Você está na página 1de 6

Nonlinear Phenomena in Complex Systems, vol. 11, no. 2 (2008), pp.

274 - 279

Spatiotemporal Chaos into the Hellenic Seismogenesis:


Evidence for a Global Seismic Strange Attractor
A. C. Iliopoulos,1 G. P. Pavlos,1 and M. A. Athanasiu1
1

Democritus University of Thrace, Department of Electrical and Computer Engineering, 67100, Xanthi, GREECE
(Received 3 January, 2008)
Self-organized criticality (SOC) is one of the most discussed concept for understanding the
earthquake process and other scale invariant physical processes but in contrast with chaos. In
this study, the concept of seismic low dimensional spatiotemporal chaos is supported. For this
reason, we used significant results obtained by the nonlinear analysis of three earthquake data sets
a)Latitude, b)Longitude and c)inter-event time intervals of earthquake events obtained in the broad
Hellenic region. Our results clearly reveal some kind of dynamical synchronization of the local seismic
active regions and the existence of a global seismic strange attractor. Moreover, a unified model of
earthquakes is given by using concepts of the modern physical theory of the far from equilibrium
statistical processes.
PACS numbers: 78.67.Pt+42.65.Ky
Keywords: earthquakes, spatiotemporal chaos, SOC, nonlinear time series analysis

1.

Introduction

phenomena in such a way that although seismicity


reveals an intrinsic randomness, earthquakes do not
occur randomly in space and time. Seismologists
have discovered that earthquakes can be related both
in near-field and at distances approaching 4000 km
[3,4]. The fractal character and the scaling laws of
the spatiotemporal seismogenesis, including global
regularity and local randomness as well as longrange correlation and space-time clustering, constitute
strong evidence for the existence of an underlying
spatiotemporal nonlinear deterministic dynamical
process. This process is unstable, complex and
sensitive to initial conditions [4,5,6]. Such a nonlinear
deterministic earthquake process, unstable and
stochastic, underlying the seismogenesis, produces the
observed space-time behavior, which is mapped into
the self-similar (fractal) properties and power-laws
profiles, the space-time clustering of seismic events
and long-range correlations. These characteristics
make earthquakes a whole [7-11]. This point of
view can be supported a) by the study of waiting
time distributions of seismic events which reveals a
unified scaling law of time process in four dimensions
(spatial coordinates and magnitude) [2,12-15),b) by
the concept of the seismic cycle supported by
Fedotov [16], c) the "self-similar"character of the
whole series of the seismic events as an entire
mainshock/ aftershock event can be nested inside a
larger aftershock sequence. That is, aftershocks can
produce their own aftershock series with the same
properties. This is a typical self-similar characteristic
of the nonlinear dynamical systems [17-20]. Until
now, several conceptual models have been proposed
to describe and explain the seismogenesis and the
properties of the observed seismicity. These models
are a) deterministic slider-block models [21-23], b)
cellular automata models [24-27] and c) statisticalprobabilistic models [28-32]. Recently, the concept of
low dimensional chaos in the earthquake process has
also been supported in many studies both theoretical
and experimental. For a review see Pavlos [33]. The
concept of low dimensional chaos into the Hellenic

According to Richter [1] and his contemporaries,


earthquake occurrence seemed to be random and
best modeled by a completely stochastic stationary
poisson process with no memory and no internal
dynamics. That is, no regular connection was apparent
between one large earthquake and another one. In
this direction, the most extended view is that of
two separated processes, one for main-shock, which
ought to follow a poisson distribution and one for
an independent process to generate aftershocks. In
contrast with this reductionistic point of view, the
alternative perspective is to look at the seismogenesis
as a whole dynamic process and place all the seismic
events on the same dynamical level [2]. Swarms,
mainshocks and aftershocks are closely related

FIG. 1. Earthquake Epicenters in Greece during the


twentieth Century for all focal depths (According to
Burton et al [45]).

E-mail:

ailiopou@gmail.com

274

Spatiotemporal Chaos into the Hellenic Seismogenesis . . .


44

(a)

800
600
500
400
300

40

36

200
100
0

32
0

1000

2000

3000

4000

5000

6000

Events
Autocorrelation Coefficient r(k)

(C)

28
24
20
16
0

1000

2000

3000

Events

1000

2000

3000

4000

5000

6000

Events

32

Longitude

(b)

700

Latitude

Interevent Times (hr)

900

4000

5000

6000

0.3
In te r e v e n t T im e S e r ie s
L a t it u d e T i m e S e r i e s
L o n g itu d e T im e S e r ie s

0.2

(d)

0.1

0
0

8 10 12
Lag Time (k)

14

16

18

20

FIG. 2. (a) Data set of time differences between


sequentially earthquake events with magnitude over 4
Richter, occurred in Greece, for the period of 1968-2004.
(b)Data set of Latitude of each earthquake event with
magnitude over 4 Richter, occurred in Greece, for the
period of 1968-2004. (c)Data set of Longitude of each
earthquake event with magnitude over 4 Richter, occurred
in Greece,for the period of 1968-2004. (d)Autocorrelation
Coefficient estimated for the three seismic Data sets.

seismogenesis has been supported, for the first time,


by Pavlos et al [34]. In this study, we extend our
previous results and we support the concept of
spatiotemporal chaos for the seismic process and the
existence of a global earthquake strange attractor, as
the basic physical process for the seismogenesis of
the Hellenic broad region. Therefore, we use methods
of nonlinear time series analysis, which are applied
simultaneously at three seismic data sets, firstly
for the inter-event time intervals and secondly for
the corresponding Latitude and Longitude sequences,
obtained at the Hellenic region. In section 2, we
introduce some theoretical concepts, which concern
the physical system underlying the seismogenesis. In
section 3, we present the results of the nonlinear
analysis applied at the seismic time-series and their
surrogate data. Finally, in section 4 we introduce the
concept of the chaotic active walker in relation with
the strange attractor process.

2.

The seismic physical system

Earthquakes can be viewed as a relaxation


process of the lithospheric crust loaded by stresses.
However, the occurrence of a particular earthquake
cannot be completely isolated from the dynamics
of the whole lithosphere. Itv may range from
global features, as the motion of tectonic plate,
mantle convection, earth rotation, to highly local
phenomena as frictional stress and transmission
of stress [35]. There is considerable evidence that
faults and earthquakes interact on a range of
scales from millimeters or less to thousands of
kilometers [36]. Although most of earthquakes occur
along preexisting faults, when the stress overcomes
a threshold of frictional resistance, the nature of
earthquake dynamics is a complex and chaotic

275

spatiotemporal critical phenomenon. The loading


of faults with stresses is caused by the internal
thermal engine of the earth that drives also the
mantle convection. According to this point of
view, earthquakes is a loading-unloading inputoutput process, whereas the lithospheric dynamics
is an externally driven and far from equilibrium
process. When an earthquake (relaxation unloading
event) occurs, the whole local distribution of the
stress threshold is reached [35,37-40]. The intrinsic
randomness of seismicity means that case studies are
insufficient and statistical methods must be used to
explore earthquake occurrence. The scaling properties
(scale invariance) of earthquake population, related to
multiple spatiotemporal power laws and long-range
correlations, show remarkable similarities to those
observed among the critical phenomena of magnetic
or other complex systems in statistical physics.
Furthermore, the extreme randomness and scaling
properties of the earthquakes allow us to compare
seismicity with the state of turbulence in distributed
and far from equilibrium continuum systems with
infinite numbers of degrees of freedom. Turbulence
is characterized by an energy transport cascade from
large-scale to smaller structures with a multiplicity
of spatial and temporal scales. This character makes
turbulence look equivalent to the critical point
dynamics and to the driven threshold dynamics,
which are observed in many self-organizing systems in
nature. Such a kind of critical process, may include
different kind of systems like neuron nets, domain
rearrangements in flowing foams, charge density waves
in semiconductors, earthquake fault networks, the
world wide web, as well as many ecological social or
political systems. Both phenomena of seismicity and
turbulence share inherent randomness, intermittency
in time and space, stochastic scale invariance and
hierarchically organized scale invariant structures.
For seismicity the scales are extended over many
orders of magnitude, from about a millimeter to
thousands of kilometers. Moreover, in turbulence
and critical phenomena generally the fluctuations
which are associated with correlations in space
and time are inherent properties characterized by
correlation lengths and correlation times respectively.
The statistical physics approach to the description
of multi-scale dynamics of the earthquake fault
systems have led to the conclusion that the observed
scaling laws of seismicity can be the result of a
generalized phase transition process, with avalanches
(clusters of failed sites), nucleation events and critical
exponents. According to the critical state phase
transition approach, the earthquake fault system
can be modeled by slider block critical models,
SOC models, percolation models or self-organized
spinodal (SOS) models [29,31,41-43]. In this study,
we present strong evidence for low dimensionality
and chaoticity of the seismic process. Our results
support the theoretical concept of spatiotemporal and
synchronized chaos, which must be included also to
the theoretical description of the earthquake system.
In accordance with the above statistical approach the
phase transition critical state dynamics ought to be
unified with the stochastic description of chaos and
general theory of the far from equilibrium stochastic
dynamics, presented by Pavlos et al [44].

Nonlinear Phenomena in Complex Systems Vol. 11, no. 2, 2008

A. C. Iliopoulos, G. P. Pavlos, and M. A. Athanasiu

6
(a)

Interevent Time Series


w=10, m=3-9

4
Slopes

5
4

2
1

2
1

-1
2

2.5

3.5

4
Ln(r)

4.5

5.5

2.5

3.5 4
Ln(r)

4.5

5.5

Latitude Tim e Series


w=100, m=3-8

(c)

6
Slopes

4
3

4
3

1
-4

-3

-2

-1

(e)

5
Slopes

4
3

1.5

3
2

1
0
-4

-3

-2

-1
0
Ln(r)

-0.5

0.5

1
Ln(r)

1.5

2.5

FIG. 3. (a) Slopes of the correlation integrals estimated


for the Interevent time series for delay time =5 and
Theiler parameter w =10 and for embedding dimensions
m=3-9, as a function of Ln(r ). (b) Slopes of the correlation
integrals estimated for the Interevent time series and its 30
surrogates series estimated for embedding dimension m=6
and delay time =10, as a function of Ln(r ). (c) Slopes of
the correlation integrals estimated for the Latitude time
series for delay time =1 and Theiler parameter w =100
and for embedding dimensions m=3-8, as a function of
Ln(r ). (d) Slopes of the correlation integrals estimated
for the Latitude time series and its 30 surrogates series
estimated for embedding dimension m=7 and delay time
=10, as a function of Ln(r ). (e) Slopes of the correlation
integrals estimated for the Longitude time series for
delay time =10 and Theiler parameter w =100 and for
embedding dimensions m=3-8, as a function of Ln(r ).
(f) Slopes of the correlation integrals estimated for the
Longitude time series and its 30 surrogates series estimated
for embedding dimension m=7 and delay time =10, as a
function of Ln(r ).

3.

5000

1000

2000
3000
Events

4000

5000

(c)

0.6
0.5

1000

2000
3000
Events

4000

5000

2
0

0.5
Ln(r)

(f)
m= 6
Lo ngitude Tim e Series
Su rro gate s

4000

m=6
Longitude Time Series
Surrogates

-0.5

Longitude Tim e Series


w=100, m =3-8

2000
3000
Events

0.3

Ln(r)
8

0.3
1000

0.4

-1

0.5
0.4

0.7

0.6

0.8

m =6
(d)
L atitude Tim e S eries
S urro gates

0.4

0.6

m =6
(b)
L a titu de T im e S erie s
S urrog ates

0.7

1.5

0.8

(a)

0.2

m =6
In te re v e n t T im e S e rie s
S u rro g a te s

0.8

Slopes

(b)

Lmax

Slopes

Slopes

m=6
Interevent Time Series
Surrogates

Lmax

Lmax

276

Data analysis and results

In this section significant results obtained by


applying nonlinear analysis to three seismic data sets
are presented. The data series correspond to 5400
seismic events that have occurred in the Hellenic
region, using a threshold local magnitude of 4 Richter
and were observed during the time period of 19682004, within the broad Hellenic area 34.00 42.00o
N and 18.00 30.00o E. These data series, which
correspond to the basic focal parameters (time and
space) of earthquakes, were constructed by the
bulletins of the National Observatory of Athens.

FIG. 4. (a) Show the largest Lyapunov exponent (Lmax )


as a function of the seismic events, for the Interevent time
series for delay time =1 and for embedding dimension
m=6. (b) The largest Lyapunov exponent (Lmax ) for
the surrogate data and the Interevent time series as a
function of events estimated for delay time =1 and
embedding dimension m=6. (c)The largest Lyapunov
exponent (Lmax ) as a function of the seismic events, for
the Latitude series for delay time =1 and for embedding
dimension m=6. (d) The largest Lyapunov exponent
(Lmax ) for the surrogate data and the Latitude series as
a function of events estimated for delay time =1 and
embedding dimension m=6. (e) The largest Lyapunov
exponent (Lmax ) as a function of the seismic events,
for the Longitude series for delay time =1 and for
embedding dimension m=6. (f) The largest Lyapunov
exponent (Lmax ) for the surrogate data and the Longitude
series as a function of events estimated for delay time =1
and embedding dimension m=6.

3.1.

The seismic time series

Fig. 1, shows a representative map that contains


all the epicenters of earthquakes concerning the Greek
catalogue for the period of 1900-1999, regardless of the
focal depth and magnitude range using a threshold of
4 Richter, within the area of 33.00 43.00o N and
18.00 30.99o E [45].
In Figure 2(a-d) the three time series of seismic
events are illustrated. In particular, the interevent
series (time intervals between successive earthquakes
shown in Fig. 1a), was created following the concept
of Pavlos et al [34] that the earthquake process
reveals similarity with the chaotic dynamics of the
dripping faucet system according to Shaw [46]. The
Latitude (Fig. 1b) and the Longitude data sets
(Fig. 1c) correspond to the spatial distribution of
earthquakes in the Hellenic region. In these figures,
it can be observed that the signals maintain a strong
stationary profile. The autocorrelation coefficient for
the interevent, latitude and longitude time series
are shown in Fig. 2d. As we can observe in this
figure, during 5-10 lag times the autocorrelations
decay at a plateau value. In the same figure, the
autocorrelation coefficient corresponding to longitude,
take higher values than the corresponding values of
the latitude correlation coefficient. What is more,

. 11, 2, 2008

Spatiotemporal Chaos into the Hellenic Seismogenesis . . .


it is observed that interevent time series correlation
coefficient values remain lower than the values of
the correlation coefficient of the spatial time series.
The profiles of the autocorrelation coefficients shown
in Fig. 2d, indicate the chaotic and non-periodical
character of the seismic time series, in the three cases
that are studied here and are shown in Fig 2(a-c).

3.2. Correlation dimension of the seismic


time series and their surrogate data

Figures 3 illustrates the slopes of the correlation


integral estimated for the seismic timeseries and their
surrogate data. In particular, Fig. 2a presents the
slopes of the correlation integrals estimated for the
Interevent time series using delay time =5, Theiler
parameter w =10 and embedding dimensions m=3-9,
as a function of Ln(r ). These slopes reveal efficient
scaling and low saturation value, D 3, at low
values of Ln(r ). According to the embedding theory
of Takens [47], the dynamical degrees of freedom can
not be higher than 2D + 1 7 and lower than
Dint 4 where Dint is the first integer value of
the slopes. Moreover, Fig. 2b shows the slopes of the
correlation integrals estimated for the Interevent time
series and its 30 surrogates series using parameters
embedding dimension m=6 and delay time =1, as
a function of Ln(r ).The low value saturation of the
slopes in the reconstructed phase space as well as the
clear discrimination, of the Interevent signal from its
surrogate data, at the scaling region ln(r )= 2.75 4.25, clearly reveal that the underlying dynamics of
the Interevent earthquake time series correspond to a
low dimensional and non linear deterministic process.
Fig. 3c presents the slopes of the correlation
integrals estimated for the Latitude data series using
delay time =10, Theiler parameter w =100 and
embedding dimensions m=3-8, as a function of Ln(r ).
These slopes reveal efficient scaling and low saturation
value, D 2.5, at low values of Ln(r ). In this case, the
dynamical degrees of freedom can not be higher than
2D + 1 6 and lower than Dint 3. Fig. 2d shows
the slopes of the correlation integrals estimated for the
Latitude time series and its 30 surrogates series using
parameters embedding dimension m=7 and delay time
=10, as a function of Ln(r ). The significants of the
statistics was found to be higher than 14 sigmas. The
low saturation value of the slopes in the reconstructed
phase space as well as the clear discrimination of
the Latitude signal from its surrogate data, at the
scaling region ln(r )= 0 - 1.2, clearly reveal that
the underlying dynamics of the Latitude earthquake
time series correspond to a low dimensional non linear
deterministic process.
Fig. 3e presents the slopes of the correlation
integrals estimated for the Longitude data series
using delay time =10, Theiler parameter w =100 and
embedding dimensions m=3-8, as a function of Ln(r ).
These slopes reveal efficient scaling and low saturation
value, D 2.5, at low values of Ln(r ). In this case, the
dynamical degrees of freedom can not be higher than
2D + 1 6 and lower than Dint 3 where Dint is the
first integer value after the saturation value D 2.

277

Moreover, Fig. 2f shows the slopes of the correlation


integrals estimated for the Longitude time series and
its 30 surrogates series using parameters embedding
dimension m=6 and delay time =10, as a function of
Ln(r ). The significants of the statistics was found to
be higher than 12 sigmas. The low saturation value of
the slopes in the reconstructed phase space as well as
the clear discrimination, of the Longitude signal from
its surrogate data, at the scaling region ln(r )= 0 1, clearly reveal that the underlying dynamics of the
Latitude earthquake time series correspond to a low
dimensional non linear deterministic process.
3.3. Lyapunov exponents for the seismic
signals and their surrogate data

In Figure 4 we present the estimation of the


largest Lyapunov exponents for the seismic timeseries
and their surrogate data. Fig. 4a presents the
largest Lyapunov exponent (Lmax) for the Interevent
time series and the corresponding surrogate data
as a function of events estimated using delay time
=10 and embedding dimension m=6. The largest
Lyapunov exponent (Lmax ) for the Interevent time
series was found to be positive, attaining a value of
Lmax 4.2 bit/event clearly discriminated from the
values of the (Lmax ) estimated for the surrogate data,
which take values higher than Lmax 4.2 bit/event.
The significance, S, of the statistical test fluctuates
at value of 3 sigmas and thus, clearly permits the
rejection of the null hypothesis with confidence greater
than 95%. These results clearly indicate existence of
sensitivity to initial conditions and so the chaotic
character of the temporal earthquake dynamics. Fig.
4b presents the largest Lyapunov exponent (Lmax)
for the Latitude time series and the corresponding
surrogate data as a function of events estimated
using delay time =10 and embedding dimension
m=6. The largest Lyapunov exponent (Lmax ) for
the Interevent time series was found to be positive,
attaining a value of Lmax 5.3 bit/event clearly
discriminated from the values of the (Lmax ) estimated
for the surrogate data, which take values higher
than Lmax 5.3 bit/event. The significance, S, of
the statistical test fluctuates at value of 5 sigmas
and thus, clearly permits the rejection of the null
hypothesis with confidence greater than 95%. These
results clearly indicate existence of sensitivity to
initial conditions and so the chaotic character of
the earthquake dynamics mapped at the Latitude
time series . Fig. 4c presents the largest Lyapunov
exponent (Lmax) for the Longitude time series and the
corresponding surrogate data as a function of events
estimated using delay time =10 and embedding
dimension m=6. The largest Lyapunov exponent
(Lmax ) for the Interevent time series was found to
be positive, attaining a value of Lmax 5.3 bit/event
clearly discriminated from the values of the (Lmax )
estimated for the surrogate data, which take values
higher than Lmax 5.3 bit/event. The significance,
S, of the statistical test fluctuates at value of 5.5
sigmas and thus, clearly permits the rejection of the

Nonlinear Phenomena in Complex Systems Vol. 11, no. 2, 2008

278

A. C. Iliopoulos, G. P. Pavlos, and M. A. Athanasiu

null hypothesis with confidence greater than 95%.


These results clearly indicate existence of sensitivity
to initial conditions and so the chaotic character of the
earthquake dynamics mapped at the Longitude time
series .

4.

Summary and discussions

Seismic events are obtained randomly in space


and time. However, the results presented previously,
indicate that the seismic randomness is caused
by a low dimensional deterministic and chaotic
spatiotemporal process. The spatial and temporal
trace of the seismogenesis, as it has shown by the
analysis of spatial and temporal time series, can
be related with a low dimensional chaotic walker,
in contrast with the purely random walker process
[48-50]. The hypothesis of seismogenesis, as an
active chaotic walker, corresponds to a point process
realized in the continuously extended lithospheric
system. Moreover, the results presented in this study
indicate that the lithospheric system is in a state
of weak turbulence, where the existence of the
finite dimensional strange attractor is possible. Weak
turbulence is the state of early turbulence in the
distributed system characterized by the existence of
low effective dimensionality and spatial long range
correlations. Generally, as the control parameters
of the system change, the effective dimensionality
may increase. As the system arrives at the state of

developed turbulence, then, long-range correlations


are absent and distant elements of the medium
are statistically independent, while the effective
dimensionality becomes practically infinite [51-54].
The different possible states of turbulence in a
distributed active system (early or developed phases of
turbulence) must also be related with different phase
transition critical states. According to this point of
view, the Hellenic lithospheric system appears to be
in a state of early turbulence with a low dimensional
universal attractor. Other regions of the earth
lithospheric crust can exist in different turbulence
state with different effective dimensionality. The
seismic chaotic walker model supported by the results
of this study could be related with the chaotic
wandering of defects in the turbulence states of the
lithospheric medium, according to the theoretical
models of distributed and interacted maps [53-55] .
Avalanche events or nucleation events in the phase
transition critical state of the lithospheric system,
are related also with the dynamics of defects in the
lithospheric turbulence state [43,44].

Acknowledgment
We would like to thank undergraduate student
of physics Eugene Pavlos for technical support during
the preparation of this study.

References
[1] Ch.F. Richter (1958) Elementary Seismology, San
Francisco (1958).
[2] A. Corral (2004), Physical Review Letters 92, 108501
(2004).
[3] D.P. Hill (2002), Phys. Today 55, 41-7 (2002).
[4] D.L. Turcotte (1999), Rep. Prog. Phys. 62, 1377-1429
(1999).
[5] V.I. Keilis-Borok (1999), Physics of the Earth and
Planetary Interiors 111, 179-185 (1999).
[6] F.F. Evison (2001), Tectonophysics 338, 207-215
(2001).
[7] R. Lasaponara, A. Santulli, L. Telesca (2005), Chaos
Solitons and Fractals 24, 139-149 (2005).
[8] L. Telesca, V. Cuomo, V. Lapenna, F. Vallianatos,
(2000), Tectonophysics 321, 179-188 (2000).
[9] L. Telesca, V. Cuomo, V. Lapenna, F. Vallianatos,
G. Drakatos, (2001), Tectonophysics 341, 163-178
(2001).
[10] L. Telesca, V. Lapenna, M. Macchiato, (2004), Chaos,
Solitons and Fractals 20, 195-203 (2004).
[11] V. Lapenna, M. Macchiato, L. Telesca (1998), Physics
of the Earth and Planetary Interiors 106, 115-127
(1998).
[12] P. Bak, K. Christensen, L. Danon, T. Scanlon (2002),
Phys. Rev. Lett. 88, 178501 (2002).

[13] A. Corral (2005), Phys. Rec. Lett. 95, 159801 (2005).


[14] J. Davidsen, C. Goltz (2004), Geophys. Res. 31,
L21612 (2004).
[15] M. Lindman, K. Jonsdottir, R. Roberts, B. Lund,
R. Bodvarsson (2005), Phys. Rev. Lett. 94, 108501
(2005).
[16] S.A. Fedotov (1968), Izdatelstvo Nauka,Moscow, 133166 (1968).
[17] T. Utsu (1970), Geophysics 3, 197-226 (1970).
[18] V.I. Keilis-Borok, A.A. Soloviev (2003) Nonlinear
Dynamics of the Lithosphere and Earthquake
Prediction, Springer-Verlag (2003).
[19] F.F. Evison (2001), Tectonophysics 338, 207-215
(2001).
[20] A.M. Gabrielov, T.A. Levshima, I.M. Rotwain (1990),
Phys. Earth. Planet. Int. 61, 18-28 (1990).
[21] R. Burridge, L. Knopoff (1967), Bull. Seismol. Soc.
Am. 57, 341 (1967).
[22] J.M. Carlson, J.S. Langer (1989), Phys. Rev. Lett. 62,
2632-2635 (1989).
[23] Maria de Sousa Vieira (1995), Physical Review E 53,
1441 (1996).
[24] P. Bak, C. Tang (1989), J. Geophys. Res. 94, 1563515637 (1989).
[25] P. Bak, K. Chen (1991), Scientific American 26-33

. 11, 2, 2008

Spatiotemporal Chaos into the Hellenic Seismogenesis . . .


(1991).
[26] C.G. Sammis, S.W. Smith (1999), Pure Appl.
Geophys. 155, 307-334 (1999).
[27] W.I. Newman, D.L. Turcotte (2002), Nonlinear
Processes in Geophysics 9, 453-461 (2002).
[28] Y. Ben-Zion, M. Eneva, Y. Liu (2003), Journal of
Geophysical Research 108, 2307 (2003).
[29] W.I. Newman, A.M. Gabrielov, T.A. Durand, S.L.
Phoenix, D.L. Turcotte (1994), Phusica D 77, 200216 (1994).
[30] A. Saichev, D. Sornette (2006), Eur. Phys. J. B.49,
377-401 (2006).
[31] J.B. Rundle, D.L. Turcotte, R. Shcherbakov, W.
Klein, C. Sammis (2003), Rev. Gephys. 41, 1019
(2003).
[32] W. Klein, J.B. Rundle, C.D. Ferguson (1997), Phys.
Rev. Lett. 78, 3793-3796 (1997).
[33] G.P. Pavlos, A.C. Iliopoulos, M.A. Athanasiu
(2007) Nonlinear Dynamics in Geoscience, Self
Organized Criticality and/or Low Dimensional Chaos
in Earthquake Processes: Theory and Practice in
Hellenic Region, Springer, New York, 235-259 (2007).
[34] G.P. Pavlos, D. Diamadidis, A. Adamopoulos, A.G.
Rigas, I.A. Daglis, E.T. Sarris (1994), Nonlin. Proc.
Geophys. 1, 124-135 (1994).
[35] V.I. Keilis-Borok (1990), Physics of the Earth and
Planetary Interiors 61, 1-7 (1990).
[36] V.G. Kossobokov, V.I. Keilis-Borok, D.L. Turcotte,
B.D. Malamud (2000), Pure Ap. Geophys. 157, 23232349 (2000).
[37] C. Codano, F. Pingue (2005), Geophys. Res. Lett. 32,
L18302 (2005).
[38] D. Sornette, V. Pisarenko (2003), Geophys. Res. Lett.
30, 1105 (2003).

279

[39] D. Kiyashchenko, N. Smimova, N. Troyan, E. Saenger,


F. Vallianatos (2004), Phys. Chem. Earth 29, 367-378
(2004).
[40] C.H. Scholz, A. Gupta (2000), Journal of Geophysics
29, 459-467 (1990).
[41] P. Bak, C. Tang, K. Wiesenfeld (1988), Physical
Review A 38, 1 (1988).
[42] W. Klein, M. Anghel, C.D. Ferguson, J.B. Rundle,
J.S.S. Martins (2000), Geophys. Monogr. Ser. 120,
43-72 (2000).
[43] W. Klein, H. Gould, N. Gulbance, J.B. Rundle, K.
Tiampo (2007), Physical Review E 75, 031114 (2007).
[44] G.P. Pavlos, (2007), (Submitted for Publication)
(2007).
[45] P.W. Burton, Y. Xu, C. Qin, G-Akis Tselentis, E.
Sokos (2004), Tectonophysics 390, 117-127 (2004).
[46] R. Shaw (1984) The dripping faucet as a model
Chaotic System, Aerial Santa Cruz (1984).
[47] F. Takens (1981), Lectures Notes in Mathematics
898, 366-381 (1981).
[48] G.H. Weiss (1994) Fractals in Science, A Primer of
Random Walkology, Springer, New York 119 (1994).
[49] Lui Lam, Rocco Pochy (1995), Computers in Physics
7, 207 (1995).
[50] A. Helmstetter, D. Sornette (2002), Phys. Rev. E 66,
061104 (2002).
[51] H. Chate (1995), Physica D 86, 238-247 (1995).
[52] P.C. Hohenberg, B.I. Shraiman (1989), Physica D 37,
109-115 (1989).
[53] T. Yanagita, K. Kaneko (1995), Physica D 82, 288313 (1995).
[54] A.S. Mikhailov, A.Yu Loskutov, Foundations of
Synergetics II (1991), Springer-Verlag (1991).
[55] K. Kaneko (1989), Physica D 34, 1-41 (1989).

Nonlinear Phenomena in Complex Systems Vol. 11, no. 2, 2008

Você também pode gostar