www.geologicacarpathica.sk
GEOLOGICA CARPATHICA, AUGUST 2009, 60, 4, 295—305 doi: 10.2478/v10096-009-0021-4
Introduction
Sedimentologists in their field work usually provide records
of sedimentary rocks. One of the routine ways to make
a sedimentary record is to plot the studied sedimentary rocks
as a vertical section of successive facies. Sediment facies
represent sediment or sedimentary rock that display distinc-
tive physical, chemical and/or biological characteristics
(Stow 2005). There are descriptive and genetic facies. While
descriptive facies are defined purely on sedimentary charac-
teristics, such as muddy sandstone, laminated mudstone,
etc., genetic facies are determined by the comparison with
standard facies models and interpreted in terms of deposi-
tional processes, such as turbidite, contourite, lahar deposit,
etc. The key to the interpretation of the sedimentary record is
to study facies in association, in particular their relative ver-
tical and lateral positions (Graham 1988). Comparison of de-
scriptive facies from the study area with standard facies
models for different depositional systems is a useful practice
(Stow 2005). However, the balance of data being interpreted
against models tempts us to look for some kind of arrange-
ment, even if there is no order (Carr 1982). Application of
statistical techniques provides an opportunity to the reduce
subjective impact on interpretations of the sedimentological
record. Some simple statistical methods have been applied to
detect the possible presence of vertical order in sedimentary
successions. Almost all use the probability matrices and em-
ploy the idea of Markov Chains. In this study, the Markov
property was applied to validate the presence of order in the
sequence of structures or descriptive facies in the turbiditic
formation from the Outer Western Carpathians. The main
goal of this study was to find statistically significant succes-
sions of descriptive facies in the investigated strata and com-
pare them to the facies models of deep-water deposits with
the purpose of supporting sedimentological interpretations
Markov Chain analysis of turbiditic facies and flow dynamics
(Magura Zone, Outer Western Carpathians, NW Slovakia)
SIDÓNIA STAŇOVÁ, JÁN SOTÁK and NORBERT HUDEC
Geological Institute of the Slovak Academy of Sciences, Severná 5, 974 01 Banská Bystrica, Slovak Republic;
stanova@savbb.sk; sotak@savbb.sk
(Manuscript received April 17, 2008; accepted in revised form December 18, 2008)
Abstract: Methods based on the Markov Chains can be easily applied in the evaluation of order in sedimentary se-
quences. In this contribution Markov Chain analysis was applied to analysis of turbiditic formation of the Outer West-
ern Carpathians in NW Slovakia, although it also has broader utilization in the interpretation of sedimentary sequences
from other depositional environments. Non-random facies transitions were determined in the investigated strata and
compared to the standard deep-water facies models to provide statistical evidence for the sedimentological interpreta-
tion of depositional processes. As a result, six genetic facies types, interpreted in terms of depositional processes, were
identified. They comprise deposits of density flows, turbidity flows, suspension fallout as well as units which resulted
from syn- or post-depositional deformation.
Key words: Outer Western Carpathians, facies, density flows, Markov Chains, turbidite.
of depositional processes. The Markov Chain analysis has
commonly been used in the past for the analysis of ordered
sequences from a wide range of depositional environments.
However, the calculations in the analysis of ordered succes-
sion of facies with the assistance of the Markov Chains are
time consuming. As a consequence, a simple computer pro-
gram entitled “phpSedistat” was devised for this study to
make the calculations easier.
General approach
A Markov process describes a sequence of states, or
events, for which the occurrence of one state may exhibit a
dependence on a previous state or states (Powers & Easter-
ling 1982). Development of methods based on the Markov
property which can be used in geology started at the end of
the 60’s. They have been used by a number of investigators
(e.g. Potter & Blakey 1968; Gingerinch 1969; Read 1969;
Doveton 1971; Miall 1973; Ethier 1975) to validate the pres-
ence of ordered and cyclic successions of facies. Stratigraph-
ic “order” may be indicated by repeated patterns of either
thickness and/or lithological variation, but it should also be
noted that the presence of lithological order alone makes no
compelling argument for periodic accumulation (Wilkinson
et al. 1997). On that account, the Markov property proved to
be of little use for characterizing sedimentary cyclicity
(Weedon 2005). However, it was considered useful, when
for example, the particular order of lithologies helps in the
description of sedimentological processes (Wilkinson et al.
1997). This is why the Markov property has been used in this
study, which is focused on the analysis of facies and flow
dynamics of turbiditic formation.
The presence of Markov process implies a degree of order
in a system (Carr 1982). In most studies, the raw data consist
296
STAŇOVÁ, SOTÁK and HUDEC
of an observed number of upward facies transitions which
are plotted in matrix form (transition matrix) and compared
to one generated by random methods (Graham 1988). The
initial works in this field proceeded from a mistaken mathe-
matical technique of generating an independent trial matrix
and resulted in inappropriate conlusions. The transition ma-
trix can be determined in two ways (Powers & Easterling
1982). One method is to make observations at regular inter-
vals of thickness. The advantage of this method is that the di-
agonal frequencies of the transitional matrix carry
information about thickness, but a disadvantage is that ob-
served transition rates depend on the sampling interval. An
alternative method is to construct a sequence of lithologies
by recording the transitions of lithologies as they occur,
withouth regard to thickness. Since the transition from a bed
of one lithology (facies) to another case of the same type is
assumed to be objectively non-recordable, the diagonal fre-
quencies are constrained zero. Transition matrices contain-
ing structural zeroes are called embedded matrices. Then
observed facies transitions are compared to wholly random
facies transitions. It was pointed out by Schwarzacher (1975)
that the presence of previously defined zeroes was a major
obstacle for accurate analysis (Graham 1988). Attempting to
overcome this problem, both Carr (1982) and Powers &
Easterling (1982) independently recommended a method (ef-
fectively the same one) for the estimation of wholly random
facies transitions, termed quasi-independence, by employing
a technique initially developed by Goodman (1968). The ex-
pected transition frequencies under a model of quasi-inde-
pendence are generated by an iterative method (Powers &
Easterling 1982). In the identification of significant transi-
tions, the observed and expected cells are used to derive a
difference matrix (Gingerich 1969; Carr 1982). The positive
values in the difference matrix were regarded as non-ran-
dom, while negative values were regarded as random, which
led to some questionable conclusions in initial works. Con-
sequently it has been suggested that all positive values in the
difference matrix have to be statistically tested for random-
ness (Harper 1984).
Geological settings
The research area extends in the NW, Slovak part of the
Outer Western Carpathians (Fig. 1), contouring the northerly
convex arcuate shape of the Western Carpathian orocline.
The Western Carpathians create the northernmost, W-E
trending orocline of the European Alpides, linked through
the Vienna Basin to the Eastern Alps in the West and to the
Eastern Carpathians in the East (Plašienka et al. 1997). The
present structural pattern of the Western Carpathians was
formed by the Late Jurassic-Tertiary subduction-collision
orogenic processes in the Tethyan mobile belt between the
stable North European Platform and drifting Apulia/Adria
related continental fragments. The Outer Western Car-
pathians contain sediments of the late Tertiary Carpathian
Foredeep deposited on the southern flanks of the North Eu-
ropean Platform and a broad Flysch Belt. Several thousands
of meters thick Upper Jurassic to Lower Miocene, mostly si-
liciclastic, flysch deposits built up several nappes, subhori-
zontally overthrusted onto Miocene deposits of the Car-
pathian Foredeep or directly onto Precambrian, Paleozoic or
Mesozoic rocks of the Carpathian foreland (Oszczypko et al.
2003). They are completely uprooted from their basement
and separated from the Central Carpathians by the Pieniny
Klippen Belt suture zone. Little is known about the nature of
the Flysch Belt basement substratum which was shortened
and underthrust below the Central Western Carpathians dur-
ing the Tertiary (Plašienka et al. 1997).
The Flysch Belt represents the Tertiary accretionary wedge
of the Carpathian orogen. The flysch nappes are generally ar-
ranged into the Silesian—Moldavian (or “Krosno-Menilite”)
group of nappes in the North, and the Magura thrust system in
the South (Plašienka et al. 1997). In the research area, there
are several large thrust units of the Magura Group, consisting
of the Rača, Bystrica and Oravská Magura (Krynica) Units.
All the studied sedimentary sections expose the Kýčera
Member of Rača Unit, deposited in the Magura Basin during
the Late Eocene. On the basis of the field sedimentary re-
search, the Kýčera Member represents a sandstone-dominat-
ed sequence alternating with mudstone rich intervals. The
sandstones mostly form thin to thick beds, although some
very thick beds are present as well. Most sandstone beds
have parallel or slightly wavy bases. Amalgamation of beds
is common. It is sometimes indicated by clustered amalgam-
ation clasts (Fig. 2D). Generally, mud clasts float in the up-
per part of a bed (Fig. 2H,J). Medium to very thick beds of
structureless sandstones and pebbly sandstones with normal
gradation of pebbles are present. A lot of them have T
b—d
Bouma divisions in the upper part of beds (Fig. 2A,B) or
most frequently they have only a thin laminated mudstone
top directly above T
a
Bouma interval. These thin laminated
tops are often bioturbated (Fig. 2C). The upper parts of sand-
stone beds are frequently eroded by the next current or re-
worked by bottom currents. Mudstone rich intervals contain
thick claystones/silty claystones or thick beds of claystone/
silty claystone with mm to cm silt/very fine sandstone lami-
nae (Fig. 2K). The Kýčera Member represents either channel
systems (or upper parts of lobes) alternating with interchan-
nel areas (Starek & Pivko 2001) or depositional lobes alter-
nating with interlobe deposits (Staňová & Soták 2007). The
Kýčera Member belongs to the Zlín Formation, which termi-
nated the deposition in the Rača Unit. The Kýčera Member is
considered equivalent to the muscovite sandstone facies of
the “Magura Sandstone” of the Polish Western Carpathians
(Ksiazkiewicz 1953; Pivko 2002).
Methodology
Detailed field sedimentary research was carried out in
a series of quarries and road-cut exposures in the NW Slovak
part of the Magura Group of the Outer Western Carpathians
(Figs. 1, 2). In the field research extended attention was giv-
en to detailed study of sedimentary structures. Visual com-
parators were used in the field to determine grain sizes. The
starting point for data analysis and interpretation is the verti-
cal section of sedimentary sequence. The main members of
297
MARKOV CHAIN ANALYSIS OF TURBIDITIC FACIES AND FLOW DYNAMICS (SLOVAKIA)
sedimentary sequences – descriptive facies were defined on
the basis of sedimentary characteristics recorded in the field
and marked with facies codes.
The presence of order in the studied successions of facies
was validated using the statistical methods based on the
Markov property. The procedures for the calculation of sig-
nificant facies transitions were incorporated into a self-de-
signed computer program, entitled “phpSedistat”, allowing
automation of the calculations (see Appendix for details). As
the calculations are practically provided using a computer
program, attention is given to the input data for calculation
and to the form and interpretation of the output data. The in-
put data for the calculation is a vertical sequence of facies,
expressed as a column of facies codes representing descrip-
tive facies and saved in a .txt file. Interruptions or absence of
sedimentary record is marked with a dividing symbol (/).
The output data are: (1) a list of facies codes and short de-
scription of associated facies, (2) a matrix of observed facies
transitions and (3) a matrix of significant facies transitions.
The matrix of observed facies transitions contain the number
Fig. 1. A – Geological map of investigated area showing the locations of the outcrops of the Kýčera Member. 1—5 – Magura Group: 1 – Vsetín
Member, 2 – Kýčera Member, Babia Góra Sandstone, 3 – Bystrica Member, 4 – Beloveža Formation (“Variegated beds”), 5 – Soláň Forma-
tion; 6—8 – Silesia Group: 6 – Istebná Member, 7 – Krosno Formation, 8 – Hieroglyph Formation; 9 – Pieniny Klippen Belt; 10 – geo-
logical boundaries: determined, expected; 11 – fractures: determined, expected; 12 – lines of the nappes and thrusts: determined, expected;
13 – localities; 14 – state boundary (modified from Biely et al. 1996). B – Location of the research area on the tectonic scheme of the Outer
Western Carpathians (modified from Żytko K. et al. in Poprawa & Nemčok 1989).
298
STAŇOVÁ, SOTÁK and HUDEC
Fig. 2. Photographs showing some general views and structures of investigated outcrops of the Kýčera Member. See the outcrop locations
on Fig. 1A. A – Detailed view from Krásno nad Kysucou outcrop showing the turbidite deposit, described by the Bouma model, consisting
of massive sandstone, T
a
at the base passing to the lower parallel lamination, T
b
, wave lamination, T
c
and upper parallel lamination, T
d
at
the top. B – Detailed view from the outcrop in Lysá pod Makytou quarry showing massive sandstone followed by laminated sandstone
(T
ab
Bouma sequence). C – Field photograph of Taphrhelminthopsis traces in laminated siltstones on the top of bed from the outcrop be-
tween Klubina and Zborov nad Bystricou villages. D – Detailed view from the Krásno nad Kysucou outcrop showing clustered amalgam-
ation clasts. E – Detailed view from the quarry between Klubina and Zborov nad Bystricou villages showing post-sedimentary
deformation of claystone/silty claystone with mm to cm silt/very fine sandstone laminae. F – Detailed view from the Krásno nad Kysucou
outcrop showing submarine slump. G – General view of the quarry between Klubina and Zborov nad Bystricou villages. H – Detailed
view from the outcrop in Lysá pod Makytou quarry showing imprints of shale clasts at the top of a sandstone bed. I – General view of a
part of the quarry near Ve ké Rovné village. J – Detailed view from the quarry between Klubina and Zborov nad Bystricou villages show-
ing imprint of shale clasts at the top of sandstone beds. K – Detailed view from the quarry between Klubina and Zborov nad Bystricou vil-
lages showing claystone/silty claystone with mm to cm silt/very fine sandstone laminae (Facies A7).
299
MARKOV CHAIN ANALYSIS OF TURBIDITIC FACIES AND FLOW DYNAMICS (SLOVAKIA)
Fig. 3. Sedimentary logs of representative parts of studied sedimentary sequences of the Kýčera Member from: A – the quarry near Lysá pod
Makytou village. B – the quarry near Ve ké Rovné village. C – the quarry between Klubina and Zborov nad Bystricou villages. D – the
road-cut near Krásno nad Kysucou village. See the outcrop locations on Fig. 1A.
of observed transitions of facies. The matrix of significant
facies transitions summarizes the possibilities of non-ran-
dom facies transitions, calculated at a significance level of
0.5 and expressed by the number of percent.
The vertical arrangement of different facies can be pre-
sented by a variety of facies relationship diagrams. Their aim
is to present either the raw data or statistics based on those
data in a format which aids interpretation (Graham 1988).
Non-random facies transitions were presented by the scheme
of significant facies transitions. This scheme was compared
to the facies successions in deep-water facies models allow-
ing the interpretation of depositional processes.
Results
Significant facies transitions
There were eight descriptive lithofacies A1, A2, ..., A8
(Table 1), distinguished in the studied sedimentary succes-
300
STAŇOVÁ, SOTÁK and HUDEC
sions (Fig. 3) during the field research, tested for ordered ar-
rangement using the computer calculation of Markov Chains.
The facies transitions A1—A3, A1—A5, A1—A7, A1—A8,
A2—A3, A2—A5, A3—A4, A4—A5, A5—A1, A5—A2, A5—A6,
A5—A7, A6—A1, A6—A2, A6—A8, A7—A1, A7—A2 reached the
probability higher than 95 %, that is they are non-random with
the probability higher than 95 % (Table 3). These transitons
were considered regular at significance level of
α=0.5. All of
these significant facies transitions are illustrated in the scheme
of significant facies transitions (Fig. 4).
Discussion
Analysis of facies and flow dynamics
In the interpretation of significant facies transitions it is
important to keep in mind that the calculated significant
transitions represent the most probable facies transitions, but
not their frequency in the studied sedimentary sequences.
The real frequencies of facies transitions are written down in
the matrix of observed facies transitions (Table 2). There-
Table 1: Lithofacies.
Facies Mean
Code Thickness
(cm)
Description
A1
77
structureless (massive) sandstone
A2
118
pebbly sandstone with normally graded pebbles
A3
21
parallel laminated sandstone
A4
9
rippled, wavy laminated very fine sandstone to siltstone
A5
7
laminated, very fine sandstone to siltstone
A6
47
mudstone
A7
78
claystone/silty claystone with mm to cm silt/very fine sandstone laminae
A8
262
chaotic, deformed heterolithic units
Table 2: Matrix of observed facies transitions.
Facies A1 A2 A3 A4 A5 A6 A7 A8
Sum
of
row
A1
–
19
65
0
160
60
12
2
318
A2
18
–
13
0
33
14
0
0
78
A3
5
0
–
74
1
1
0
0
81
A4
5
3
0
–
65
12
0
0
85
A5
144
24
0
5
–
97
6
0
276
A6
126
27
2
6
25
–
3
1
190
A7
14
4
1
0
0
3
–
0
22
A8
1
0
0
0
1
1
0
–
3
Sum of column
313 77 81 85 285
188 21 3
Fig. 4. Scheme of significant facies transitions. Facies codes are described in Table 1.
301
MARKOV CHAIN ANALYSIS OF TURBIDITIC FACIES AND FLOW DYNAMICS (SLOVAKIA)
fore, when interpreting sedimentary successions it is useful
to consider both statistically significant and real facies tran-
sitions in order to better understand their significance and
real occurrence in the studied sedimentary record.
The calculated significant successions of facies A1—A3—
A4—A5—A6 and A2—A3—A4—A5—A6 (Fig. 4) remarkably re-
semble a Bouma sequence (Bouma 1962). This succession of
lithofacies is considered a typical sequence deposited by tur-
bidity currents. It comprises a massive or graded sandstone in-
terval T
a
(facies A1 or A2) at the base, followed upwards by
laminated sandstone interval T
b
(facies A3), then passing to a
rippled/wavy interval T
c
(equivalent to A4 facies), upper lam-
inated siltstone interval T
d
(facies A5) and terminating with
mud interval T
e
(facies A6). The real occurrence of this suc-
cession of facies in the sedimentary record (Fig. 2A,B) was
considered with reference to the matrix of observed facies
transitions (Table 2). There are 396 A1/A2—AX facies transi-
tions in the matrix of observed facies transitions, but only 78
of them (= 19.70 %) are A1/A2—A3 transitions resembling T
a—b
intervals of the Bouma sequence (Fig. 2B). There are also 74
A3—A4 facies transitions (equivalent to T
b—c
Bouma intervals),
65 A4—A5 facies transitions (equivalent to T
c—d
Bouma inter-
vals) and 97 A5—A6 facies transitions (equivalent to T
d—e
Bou-
ma intervals) in the matrix of observed facies transitions
(Table 2). This means that only 19.70 % of the sandy intervals
beginning with the A1 or A2 facies constitute either a com-
plete or a basal part of the classical Bouma sequence. In addi-
tion to A1/A2—A3 facies transitions, the A1 or A2 facies can
also be followed by the A5, A6, A7, A8 facies as well as by
the A1 or A2 facies. The A1 or A2 facies are most frequently
followed by facies A5. There are 193 transitions of A1/A2—A5
facies in the matrix of observed facies transitions (Table 2). It
is 48.74 % of the whole 396 transitions of A1/A2—AX facies
in the matrix of observed facies transitions. While the previ-
ously discussed transitions of facies A1/A2—A3 can be com-
pared to the T
ab
Bouma divisions, the A1/A2—A5 transitions
resemble rather the T
ad
Bouma divisions. According to the
field record, the units of facies A5 are usually much thinner
than those of the A1 or A2 facies. This may suggests that apart
from the deposits of turbidity currents (Bouma sequence), per-
haps the most significant part of the massive or graded sand-
stone beds with thin, fine laminated tops in the investigated
sections are products of probably the most disscussed deep-
water sediment gravity flows, frequently interpreted as high-
density turbidity currents (Lowe 1982), concentrated density
flows (comp. Mulder & Alexander 2001) or sandy debris
flows (Shanmugam 1997), while their upper laminated tops
were deposited from dilute current. These tops contain pieces
of mica and fossil plants and they are frequently bioturbated
(Fig. 2C). They could be reworked by bottom currents. The
A1/A2—A6 facies transitions have a similar explanation. At
the top of the A1 or A2 facies there are sometimes shale clasts.
They apparently represent rip-down clasts (Fig. 2H,J) that in-
dicate post-depositional liquefaction (Stow & Johansson
2000). Sometimes, amalgamated intervals in the investigated
sedimentary section are indicated by the presence of clustered
amalgamation clasts (Fig. 2D) formed by erosional break-up
of a thin intervening shale layer between successive sand de-
posits (see Stow & Johansson 2000).
Facies A7 most probably follows the A1 or A5 facies
(Fig. 2K). Facies A7 is considered to represent repetitive
A5—A6 facies transitions and it is interpreted as representing
deposition from low-concentration turbidity currents. Ac-
cording to the matrix of observed facies transitions (Ta-
ble 2), the facies A6 follows all the defined facies. It is most
likely to follow A2, A5 or A8 facies (Table 3). Thick inter-
vals of A6 facies were deposited from suspension fallout.
Facies A8 represents chaotic, contorted heterolithic units,
which commonly overlie A1 or A6 facies (Table 3, Fig. 4).
The A8 facies is apparently formed by submarine slumps
(Fig. 2F). In some cases, they could be the result of post-sed-
imentary deformation (Fig. 2E).
Genetic facies definition
The examination of both the calculated significant facies
transitions and the original sedimentary records allowed us
to define six genetic facies types, which were compared to
the standard deep-water facies models and schemes of deep-
water gravity-driven deposits. The genetic facies, when ar-
ranged laterally according to the declining grain size,
provide the evidence of deposition from increasingly diluted
and more mature flows (Fig. 5). Such facies arrangement re-
flects depositional settings from proximal to distal. The most
proximal and usually the coarsest facies are chaotic and de-
formed units (Facies no. 1 in Fig. 5), which apparently de-
veloped on the slope of depositional lobes. However, some
of these units may developed as post-sedimentary deforma-
tions. The next facies represent thick- to medium-bedded
sandstones with normally graded granules and pebbles (Fa-
cies no. 2 in Fig. 5) and thick- to medium-bedded structure-
less sandstones (Facies no. 3 in Fig. 5). These facies reflect
the deposits of density flows (Alexander & Mulder 2001) or
dense flows (sensu Mutti et al. 2003) or high-concentration
turbidity currents (Lowe 1982; Pickering et al. 1986). Both
these facies types may have thin laminated tops, which de-
veloped as the flows were progressively diluted by the en-
trainment of surrounding water. Normal gradation developed
if the dilution of water was sufficient to allow particle fall
out within the flow. The next facies (Facies no. 4 in Fig. 5)
are characterized by the Bouma sequence of structures, rep-
resenting deposits from turbidity currents. The next facies
(Facies no. 5 in Fig. 5) are claystone/silty claystone with
mm to cm silt to very fine sandstone laminae. These facies
are compared to the deposits of low-concentration turbidity
currents. The last and the most distal facies are claystones or
Table 3: Matrix of significant facies transitions.
Facies
A1 A2 A3 A4 A5 A6 A7 A8
A1
– – 99.99 – 99.99 – 99.99
99.99
A2
– – 99.82 – 98.38
52.02
– –
A3
– – –
99.99
– – – –
A4
– – – –
99.99
– – –
A5
99.99
99.99
– – –
99.99
99.99
–
A6
99.99
99.99
– – – – –
99.99
A7
99.36
96.02
– – – – – –
A8
– – – –
36.46
58.74
– –
302
STAŇOVÁ, SOTÁK and HUDEC
Fig. 5. List of genetic facies with their characterization compared to some deep-water facies models.
303
MARKOV CHAIN ANALYSIS OF TURBIDITIC FACIES AND FLOW DYNAMICS (SLOVAKIA)
Fig. 5. Continued.
304
STAŇOVÁ, SOTÁK and HUDEC
silty claystones (Facies no. 6 in Fig. 5), which are compared
to the deposits of low-concentration turbidity currents or sus-
pension fallout.
Conclusions
The application of Markov Chain analysis allowed us to
identify ordered succession of descriptive facies or structures
in a turbiditic formation from the Magura Zone of the Outer
Western Carpathians. Significant facies transitions were de-
termined providing the evidence that classical turbidites con-
stitute the important part of the studied sedimentary
sequences of the Kýčera Member. In addition to turbidite
currents, the Kýčera Member facies were also deposited by
density flows, suspension fallout or they were reworked by
bottom currents or resulted from syn- or post-depositional
deformation.
In this paper, the Markov Chain analysis has been applied to
calculate the arrangement of facies in turbiditic formation, al-
though it has been utilized in the past for the analysis of or-
dered sequences of facies in other environments (e.g.
meandering river, shallow-marine, beach-barrier etc.). The
calculation aspects of Markov analysis are reviewed in
a concise and instructive form, therefore this study can serve
as a case study for the application of the Markov Chains in
the interpretations of sedimentary sequences in the field or
from boreholes. A subsequent contribution of this study was
the creation of a self-designed computer program, named
phpSedistat. Computer-based calculation of regular facies
transitions noticeably simplified and hastened the utilization
of Markov Chains.
Acknowledgments: This study was supported by the Slovak
Research and Development Agency (APVV 51-011305
Project) and by VEGA Grant GA 6093. Official reviews by
Prof. Dr. Nestor Oszczypko, Dr. Ivan Baráth and Prof. Dr.
Szczepan Porębski provided insightful comments that helped
improve the original manuscript.
References
Biely A. (Ed.), Bezák V., Elečko M., Gross P., Kaličiak M.,
Konečný V., Lexa J., Mello J., Nemčok J., Potfaj M., Rakús
M., Vass D., Vozár J. & Vozárová A. 1996: Geological map of
Slovakia, 1 : 500,000. Ministry of Environment SR & State Geo-
logical Institute of Dionýz Štúr, Bratislava.
Bombita G.
& Pop G. 1991: Mesozoic formations from Poiana Bot-
izii, Pieniny Klippen Belt of Romania. Geol. Carpathica 42, 3,
139—146.
Bouma A.H. 1962: Sedimentology of some flysch deposits. Elsevi-
er, Amsterdam, 1—168.
Carr T.R. 1982: Log-linear models, Markov chains and cyclic sedi-
mentation. J. Sed. Petrology 52, 905—912.
Dasgupta P. 2003: Sediment gravity flow – the conceptual prob-
lems. Earth Sci. Rev. 62, 265—281.
Doveton J.H. 1971: An application of Markov Chain analysis to the
Ayrshire Coal Measures succession. Scott. J. Geol. 7, 11—27.
Eliáš M., Schnabel W. & Stráník Z. 1990: Comparison of the Flysch
Zone of the Eastern Alps and the Western Carpathians based
on recent observations. In: Minaříková D. & Lobitzer H.
(Eds.): Thirty years of geological cooperation between Austria
and Czechoslovakia. Geol. Surv., Praha, 37—46.
Ethier V.G. 1975: Application of Markov analysis to the Banff For-
mation (Mississippian), Alberta. Math. Geol. 7, 47—61.
Gingerich P.D. 1969: Markov analysis of cyclic alluvial sediments.
J. Sed. Petrology 39, 330—332.
Goodman L.A. 1968: The analysis of cross-classified data: indepen-
dence, quasi-independence and interactions in contingency ta-
bles with or without missing entries. J. Amer. Statist. Assoc.
63, 1091—1131.
Graham J. 1988: Collection and analysis of field data. In: Tucker
M. (Ed.): Techniques in sedimentology. Blackwell Sci. Publ.,
1—394.
Harper C.W. Jr. 1984: Improved methods of facies sequence analy-
sis. In: Walker R.G. (Ed.): Facies models. Geosci. Can Reprint
1, 11—13.
Lowe D.R. 1982: Sediment gravity flows: II. Depositional models
with special reference to the deposits of high-density turbidity
currents. J. Sed. Petrology 52, 279—297.
Miall A.D. 1973: Markov Chain analysis applied to an ancient allu-
vial plain succession. Sedimentology 20, 347—364.
Mulder T. & Alexander J. 2001: The physical character of subaque-
ous sedimentary density flows and their deposits. Sedimentolo-
gy 48, 269—299.
Mutti E. 1992: Turbidite sandstones. Inst. Geol. Univ. Parma, 1—275.
Mutti E. & Ricci Lucchi F. (1972): Le torbiditti dell’Appenino set-
trentrinale, introduzione all´analisi di facies. Soc. Geol. Ital.,
Mem. 11, 161—199.
Mutti E., Tinterri R., Benevelli G., di Biase D. & Cavanna G. 2003:
Deltaic, mixed and turbiditic sedimentation of ancient foreland
basins. Mar. Petroleum Geol. 20, 733—755.
Oszczypko N. 1992: Late Cretaceous through Paleogene evolution
of Magura basin. Geol. Carpathica 43, 6, 333—338.
Oszczypko N., Golonka J., Malata T., Poprawa P., Słomka T. & Uch-
man A. 2003: Tectono-stratigraphic evolution of the Outer Car-
pathian basins (Western Carpathians, Poland). XVIIth Congress
of the Carpathian-Balkan Geological Association, Bratislava,
Post-Congress Proceedings. Miner. Slovaca 35, 17—20.
Pickering K., Stow D., Watson M. & Hiscott R. 1986: Deep-water
facies, processes and models: a review and classification
scheme for modern and ancient sediments. Earth Sci. Rev. 23,
75—174.
Plašienka D., Grecula P., Putiš M., Kováč M. & Hovorka D. 1997:
Evolution and structure of the Western Carpathians: an over-
view. In: Grecula P., Hovorka D. & Putiš M. (Eds.): Geologi-
cal evolution of the Western Carpathians. Miner. Slovaca –
Monogr., Bratislava, 1—24.
Potter P.E. & Blakey R.F. 1968: Random processes and lithologic
transitions. J. Geol. 76, 154—170.
Powers D.W. & Easterling R.G. 1982: Improved methodology for
using embedded Markov chains to describe cyclical sediments.
J. Sed. Petrology 52, 913—923.
Read W.A. 1969: Analysis and simulation of Namurian rocks of
Central Scotland using a Markov-process model. Math. Geol.
1, 199—219.
Schwarzacher W. 1975: Sedimentation models and quantitative
stratigraphy. Elsevier, Amsterdam, 1—382.
Shanmugam G. 1997: The Bouma sequence and the turbidite mind
set. Earth Sci. Rev. 42, 201—229.
Shanmugam G. 2002: Ten turbidite myths. Earth Sci. Rev. 58, 311—341.
Stow D.A.V. 2005: Sedimentary rocks in the field. A colour quide.
Manson Publ., London, 1—320.
Stow D.A.V. & Johansson M. 2000: Deep-water massive sands: na-
ture, origin and hydrocarbon implications. Mar. Petrol. Geol.
17, 145—174.
305
MARKOV CHAIN ANALYSIS OF TURBIDITIC FACIES AND FLOW DYNAMICS (SLOVAKIA)
Step-by-step list of procedures incorporated into the
phpSedistat program:
1. The matrix of observed facies transitions is created.
As the quasi-independence principle was applied (see e.g.
Carr 1982; Powers & Easterling 1982 for details), the facies
transitions between the same facies were excluded. They are
usually hardly identified in the field.
2. The matrix of random transitions is calculated.
Row/column sums of the matrix of observed facies transi-
tions and parameters a
i
, b
j
are utilized to calculate the matrix
of random transitions. Parameters a
i
, b
j
are estimated using
an iterative solution (see Powers & Easterling 1982).
The first iteration of a
i
and b
j
is calculated as follows:
, i = 1, 2, ..., m
(1)
, j = 1, 2, ..., m
(2)
where m is a number of facies members, n
i+
is an i
th
row
sum and n
+j
is a sum of j
th
column.
Then the i
th
iteration is calculated as follows (3, 4):
, i = 1, 2, ..., m
(3)
, j = 1, 2, ..., m
(4)
The iteration is continued until (5, 6):
|a
i
(I)
— a
j
(I—1)
| < 0.01 a
i
(I)
, for i = 1, ..., m (5)
and
|b
j
(I)
—b
j
(I—1)
| < 0.01 b
j
(I)
, for j = 1, ..., m (6)
Let a’ and b’ be the last iterations of a
i
(I)
and b
j
(I)
, then the
estimated random transitional frequencies are calculated as
follows (7):
E
ij
= a
i
’
· b
i
’
i = 1, 2, ..., m; j = 1, 2, ..., m; i
≠j
(7)
3. The matrix of observed possibilities of facies transitions
and the matrix of random possibilities of facies transitions
are calculated.
The matrix of observed transition frequencies and the ma-
trix of random transition frequencies are recalculated through
their row sums, e.g. the first row is recalculated as follows:
a
12
/n
1+
, a
13
/n
1+
, ... a
1m
/n
1+
(8)
The result of these recalculations is the creation of the ma-
trix of observed possibilities of facies transitions and the ma-
trix of random possibilities of facies transitions.
4. Calculation of a difference matrix.
The difference matrix is calculated as the matrix of ran-
dom possibilities of facies transitions is substracted from the
matrix of observed possibilities of facies transitions.
5. Positive values in a difference matrix are tested.
All positive values in the difference matrix are tested for
randomness at a selected significance level. Testing proce-
dures within the phpSedistat computer program are provided
at significance level 0.5. It means that all computed values
greater than 0.05 are non-random with the possibility less
than 5 %, i.e. random with the possibility greater than 95 %
and all computed values less than 0.05 are non-random with
the possibility greater than 5 %. The testing criterion is as
follows (see Harper 1984):
(9)
where N = relevant row sum in the matrix of observed tran-
sition frequencies, n
cont
= observed number of a specific tran-
sition, p = the possibility of a specific transition in the matrix
of random possibilities of transition frequencies and q = 1—p.
6. The matrix of regular facies transitions is created.
The values computed in the testing procedure are multiplied
by 100 and the matrix of regular facies transitions is created,
where the regular facies transitions are calculated at the signif-
icance level 0.5 and expressed in the number of percent. The
formulation of the resulting possibilities of facies transition in
the number of percent simplifies their evaluation.
The program phpSediStat was programmed under a GNU
lincence as an open source software product. It can be down-
loaded from http://phpsedistat.sourceforge.net.
Appendix
Weedon G.P. 2005: Time-series analysis and cyclostratigraphy: Ex-
amining stratigraphic records of environmental cycles. Cam-
bridge University Press, 1—273.
Wilkinson B.H., Drummond C.N., Rothman E.D. & Diedrich N.W.
1997: Stratal order in peritidal carbonate sequences. J. Sed.
Res. 67, 1068—1082.
Żytko K., Zając R., Gucik S., Ryłko W., Oszczypko N., Garlicka I.,
Nemčok J., Eliáš M., Menčík E. & Stráník Z. 1989: Map of the
tectonic elements of the Western Outer Carpathians and their
Foreland. In: Poprawa D. & Nemčok J. (Eds.): Geological atlas
of the Western Outer Carpathians and their foreland. PIG
Warszawa/GÚDŠ Bratislava/ÚÚG Praha.