2, 3 and 4 for
mire facies (5, 6, 7). Line colours as per Figure 12. The Springbok curves are
commonly noisy due to the relative lack of data and lower proportion of mire and
overbank facies. .....................................................................................................................23 Figure 18. Normalised border effects matrix – leaving mire facies for channel and
overbank facies. Line colours as per Figure 12. ....................................................................24 Figure 19. Normalised border effects matrix – leaving a mire facies for another
mire facies. Line colours as per Figure 12. ............................................................................25 Figure 20. Normalised border effects curves for a Carb. mdstSandstone
transition. The red line is the sill............................................................................................26 UniQuest File Reference: 00572 addendum
Page iv
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
Figure 21. Normalised border effects curves for a Shaley coalSandstone
transition. The red line is the sill............................................................................................26 Figure 22. Distribution of coal-rich and coal-poor intervals from north to south.
Hypothesis: prograding sequence with thicker coals and thicker sandstones upsection and to the north (modified from Sliwa & Esterle 2008). ...........................................28 Figure 23. Normalised border effects curves for a Silty sandstoneSiltstone
transition. The red line is the sill............................................................................................29 Figure 24. Normalised border effects curves for a Shaley coalMudstone
transition. The red line is the sill. The blue arrow highlights an excess of
mudstone beyond the range, reflecting the need to pass through carbonaceous
mudstone first.........................................................................................................................30 Figure 25. Normalised border effects curves for a Silty sandstone’Clean’
sandstone transition. The red line is the sill. ..........................................................................30 Figure 26. Correlation section showing the Taroom Coal Measures, Durabilla
Formation and the upper part of the Hutton Sandstone. ........................................................31 Figure 27. Map of lithofacies proportions per well in the Durabilla Formation. Pie
chart colours: ‘clean’ sandstone is yellow and silty sandstone is orange. .............................31 LIST OF TABLES
Table 1 Density and gamma value cut offs used to develop a lithofacies
interpretation. Preference given to density, followed by gamma value. ................................11 LIST OF EQUATIONS
Equation 1......................................................................................................................................... 16
Equation 2......................................................................................................................................... 18
UniQuest File Reference: 00572 addendum
Page v
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
1.
INTRODUCTION AND OBJECTIVES
Activity 1.3 of the Healthy Head Waters Coal Seam Gas (CSG) Water Feasibility Study aims to develop a platform in which issues of geological heterogeneity, and their potential impact on flow pathways and aquifer connectivity, can be explored. Specifically, this work aims to develop a static geological model of the Walloon Subgroup in the eastern Surat Basin. The present report builds on the static modelling work of Esterle et al. (2013) by investigating lithofacies transition probabilities in the Walloon Subgroup. The deliverables of the project are:  Determination of lithofacies transition probabilities;  A preliminary geological interpretation of lithofacies transition probabilities;  Documented geostatistical workflow(s), with associated assumptions and interpretations. 1.1
Context: Geological heterogeneity within the Walloon Subgroup
Pathways for the flow of gas and water within coal measures are controlled by: (1) the lateral continuity and connectivity of the coals and intervening rock strata; (2) their reservoir properties; and (3) their relationship to structural discontinuities (faults and fractures) and tectonic stress. Fracture networks within the coals provide the dominant pathways for flow of gas and fluids. The permeability of the intervening strata, mainly low porosity lithic sandstones (generally <2 millidarcies or mD; Geological Survey of Queensland 2011a) and mudstones, is considered to be low. The lateral continuity of coals in the Walloon Subgroup is variable and stratigraphic marker horizons are rare, resulting in low confidence in basin‐scale correlations of individual coal seams. Correlation in areas of high data density suggests that some seams split or thin over short distances (<5 km), and this is corroborated by the general size of open cut coal mines. As a result, it is difficult to determine the spatial heterogeneity of the Walloon Subgroup by connecting the different rock types between wells. Consequently, dynamic groundwater flow simulations are performed on up‐scaled stochastic geological models, wherein lithofacies and their associated properties are modelled using a range of statistical approaches. Without a good understanding of the overall spatial heterogeneity of the system, flow within the Walloon Subgroup is likely to be poorly understood (Figure 1). As a consequence, the response and vulnerability of overlying and underlying aquifers to their depressurisation as a result of CSG extraction can be uncertain. The objective of this study is to further improve our understanding of the spatial heterogeneity of the Walloon Subgroup, by investigating vertical lithofacies transition UniQuest File Reference: 00572 addendum
Page 6
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
probabilities. Lateral lithology transition probabilities reflect vertical transition probabilities. Interpreting the structural characteristics of lithofacies well log data relative to depositional models enhances knowledge of the depositional system, thus aiding in its conceptualisation. Figure 1. Conceptualised flow through isotropic vs anisotropic reservoirs. The vertical and lateral continuity and associations of sedimentary lithofacies will influence the connectivity of coal reservoirs, aquitards and aquifers, and the way that water and gas flow through the sequence. 2.
SCOPE
The first part of this activity (Esterle et al., 2013) produced an internally consistent, static, geological model of the lithological heterogeneity in the Walloon Subgroup in the eastern Surat Basin. The present contribution investigates the vertical stacking characteristics (vertical facies transitions) of the Walloon Subgroup across the core region of CSG production and greatest data density (Figure 2). Unlike natural outcrop and manmade exposures, where strata can be directly mapped, 3D subsurface models rely on interpretation and interpolation of drilling data. As such, lithofacies models draw on the principle of Walther’s Law (Walther, 1894), i.e., that a conformable vertical succession reflects the lateral distribution of facies, assuming no unconformity. Information on vertical lithofacies transition probabilities assists in determining the lateral continuity of the coals and their relationship to sandstones, which potentially act as flow pathways for water, and mudstones, which act as aquitards. If vertical stacking patterns are determined to be non‐random, these patterns can be used to guide realistic stochastic simulations in future. For the purpose of this activity, up‐ and down‐hole transition probabilities for lithofacies (clean sandstone, silty sandstone, siltstone, mudstone, carbonaceous mudstone, shaley coal, clean coal) combination pairs are presented graphically using variograms and transition probability matrices. UniQuest File Reference: 00572 addendum
Page 7
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
Figure 2. Map showing location of the Surat Basin and related Mesozoic basins, Surat Basin structure, Walloon coal outcrop/subcrop and the area of this study. Outlier of the Bowen Basin south of Dalby based on Day et al. (2008); structure modified from Day et al. (2008) and Geological Survey of Queensland (2011a). 3.
GEOLOGICAL SETTING
3.1
Depositional setting
In a local study of the coal fields in the Clarence‐Moreton Basin, Fielding (1993) interpreted the Walloon Subgroup as as a series of major channel, floodplain and mire (swamp) deposits (Figure 3). To date, this depositional model is the most widely applied by CSG companies operating in the Surat and Clarence‐Moreton basins. Utilising subsurface data and comparison to coal mine information, Sliwa & Esterle (2008) proposed a southerly‐prograding fluvial model for the Walloon Subgroup in the eastern Surat Basin. They identified up‐section stratigraphic changes in coal depositional style, from laterally continuous seams, to splitting and converging seams to pod‐like forms, reflecting progressively more upstream depositional environments. UniQuest File Reference: 00572 addendum
Page 8
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
In this context, coals can occupy abandoned channels, or occur in interfluves, similar to modern analogues. The question being raised in this study is: to what extent do coals cap coarsening or fining up sequences or blocky sands, and how does this vary up‐
sequence and across the basin. Figure 4 illustrates the variety of patterns that could be observed in vertical lithofacies transitions within a stratigraphic sequence. Figure 3. Block diagram of a meandering river system, showing similar depositional environments to those proposed for the Walloon Subgroup (from Draper in Beeston & Gray 1993). UniQuest File Reference: 00572 addendum
Page 9
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
Figure 4. Schematic illustration of various coal type associations. 3.2
Stratigraphy
A regionally‐consistent stratigraphic framework model forms the foundation to the study, interpreted from open‐file digital wireline log data (see Esterle et al., 2013). The Middle Jurassic Walloon Subgroup is strongly heterogeneous, comprising an upper (Juandah) and lower (Taroom) coal measures separated by a relatively coal‐barren unit (Tangalooma Sandstone), with a transitional unit at the base (Durabilla Formation). Within these units lateral variation in coal character is high, precluding a regionally agreed coal group or seam correlation. Nonetheless, the units are laterally extensive, facilitating an assessment of lithofacies transition probabilities in a regional stratigraphic context. For more detailed information on the geological setting see Esterle et al. (2013). UniQuest File Reference: 00572 addendum
Page 10
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
Figure 5. Simplified lithostratigraphy of the Walloon Subgroup in the study area (modified from Hamilton et al., 2012). 4.
DATA
4.1
Lithofacies interpretation
Density and gamma‐ray logs were used to assign lithofacies (sandstone, silty sandstone, siltstone, mudstone, carbonaceous mudstone, shaley coal, coal) down each well (Table 1). Table 1 Density and gamma value cut‐offs used to develop a lithofacies interpretation. Preference given to density, followed by gamma value. Priority was given to the density value, for identification of coal lithofacies, followed by the gamma values to identify sandstones, siltstones and mudstones. A query was run on the lithofacies logs, which allowed the rock type to be contiguous from the first code UniQuest File Reference: 00572 addendum
Page 11
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
occurrence to the first different code occurrence. This query compounded lithofacies less than 30 cm thick (i.e. any beds thinner than 30 cm were included with the predominant lithofacies).
The bivariate distribution resulting from the number of log types and cut‐off values is depicted graphically in Figure 6. This schematic lithotype chart or ‘flag’ is a qualitative representation of the contact relationships between the different lithofacies, which are shown as different coloured rectangles. The x‐axis (density) corresponds to the identification of coal and non‐coal lithofacies, and the y‐axis (gamma ray) to the sequential deposition of sandstone, silty sandstone, siltstone and mudstone (or the inverse, both representing simple continuous patterns of sediment deposition under decreasing or increasing energy regimes). This flag implies that silty sandstone cannot be in direct contact with mudstone, whereas carbonaceous mudstone can be in contact with any of the channel or overbank lithofacies (1, 2, 3 and 4). Figure 6. Schematic chart of contact relationships between the different lithofacies, based on the density and gamma ray thresholds given in Table 1. Digital wireline log data were sourced from the Queensland Petroleum Exploration Database (QPED; Geological Survey of Queensland, 2011b) and the Queensland Digital Exploration Reports System (QDEX, Geological Survey of Queensland 2011c). For detailed information on our lithofacies identification, wireline log normalisation and cleaning procedures see Esterle et al. (2013). 4.2
Diagnostics: lithofacies thickness distributions, proportions and
univariate variograms
The relative thickness distribution of the different lithofacies is shown in Figure 7. Coal and coaly lithofacies are thin (less than 1 m thick). There is an increase in the bed thickness from fine grained mudstones and siltstone to sandstone beds, with thicker sandstone units possibly representing single or stacked sandstones with a mode of 16 to UniQuest File Reference: 00572 addendum
Page 12
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
32 m. The thickness distribution of the lithofacies reflects the relatively heterogeneous and thin‐bedded nature of the Walloon Subgroup. The general thickness ranges of the different lithofacies are as follows:  Coal seams <1 m
 Mudstones and siltstones 4‐8 m
 Silty sandstones <2 m
 Sandstones 8‐32 m
Figure 7. Relative thickness of different lithofacies in the Walloon Subgroup as derived from wireline logs. N=350 wells and n=190299 counts of different lithofacies. These bed thicknesses are corroborated by the vertical indicator simple variograms shown in Figure 8. The variograms shown are for the Juandah Coal Measures. The indicator variogram ranges differ slightly between the Walloon Subgroup formations, but in general show shorter ranges for the coal and finer‐grained lithofacies compared to the sandstones. UniQuest File Reference: 00572 addendum
Page 13
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
SANDSTONE
SILTY SANDSTONE
SILTSTONE
MUDSTONE
CARBONACEOUS MUDSTONE
SHALEY COAL
CLEAN COAL
Figure 8. Vertical indicator experimental variograms per lithofacies for the Juandah Coal Measures. UniQuest File Reference: 00572 addendum
Page 14
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
Figure 9. Example stratigraphic correlation through the Walloon Subgroup. The gamma ray log is colour‐coded by the different lithofacies determined by the cut‐off values. 5.
LITHOFACIES CONTACT ANALYSIS
5.1
Border effects calculation
Lithofacies contacts were analysed by formation using border effects curves. A sliding scale of various lag lengths was applied to each lithofacies log, and used to calculate transition probabilities according to the example formula below (Equation 1): UniQuest File Reference: 00572 addendum
Page 15
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
Equation 1 Equation 1 says that for a given lag distance (h), the probability of leaving sandstone for siltstone is equal to: the total core length of leaving sandstone for siltstone, divided by the total core length of leaving sandstone for any non‐sandstone lithofacies. As such, the transition probability is a ratio of lengths, generated by sliding a lag bar (h) up and down each lithofacies log and blocking out sections of core over which the transitions are possible. This concept is illustrated in Figures 10 and 11 below. 1.
2.
3.
4.
Figure 10. Sequential diagram illustrating an up‐hole border effects calculation for a transition from Lithofacies A to Lithofacies B. The resultant transition probability is a ratio of lengths, generated by sliding a lag bar (h) up each lithofacies log and blocking out sections of core over which the transitions are possible. UniQuest File Reference: 00572 addendum
Page 16
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
Figure 11. Schematic diagram illustrating an up‐hole border effects calculation for a transition from Lithofacies A to Lithofacies B. The resultant transition probability is a ratio of lengths, generated by sliding a lag bar (h) up each lithofacies log and blocking out sections of core over which the transitions are possible. It is possible to land in siltstone (brown) when leaving sandstone (yellow) with a step h, when we are leaving intervals 1 and 4. Lithofacies transition probabilities were calculated in both and up‐ and down‐hole directions for every well and every possible lithofacies combination. The different formations were analysed both independently and in combination. The resulting border effects curves represent the combination of two geostatistical structural tools: the negative ratio between the indicator non‐centred covariance, and the indicator variogram of the lithofacies we are leaving. For a given transition, the transition probability varies with increasing lag distance. Beyond a certain lag distance, the variation in transition probability values is no longer spatially correlated (no border effects or the range has been reached). 5.2
Example output
A single border effects curve was generated for each formation. The curves may be presented either un‐normalised or normalised for proportions. Figure 12 is an example output where the border effects curves have been normalised. The ‘normalised ratio’ values were generated by dividing the initial probability of landing in silty sandstone (2) by the probability of not being in sandstone (1) (Equation 2). They represent the probability of landing in silty sandstone, knowing we are leaving sandstone, irrespective of the lag distance. UniQuest File Reference: 00572 addendum
Page 17
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
Equation 2. The border effects curves are symmetrical, indicating that there is an equal probability of leaving sandstone for silty sandstone in both the up‐ and down‐hole directions over short distances. However, there are minor differences between the formations that bear consideration. In the Taroom Coal Measures we are 5 times more likely to land in silty sandstone when leaving sandstone, compared to 3.75x in the Juandah Coal Measures.
Figure 12. Example output. A single border effects curve was generated for each formation. The different formations are represented using the colour scheme shown left. The red border effects curve was generated from all the study data across all formations. These colours are used in subsequent figures. 6.
RESULTS
6.1
Interpreting the border effects curves: Durabilla Formation example
From Figure 13 it can be seen that the Durabilla Formation border effects matrix contains both symmetrical and asymmetrical curves. For a given column, the border effects curves represent the probability of landing in that lithofacies when leaving the lithofacies with an index displayed on the diagonal. Initially the border effects matrices were interpreted against the vertical proportion curves (VPC’s) (1st order trends), to UniQuest File Reference: 00572 addendum
Page 18
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
assess the effect of vertical non‐stationarity at the formation scale. The Durabilla Formation VPC displays an overall fining upward trend. Figure 13. Vertical proportion curve and un‐normalised border effects matrix for the Durabilla Formation. Border effects matrix ‐ The red lines represent the probability of being in that lithofacies, knowing we are not in the diagonal lithofacies, irrespective of lag distance. The pink rectangle is expanded in Figure 14. For a given column, the border effects curves represent the probability of landing in that lithofacies when leaving the lithofacies with an index displayed on the diagonal. The X axis is the transition probability and the Y axis is the lag distance (range ‐8 to +8 m) in the up and down‐hole directions. Figure 14 is an example interpretation key when leaving sandstone. The initial probability at the contact (lag distance = 0 m) is a function of the wireline log thresholds given in Table 1. The remainder of the border effects curve shows how the transition probability varies with changing lag distance. The orange border effects curve for P(12) is positive, implying that it is possible to transition from sandstone to silty sandstone; this reflects the lithotype chart in Figure 6. The initial probability of leaving sandstone for silty sandstone is 0.9, and decreases with increasing lag distance, and the increased likelihood of skipping over a silty sandstone bed. This relates back to bed thickness. Conversely, the border effects curves for P(13) and P(14) are inverted, implying separation. It is not possible to transition from sandstone to siltstone or sandstone to mudstone without passing through the intermediate lithofacies. It is possible to transition directly from sandstone to carbonaceous mudstone, but this is less likely, due to the lower proportion of carbonaceous mudstone. UniQuest File Reference: 00572 addendum
Page 19
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
1
INITIAL PROBABILITY OF LEAVING SANDSTONE FOR SILTY SANDSTONE = 0.9. DECREASES WITH INCREASING LAG DISTANCE. INITIAL PROBABILITY OF LEAVING SANDSTONE FOR SILTSTONE = 0. MUST PASS THROUGH SILTY SANDSTONE.  SEPARATION (C.F. LITHOTYPE FLAG, FIGURE 6). INITIAL PROBABILITY OF LEAVING SANDSTONE FOR MUDSTONE= 0. MUST PASS THROUGH 2 AND 3. NO SEPARATION.
INITIAL PROBABILITY OF LEAVING SANDSTONE FOR CARB. MUD LOW (~0.05). Figure 14. Example Durabilla Formation interpretation key when leaving sandstone (Lithofacies 1). Border effects curves are un‐normalised. Inverted border effects curve = separation. The initial probabilities are a function of the thresholds (wireline log cut‐off values). The red lines represent the probability of being in that lithofacies, knowing we are not in sandstone, irrespective of lag distance. When the border effects curves are normalised by conditional proportions, it becomes clearly evident that some lithofacies transitions show strong asymmetry, e.g., when we are leaving mire facies (5, 6, or 7) for sandstone (1) (Figure 15). Asymmetrical border effects curves imply excesses or deficits of a certain lithofacies in either the up‐ or down‐hole direction. These are variably compensated; for example in Figure 15, when leaving Lithofacies 5, down‐hole deficits of 1 and 2 are compensated by excesses of 3, 4, 6 and 7.
UniQuest File Reference: 00572 addendum
Page 20
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
Border effects matrix (normalised)
Figure 15. Example of a Durabilla Formation border effects matrix where the curves have been normalised for proportions. The X axis is the normalised transition probability ratio and the Y axis is the lag distance (range ‐5 to 5 m) in the up and down‐hole directions. 6.2
Leaving lithofacies 1-4
Figure 16 presents border effects curves for each formation when transitioning from one channel or overbank facies to another. The majority of the border effects curves show similar behaviour across all of the Walloon Subgroup formations. The Springbok Sandstone shows quite different behaviour. The most obvious forms of asymmetry in this matrix are discussed later. UniQuest File Reference: 00572 addendum
Page 21
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
Figure 16. Normalised border effects matrix – leaving lithofacies 1, 2, 3 and 4 for other channel and overbank lithofacies. Line colours as per Figure 12. Figure 17 presents border effects curves for each formation when leaving channel or overbank facies for mire facies. The most noticeable feature is the excess of coal in the down‐hole direction when leaving sandstone (row 1). We are more likely to land in carbonaceous mudstone, shaley coal or coal when leaving sandstone in the down‐hole direction over short distances. Another noticeable feature is the sharp deficit of coal at the origin (negative inflection) (i.e. decreased probability to land in shaley coal or coal when leaving channel or overbank facies). These trends are discussed later in Section 6.4.
UniQuest File Reference: 00572 addendum
Page 22
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
Figure 17. Normalised border effects matrix – leaving lithofacies 1, 2, 3 and 4 for mire facies (5, 6, 7). Line colours as per Figure 12. The Springbok curves are commonly noisy due to the relative lack of data and lower proportion of mire and overbank facies. UniQuest File Reference: 00572 addendum
Page 23
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
6.3
Leaving mire lithofacies 5, 6, 7
Figure 18 presents border effects curves for each formation when transitioning from a mire lithofacies to a channel or overbank facies. In the Durabilla Formation example, the transition from mire facies (5, 6, or 7) to sandstone (1) was shown to be strongly asymmetric (see Section 6.1). The first column of Figure 18 reveals that this trend is common to the other Walloon Subgroup formations. Each of these curves was generated independently using disparate datasets, suggesting they represent strong trends in the input data and thus cannot be noise. Figure 18. Normalised border effects matrix – leaving mire facies for channel and overbank facies. Line colours as per Figure 12. Of the four border composite border effects matrices presented in this section, Figure 19 is the only example where the transitions are symmetrical for every formation of the Walloon Subgroup. There is an equal probability of transitioning from one mire facies to another in both the up‐ and down‐hole directions over short distances. UniQuest File Reference: 00572 addendum
Page 24
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
Figure 19. Normalised border effects matrix – leaving a mire facies for another mire facies. Line colours as per Figure 12. 6.4
Close-up examples of asymmetry: interpretation
The previous section presented four composite border effects matrices representing all possible lithofacies transitions along the study wells. This section examines the key examples of asymmetry in detail by comparing the up and down ranges, and interprets possible causes for the variation between formations. In Figure 20 below, there is a deficit of sandstone in the down‐hole direction. The ‘normal rate’ of sandstone, or the ‘sill’, is reached more quickly in the up‐hole direction than the down‐hole direction. This suggests that we are more likely to land in sandstone when leaving carbonaceous mudstone when going up. This tendency for carbonaceous mudstone to be overlain by sandstone is consistent with autocompaction of peats and channel attraction, or channel incision. For this particular example, the shorter the up‐hole range, the stronger the asymmetry. In this context, the Tangalooma Sandstone and Juandah Coal Measures curves exhibit the weakest asymmetry, which may suggest an upward decrease in channel switching and fining up cycles. UniQuest File Reference: 00572 addendum
Page 25
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
Figure 20. Normalised border effects curves for a Carb. mdstSandstone transition. The red line is the sill. Similarly, the sill is reached quicker when leaving shaley coal for sandstone in the up‐
hole direction (Figure 21). Figure 21. Normalised border effects curves for a Shaley coalSandstone transition. The red line is the sill. UniQuest File Reference: 00572 addendum
Page 26
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
These trends may be considered consistent with progradation of a fluvio‐lacustrine system (Figure 22), with up‐section stratigraphic changes in coal type associations from fining up in the Durabilla Formation and Taroom Coal Measures, to fining and coarsening up in the Tangalooma Sandstone, to fining and coarsening up and blocky in the Juandah Coal Measures (see Figure 4). In Figure 23, there is an excess of siltstone in the up‐hole direction. Unlike the previous two examples, the border effects curves for a transition from silty sandstone to siltstone are not inverted. As such, the transition is favoured in the direction for which the excess or ‘above‐normal rate’ of siltstone is maintained the longest. The suggestion that we are more likely to land in siltstone when leaving sandy siltstone in the up‐hole direction offers further support for strong fining up cycles.
UniQuest File Reference: 00572 addendum
Page 27
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
Isatis VPC’s
Upper Walloon streams more stable
 Less avulsion, thicker coaly units
NORTH
SOUTH
Figure 22. Distribution of coal‐rich and coal‐poor intervals from north to south. Hypothesis: prograding sequence with thicker coals and thicker sandstones up‐section and to the
north (modified from Sliwa & Esterle 2008). UniQuest File Reference: 00572 addendum
Page 28
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
Figure 23. Normalised border effects curves for a Silty sandstoneSiltstone transition. The red line is the sill. Figure 24 says that we are 1.5 times more likely to land in mudstone when leaving shaley coal in the down‐hole direction. The deficit of mudstone at the origin (negative inflection) (i.e. decreased probability to land in mudstone over short distances) likely reflects the necessity to transition through a carbonaceous mudstone bed before landing in mudstone. The magnitude of this excursion is small (<1 m) and most likely relates back to bed thickness (c.f. Figure 7). The Walloon Subgroup border effects curves in Figure 25 show similar asymmetrical behaviour, except for the Durabilla Formation, which is symmetrical. Leaving silty sandstone, there is an excess of ‘clean’ sandstone down‐hole in the Taroom Coal Measures and the Tangalooma Sandstone. This asymmetry is less obvious in the Juandah Coal Measures, which is known to contain thick, blocky sands (see Figure 9) deposited during a time of increased channel stability. The symmetrical Durabilla Formation curve may reflect combination coarsening and fining up sequences (Figure 26). Such patterns are consistent with distributary channel behaviour during transitional reorganisation and platform development. Some of the Walloon Subgroup border effects curves don’t reach the sill, indicating uncompensated excesses/deficits. The Durabilla Formation in Figure 25 is a good example of this. Possible causes include: (1) lateral non‐stationarity; or (2) the two lithofacies rarely occur together in a drill‐hole. In this case it is likely the former (Figure 27). UniQuest File Reference: 00572 addendum
Page 29
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
Figure 24. Normalised border effects curves for a Shaley coalMudstone transition. The red line is the sill. The blue arrow highlights an excess of mudstone beyond the range, reflecting the need to pass through carbonaceous mudstone first. Figure 25. Normalised border effects curves for a Silty sandstone’Clean’ sandstone transition. The red line is the sill. UniQuest File Reference: 00572 addendum
Page 30
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
Figure 26. Correlation section showing the Taroom Coal Measures, Durabilla Formation and the upper part of the Hutton Sandstone. Figure 27. Map of lithofacies proportions per well in the Durabilla Formation. Pie chart colours: ‘clean’ sandstone is yellow and silty sandstone is orange. UniQuest File Reference: 00572 addendum
Page 31
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
6.5
Overview of results
The results of this preliminary transition probability assessment imply that vertical stacking patterns in the Walloon Subgroup are non‐random and should be modelled as such. The border effects curves are consistent with prior geological knowledge, e.g., from open cut coal mines and core logging, suggesting that they represent a good test of the cut‐off values and lumping criteria used in the lithofacies identification procedure. The range of border effects appears to be quite short (~<10 m), reflecting the relatively heterogeneous and thin‐bedded nature of the Walloon Subgroup. As such, asymmetrical border effects curves are more easily explained by autogenic (local intrabasinal) sedimentary processes rather than formation‐scale VPC’s; the majority of the border effects curves showed similar patterns across all the Walloon Subgroup formations, which have contrasting VPC’s. Based on this work, there is a strong case for varying the structural characteristics of the input data to future geostatistical models and modelling the Springbok Sandstone and Durabilla Formation separately. The Taroom Coal Measures and Tangalooma Sandstone could plausibly be combined, and the Juandah Coal Measures separated out.
7.






CONCLUSIONS
Border effects curves represent a useful tool for quantifying geological heterogeneity, testing depositional models, and interpreting the structural characteristics of the input data to geostatistical models.
Determination of lithofacies transition probabilities, in both the up‐ and down‐hole directions, revealed that vertical stacking patterns in the Walloon Subgroup are non‐
random and show trends that can be used to guide realistic stochastic simulations. Many lithofacies transitions are symmetrical, i.e., there is an equal probability of leaving, say, sandstone for siltstone in both the up‐ and down‐hole directions over short distances. Some transitions are asymmetrical and likely reflect autogenic sedimentary processes.
The majority of the border effects curves show similar behaviour across all of the Walloon Subgroup formations. Conversely, the Springbok Sandstone shows quite different behaviour.
Differences between the ranges of the border effects curves for the constituent formations of the Walloon Subgroup are consistent with a prograding fluvio‐
lacustrine system.
– Asymmetrical border effects curves suggest strong fining up sequences that get abandoned and capped by mudstones then coals.
– Coals tend to be overlain or cut by sandstones. This may reflect autocompaction of peats and channel attraction, or channel incision.

There appears to be a strong case for varying the structural characteristics of the input data to future geostatistical models.
UniQuest File Reference: 00572 addendum
Page 32
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures

The Springbok Sandstone and Durabilla Formation should be modelled separately. The Taroom Coal Measures and Tangalooma Sandstone could plausibly be combined, and the Juandah Coal Measures separated out.

Not all asymmetry can be reproduced by current stochastic modelling methods; however, its geological interpretation enhances understanding of the depositional system and assists in its conceptualisation. Future work would be best focused on developing simulation parameters that can reproduce the identified input asymmetry.
UniQuest File Reference: 00572 addendum
Page 33
Lithofacies transition probabilities within the Walloon Subgroup and its coal measures
8.
REFERENCES
BEESTON, J. & GRAY, A. 1993. The ancient rocks of Carnarvon Gorge. Department of Minerals and Energy, Queensland, 48pp. DAY R. W., BUBENDORFER P. J. & PINDER B. J. 2008. Petroleum potential of the easternmost Surat Basin in Queensland. In: Blevin J. E., Bradshaw B. E. & Uruski C. eds. Eastern
Australasian Basins Symposium III, pp. 191‐199. Petroleum Exploration Society of Australia, Special Publication Petroleum Exploration Society of Australia, Special Publication. ESTERLE J., HAMILTON S., WARD V., TYSON S. & SLIWA R. 2013. Scales of geological heterogeneity within the Walloon Subgroup and its Coal Measures. Report by School of Earth Sciences, University of Queensland for Queensland Department of Natural Resources and Mines, Brisbane. 50pp. FIELDING C. R. 1993. The Middle Jurassic Walloon CM in the Type Area, the Rosewood–
Walloon Coalfield, SE Queensland. Australian Coal Geology 9, 4‐15. GEOLOGICAL SURVEY OF QUEENSLAND 2011a. Digital Geological Mapping Data: Regional and 1:100 000 Sheet Areas (DVD). Geological Survey of Queensland, Brisbane. GEOLOGICAL SURVEY OF QUEENSLAND 2011b. Queensland Petroleum Exploration Database (QPED) (DVD). Geological Survey of Queensland, Brisbane. GEOLOGICAL SURVEY OF QUEENSLAND, 2011c. Queensland Digital Exploration Reports (QDEX) System (Online). Geological Survey of Queensland, Brisbane. HAMILTON S.K., ESTERLE J.S. & GOLDING S.D. 2012. Geological interpretation of gas content trends, Walloon Subgroup, eastern Surat Basin, Queensland, Australia. International
Journal of Coal Geology 101, 21‐35. SLIWA R. & ESTERLE J. 2008. Re‐evaluation of structure and sedimentary packages in the eastern Surat Basin. In: Blevin J. E., Bradshaw B. E. & Uruski C. eds. Eastern Australasian
Basins Symposium III p. 527. Petroleum Exploration Society of Australia, Special Publication. WALTHER J. 1894. Einleitung in die Geologie als historische Wissenschaft — Beobachtungen über die Bildung der Gesteine und ihrer organischen Einschlüsse. G. Fischer, Jena (1894), pp. 535–1055. UniQuest File Reference: 00572 addendum
Page 34