JESS-D-14

Manuscript
Click here to download Manuscript: ramdani2.docx
Click here to view linked References
1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
Seismic link at plate boundary
1
Faical Ramdani, Omar Kettani, Benaissa Tadili
Mohamed V- University, Scientific Institute, Physics of the Earth Laboratory
Rabat, Morocco
2
3
Abstract
4
5
Seismic triggering at plate boundaries has a very complex nature that includes seismic events
6
at varying distances. The spatial orientation of triggering cannot be reduced to sequences from
7
the main shocks. Seismic waves propagate at all times in all directions, particularly in highly
8
active zones. No direct evidence can be obtained regarding which earthquakes trigger the
9
shocks. The first approach is to determine the potential linked zones where triggering may
10
occur. The second step is to determine the causality between the events and their triggered
11
shocks. The spatial orientation of the links between events is established from pre-ordered
12
networks and the adapted dependence of the spatio-temporal occurrence of earthquakes.
13
Based on a coefficient of synchronous seismic activity to grid couples, we derive a network
14
link by each threshold. The links of high thresholds are tested using the coherence of time
15
series to determine the causality and related orientation. The resulting link orientations at the
16
plate boundary conditions indicate that causal triggering seems to be localized along a major
17
fault, as a stress transfer between two major faults, and parallel to the geothermal area
18
extension.
19
20
21
Keywords: seismic link, time series, distant earthquakes, causality, plate boundary
22
Plate boundaries are the zones where most Earth dynamics are focused. The complexity of
23
tectonic boundaries draws attention to them as the largest earthquakes are felt in these areas
24
and they elicit the natural hazard of seismic activity. However, the sequences of the main
25
shocks and the triggering process constrain the seismic hazard assessment. Some frequent
26
issues remain as to whether the seismicity of the interplate zone with a high strain rate has
27
random aspects and may be associated with a triggering mechanism, and how the seismicity
28
can be associated with fault patterns and plate motion. Triggering causes changes in the
29
Coulomb stress on a specified fault, which is independent of regional stress but which
30
depends on the fault geometry, the sense of slip, and the coefficient of friction (King et al.,
31
1994). Restricting the causality connection to a single predecessor or to an arbitrary
1. Introduction
2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
32
mainshock-aftershock scenario may not be enough (Krishna Mohan and Revathi, 2011). This
33
shows that the link between events is a complex feature linked to varying network models
34
(Abe and Suzuki, 2006). Links limited to individual events have been related to low strain
35
rate environments (Hough et al., 2003), visco-elastic relaxation (Lorenzo-Martin et al., 2005),
36
and solid Earth tides (Cochran et al., 2004). By analyzing earthquake pairs over a period of
37
many years, Wan et al. (2004) suggested that Coulomb stress triggering may be observed for
38
thrust earthquakes, while McKernon and Main (2005) limited triggering process to about 150
39
km. Extending earthquake activity to a spatial network link shows that the alignment of the
40
links is parallel to the Honshu Trench azimuth in Japan (Tanenbaum et al., 2012) and the
41
direction of the San Andreas Fault (SAF) (Jimenez et al., 2008). In the Ibero-Moroccan
42
region, main shocks and their aftershocks are investigated as local features due to fault
43
mechanisms except for some historical event studies related to triggering in the Catalan and
44
NE Iberian regions (Perea, 2009). As zones of permanent activity make it difficult to detect
45
causal triggering, it is important to search for the statistical flow of triggering. The use of
46
individual events is complex as varying fault systems may be in a critical state of failure
47
before the passage of stress from the driver event. Thus, it is useful to consider a zoning
48
characterized by a set of events over a period of many years before searching for triggering.
49
This provides inter-dependent zones of coeval activity in which causal triggering may be
50
estimated. We test a seismic link network from catalogs of Turkey and California, Gibraltar as
51
zones of collision, and the Philippines, New Zealand, and Japan as subduction zones. Short
52
time triggering and a long-lived process of two decades are included, and the main objective
53
is to first establish zones of spatial seismic dependence. The seismic grids represented by time
54
series are further tested to determine the drivers from recipient zones based on the coherence
55
of the time series. Comparing high linked zones with stress field azimuths and local tectonics
56
provides insight into some possible widespread behavior of seismic triggering in the collision
57
zones.
58
2. Dataset and Methods
59
The databases used in this study include events that occurred in a relatively large area of the
60
plate boundary, but they are limited in space because linked earthquakes of major events may
61
traverse the entire Earth. Since a set of shocks may trigger events far away from the main
62
shock areas, they must be compiled during a relatively important time window. In turn,
63
triggered events may also trigger events in the main shock source zone. For this reason, the
64
activity measured between zones represents a cross-correlation in both space and time.
3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
65
Sequences from aftershocks are not the only factor controlling the threshold link owing to the
66
wide time span and distant zones within the plate boundaries. Individual shocks cannot
67
represent the source zones of the triggering in zones of high activity where shocks occurred in
68
every zone along the plate boundary. In those particular areas, it is more convenient to
69
introduce the triggering flow direction that considers the link between active zones; then,
70
when a threshold link is reached, the orientation of flow may be estimated to delimit the
71
direction from the driver to the recipient zone in terms of statistical activity. However, many
72
plate boundaries include a considerable number of events that are difficult to process as we
73
are limited by the running time. In this regard, a specific time window was adopted for each
74
of the plate boundary catalogs. Pacific zones of subduction present abundant background
75
seismicity ranging with various magnitudes, such as the Japan and New Zealand regions. We
76
adopted a 10-year time window because it is a reasonable period to search for a link between
77
distant regions. The catalogs have to follow Gutenberg-Richter regression from a specific
78
magnitude: the Magnitude of Completeness (Mc). Mc is calculated from Zmap by using
79
maximum likelihood solution. However, the use of Mc cutoff in these catalogs shows that the
80
remaining event number is considerable so we then increased our minimum magnitude to 3
81
for Japan and New Zealand. In turn, Philipppines regions shows Mc4.5 which reduced too
82
much the database, so in this case the lower magnitude cutoff is reduced to 3.The data was
83
obtained from the Japan University Network (JUNEC) Earthquake Catalogue and GeoNet of
84
New Zealand for the period spanning from 1988–1998. For the other regions (the Philippines,
85
Turkey, California), the catalogs were obtained from the US Geological Survey with a time
86
window of 25 years (1988–2012) and the Istituto Geográfico Nacional (Spain) for the
87
Gibraltar region. The Magnitude cutoff was adopted (Table 1) as we investigated the long-
88
distance link of the shocks, and the links related to the lower magnitudes were not included.
89
Many earthquakes may occur after wave arrivals, so we then enlarged the period necessary for
90
triggering and we also made the distance as large as possible around the plate boundaries in
91
order to include most of the events. The plate boundaries were partitioned into a 1°×1° grid.
92
The grids are described by a time series of event numbers to each temporal sample, and the
93
activity within a grid covers a period sample of 90 days, and the next sample will cover the
94
next 90-day non-intersecting time period. The prominent factor in determining a link between
95
distant regions is that there is at least one period during a 25-year span where both cells are
96
seismically active. The seismic dependence between the cells increases linearly when the
97
number of simultaneous active periods increases. Standard cross-correlation includes
98
simultaneous passive periods that may not fit the correlation, e.g., Pearson’s correlation;
4
99
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
instead, we suggest a method to compute a coefficient based on canceling periods free from
100
earthquakes and we consider only the number of events in the same temporal sample. The
101
number of synchronized active periods is compiled in regard to the total number of active
102
temporal samples (T) in the cell couple i, j. We use the relationship R of two time series in the
103
form:
104

105
R = α . (e -1)
106
with  = ∑(kij/Tij), α= (1-1/e)/e
107
where kij is the number of synchronized active periods in grids i j, Tij is the total number of
108
active periods in both grids, e is the Napier constant, and A is a constant of normalizing R to
109
boundary conditions. The relation (1) provides a way to compile the synchronized active
110
periods at two grids with respect to the entire activity that occurred in the grid couple. The
111
relationship is computed independently from the activity in the region as the two cells are
112
considered to be individual regions. Figure 1a shows the variation of R with the values of 
113
for all regions. When the coefficient R is above a threshold of 0.5, we recorded the associated
114
link number. The number of links was compiled by successive thresholds and by region
115
(Figure 1b). When the links of high correlated regions were established to each grid, we tested
116
the time series of the grid couple (i,j) in order to estimate the direction of flow. The goal was
117
to obtain information on the causality between the well-correlated grid couple, which shows
118
the direction of triggering. The method is based on a frequency average of the slope of the
119
phase coherence with respect to the instantaneous mixture of the independent source (Notle et
120
al.,2008a). This method, called the Phase Slope Index (PSI), includes the non-linear
121
interaction between the zones, which incorporates waves coming from other seismic sources
122
independent from the two zones under the link. The PSI procedure is more appropriate for a
123
time series than Granger’s causality method (Notle et al., 2008b) particularly when time
124
series are inferred from seismic activity. The Phase Slope Index is defined by:
(1)
125
126
∑
127
()
(
)
(
)
()
(2)
128
( ) is the complex coherence, (f) is the phase spectrum linear and equal to 2 f , f
129
where
130
is the frequency, and  is the delay time. The slope of (f) indicates the causal direction from
131
grid 1 to grid 2, if it is positive, or from grid 2 to grid 1, if it is inversely negative. The PSI
5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
132
code by MATLAB input requires epoch and segment length values, as the time series is
133
sampled by epoch (epleng) and each epoch is divided by segment length (sgleng). Since we
134
have postulated that the activity by grid is sampled over 90 days, this provides the value of the
135
epoch sampling data running at 25 years. The sgleng parameter is determined according to the
136
scores obtained by varying sgleng from 1 to 90. A sgleng of 90 days equal to the epoch value
137
provides the best scores for the regions where the 25-year time span is adopted (Figure 2).
138
However, when only 10 years are used the sample number is reduced and, in that case, the
139
score is 0 and no PSI results are obtained. The reason for this is that the sampling number of
140
the time series has to be at least 100 because of the 1/100frequency limit. To achieve this
141
sampling rate of the subduction zone (Japan and New Zealand), we re-sampled the time series
142
into 30 days for 10 years of datasets. In this case, there would be a total of 120 samples and
143
the score and the PSI results can be obtained. We then calculated the causality by grid couple
144
using sgleng 90 or sgleng 30 for time series with high thresholds.
145
By investigating the activity by grid couples we found that several grid couples present
146
significantly reduced activity (one or two events only) during one period of 90 days over the
147
course of 25 years. This particular single, but synchronous, activity to both cells reaches
148
correlation 1 according to Equation (1). As shown in Figure 3a, this Dirac type of time series
149
is found in all of the studied regions except Turkey. Those will be distinguished from
150
continuous seismic activity by the other grids (figure 3b). One period of activity and multiple
151
periods of activity are then called ts1 and ts2, respectively. The multi-period activity shown
152
in Figure 3b presents the complex interactions between the grid couple at a large distance. PSI
153
coherence is computed to suggest the orientation of causality that predominates the
154
interactions between grids. When causality indicates orientation from grid 1 to grid 2, it
155
cannot mean that grid 2 has no impact on grid 1. Using cases of ts1, we assigned causality
156
orientation by recording the timing of the events in both grids. The grid where the first event
157
occurred is the driver. By using PSI we compute the score obtained by each timeseries and the
158
score we found vary. The best score obtained by each maximal threshold related to distance
159
between grids is shown in table 1.
160
161
3. Results
162
The volume of events varies at each plate boundary zone, which may have an impact on the
163
resulting number of links. Figure 1 shows the link number related to the coefficient R that
164
exhibits a power law variation as commonly stated for several statistical seismic correlation
6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
165
distributions. Such a high correlated link between two grid zones shows evidence for possible
166
triggering, but the orientation of triggering is hard to assess in the presence of multiple
167
sources of earthquakes continuously active at the plate boundary setting.
168
169
170
At the 0.6 threshold a considerable link is found in all regions. These links disappear at a
171
threshold 0.7 and 0.9, revealing a reduced closed zone formed by closed structure in triangles
172
such as on the San Andreas Fault. The Turkish network link model does not appear along the
173
Anatolian Fault; instead, it is limited to the western off-shore part of the Anatolian plate
174
(Figure 5) and it seems that it does not follow the EW-oriented Anatolian Fault but is normal
175
to it. The major EW faults, North Anatolian and Hellenic Arc, are then NS linked in the
176
western sections at the place where they are in contact with the continental margin. The Ibero-
177
Moroccan region shows that the network link lies around the Gibraltar Arc (Figure 5). In
178
subduction zones, the network links appear parallel to the trenches, as shown in Japan.
179
Similarly, the Philippines are located between two major faults, the Manila Trench and the
180
Philippine Trench, and we found that a high threshold link is obtained in the junction of these
181
major faults, but the intersecting zone coincides with a secondary transform fault passing near
182
the volcanic zone of Mayon (Figure 5a). More large networks of linked zones in the Ibero-
183
Moroccan region seem to follow the Gibraltar Arc configuration. The W direction of causality
184
points to the west along the Açores-Gibraltar Transform Fault (Figure 5b). A triangular
185
structure showing the causal direction of the link is found in California on the San Andreas
186
Fault pointing to SW direction (Figure 6a) that indicates zones of flow interaction. 18
187
individual sites of unique synchronized event are observed with varying direction. No such
188
individual synchronized events are observed in Turkey. The EW-oriented faults in this zone
189
might generate an EW link that is not observed. Instead, we observed two closed structures,
190
one located to the north and one located to the south along the Hellenic Arc (Figure 6b).
191
The Japan model of oriented links based on PSI outlines NW margin of Honshu (Figure 7a)
192
with two additional individual synchronized sites. In the presence of the subduction zone,
193
New Zealand exhibits most links along the Alpine Fault, oriented NE-SW, and another cluster
194
link to the northern subduction zone (Figure 7b). In these regions, the link has a preferred
195
orientation along the transform faults and it is normal to the plate motion driven by
196
subduction.
197
4. Discussion and Conclusions
7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
198
Triggering processes at a plate boundary exhibit a preferred direction for the triggering. Zones
199
that trigger far away zones through a preferred direction of triggering will be recipients of the
200
stress coming from the regions they have triggered, so the system will be closed. While this
201
process works repetitively in relatively small scale areas, cluster zones of causal triggering
202
can be observed. The preferred direction of triggering is a direction where the amplitude of
203
the stress transfer is increased. As the causality is measured through many years of activity
204
and the faulting system within a grid is not specified, the causality only includes the far away
205
spatio-temporal stress transfer in statistical terms. The time delays for static, dynamic, or
206
viscoelastic relaxation are not distinguished, but the preferred direction of the stress transfer is
207
one of the main parameters in the triggering process. The correlation between grids shows
208
that an earthquake primarily presents an instability setting of a zone or a fault. This event can
209
also trigger another far away fault and it can also receive stress from another event and
210
become a second order of failure. Because of this mixture of roles between the recipient and
211
the driver, it seems more appropriate to compile the interaction with multi-sources to infer the
212
dominant flows of triggering. As there is not a zero-time to decide whether this event
213
triggered some other events, and so on, we can consider that each event is due to the stress
214
increase caused by past and permanent wave propagation. Only by studying the number of
215
events in several sampling time windows we can determine if two zones may be linked. When
216
two distant and passive grid zones experienced only one event over the course of many years,
217
and that event coincided in both grids within the same sampling time of 90 days, the event
218
may be assumed to be a particular triggering in aseismic region. Since the time delay between
219
the two events may be recorded, it can be interpreted as a real time delay for that particular
220
triggering. However, both events may be caused by a third previous event, and the time delay
221
between these two events is not as long as the time it takes for each zone to reach critical
222
failure. Otherwise, tidal deformation may trigger events occurring in passive regions.
223
In a subduction zone, stress transfer is dominantly guided by the direction normal to
224
subduction and parallel to the trench zones or the volcanic lines, as it is in Japan. This
225
evidence seems to be contrary to the absence of widespread triggering in Japan reported by
226
Harrington and Brodsky (2006), but it is in agreement with the evidence for tidal triggering on
227
reverse faults in Japan (Tanaka et al., 2004), and low frequency deep triggering (Miyazawa
228
and Mori 2005). NE-SW oriented triggering flow appears more pronounced in SW direction
229
than the NE azimuth. This zone is parallel to Japan trench and volcanic lines which favors
230
fluid dynamic released in the extension normal to subduction. At subcrustal depths,
231
earthquakes due to metastable phase changes in the slab release kinetic energy normal to slab
8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
232
penetration. It is notable that the most synchronized link stop at the Sagami trench to the
233
South, and the absence of triggered link at maximal threshold in Nankai trough. Notable
234
observation is that NE-SW oriented linked zone of northern Japan is similar to the NE-SW
235
linked zones of New Zealand even though both regions are not in the same azimuth of
236
subduction. Intriguing observation is that in both subduction zones we found consistently
237
most of causal links (70%) are SW oriented. In northern zone of New Zealand that focuses
238
driver sites is a volcanic area while in the South the Alpine fault seems to be recipient zone.
239
Crustal weakness beneath the fault made it possible repetitive triggering process along the
240
fault.
241
In the Gibraltar zone, causal direction outlines the Açores-Gibraltar fault zone strike in the W
242
direction, globally, that is oblique to the NW-SE compression, but in this special case
243
triggering is oblique to compression but parallel to the W subduction, as reported by Ramdani
244
et al. (2014). Model of fault-to-fault stress transfer at the regional scale including a closed
245
structure of causality is shown in the trenches in the Hellenic Arc and the eastern boundary of
246
the Aegean Sea plate and the southern Anatolian Fault. In between, these closed structures are
247
separated by the EW volcanic arc, which shows that the triggered flow in these regions is
248
compression parallel, but not parallel to the magma fluid dynamics. The presence of
249
Anatolian and Hellenic Fault arcs probably predominate the causal flow rather than the
250
volcanic extension EW oriented. The California causal directions also present closed
251
structures along the northern parts of San Andreas Fault in the SE direction from 40°N to
252
36°N. This causal structure of the link is parallel to the SE-oriented volcanic line. However,
253
California triggered zones seem dominated at this maximum threshold by many individual
254
triggering dispersed in the region. Some sites are repetitively triggered from several sites and
255
sometimes at the same period sample. This may not be evidence for direct triggering as it may
256
be explained by the fact that many sites may be triggered at varying timing within the same
257
sample of 90 days. However, the individual earthquake in a passive site is the clear example
258
for triggering. The orientation of these individual triggering at such large distance seems not
259
related to stress field environment or a particular dynamic process. The passive site
260
experienced permanent passage of seismic waves without any response, and sudden
261
occurrence of an event may be caused by a driver earthquake with particular characteristics.
262
Frequency content or Tidal effects may be source of the reaction of a passive zone that returns
263
passive just after the event has occurred.
264
265
9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
266
The Philippines lying between EW-opposed subduction (the Manila and Philippines
267
Trenches) shows that the stress transfer switches between zones lying in the Manila Fault to
268
the west and the eastern Philippine Trench. However, along the Philippine Fault, causal
269
directions of triggering parallel this major fault that is normal to compression related
270
subduction. Triggering between these two major faults is focused in two cluster zones of
271
Manilla trench and a much larger zone of the opposed Philippines trenchs forming a triangular
272
beam. Manilla trench sites appears to be recipient of driver sites from Philippine fault zone.
273
The presence of opposed slab beneath Philippines favours fluid dynamics and increasing pore
274
pressure within the faults. As triggering flow is mainly oriented to Manilla trench this may be
275
evidence for that subduction from Philippines provides more stress transfer than the opposed
276
subduction beneath Manilla trench.
277
In New Zealand, the causal direction is NE-SW oriented along the Alpine Fault and the
278
localized zones in the northern subduction zone. However, in this case, triggering only
279
reflects the partial possibilities of the links since only grids that are separated by a distance
280
greater than 2° and a magnitude M>3 are used. Lower magnitudes of the near field triggering
281
are not included. In the case of New Zealand, both the stress transfers along the fault to the
282
south and the triggering in the northern volcanic zones are viable. In terms of triggering
283
flows, the predominant extension setting, may first be explained by the fact that extension
284
induces a widespread direction of stress transfer, which is larger than what occurs with
285
undirected stress from compression. Secondly, dynamic fluid relaxation in the lower crust
286
provides long-lived stress in volcanic eruption zones. Repetitive triggering is estimated by the
287
causal flows of zones of high seismic dependence. This result is not due to the PSI method
288
since the procedure does not take into account the geographic location of the grids. Reversed
289
flow directions parallel to volcanic arc is observed in subduction zones.
290
The time span and distance scale used in this study seem sufficient to observe some aspects of
291
the complex earthquake interactions at plate boundaries, and the associated correlated events
292
are shown from a reasonable period during which multi-source triggering may have occurred.
293
This approach that compares one grid to all of the grids of the region is crucial to establishing
294
the regional seismic interdependence. It is justified by the unknown direction of possible
295
triggering that is confined to multiple sources of permanent seismic activity at plate
296
boundaries. Individual earthquakes in a region provide additional stress that may be sufficient
297
to meet the critical failure at varying distances and times. In expanding the process to a period
298
as large as 25 years over a distance of more than 1000 km we can expect to determine how
299
triggering operates in the presence of multiple source events. Transfer from fault-to-fault is
10
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
300
observed in the presence of major faults and is limited along the fault in the presence of a
301
unique regional fault. The question of why extension appears to fit the flow direction of
302
triggering is consistently asked. Crustal relaxation with volcanic eruptions in geothermal areas
303
provides an extension field for triggering in the subduction zones. However, in the presence
304
of two major faults, the triggering may be compression parallel, as is observed in the Aegean
305
Sea and the Philippines. It is concluded that causal triggering at varying plate boundaries
306
occurs primarily along a major unique fault or is due to the stress transfer bi-directed between
307
two major faults. It also occurs along a geothermal volcanic arc when no parallel major fault
308
exists in the enlarged proximity.
309
310
Acknowledgments
311
This work has been supported by Mohamed V-University founding from PU project.
312
P. Dewangan and anonymous reviewers are thanked for reviews.
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
11
331
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
332
References
333
Abe S and Suzuki N 2006 Complex-network description of seismicity; Nonlin.
334
Processes Geophys., 13, 145–15
335
Krishna Mohan T R and Revathi P G 2010 Earthquake correlations and networks: A
336
comparative study; Phys. Rev. E 83, 046109
337
Cochran E S, Vidale J E and Tanaka S 2004 Earth tides can trigger shallow thrust fault
338
earthquakes; Science 306:1164-1166
339
Harrington, R. M. and Brodsky E E 2006 The absence of remotely triggered seismicity
340
in Japan; Bull Seismological Society America 96: 871-878.
341
Hough S E, Seeber L and Armbruster J G 2003 Intraplate Triggered Earthquakes:
342
Observations and Interpretation; Bulletin of the Seismological Society of America 93
343
5:2212–2221
344
Jimenez A, Tiampo K F and Posadas A M 2008 Small world in a seismic network: the
345
California case; Nonlin Processes Geophys 15:389–395
346
King G C P, Stein R S, and Lin J 1994 Static Stress Changes and the Triggering of
347
Earthquakes; Bulletin of the Seismological Society of America 84 3: 935-953
348
Lorenzo-Martin F, Roth F and Wang R J 2006 Elastic and inelastic triggering of
349
earthquakes in the North Anatolian Fault zone; Tectonophys 424:3-4
350
Miyazawa, M., J. Mori 2005 Detection of triggered deep low-frequency events from
351
the 2003 Tokachi-oki earthquake; Geophys. Res. Lett 32: L10307; doi:
352
10.1029/2005GL022539
353
Nolte G, Ziehe A, Nikulin V, Schlogl A, Kramer N, Brismar T and Muller K-R 2008a
354
Robustly Estimating the Flow Direction of Information in Complex Physical Systems;
355
Phys Review Letters 100:234101
356
Nolte G, Ziehe A, Krämer N, Popescu F and Müller K R 2008b Comparison of
357
Granger Causality and Phase Slope Index; JMLR Workshop and Conference
358
Proceedings 6: 267–276
359
Perea, H 2009 The Catalan seismic crisis [1427 and 1428; NE Iberian Peninsula]:
360
Geological sources and earthquake triggering; Journ of Geodynamics 47:259–270
12
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
361
Ramdani F Kettani O and Benaissa Tadili B 2014 Evidence for subduction beneath
362
Gibraltar Arc and Andean regions from k-means earthquake centroids J Seismol DOI
363
10.1007/s10950-014-9449-9
364
365
Tanaka S Ohtake M and Sato H 2004 Tidal triggering of earthquakes in Japan related
366
to the regional tectonic stress; Earth, Planets, and Space 56: 511-515
367
Tenenbaum J, Halvin S and Eugene Stanley H 2012 Earthquake networks based on
368
similar activity patterns; Phys Rev- E 86:046107
369
Wan Y G, Wu Z L and Zhou G W 2004 Focal mechanism dependence of static stress
370
triggering of earthquakes; Tectonophys 390 164: 235-243
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
13
395
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
396
Figure Captions
397
398
399
Figure 1. The variations in the number of links by correlation threshold and by region indicate
400
the power law distribution in both the subduction and collision regions. In (b) the variation of
401
R versus the relative synchronous activity of the two grids.
402
403
Figure 2. The PSI scores obtained for time series (ts) of Turkey (a) and Japan (b) up to a
404
correlation of 0.6 sampled by an epoch (epleng) of 90 days for Turkey and 30 days for Japan
405
by varying sgleng parameters. It appears that the best scores are found when sleng and epleng
406
are equal for both regions.
407
408
Figure 3. The time series of threshold link 0.7 and 1 from the Turkey and California (a) and
409
show variations of seismicity sampled over quarter (90 days) of a period of 25 years. The
410
coordinates and distances between the two cells are indicated at the top left and the arrow
411
shows the direction of causality from the PSI. In (b), time series of rare synchronous activity
412
in passive zones from Philippines and New Zealand regions. Coordinates of grids separated
413
by a distance (D) are indicated. Causality (arrows) is determined directly from arrival times of
414
events.
415
416
Figure 4. Link network up to threshold 0.7 in Japan (a) and grid sampling used. Link network
417
obtained from threshold 0.5 in Turkey in (b) and Gibraltar zone (c). The grids of the region
418
are shown.
419
420
Figure 5. Most of the 34-oriented links at the 0.7 threshold of the Philippines (a) represent
421
stress transfer related to the Manila Trench (MT) and the Philippines Trench (PT).
422
The Ibero-Moroccan region (b) shows a link network at threshold 0.65 with a PSI causal
423
direction (red link) and blue link of individual synchronized event. The arrows show the
424
direction of compression in the region.
425
426
Figure 6. Network and 25 are obtained at threshold 1 in California (a). Localized zones of
427
causality flows are shown on the San Andreas Fault (SAF) including 18 links of individual
14
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
428
synchronized event. Turkish network provides 21 links determined at threshold 0.7 from PSI
429
with causal direction shows two closed structures located south from the Anatolian Fault and
430
north from the Hellenic Arc (b).
431
432
Figure 7. Japan causal determinations includes 22 from PSI and two inferred directly from the
433
timing of individual events shown by red arrows (a). New Zealand link network of 30 links
434
(b) of threshold 1 lie along Alpine fault and the Pyusegur Trench (PT).
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
15
462
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
463
464
465
466
467
468
Table1. The data used in the link network processing, the magnitude cutoff, score obtained by
469
maximal coefficient R and related distance
Region
Jap
NZ
Gibr
Cal
Phil
Turk
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
Period
Nb
events
M-cutoff
Rmax
scorePSI
Distance,
km
1988-1998
2002-2012
1988-2012
1988-2012
1988-2012
1988-2012
39403
40091
6145
23350
9831
18136
3
3
2.5
2.8
3
2.7
1
1
0,75
1
0.8
0,7
1,6743
3,1282
33,09
22,187
35,523
31,113
143
1072
631
620
332
88
16
489
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
490
491
492
493
(a)
(b)
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
Figure 1
17
515
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
Figure 2
18
537
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
538
539
540
541
542
(a)
(b)
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
Figure 3
19
559
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
560
561
562
563
564
565
566
567
568
569
570
(a)
(b)
(c)
571
572
573
574
575
576
577
578
579
Figure 4
20
580
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
581
582
583
584
585
586
587
588
589
590
591
592
(a)
(b)
593
594
595
596
597
598
599
600
601
602
603
604
605
Figure 5
21
606
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
607
608
609
610
611
612
613
614
(a)
(b)
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
Figure 6
22
632
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
633
634
635
636
637
638
639
640
(a)
(b)
641
642
643
644
645
646
647
648
649
650
651
Figure 7