Mathematical Geology, VoL 8, No. 5, 1976
S e q u e n c e in A u s t r a l i a n C o a l S e a m s 1 M. Smyth 2 and A. C. Cook 3 An embedded Markov model is used to test microfithotype analyses o f subsections o f a wide range o f Australian coal seams for the presence o f nonrandom sequences o f lithologies. The data for individual seams, transformed to give five states (four states i f dirt bands are excluded), were summed into geologically and geographically distinct groupings. The results suggest that dirt bands form an essential part o f the sequences and that partial or complete cyclicity is present in many seam groupings. The cyclicity is either asymmetric or partially symmetric with the vitrite + clarite content o f the coal decreasing upwards within each cycle. A new cycle is marked either by a sharp reversion, or by a slightly gradational reversion, to a vitrite + clariterich lithology. This reversion may or may not be preceded by a dirt band. In virtually all groupings, a vitrite + clarite-rieh lithology is the most likely type after a dirt band. The sequences are similar to those that have been described in European coals and it seems probable that the presence o f intraseam, cyclic sequences is a normal, rather than an unusual condition, within coal seams. This cyelicity is a response to changes in the sedimentation balance. These changes are probably due in large part to processes originating within the peat-forming environment but processes external to this environment are also likely to produce cyclic sequences o f coal lithologies. KEY WORDS: cycles, Markov processes, cluster analysis, time-trend analysis,
coal petrology, sedimentology.
INTRODUCTION Smith (1962, 1968) established the existence of a characteristic mircofloral succession within the Westphalian coal seams of Britain and the association of the microfloras with a related succession of coal lithologies. These sucessions are summarized in Table 1. Smith showed that there is, within the coal seams, a n o r m a l lithological sequence. Cyclicity also is recognized in the sediments with which the coals are associated (Wanless and Weller, 1932; Duff and Walton, 1962; Read and Dean, 1967, 1968). No detailed work on the vertical variation of spore assemblages has been published for Australian coals. Australian coals generally have a low content of spores and some are virtually devoid of extractable spores. HowManuscript received 17 October 1975; revised 8 February 1976. 2 CSIRO Division of Mineralogy, North Ryde, New South Wales, Australia. 3 Department of Geology, University of Wollongong, Wollongong, New South Wales, Australia. 529 ~) 1976 Plenum Publishing Corporation, 227 West 17th Street, New York, N.Y. I0011. N o part o f this publication m a y be reproduced, stored in a retrieval system, or transmitted, in any form_ or by any means, electronic, mechanical, photocopying, microfilming, recording, or otherwise, without written permission o f the publisher.
530
M. Smyth and A. C. Cook
Table 1. Spore and Lithological Successions (After Smith, 1968)" Spore assemblage p h a s e s 5. lycospore 4. transition 3. densospore 2. transition Base-1. lycospore
Associated lithology vitrite+ clarite duroclarite, clarodurite crassidurite clarodurite, duroclarite vitrite+ clarite
"An incursion phase of tenuidurite with a spore flora may occur at any stage. ever, there is a large amount of data available on the vertical variations in coal lithology in the form of microlithotype 4 or maceral 4 analyses of subsection (ply) samples. Using some of these data, Smyth (1970, 1972) showed that the form of what she has termed "standard profiles" departs from that to be expected if their form were determined by random processes. This study concerns the nature of the vertical compositional variation in Australian coal seams using a first-order embedded Markovian model as a basis for the analysis of the original subsection data on coal lithology. The properties of Markovian models were described by Kemeny and Snell (1960) and Harbaugh and Bonham-Carter (1970). Krumbein (1967, 1968) described the use of Markovian models using both fixed-interval and lithology transition sampling. Krumbein and Dacey (1969) considered the properties of embedded Markov chains. The approach used here follows closely that adopted by Gingerich (1969) in a study of cyclicity in a coalbearing sequence. NATURE OF DATA USED The majority of the coal seams studied are Permian in age and come from either the Sydney Basin or the Bowen Basin. Some Triassic and Jurassic coals from other areas also are included. The locations of the samples are shown in Figure 1 ; the general stratigraphic relations of the coal measures in Table 2. In all, 10 different coal measure sequences, 105 different coal seams, and 324 individual sampling locations form the basis for the data. The total number of coal subsections is 2223 and additionally a further 1003 dirt bands were included in part of the study. 4 Coal petrology terms in this paper follow the definitions agreed on by the International Committee for Coal Petrology and published by the Committee as a Glossary (ICCP, 1963).
Sequence in Australian Coal Seams
1~
531
" ~ O W N S V ILLE • Mt. I SA
9 ~ . . . . .~~ l ~ k uuwEN "^'"
Q U E E N SL A N O
8 "~'~'~ ~
ROCKHAMPTON
!?
_
{ 10
'
I J ~ . P, P,R,,
ca" 2,~
5
~
_, 3 4 [
sou,.
.s~t.//,,o~,
w, LEs
r--~,
¢ " Z ADELAIDE \
2
~
SYDNEY
1
i - Newcastle Ct~ 3 - Singleton " - Walloon " 7 - Callide 9 - Collinsville " CM
BRISBANE
2 468 10 -
100 300 500 km I I I I I
Iliawarra Greta Ipswich Bara]aba Leigh Creek
Ct~ " ' •
Coal Measures
Figure I. Location of samples used in study. The sampling formed part of a continuing survey of Australian coal resources by the CSIRO Division of Mineralogy. Wherever possible, pillar samples were obtained, but in difficult conditions channel samples were taken. For analysis the full seam section is subdivided into plies or subsections; dirt bands were sampled separately. Subdivision of the coal is made upon its visual appearance--in practice, a subjective estimate of factors such as the brightness of the coal and the nature of the layering of coal lithotypes. Shadow radiographs also are used in some situations to decide where a section should be divided. In general, there is a lower limit of 0.1 m on the thickness of a coal subsection and an upper limit of 0.9 m. The actual sampling, subsampling, and therefore subsection definition was not performed by the authors and there seems to be no reason to suspect that any particular structuring of the data occurred due to sampling methods. It seems more probable that the sampling may have obscured some real structure. Continuous profiling would be a more suitable method of collecting petrographic data for testing for Markovian properties. However, this method is extremely time-consuming and it seems preferable to deal with a large number of profiles using subsection analyses, rather than obtaining greater detail on a small number of profiles.
M. Smyth and A. C. Cook
532
Table 2. Groupings of Coal Seams Used in Analysis of Transition Paths
Age New South Wales Permian
Stratigraphic unit Newcastle Coal Measures
Illawarra Coal Measures
Singleton Coal Measures Greta Coal Measures Queensland Jurassic Triassic
Walloon Coal Measures Ipswich Coal Measures
Permian
Callide Coal Measures Baralaba Coal Measures
District or formation Newcastle [Moon Island Beach Subgroup only (MIB)I Newcastle (excluding MIB) Illawarra and Burragorang District Lithgow District Singleton District
Rosewood-Walloon District Blackstone Formation Tivoli Formation Baralaba District Blackwater District Moura District Nipan District Theodore District
Collinsville Coal Measures South Australia Triassic
Leigh Creek Coal Measures
In order to obtain the data in a form suitable for the compilation of tally matrices, the percentage compositional data have to be transformed from percentages into multistate form. The most widely available petrographic analyses were microlithotype analyses performed by the selon la ligne method (ICCP, 1963). The dominant microlithotype is usually vitrite (>95~oo of the maceral vitrinite, and is usually, but not always, of bright appearance in the hand specimen). Clarite may be included with vitrite in presenting petrographic data, and the coal was assigned to one of four states by taking the extremes of the vitrite + clarite range in any given profile and then dividing this range into four (Table 3). The coal lithologies were split into four states after some experimentation. Real differences could be masked if fewer states were distinguished. The recognition of more than four states seemed to give a large number of trivial transitions, yielded unacceptably sparse matrices, and, in some situations, would not be warranted if the
Sequence in Australian Coal Seams
533
Table 3. Transformation of Data from Percentages into Multistate Form, Bayswater Coal, Singleton Coal Measures ~
Seam subsections A B C D E F G H J K L M N O P Q R S T U V W X Y Z AA BB CC DD EE FF GG HH JJ KK
LL MM
Vitrite + clarite
State
dirt band 58 58 dirt band 9 dirt band 16 dirt band 11 13 dirt band 22 dirt band dirt band 1 trace trace 1 dirt band 1 dirt band 1 dirt band 12 dirt band 11 dirt band 13 34 24 dirt band 23 13 14 41 42 dirt band
5 4 4 5 1 5 2 5 1 1 5 2 5 5 1 1 1 1 5 1 5 1 5 1 5 I 5 1 3 2 5 2 1 1 3 3 5
(Vm.~-- Vmi.)/4 = (58--0)/4
= 14} 1 2 3 4 5
state state state state state
0-14 15-29 30-44 45-58 dirt band
" Vmax = the maximum ~ value of vitrite+clarite; Vmi. = the minimum value of vitrite+ clarite.
534
0
M. Smyth and A. C. Cook
50-
Vt + Cl
0
50 I
0
50 Du
0
50 Fu
Figure 2. Petrographic profile of Main Lower seam, Blackwater, showing decline in overall content of vitrite+ clarite towards top of seam. Vt: vitrite; I: intermediates; CI: clarite; Fu: fusite; Du: durite (including microite).
precision of point-count analyses is taken into account. Material having an ash yield over 35 percent was distinguished separately as a "dirt band," in dividing the sections into plies, with a minimum thickness of 10 mm. The vast majority of the dirt bands are clay-rich (mainly kaolinite) so no further subdivision on dirt-band lithology was attempted. These dirt bands therefore constituted a fifth lithology (state 5, Table 3). The data were processed including the dirt bands as a five-state system, and excluding the dirt bands as a four-state system. The coal seams as defined here contain no absorbing states. Visual inspection of the profiles suggests that some do not fully comply with the requirement for the use of a Markovian model that the system be stationary in time. Some profiles show a tendency for cyclical changes in vitrite+ clarite content, but with a tendency for the overall content of vitrite + clarite to decline towards the top of the seam (Fig. 2). In profiles with a large number of plies it would be possible to remove this time trend, but it would be a highly subjective procedure for many of the profiles owing to the small number of plies. The effect of leaving a time trend in the data would be to increase the frequency of oscillatory transitions (ABAB) and to obscure cyclical patterns involving more than two states. It was decided that by disregarding the possibility of time trends the increase in the number of profiles and therefore the number of transitions which could be considered outweighed any loss of precision and so no attempt was made to remove time trend from the data. Tally, independent trials, transition, and difference matrices were prepared using both five-state (including dirt bands) and four-state data (excluding dirt bands) for each profile. The matrices were summed for each seam,
535
Sequence in AustralianCoal Seams Table 4. Transitions with Highest Values in Difference Matrices for Groups of Seams, Dirt Bands Included Groups of seams
Sequence
Z~
p
54552 54553 21433 51154 55554 55554 51154 51254 53154 53254 51134 51234 51234 51134 51424 51214 54514 55521
37 37 104 20 141 89 27 27 27 27 73 73 40 50 19 37 196 29
< 0.01 < 0.01 < 0.01 > 0.30" < 0.01 < 0.01 < 0.20~ < 0.20" < 0.20" < 0.20" < 0.01 < 0.01 < 0.01 < 0.01 > 0.30" < 0.01 < 0.01 < 0.10a
55514 55154 55554 53154
56 87 136 59
< 0.01 < 0.01 < 0.01 < 0.01
Walloon Coal Measures Leigh Creek Coal Measures Callide Coal Measures Blackstone Formation Tivoli Formation Baralaba District
Blackwater District Moura District Nipan District Theodore District Collinsville Coal Measures Illawarra Coal Measures (ICM) Lithgow District Newcastle Coal Measures [Moon Island Beach Subgroup (MIB)] Newcastle Coal Measures (excl. MIB) Singleton Coal Measures Greta Coal Measures
"Sequence does not depart significantly from random. for each geologically distinct set of coal measures, for related sets of coal measures, a n d for all profiles. The matrices for individual profiles are too sparse to be significant, so that only the results for groups of profiles will be considered here.
RESULTS The results for the groupings adopted are summarized in Tables 4 to 8 a n d illustrated in Figures 3 a n d 4, b y giving a five- or four-figure sequence to indicate the t r a n s i t i o n with the highest positive values in the difference matrices. T h u s the transitions with the highest m a r g i n a l probability of the sum for all the profiles (Table 6) using five-state data (51554) are: State 1 (lowest v i t r i t e + c l a r i t e content) State 2
--,dirt b a n d 5 ~state 1
536
M. Smyth and A. C. Cook
State 3 State 4 (highest vitrite + c l a r i t e content) State 5 (dirt b a n d )
~dirt band 5 ~dirt band 5 ~ state 4
The degree to which the difference matrices d e p a r t e d significantly f r o m a null hypothesis expectation o f i n d e p e n d e n t events has been assessed using a Z2 test with Xz c o m p u t e d a c c o r d i n g to the m e t h o d o f A n d e r s o n and G o o d m a n (1957). F l o w d i a g r a m s for some g r o u p i n g s are shown in F i g u r e 3. D i r t b a n d s are the m o s t a b u n d a n t o f the five lithologies d i s t i n g u i s h e d - 1003 o u t o f a t o t a l o f 3226 plies. I n o r d e r to test the possibility t h a t the dirt b a n d s might obscure i m p o r t a n t relationships b e t w e e n the coal plies because o f their a b u n d a n c e , the dirt b a n d s were o m i t t e d to give the four-state d a t a set. The values for X2 for the difference matrices generally exceed the a p p r o p r i a t e critical test values b y a greater m a r g i n for any given set when dirt b a n d s are included t h a n when they are excluded. This suggests t h a t the dirt b a n d s
Table 5. Transitions with Highest Values in Difference Matrices for Groups of Seams, Dirt Bands Excluded Groups of seams Walloon Coal Measures Leigh Creek Coal Measures Callide Coal Measures Blackstone Formation Tivoli Formation Baralaba District Blackwater District Moura District Nipan District Theodore District Collinsville Coal Measures Illawarra Coal Measures Lithgow District Newcastle (MIB) Newcastle (excl. MIB) Singleton Coal Measures Greta Coal Measures
Sequence
Z2
2421 2143 2112 2142 2423 2121 2141 2313 2143 4123 4413 3142 4142 4121 4441 4311 4441 4112 4143 3411 3413
11 91 14 14 45 29 29 15 45 33 20 10 10 24 94 9 27 18 47 41 41
a Sequence does not depart significantly from random.
p > 0.30" < 0.01 > 0.20" > 0.20" < 0.01 < 0.01 < 0.01 > 0.10" < 0.01 < 0.01 < 0.05 > 0.50 a > 0.50 < 0.02 < 0.01 > 0.50" < 0.01 < 0.10" < 0.01 < 0.01 < 0.01
Sequence in AustralianCoal Seams
537
Table 6. Sequences for Major Stratigraphic Groupings, Dirt Bands Included Supergroups of coal measures Baralaba Coal Measures (Baralaba, Moura, Nipan, Theodore, and Blackwater Districts) Illawarra (incl. Lithgow District) and Newcastle (Moon Island Beach) Coal Measures Ipswich Coal Measures (Tivoli and Blackstone Formations) All coals combined
Sequence
Z2
p
51234
166
< 0.01
54524
127
< 0.01
55554 51554
228 911
< 0.01 < 0.01
f o r m an essential element within the succession o f coat lithologies. A further indication of the significance of the dirt bands is the marked preference for vitrite-rich coals to follow a dirt band. The Leigh Creek data difference matrix shows a positive preference for either o f the vitrite-rich lithologies (States 4 and 3) to follow a dirt band (Tables 4 and 8). The few groups o f seams which do not show this preference show some other atypical features. The Walloon Coal Measures have an unusually high proportion o f dirt bands (30 of 66) and the range o f vitrite + clarite content is relatively restricted. The Illawarra Coal Measures (Lithgow District) data show a distinct preference for vitrite-poor lithologies to follow dirt bands. The main seam in this data set is the Lithgow Coal. The vitrite-poor lithology is the most a b u n d a n t
Table 7. Sequences for Major Stratigraphic Groupings, Dirt Bands Excluded Supergroups of coal measures Baralaba Coal Measures (Baralaba, Moura, Nipan, Theodore, and Blackwater Districts) Illawarra (incl. Lithgow District) and Newcastle (Moon Island Beach) Coal Measures Ipswich Coal Measures (Tivoli and Blackstone Formations) All coals combined
Sequence
Z2
p
2143
100
< 0.01
4441
126
< 0.01
2123 2143
74 491
< 0.01 < 0.01
i 1= c
_
_~_
..~
"6
7i
>_-
,.,,~ ~
~
~
R
I' ÷
7 o z
,~,,'7"---=
', Y:t ,.,.....,,,,~
t
m
!
2,
.i II
!
s ,, ~
\//
~
°i '
l'
i
o z
<~', ///
il
1\
,Is,
"':
Li '7'
'
Ii
ltl u~
~\/,/
8
Sequence in Australian Coal S e a m s
l'
539
= g,
o!
@
i/.° g
u~
i #~
,.o
o
z
i
",yl
i
m
m
/~,
/\
@
~o .o
=
~--
>
c: >
.!::
F.
.~
:_~
~,.~0 o z
_
,-.d ~0
'
;t/
~t
540
M. Smyth and A., C. Cook
o~
n •u
~
I
<"> '-'
i
I
~,7 .~
-d
~
-'d
t
I
u
b
u
~
°~
~<-> ~
~<->'
<",'
7''
~/T ~<-'
~,
e-,
ii';(
"I
;;
°~
>,,LI
°'i7
~
.~-
i t > , > , I ~ -~
7 7
L2
>,
-~
u~ -~
>,
~
~
\t
,T e,
-
1l,
1l,
l!,
~
'-'
1l,
~
"'
"
~
~
~ ~
'-'
'-'
~
~
~
u
qo!A~sdI
[~oD ~ q ~ l ~ a
~z~A~III
~
1l,
11,
11,
~
"'
"
"'
Sequence in Australian Coal Seams
"//////I~ "//YY//'~ 4
541
First preferences, Biaekwater gist. Noura Dist. Second preferences:Leigh Creek NIpan Dist.
Decreasing I
//////~~/.-;'.
Baralaba Dist.
5
~////~/~/5 Y~/.:/S~ Alternating
Second preferences:Callide
~,×1 •
7////2ii ~; ~>',/; //Z//D~ Y/////> I 4 1
Second preferences:Greta
irregular
~/////. ~/~//~'5'~///~////~S
Secondpreferences:TheodoreDist.
Basic
,5(:X5~5 ~/55;7,. Figure 4. Profiles from difference matrices (alter Smyth, 1972). coal lithology (16 of 36 coal plies) and the result may not be reliable because of the uneven and sparse nature of the matrix. Because dirt bands form an integral part of the sequence, the results that include dirt bands may be a more meaningful representation of the successions than those which exclude dirt bands. Nevertheless there are situations where the abundance of dirt bands dominates the matrix in such a manner as to mask possible relationships between coal plies. Therefore the results for the four-state data will be included in the discussion, even though we consider that the dirt bands form an integral part of the lithological sequence in Australian coal seams. Between profiles the results show a wide range in the form of the most probable sequences (Figs. 3 and 4). However, a number of features is common to many of the data sets for sequences generated by the most likely transitions and, where oscillations occur, the next most likely transitions (second preferences). Dirt bands exhibit a marked tendency (95% of the groups of coal seams) to be the most likely rock-type after a ply which has a low vitrite+ clarite content and also there is a marked tendency (82% of the groups of coal seams) for a vitrite +clarite-rich ply to overlie a dirt band. Further, within the coal to coal transitions, for sequences which show cyclicity, there is a predominance of the sequence profiles termed "de-
542
M . Smyth and A. C. Cook
creasing" and "basic" by Smyth (1970, 1972). In the "decreasing" profile the proportion of vitrite + clarite decreases upwards until the next vitrite + clarite ply is reached with a sharp change in lithology. This is an asymmetrical type of cycle. The "basic" profile is semisymmetrical in that there is a transition at the top of each cycle from vitrite + clarite-poor coal to vitrite + clarite-rich coal via an intermediate lithology. Few sequences show the "alternating" and "irregular" profiles of Smyth (1970, 1972). No sequences are devoid of evidence for cyclicity if the results for both the five-state and four-state data are considered. Most of those which show only partial cyclicity tend to show preferred transitions having the same sense as those in the "decreasing" profile with an upwards decrease in vitrite+clarite followed by a reversion to a vitrite + clarite-rich lithology. In general the sequence is not stable between the five-state and the fourstate models. However, examination of the results shows that, whereas the overall sequence is not stable with respect to the number of states used, the same sense of transitions is usually present in both sets of data. Major groupings of some of the coal measures were made on general stratigraphic and geographic affinities to yield the data in Tables 6 and 7. It was considered desirable to use cluster analysis to test for the presence of any statistical groupings of difference matrix characteristics. The cluster analysis was performed using a single-linked unweighted method following I
I
Illawarra Coal Measures Newcastle (MIB)
I
I
I
I
L
L
Greta Collie Leigh Creek Callide Theodore Blackwater Barala~ MOU ro Nipan Lithgow Newcastle (excl. MIB) Walloon Singleton Tivoli Blackstone
I 1.0
0.9
0.8
I
J
L
I
I
0.7
0"6
0.5
0-4
0.3
Similarity coefficient
Figure 5. Cluster analysis: five-state difference matrices.
Sequence in Australian Coal S e a m s I
l
Greta Newcastle (MI B) lllawarra CM Walloon
I
I
I
543 i
I
}--
Blackstone Collie Tivoli Theodore
I
Blackwater Singleton Leigh Creek
Moura Callide Baralabe Lithgow
]--
Nipan Newcastte (excl. MIB) 1.0
t
I
o.g
0.8
I. . . . 0.7
I
I
I
I
0-6
0.5
0-,4
0.3
Similarity coefficient
Figure 6. Cluster analysis: four-state difference matrices.
McCammon and Wenninger (1970). The availability of both five-state and four-state data provided a test of the stability of difference matrix characteristics in relation to the removal of the dirt bands, and of the stability of the clusters to the different types of data. The results of the cluster analysis are shown in Figures 5 and 6. The clustering level is shown in terms of the similarity of correlation coefficients for the difference matrices. The distances between the lines also are a measure of similarity in terms of between-group variation. Both five-state and four-state data give rise to two major clusters which are linked at the low level of approximately 0.2. Each cluster has a number of elements which are stable in relation to the deletion of dirt bands and cluster at high levels Of similarity. The Illawarra Coal Measures (ICM), Newcastle Coal Measures (Moon Island Beach Subgroup, MIB), and Greta Coal Measures coals are clustered at similarity levels in excess of 0.7 for both data types, whereas the Newcastle Coal Measures (excluding MIB), Nipan District (Baralaba Coal Measures), and Lithgow District (Illawarra Coal Measures) are present in a different cluster at similarity levels in excess of 0.6 for both data types. A number of the groups which change clusters according to the number of states recognized are those which are characterized by seams with large numbers of dirt bands, for example Walloon Coal Measures, C
544
M. Smyth and A. C. Cook
Blackstone Formation, and Tivoli Formation. Also of importance is the fact that different districts of the one set of coal measures do not necessarily cluster together. This suggests that it might be useful to attempt clustering at a lower level of geologically based grouping so that each seam could be tested and, desirably, different sections of one seam tested to explore lateral variation. Unfortunately many of the matrices are too sparse to allow this approach to be pursued with the presently available data. Johnson and Cook (1973) and Howell (1974) have determined differing characteristics for similar zones of coal-measure sequences in the northern and western parts of the Sydney Basin. It is possible that differences in the intraseam sequences may be related to those in the interseam sequences. DISCUSSION A number of factors point to the importance of dirt bands in relation to the sequence. To understand the significance of this relation it is necessary to examine the nature of the coal lithology successions which have been described (Fig. 4). The most persistent tendency is for upwards decrease of vitrite + clarite, and therefore of vitrinite, with a new sequence commencing in some but not all situations with a dirt band. Where they exist cycles may be asymmetrical or partly symmetrical. The sequence of lithologies is similar to that described by Smith (1968) to be associated with his microfloral succession. The major difference is that the Australian coals with a few exceptions have a lower exinite content than the European Westphalian coals he studied. This leads to a paucity of clarite in the Australian successions and the inertinite-rich (vitrite+clarite- and vitrinite-poor) lithologies are different from the crassidurains and to a lesser extent from the tenuidurains studied by Smith. There is considerable dispute over the origin of inertinite-rich lithologies. Some authors (e.g., Stach, 1956) consider that the finely macerated nature and the presence of disseminated epiclastic mineral matter indicate deposition under wet conditions, probably in standing water. By contrast, Smith (1968) on botanical and petrological grounds proposed an origin under relatively dry, oxidizing conditions with the high exinite content being a remanid effect. Although there are undoubted petrological differences between Australian coals and European Westphalian coals, we consider that most, if not all, of the Australian inertinite-rich coals originated under relatively dry, oxidizing conditions with the high levels of mineral matter being a remanideffect (Cook and Johnson, 1975) associated with considerable amounts of ablation of the peat. The low to very low exinite content of many Australian inertinite-rich coals could in part be due to lower rates of production of exinite by the floras, but also could be an early diagenetic feature associated
Sequence in Australian Coal Seams
545
perhaps with unusually high pH values in the peat. Thus we consider that the lithological sequences which are evident in the Australian coals are analogous to, and have a similar origin to, the successions proposed by Smith (1962, 1968). An autochthonous origin is assumed for all the coals, other than torbanites and the rare cannels which occur in some seams. This assumption is supported both by the overall stratigraphic relationships of the coals to the adjacent strata and by the microtextural features in the coals. The cause of the trend towards an upwards decrease in vitrite + clarite and vitrinite is not clear. It seems most likely to arise from an increasing tendency for oxidation to occur. This in turn could be explained by peat growth tending to outstrip subsidence, thus allowing the peat surface to rise above the water table. Individual cycles generally range between 0.1 m and 2 m. This would correspond to an original peat thickness of between 1 m and 20 m which could be sufficient to upset what was presumably a relatively delicate balance. The cause of the initiation of a new cycle of coal lithologies is even less clear than the cause of the cycle itself. The problems are much the same as those which beset researchers studying the major cyclothemic units in many coal-measure sequences. Much the same set of possible causes can be listed: eustatic changes in sea level, tectonism, and the natural progression of sedimentary processes in fluvial and deltaic regimes. Indeed it is probable that some of the coal cycles are directly analogous with full cyclothemic sequences. Such would be the situation where a decreasing profile overlies a dirt band which thickens laterally to split the seam, presenting, in the split area, a full sequence of sediments occurring between the two coal units. It seems clear from the results obtained that where dirt bands are present they form a significant part of the succession because they normally mark the initiation of vitrinite-rich coal formation. However, similar coal sequences can form without dirt bands present so that whereas the deposition of sediment will normally initiate a new cycle, the same sequence of coal lithologies can be initiated without the deposition of a dirt band. External causes such as eustatic changes in sea level would undoubtedly affect the sediment balance in a peat swamp, but it seems probable that the most important controls were within the depositional environment of the peat. Notwithstanding the extensive discussion over the possible volcanic origin of some claystones in coal seams, periodic floods are the most probable explanation for most dirt bands. Such an event may mark a change to more moist conditions, thus favoring the preservation of plant material as vitrinite. The inorganic sediment also may accelerate the compaction of the underlying peat, as well as providing a more nutritive substrate for plant growth. A return to vitrinite-rich coal, either gradually or rapidly, in the absence of a dirt band can be explained in terms of changes in water table or water supply.
546
M. Smyth and A. C. Cook
Such changes would be likely to occur concomitant with the frequent changes in the position of channels in deltaic and fluvial environments.
CONCLUSIONS The sequence of coal lithologies for a wide range of Australian coal seams shows a number of nonrandom characteristics. In particular, dirt bands are most likely to be followed by vitrite+clarite-rich coal. Overall there is a tendency for the coals to exhibit a degree of cyclicity with the percentage of vitrite+clarite decreasing upwards in each cycle. This tendency is more strongly displayed by some groups of seams than others. Attempts to cluster groups of seams on the basis of the characteristics of the difference matrix for lithology transitions suggest that there may be two rather different types of sequence with many groups of seams falling into intermediate categories. Further work to characterize these differences would probably necessitate resampling, preferably using continuous profiling methods instead of the fixed subsection method used for this study. There are notable similarities with the results obtained by Smith (1962, 1968) for Carboniferous coals, although there are marked differences in the petrology of the Australian coals as compared with those studied by Smith. It seems probable that a tendency towards cyclicity within coal seams is a normal condition. It also is probable that changes within the peat swamp itself are sufficient both to produce vertical changes in the maceral composition of a coal and to initiate new cyclical events.
ACKNOWLEDGMENTS The authors are grateful for permission to use data obtained by CSIRO Division of Mineralogy, and thank Dr. K. R. Johnson of the University of New South Wales for access to his cluster-analysis program.
REFERENCES
Anderson, T. W., and Goodman, L. A., 1957, Statistical inference about Markov chains: Annals Math. Stat., v. 28, p. 99-110. Cook, A. C., and Johnson, R. D., 1975, Seam thickness, quality and structure variation in the Bulli Coal, Sydney Basin, N.S.W. : Ann. Conf., Australas. Inst. Min. Metall., p. 537-550. Duff, P. McL. D., and Walton, E. K., 1962, Statistical basis for cyclothems. A quantitative study of the sedimentary succession in the East Pennine Coalfield: Sedimentology, v. 1, no. 4, p. 235-255. Gingerich, P. D., 1969, Markov analysis of cyclic alluvial sediments: Jour. Sed. Pet., v. 39, no. 1, p. 330-332.
Sequence in Australian Coal Seams
547
Harbaugh, J. W., and Bonham-Carter, G., 1970, Computer simulation in geology: John Wiley & Sons, Inc., New York, 575 p. Howell, D. J., 1974, Permian deltaic sedimentation, Western Coalfield, Sydney Basin: Abst. 9th Symp. on Advances in the Study of the Sydney Basin, p. 14-15. ICCP, 1963, International Handbook of Coal Petrography (2nd ed.): Centre National de la Recherche Scientifique, Paris, France. Johnson, K. R., and Cook, A. C., 1973, Cyclic characteristics of sediments in the Moon Island Beach Subgroup, Newcastle Coal Measures, New South Wales: Jour. Math. Geology, v. 5, no. 1, p. 91-110. Kemeny, J. C., and Snell, J. L., 1960, Finite Markov chains: Van Nostrand, Princeton, New Jersey, 210 p. Krumbein, W. C., 1967, F O R T R A N IV computer programs for Markov chain experiments in geology: Kansas Geol. Survey Computer Contr. 13, 38 p. Krumbein, W. C., 1968, FORTRAN IV computer program for simulation of transgression and regression with continuous-time Markov models : Kansas Geol. Survey Computer Contr. 26, 38 p. Krumbein, W. C., and Dacey, M. F., 1969, Markov chains and embedded Markov chains in geology: Jour. Math. Geology, v. 1, no. 1, p. 79-96. McCammon, R. B., and Wenninger, G. 1970, The dendrograph: Kansas Geol. Survey Computer Contr. 48, 28 p. Read, W. A., and Dean, J. M., 1967, 1968, A quantitative study of a sequence of coalbearing cycles in the Namurian of Central Scotland: Part 1, Sedimentology, v. 9, no. 2, p. 137-156; Part 2, Sedimentology, v. 10, no. 2, p. 121-136. Smith, A. H. V., 1962, The palaeoecology of Carboniferous peats based on the miospores and petrography of bituminous coals : Proc. Yorks. Geol. Soc., v. 33, p. 423-474. Smith, A. H. V., 1968, Seam profiles and seam characters, in Coal and coal-bearing strata: Oliver & Boyd, Edinburgh, 418 p. Smyth, M., 1970, Type seam sequences for some Permian Australian coals; Proc. Australas. Inst. Min. Metall., no. 233, p. 7-16. Smyth, M., 1972, A petrographic study of the stratigraphy of Australian coal seams: unpubl, masters thesis, Univ. New South Wales, 142 p. Stach, E., 1956, La sclerotinite et son importance pour l'origine de la durite: Proc. Int. Comm. Coal Pet., no. 2, p. 56-58. Wanless, H. R., and Weller, J. H., 1932, Correlation and extent of Pennsylvanian cyclothems: Geol. Soc. America, Bull., v. 43, no. 4, p. 1003-1016.