Cannata Et Al., 2013 GR

8/13/2019 Cannata Et Al., 2013 GR

1/15

This article appeared in a journal published by Elsevier. The attached

copy is furnished to the author for internal non-commercial research

and education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling or

licensing copies, or posting to personal, institutional or third partywebsites are prohibited.

In most cases authors are permitted to post their version of the

article (e.g. in Word or Tex form) to their personal website or

institutional repository. Authors requiring further information

regarding Elseviers archiving and manuscript policies are

encouraged to visit:

http://www.elsevier.com/authorsrights
http://www.elsevier.com/authorsrightshttp://www.elsevier.com/authorsrights

8/13/2019 Cannata Et Al., 2013 GR

2/15

Author's personal copy

Repeating volcano-tectonic earthquakes at Mt. Etna volcano (Sicily, Italy) during19992009

Andrea Cannata , Salvatore Alparone, Andrea Ursino

Istituto Nazionale di Geosica e Vulcanologia, Osservatorio Etneo Sezione di Catania, Piazza Roma 2, 95123 Catania, Italy

a b s t r a c ta r t i c l e i n f o

Article history:

Received 21 June 2012

Received in revised form 4 February 2013

Accepted 25 February 2013

Available online 22 March 2013

Handling Editor: A.R.A. Aitken

Keywords:

Volcano tectonic earthquakes

Repeating earthquakes

Mt. Etna volcano

Pernicana fault

Repeating volcano-tectonic (VT)earthquakes, taking place at Mt. Etna during 19992009, were detected and an-

alyzed to investigate their behavior. We found 735 families amounting to 2479 VT earthquakes, representing

~38% of all the analyzed VT earthquakes. The number of VT earthquakes making up the families ranges from 2

to 23. Over 70% of the families comprise 2 or 3 VT earthquakes and only 20 families by more than 10 events.

The occurrence lifetime is also highly variable ranging from some minutes to ten years. In particular, more

than half of the families have a lifetime shorter than 0.5 day and only ~10% longer than 1 year. On the basis of

these results, most of the detected families were considered burst-type, i.e., show swarm-like occurrence,

and hence their origin cannot be explained by a temporally constanttectonic loading. Indeed, since the analyzed

earthquakes take place in a volcanic area, the rocks are affected not only by tectonic stresses related to the fairly

steady regional stress eld but also by local stresses, caused by the volcano, such as magma batch intrusions/

movements and gravitational loading. We focused on theve groupsof familiescharacterizedby thelongestre-

peatability over time, namely high number of events and long lifetime, located in the north-eastern, eastern and

southern anks ofthe volcano. Unlike the rst fourgroups,which similarly to most of the detected familiesshow

swarm-like VT occurrences, group v, located in the north-eastern sector, exhibits a more tectonic behavior

with the events making up such a group spread over almost the entire analyzed period. It is clear how both

occurrence and slip rates do not remain constant but vary over time, and such changes are time-related to the

occurrence of the20022003 eruption. Finally, by FPFIT algorithm a good agreement between directions identi-

ed by nodal planes and the earthquake epicentral distribution was generally found.

2013 International Association for Gondwana Research. Published by Elsevier B.V. All rights reserved.

1. Introduction

Multiplets, also called repeating earthquakes (e.g. Chen et al., 2008)

or earthquake families (Tsujiura, 1983), are earthquakes with similar

waveforms. A high degree of waveform similarity implies the same

source mechanism and locations, varying roughly within one fourth of

the dominant wavelength of the events (Geller and Mueller, 1980).

However, the exact fraction of thewavelength depends on many factorsincluding the heterogeneity of the velocity structure around the source

(e.g. Nakahara,2004). Multiplets have been observedat many active vol-

canoes such as Soufriere Hills (Rowe et al., 2004), Redoubt (Buurman et

al., in press), Mt. St. Helens and Bezyamianny (Thelen et al., 2011) and

Mt. Etna (Alparone and Gambino, 2003), as well as in non-volcanic

areas (for instance along transform faults and subduction zones; e.g.

Nadeau et al., 1995; Igarashi et al., 2003).

Because of their particular characteristic of having repeatable sources

at the same spot but at different times (Schaff and Beroza, 2004), multi-

plets have many seismological applications: highly precise locations to

highlight seismic structures at depth (e.g.Waldhauser et al., 2004); de-

tection of temporal variations of attenuation (e.g.Antolik et al., 1996),

shear wave splitting (e.g. Zaccarelli et al., 2009; Johnson et al., 2010)

and medium velocity (e.g. Schaff and Beroza, 2004; Cociani et al.,

2010); to acquire information on the dynamics of active faults and

their slip rate (e.g.Chen et al., 2008); nally, in volcanic environments,

evaluation of the volcano conditions (e.g. Green and Neuberg, 2006;

Thelen et al., 2011).Several papers have dealt with multiplets detected at Mt. Etna. For

instance, Alparone and Gambino (2003) performed high precision

location analysis of multiplets of volcano-tectonic (VT) earthquakes

recorded during 2001. Brancato and Gresta (2003) analyzed the

multiplets accompanying the beginning of the 19911993 eruption.

Zaccarelli et al. (2009)applied coda wave interferometry and shear

wave splitting techniques on multiplets of VT earthquakes to evaluate

wave propagation effects during the waning phase of the 20022003

eruption.

The structural features of Mt. Etna appear rather complex. On the

volcano surface different fault and ssure systems can be recognized.

The most outstanding tectonic features at Mt. Etna are clearly recogniz-

able on the eastern and south-easternanks of the volcano, where the

clearest morphological evidence of active faulting exists (Azzaro et al.,

Gondwana Research 24 (2013) 12231236

Corresponding author. Tel.: +39 095 7165843; fax: +39 095 7165826.

E-mail addresses:[email protected](A. Cannata),

[email protected](S. Alparone),[email protected](A. Ursino).

1342-937X/$ see front matter 2013 International Association for Gondwana Research. Published by Elsevier B.V. All rights reserved.

http://dx.doi.org/10.1016/j.gr.2013.02.012

Contents lists available at ScienceDirect

Gondwana Research

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / g r

8/13/2019 Cannata Et Al., 2013 GR

3/15


2012). Here, seismogenic faults can be related to the NNWSSE Malta

Escarpment, that is the main lithospheric structure in the eastern Sicily

(Scandone et al., 1981), to which the Timpe Fault System is closely asso-

ciated(Fig. 1; Azzaroet al., 2012). Other seismogenetic faults, though not

recognizable on the surface, may be linked to the NESW, ENEWSW

fault systems that control the tectonic evolution of the northern margin

of the Hyblean Plateau (Torelli et al., 1998).Seismic data analysis has shown that ~50% of the VT earthquakes

at Mt. Etna are shallow (focal depth b5 km b.s.l.) and mainly located

in the eastern ank (Patan et al., 2004). Similar to what has been

observed in many volcanic areas (McNutt, 2005), these earthquakes,

showing magnitude generally lower than 4.0 (e.g.,Ferrucci and Patan,

1993), mostlyoccur in the formof swarms(Patan et al., 2004). The east-

ernank of Mt. Etna is characterized by frequent shallow seismic activity

(depth b7 km b.s.l.; Alparone et al., 2011). Conversely, the westernank

of Mt.Etna, normallycharacterized by a deeperseismicity (depth > 5 km

b.s.l.), is considered the most stable sector of the volcano. According

to Patan et al. (2004), three main source processes are involved in

the generation of VT seismicity at Mt. Etna: i) regional tectonic stresses,

which induce shear failure on fracture or fault planes; ii) local

stresses generated by the migration of magma in the crust; iii) localstresses due to the ination/deation of the volcano edice.

Different kinds of activity have characterized the eruptive activity

at Mt. Etna over the past decade. Two major eruptions, characterized

by very intense explosive activity, took place in 2001 and 20022003

in the southern and northeastern anks of the volcano, producing

~30 106 m3 and ~50 106 m3 of lava/tephra (dense rock equivalent,

DRE)(e.g.Allard et al., 2006). Successively, after about 20 months of qui-

escence,on 7 September2004an eruption took placeat twoventswithin

Valle del Bove, emitting essentially degassed magma (~40 106 m3

DRE; Burton et al., 2005; Allard et al., 2006). After a 15-month-long

period, mainly characterized by degassing, the eruptive activity re-

sumed in late 2006 with strombolian activity, lava fountaining andlava overows, producing ~25 106 m3 of lava/tephra (DRE;Behncke

et al., 2009). After 7 lava fountain episodes, the last eruption started

on 13May 2008 froman eruptivessure that openedeast of thesummit

area (e.g.Cannata et al., 2009; Bonaccorso et al., 2011). This eruption,

ending on 6 July 2009, was characterized by botheffusive and explosive

activities (Bonaccorso et al., 2011; Cannata et al., 2011) and emitted

~77 106 m3 of lava/tephra (DRE;Neri et al., 2011).

The aim of this work is to detect and characterize the multiplets

taking place at Mt. Etna during 19992009, as well as to investigate

the dynamics of the seismogenic structures generating them and in-

teraction processes between eruptions and particular seismogenic

structures. This is therst work dealing with such a complete dataset

of VT earthquakes with the purpose not only of investigating Mt. Etna

multiplets, butalso, more generally, of studying thebehavior of VT earth-quakes in volcanic areas as well as eruptionearthquake interactions.

2. Data analysis

2.1. Seismic network

The seismic network is made up of ~100stations, located in a wide

area, comprising eastern Sicily, the Aeolian Islands and southern

Calabria, most of which are clustered in the Etnean region (Fig. 1a).

This network was managed by Istituto Internazionale di Vulcanologia

and Sistema Poseidon until 2001, andfrom then on by Istituto Nazionale

di Geosica e Vulcanologia, Osservatorio Etneo Sezione di Catania

(INGV-CT). From 1999 to 2009, the seismic network was progressively

improved both by the installation of a higher number of stations and

by the replacement of analog, one-component, short-period (1 s) seis-

mometers with digital three-component, broadband (40 s) seismome-

ters. This means that not only were all the stations not equipped with

the same seismometers, but also the sensor equipping a single station

changed over time.

2.2. VT earthquake dataset

During the period August 1999December 2009, the number of

earthquakes located was equal to 6464. Their time distribution and

cumulative seismic strain release as well as the main eruptive periods

are reported inFig. 2a. The average duration magnitude (Md) is equal

to 1.9 with minimum and maximum values of 0.5 and 4.4, respectively

(Fig. 2b).

The locations of these VT earthquakes were obtained from the cat-alog compiled byGruppo Analisi Dati Sismici (2012), belonging to

INGV-CT. Such locations were calculated by using HYPOELLIPSE algo-

rithm (Lahr, 1999) and the 1D velocity model ofHirn et al. (1991),

modied as reported byPatan et al. (1994). The average values of

gap, RMS, horizontal and vertical errors (136, 0.14 s, 0.78 km and

0.67 km, respectively) testify the high quality of such locations. The

space distribution of VT earthquakes is plotted in the map and section

in Fig. 3a. Most of the VTs are located at shallow depth (b7 km b.s.l.) in

the eastern ank of the volcano. On the other hand, the deep seismicity

(focal depth > 15 km b.s.l.) mainly affects the western ank.

2.3. Waveform classication method

We looked for repeating VTs occurring during 19992009 and be-longing to the dataset shown in Section 2.2. We used waveforms

recorded by thevertical componentof 10 stations (Fig. 1b), characterized

Fig. 1.(a) Map of Sicily and southern Calabria; the triangles indicate the stations used

for analytical location, and, in particular, the white triangles indicate the ten stationsalso used for the waveform classication. (b) Schematic structural map of Mt. Etna

volcano (redrawn from Neri et al., 2009) with the location of the ten stations used

for the waveform classication. SEC indicates South East Crater.

1224 A. Cannata et al. / Gondwana Research 24 (2013) 12231236

8/13/2019 Cannata Et Al., 2013 GR

4/15


by a long time recording period and showing a fairly good coverage

of the Mt. Etna area. The signals were band-pass ltered from 1 to

20 Hz. The 1-Hz high-pass lter was applied because, as mentioned

in Section 2.1, during the analyzed 10-year period the stations

were equipped with different sensors, characterized by distinct in-

strument response especially at low frequencies (b1 Hz). Further-

more, anthropic high frequency noise at the stations closest to the

towns on the Mt. Etna anks made the use of 20-Hz low-pass lter

necessary. 5-second signal windows, starting 0.5 s before the P-wave ar-

rival time, were extracted. Then, a cross correlation value was calculated

for each VT window pair. In particular, the windows were also shifted

one with respect to the other one (maximum shift equal to 1 s) to nd

the best alignment and then the highest cross correlation value. Once a

cross correlation matrix for each station was obtained, the method of

Green and Neuberg (2006)was applied to extract families of VTs withsimilar waveforms separately for each station.

A critical point of this method, as well as of all the methods based

on the cross correlation coefcient, is the choice of thecross correlation

threshold, on which the classication results heavily depend. To try to

choose such a value properly, we visualized the histograms showing

the number of VT pairs versus the values of cross correlation for all

the chosen stations (Fig. 4a). Most of the values range between 0.1

and 0.3 and showed a normal distribution as expected for random

signals(Fig. 4a). However, a small percentage of VTs, withhigher corre-

lation coefcient and then similar waveforms, deviates from the normal

distribution (Fig. 4b). As suggested by other authors (Maurer and

Deichmann, 1995; Ferretti et al., 2005; Thelen et al., 2011), the range

of acceptable cross correlation values starts where the histogram

shows a suddenattering and deviates from a pure normal distribution.Thus, the cross correlation threshold was xed to 0.8. To verify such a

value, we used the VT locations shown inSection 2.2. Then, following

Peng and Ben-Zion (2005), contour plotsin Fig. 4c,d, showing the num-

ber of VT pairs with a given value of cross correlation coefcient and hy-

pocentral distance, were drawn. Unlike the plot inFig. 4c, the one in

Fig. 4d is normalized, that is, each column, containing the number of

VT pairs with a givensmall range of crosscorrelation values, wasdivid-

ed by its maximum value.Fig. 4c shows that most of the VT pairs are

characterized by cross correlation values and hypocentral distances

ranging in 0.10.3 and 515 km, respectively. Fig. 4d highlights that

the event pairs with high cross correlation values are mostly located

very close to each other. For instance, a cross correlation value of 0.8,

which as aforementioned was chosen as the threshold, is good to reli-

ably isolate the event pairs located very close to each other and can

also be considered fairly restrictive for most of the used stations.

Once the cross correlation threshold was xed, the classication

procedure was performed separately for each station. In this methodno overlap was allowed between clusters; indeed, an event, assigned

to a family, is removed from the cross correlation matrix. Successively, to

merge the results of the 10 classications (obtained by the 10 stations),

we used the following criterion: if the event a belongs to the family

1at the station STA1and to the family 2at the station STA2, the

families 1 and 2 are unied into a single family. This procedure

might seeminsufciently restrictive to create familiescontaining repeat-

ing events. However,it was considered reliablefor the following reasons:

i) thewindow length used in thecross correlation analysis is equalto 5 s,

that is, twice the window length suggested in Schaff et al. (2004), and,in

most cases, due to the short station-hypocenter distance, it is enough to

include both P and S phases; ii) the lack of bridge events, used in the

bridging or open clustering technique (Aster and Scott, 1993) and not

in the used Green and Neuberg (2006)method, requires that all theevents belonging to a family in the single station classication have a

cross correlation coefcient with the stacked event above the threshold;

Fig. 2.(a) Time distribution of the VT earthquakes at Mt. Etna during 19992009 (histogram) and their cumulative seismic strain release curve (gray line). The red areas show the

main eruptive periods. (b) Histogram showing the number of VT earthquakes versus their duration magnitude. The seismic energy (E) was computed using the following equation:

logE(erg) = 9.9 + 1.9 M 0.024 M2 (Richter, 1958).

1225A. Cannata et al. / Gondwana Research 24 (2013) 12231236

8/13/2019 Cannata Et Al., 2013 GR

5/15


iii) the chosen cross correlation threshold is rather higher than the min-

imum value separating the similar events closeto each other from all the

remaining ones (Fig. 4d). Finally, this criterion was considered suited to

our dataset because of the long analyzed time period and consequently

the possiblelow percentage oftemporal coverage of the single stations.

After applying such a method, a detailed visual checkout was

performed both to verify the similarity of the events belonging to a

single cluster and to remove glitch-type clusters.

2.4. Waveform classication results

We found 735families thattotal to 2479VT earthquakes, representing

~38% of all the VT earthquake dataset. The number of VT earthquakes

making up the families ranges from 2 to 23 (Fig. 5a). Over 70% of the

families comprise 2 (doublets) or 3 (triplets) VT earthquakes, and

only 20 families of more than 10 events. For instance,Fig. 6shows the

waveformsof theVT earthquakes in thefamily198 recorded by the ver-

tical component of EMFO station. Also the occurrence lifetime is highly

variable, ranging from minutes to ten years (thewholeanalyzed period;

Fig. 5b). Therefore, following Igarashi et al. (2003) and Chen et al.

(2009), we can recognize burst-type and nonburst-type families.

The former are made up of VT earthquakes taking place in a short

time period (b1 year in this work), while the latter of VTs spread overa longer interval. It is worth noting that more than half of the families

have a lifetime shorter than 0.5 day and only ~ 10% longer than 1 year.

On the basis of these results and the above mentioned nomenclature,

most of the detected families can be considered burst-type. Then, we

investigated the recurrence time (hereafter referred to as Tr) of the VT

earthquakes and found that most of the events are characterized by re-

currence times shorter than 5 days (Fig. 5c). A parameter used to quan-

tify the variability of some features of the VT families is the coefcient of

variation (hereafter referred to as COV; e.g. Li et al., 2011) given by the

standard deviation divided by the mean value. It was calculated for Tr

and Md of all the families and plotted in Fig. 5d and e, respectively.Concerning Tr, COV equal to 0 implies perfect periodicity, COV close

to 0 quasi-periodicity, COV = 1 Poissonian recurrence, that means

unpredictability, and COV > 1 temporal clustering. As shown in Fig. 5d,

most of the families are characterized by COV values greater than 1

and then by temporal clustering. On the other hand, COV calculated on

Md shows values lowerthan 0.5,suggesting that eachfamily is character-

ized by VT earthquakes with similar Mdvalues (Fig. 5e).

2.5. Families with the longest repeatability over time

We focused on the families characterized by the longest repeat-

ability over time, that is, high number of events and long lifetime. In

particular, we considered a number of events higher than 6 and a life-

time longer than 1 year, and found 16 families. In Fig. 7the locationsof the VT earthquakes makingup these families are plotted andgrouped

into 5 groups. Group i comprises 2 families containing relatively deep

Fig. 3.(a) Epicentral map and WE cross-section of VT earthquakes occurring during 19992009, located by HYPOELLIPSE algorithm. (b) Epicentral map and cross-section of VT

earthquakes, belonging to the families detected in this work, located by HypoDD algorithm. The gray lines in the top plot are elevation contours at 500-m intervals. The contour

labels in the maps indicate the contour altitude in m a.s.l.


8/13/2019 Cannata Et Al., 2013 GR

6/15


VT earthquakes (depth 1015 km b.s.l.) with epicenters in the southern

ank of the volcano (called the South Rift). Groups ii, iii and iv,

made up of 4, 2 and 1 families, respectively, are located in the eastern

ank at depth 05, 310 and 810 km b.s.l., respectively. Finally,

group vcontains 7 families and is located at shallow depth (b5 km

b.s.l.) in the north-eastern sector of the volcano. This zone is affected by

an important structure called the Pernicana Fault System (e.g. Alparone

et al., in press).

InFigs. 8 and 9a the occurrence times of the events in these 5

groups are reported. While the rst 4 groups contain families mainly

taking place in a single swarm and another 12 isolated events, the

group v comprises families more spread out over the analyzed period.

This difference reects on the higher COV values, calculated on the Tr

(indicated by COV in Fig. 8 and ranging from 1.73 to 3.60), of the groupsiiv thanthe group v (indicated by COVtot in Fig. 9a and comprised

between 0.55 and 1.51).

Another interesting feature of the group vis the clear increasing

trend of the Tr, visible for almost all the families, and then the de-

creasing trend of occurrence rate (Fig. 9a). In particular, the highest

occurrence rate was observed during 20022003, at the same timeand right after an important eruption, accompanied by an intense

ank dynamics, took place (Acocella et al., 2003). In order to investi-

gate this behavior we compared COVtot values, calculated on the

whole 19992009 period, and COVpost, on the time span interval fol-

lowing the 20022003 eruption (in particular July 2003December

2009). All the 7 families, making up group v, showed signicantly

lower values of COVpostthan COVtot. In four cases (families 2, 3, 9 and

207) COVpostproved much lower than 1, suggesting that such families

after the ank eruption phases were characterized by a quasi-periodic

behavior.

Further, following the method ofNadeau and McEvilly (1999), we

determined the slip related to these specic VT families. The assump-

tion behind this method is that a repeating earthquake sequence is

caused by repeated ruptures of small asperities surrounded by a sta-bly sliding area (e.g. Uchida et al., 2003). In view of this, the used

equation, relating the seismic moment (M0expressed in dyne cm)

with the slip (diin cm) and proposed byNadeau and Johnson (1998),

is the following one:

di 102:36

M00:17

: 1

The coefcients 2.36 and 0.17 were empirically derived from

earthquake and geodetic data at Parkeld area, but were also used

to infer slip rate in other regions such as Japan subduction zone

(Matsuzawa et al., 2002; Igarashi et al., 2003; Uchida et al., 2003;

Matsuzawa et al., 2004; Uchida et al., 2006; Kimura et al., 2006;

Yamashita et al., 2012), Chihshang fault (eastern Taiwan;Chen et al.,

2008) and northern Longitudinal Valley fault (eastern Taiwan; Rau

et al., 2007). To calculate the seismic moment from Mdwe used the

following equation (Patan et al., 1993):

log M0 17:8 1:9 0:9 0:1 Md: 2

Similarly to the occurrence rate of the VT earthquakes, also the cu-

mulative slip rate did not remain steady during the analyzed period,

but showed an acceleration during 20022003 (Fig. 9c). In particular,the average slip rate of the 7 families changed from 47 11 cm/year

(calculated from 1 July 2002 to 1 July 2003) to 11 3 cm/year

(calculated from 1 July 2003 to 1 July 2009). After the high slip rate

period, the families 2, 3, 9 and 207 exhibited an almost constant slip

rate. To reliably estimate the slip along a fault patch, it is necessary

that all the generated VT earthquakes are detected and properly classi-

ed. Concerning the detection issue, the catalog completeness threshold

in theEtneanarea is equal to 1.5during the10-year analyzed period and

in the interval 20052010 it was equal to 1.3 (Alparone et al., 2010).

Since 75 out of 77 earthquakes belonging to the considered families

have magnitude higher than 1.5 (Fig. 9b; the 2 earthquakes with

Md= 1.3 took place in 2009), the dataset can be considered fairly com-

plete during the analyzed period. Further, since the completeness

threshold gradually decreased from 2005, the observed decreasingtrend of the slip rate can truly be considered reliable. Regarding the

classication problem, the used multi-station method signicantly

Fig. 4.(a,b) Number of pairs of VT earthquakes versus the cross correlation coefcient at the 10 stations used for the waveform classi cation. (c) Contour plots showing the number

of VT pairs with a given value of cross correlation coefcient and hypocentral distance at the 10 stations used for the waveform classication. (d) Normalized contour plots showing

thenumberof VTpairs witha givenvalue ofcrosscorrelation coefcient and hypocentraldistance at the 10 stations usedfor the waveform classication (see Section 2.3 fordetails). The

red dashed lines indicate the chosen cross correlation threshold.


8/13/2019 Cannata Et Al., 2013 GR

7/15


Fig. 5.(a) Histogram showing the number of families with a given number of VT earthquakes. (b) Histogram showing the number of VT earthquake families with a given lifetime.

(c) Histogram of recurrence time (Tr) for VT earthquake families. (d) COV in Tr for VT earthquake families. (e) COV in Mdfor VT earthquake families.


8/13/2019 Cannata Et Al., 2013 GR

8/15


reduces the probability of undetected/missing events. Finally, the max-

imum cumulative slips, calculated by this method for families 1 and 3

(~160 and 150 cm, respectively), have very similar values to the slip

along the Pernicana fault obtained by GPS and EDM measurements

(~140 cm during 20022005; Palano et al., 2006; Bonforte et al., 2007).

2.6. Fault plane solutions

We calculated the focal plane solutions (FPSs) using the FPFIT

algorithm (Reasenberg and Oppenheimer, 1985) and veried that the

earthquakes belonging to the same family have similar fault plane ori-

entations. Lower-hemisphere, equal area projection was used to plot

rst motion data and evaluate nodal planes and orientation of the

main strain axis. The following selection criteria were used: number

of polarities >8, number of polarity discrepancies b20%, focal plane

uncertainty b20, and unique unambiguous solutions. For the events

belonging to groups i, ii, iv, and vwe considered the overlap of

individualplanesof singlesolutions (Fig. 7). The mean and mode values

associated to the errors of nodal plane directions are 8 and 5, respec-

tively. These low values, together with the used selection criteria,assured a good reliability of the focal solutions. The nodal plane direc-

tions ranged between 6 and 67. Generally, the nodal planes overlap

along two preferential directions one of which often coincides with a

structural trend known from literature data.

In group i, located in the southern sector of the volcano called

South Rift (Fig. 1), we found a strong similarity between all the FPSs.

Forthis group,the epicentral distributiondoes notdene a clear align-

ment and therefore it is not possible to identify a direction associated

to one of the nodal planes(Fig. 7). Ingroup ii, located in the Valle del

Bove (Fig. 1b), between the two directions identied by nodal planes

(~EW and NNESSW)the ~EW directionseemsto bettert the epi-

central distribution. In literature there are no data about similar struc-

tural trends in this area. In group iii, located in the eastern ank of

the volcano, since it was not possible to calculate the individual FPSdue to the low energy of earthquakes (maximum magnitude is equal

to 2.2), we calculated a composite mechanism. This focal solution,

together with the earthquake epicentral distribution (Fig. 7), allowed

the distinguishing of the probable fault plane (NWSE), that could be

associated with theTimpeFault System(Fig.1b).In group iv, located

close to the Ioniancoastline, there is only one family with earthquakes

showing similar kinematics, and in particular the ~EW direction is

likely more consistent with the earthquake spatial distribution (Fig. 7).

Concerning group v, the nodal planes overlap is not as clear. This

could be ascribed to the heterogeneity of the kinematics present along

the Pernicana Fault System (Fig. 1b;Alparone et al., in press), as well

as to the extreme shallowness of earthquakes (Azzaro et al., 1998;Alparone et al., in press), and then to the inadequacy of the velocity

model for the surface layers leading to high instability in computing

FPSs. However, a peculiar feature linking FPSs of this group is the evi-

dence of a nodal plane approximately with EW direction, in agreement

with the epicentral distribution (Fig. 7) and surface structural alignment.

2.7. Time relation among VT earthquake families

We investigated potential relationships among the occurrencetimes

of the VT earthquakes belonging to different families by performing the

following analysis. A given couple of VT families was taken into account.

If there are at least two different 10-day long windows (shifting from

1999 to 2009) within which an earthquake per family takes place, the

two families are consideredtime-related. After analyzing all the possi-ble couples of families, we found that several families can be considered

time-related to each other. In spite of the applied 10-day long windows,

Fig. 6.Waveforms of VT earthquakes, making up the family 198, recorded at the vertical

component of EMFO station. The bottom trace is the stacked waveform.

Fig. 7.Map (top) and section (bottom) of Mt. Etna with the spatial distribution of the

ve identied groups(iv). Forgroups i, ii, iv, and v theoverlapsof thenodal

planes are reported in various colors. For the group iiia composite focal mechanism(in white and red) was calculated. The nodal planes that are roughly coincident with

the epicentral distribution are marked in black.


8/13/2019 Cannata Et Al., 2013 GR

9/15


it is worth noting that ~70% of the detected earthquake couples, each of

which is made up of events that are time-related to each other and

belonging to two different families, shows a time delay shorter

than 1 day. Then, three different sets of time-related families were

distinguished: a) Pernicana Fault System set; b) Pernicana and

time-related structures set; c) other sets located in the southern

and eastern anks of the volcano. Focusing on the rst one, it was

highlighted that many families, whose VT earthquakes were located

in the Pernicana fault area, are closely time-related to each other

(Fig. 10a). For instance, on 25 August 2009, earthquakes belonging to

nine differentfamilies tookplace. There are also cases when the familiesof earthquakes located in the Pernicana fault area are related with rela-

tively distant families located ~5 km east of the summit area (Fig. 10b).

Finally, the third set comprises several sub-sets each of which made

up of two or three families that are time-related to each other (each

subset is characterized by a different color in Fig. 10c). Also in this

case there are time-related families close to each other, such as fam-

ilies 7, 185 and 464 (yellow dots in Fig. 10c), as well as relatively

distant ones from each other, such as 376 and 568 (light green dots

inFig. 10c).

2.8. HypoDD relocation

The 2479 VT earthquakes, making up the735 families, were relocatedusing HypoDD, the double-differencing algorithm ofWaldhauser and

Ellsworth (2000) and Waldhauser (2001). As stated by Waldhauser

Fig. 8.Times of occurrence of the VT earthquakes in the families making up groups iiv(see text for details).


8/13/2019 Cannata Et Al., 2013 GR

10/15


Fig. 9.Times of occurrence, magnitude and cumulative slip of the VT earthquakes belonging to seven families located in the Pernicana area and contained in the group v.


8/13/2019 Cannata Et Al., 2013 GR

11/15



8/13/2019 Cannata Et Al., 2013 GR

12/15


(2001), this technique takes advantageof thefactthat if thehypocen-

tral separation between two earthquakes is small compared to the

event-station distance and the scale length of the velocity heterogeneity,

then the ray paths between the source region and a common station are

similar. Thus, the travel time difference for two events observed at one

station can be attributed to the spatial offset between the events with

high accuracy.The data used to perform the HypoDD relocation consisted of both

catalog and cross correlation differential times. Concerning the former,

the 30,502 P-phase and 6663 S-phase picks at 103 stations allowed cal-

culating 460,888 P and 60,173 S catalog differential times. To get these

differential times the maximum event-station distance was set as

200 km, and the maximum hypocentral separation as 4 km. The cross

correlation differential times (35,182 P and 5058 S) were computed

by comparing events belonging to the same cluster. Because of the

high number of VT earthquakes the conjugate gradient method (LSQR)

was used. The minimum number of catalog and cross correlation obser-

vations were set as 7 and 6, respectively. As suggested by Waldhauser

(2001), to properly combine thetwo datasets of differential times,strong

weights were given to thecatalogdata in theinitialiterations, andalmost

exclusively the cross correlation data were used in the later iterations. Inparticular, we apply the weight scheme shown byJohnson et al. (2010).

Using such a technique 999 earthquakes, constituting about 40% of the

dataset of clustered events, were located. The high number of discarded

earthquakes was due to both the presence of very shallow events

(airquakes) and to the paucity of picked phases per earthquake, espe-

cially during the very rst years of the analyzed period. The mean errors

of the HYPOELLIPSE locations of the VT earthquakes making up the fam-

ilies are 0.78 km and 0.67 km horizontally and vertically, respectively.

The average error, estimated by HypoDD, is ~20 m in both horizontal

and vertical directions. However, since the LSQR method has the draw-

back of underestimating the errors, to verify the reliability of the error

evaluation, 10 VT earthquake families were located by singular value

decomposition technique. The SVD error values were of the order of

20 m, and then very similar to the LSQR errors.

The results, reported inFig. 3b, show different rock volumes char-

acterized by clustered seismicity. For instance, in the eastern ank of

the volcano at 05 km b.s.l. there are areas with high seismicity,

already recognizable in plots with HYPOELLIPSE locations (Fig. 3a)

and already shown in other papers (e.g. Alparone et al., 2011). Another

volume with clustered seismicity is located in the southernank at shal-

low depth (1 to 2 km b.s.l.). Finally,thereare also deeperrockvolumes

with high seismicity, such as in the southernank at1015 km b.s.l. and

in the westernank at 2025 km b.s.l.

3. Discussion

We detected and analyzed the repeating earthquakes occurring at

Mt. Etna during 19992009. The number of families found was 735,

amounting to 2479 VT earthquakes. It represents ~38% of all theanalyzed VT earthquakes. The percentage of similar event clusters is

rather variable in literature, just as the methods used to detect them

are different. For instance, Thelen et al. (2011) analyzing repeating seis-

micity at Mount St. Helens and Bezymianny showed how the multiplet

proportion of total seismicity is very variable (ranging from 10 up to

90% of total seismicity), depending on the volcano in question, as well

as on the time interval and volcanic activity. Buurman and West

(2010), investigating volcano seismicity preceding and accompanying

the 2006 eruption of Augustine Volcano, found that the vast majority

of earthquakes during this eruption have unique waveforms. Buurman

et al. (2012), presenting an overview of the seismic activity associated

with the 2009 eruption of Redoubt Volcano, found that most of the

analyzed seismic swarms are mainly made up of repeating events clus-

tered in a few families. The percentage we found in this work suggests

that a signicant portion of the Mt. Etna seismicity is spatially clustered

and mainly affects a limited number of structures.

The analysis of lifetime, Tr and number of VTs of all the detected

families shows that most of them can be considered burst-type,

namely they take place in a short time interval (more than half of thefamilies have a lifetime shorter than 0.5 day), and contain a small num-

ber of events (over 70% of the families are made up of 2 or 3 events)

(Fig. 5ac). As stated bySchaff and Richards (2011), the occurrence ofburst-type families cannot be explained by a temporally constant tec-

tonic loading. Indeed, since the analyzed earthquakes take place in a

volcanic area, the rocks are affected not only by tectonic stresses related

to fairly steady regional stresselds but also by local stresses, caused by

the volcano, such as magma batch intrusions/movements and gravita-

tional loading (e.g. Patan et al., 2004; McNutt, 2005). These local

stresses, that at volcanoes can be abruptly variable not only over time

but also over space, can trigger seismicity by different mechanisms

such as static stress transfer (e.g.Gresta et al., 2005). Moreover, other

phenomena related to uid circulation can promote rock fracturing in

a volcanic areasuch as: i) an increase of porepressure, therebyreducingthe effective stress; ii) alteration of rock to secondary minerals, includ-

ing clays, thus reducing the shear stress required to initiate fracturing;

and iii) local gradients of temperature in the rocks, due to hot uids,

and then thermal forces (e.g.Moran et al., 2000; Cannata et al., 2012).

All these factors can account for the very high number ofburst-type

families at Mt. Etna.

The fact that most of the earthquakes are not caused by the steady

regional stress eld but by different factors that can be very variable

over time, reects on the aperiodic behavior of most families, testied

by the mainly high COV calculated on Tr values (Fig. 5d). Indeed, COV

values larger than 1, suggesting temporal clustering, are expected be-

cause in the volcanic areas seismic energy is mainly released in VT

swarms rather than in tectonic mainshockaftershock sequences

(e.g.McNutt, 2005). In light of this, the loading rate of the structures

generating the VT earthquakes at Mt. Etna can be considered a tempo-

rally variable parameter. VT families showed a much lower variability

in size, as testied by the low COV values calculated on magnitude

(generally b0.5;Fig. 5e). This is partly due to the small size variability

of VT earthquakes at Mt. Etna, where M > 3.0 earthquakes are very

rare (Fig. 2b). The relative difference between COV values calculated

on sizeand recurrence time was also observed by Li et al. (2011) analyz-

ing repeating microearthquakes occurring along the Longmen Shan

fault zone.

On the other hand, ~10% of the families, characterized by a life-

time longer than 1 year, can be considered nonburst-type. Sixteen

families belonging to this ~10% are characterized by a number of

events higher than 6. Based on their location, these families were

grouped into 5 groups related to the activity of seismogenic systems

located in the southern, eastern and north-easternanks of the volca-no (Fig. 7). In particular, groups iii and v are likely related to

well-known structural alignments recognizable at the surface,

Timpe and Pernicana Fault Systems, respectively (Figs. 1 and 7).

The other three groups are not associated to any fault visible at the

surface. Moreover, families belonging to groups iiv show

swarm-like occurrences, that, as aforementioned, are typical of volca-

nic areas.

Unlike the groups iiv, group v shows a more tectonic

behavior. Indeed, with the exception of the rst event of family 1

(taking place in 1999), the other events of this group were observed

for therst time in 20022003, and fromthentheywerespread over al-

most the entire remaininganalyzed period,with no evident swarm-like

Fig. 10.Stacking, time distribution and location of VT earthquake families time-related to each other. In particular, all the families shown in (a) and (b) are time-related to each

other,while only the families plotted with the same color in (c) are time-related to each other.


8/13/2019 Cannata Et Al., 2013 GR

13/15


behavior. Assuming that a repeating earthquake sequence is caused by

the repeated ruptures of small asperities surrounded by stably sliding

areas and following the approach ofNadeau and Johnson (1998), the

slips related to the VT families of group vwere calculated. It is clear

how both occurrence and slip rates do not remain constant but vary

over time. Indeed, after an initial phase of high occurrence and slip

rates, coinciding and following the Mt. Etna 2002

2003 eruption,these parameters decreased. In particular, the slip rate changed from

46 11 cm/year (calculated in the interval 1 July 20021 July 2003)

to 11 3 cm/year (in the interval 1 July 20031 July 2009). Further,

after the initial phase 4 families were characterized by a quasi-periodic

behavior with a COV calculated on the Tr values lower than 0.4. Varia-

tions in time of repeating earthquakes Tr and/or related slip rate have

sometimes been described in literature and have generally been attrib-

uted to the occurrence of strong earthquakes nearby the repeating

earthquake sources (e.g.Peng and Ben-Zion, 2006; Templeton et al.,

2008; Chen et al., 2009; Lengline and Marsan, 2009; Chen et al.,

2010). For instance,Lengline and Marsan (2009) interpreted the var-

iations observed in Tr of repeating earthquakes, taking place along a

~75 km portion of San Andreas Fault after the M = 6.0 Parkeld

earthquake, as resulting from the coseismic stress

eld.Chen et al.(2010) observed that Tr of repeating earthquakes subsequent to

nearby M = 4.05.0 earthquakes reduced, and suggested that dy-

namic triggering, static triggering or transient increase of the creep

rate can be responsible for this variation. In our case, the Tr/slip rate

changes of the group v are closely time-related to the Mt. Etna

20022003 eruption. To interpret such a link between eruptive and

seismic activities, it is necessary to take into account the seismogenic

system generating the earthquakes making up group v, namely the

Pernicana Fault System. This left-lateral strike slip structure trending

EW is considered one of the most active faults in the Etnean area and

plays a very important role in the dynamics of the eastern ank,

which is affected by a continuous seaward sliding (e.g. Neri et al.,

2004; Palano et al., 2006). The movement alongsucha fault system cre-

ates the space and the decompression for the magma rise along the NE

Rift, as well as accommodating the opening of the NE Rift and the east-

ward sliding of the eastern portion of the volcano (e.g. Acocella and

Neri, 2005). Thus, theacceleration in slip along the Pernicana Fault Sys-

tem, observed by repeating earthquakes occurrence and slip rate and

also conrmed by other seismological and ground deformation studies

(e.g.,Palano et al., 2006; Alparone et al., in press), wasdue to the intru-

sion leading to 20022003 eruption. Unlike the 20022003 eruption,

the 20042005, 2006 and 20082009 eruptions did not involve such

an intense ank dynamics and then did not affect recurrence and

slip rate of the analyzed Pernicana VT earthquake families. The differ-

ences in the slip rate of the 7 families of group v(Fig. 9) may simply

be due to the fact that such VTs are caused by the movement not

along a single fault plane, but rather along a fault system. Thus, for

instance the slip of a certain family, representing the movement of the

entire fault system, can be accommodated by the slip of more than afamily in another fault system portion.

Another interesting observation, regarding the Pernicana VT

earthquake families belonging to the groupv, is that all the families,

with the exception of an event belonging to family 1, were observed

for the rst time in 2002. Such an almost complete lack of events be-

fore 2002 could be due to either the incomplete earthquake catalog

during the rst part of the analyzed period or to the very slow slip rate

before the 20022003 eruption. The latter hypothesis is preferred for

the following reasons.Firstly, as suggested by GPS measurements, before

the 20022003 eruption the slip rate along the Pernicana Fault System

ranged between 2 and 3 cm/year (Palano et al., 2006). Then, on the

basis of such slip rate, a M = 2.0 earthquake (M0 and slip equal to

4 1019 dyn cm and 9 cm, respectively; Patan et al., 1993; Nadeau

and Johnson, 1998) accommodates the slip accumulated in about 43years, making the lack of events in a 3 year long interval plausible.

Secondly, the unique event preceding 2002 belongs to the family 1,

which is characterized by the highest both average slip rate and occur-

rence rate. Finally, even family 207, characterized by average magnitude

equal to 3.1 (much higher than the completeness threshold equal to 2.0

during 19891999 and 1.5 during 19992009;Alparone et al., 2010),

does not contain earthquakes preceding 2002. Therefore, the almost

complete lack of Pernicana VT families during 19992002 and their fol-

lowing activity at least up to the end of 2009, highlight the importanceand long duration of the eastern ank instability phase that began with

the 20022003 eruption.

As also observed in Parkeld area byChen et al. (2013), we found

VT earthquake families time-related to each other. In most cases,

these families at Mt.Etna are close to each other (some VT earthquake

families located along the Pernicana Fault System have this behavior)

and hence a mechanismsimilar to thechain reaction model, presented

byMatsuzawa et al. (2004), can be inferred to explain such an interac-

tion. For instance, let us take into account two asperities located close

to each other on the same fault plane, responsible for the generation of

two VT earthquake families. The rupture of the rst asperity, causing

the rst earthquake, generates its afterslip, thus loading to failure the

nearby asperity, and then triggering the second earthquake. According

to another recent study (Chen et al., 2013), the interaction amongclose-by repeating earthquake families can be caused by both static

and dynamic stress transfers. Most of the time-related earthquakes

show time delayshorter than 1 day, but only a handful shorter than

a very few minutes. Thus, followingChen et al. (2013), in most of the

considered cases static stress changes help explain such interaction

among repeatingearthquake families, while delayed dynamic stress trig-

gering cannot be ruled out.

However, also a different mechanism can be invoked. The VT

earthquake pairs that are time-related to each other can be consid-

ered not only as linked by causeeffect relations, but alternatively as

both effects of an external cause, possibly the dynamics of a volcano

sector. Such second mechanism may also be responsible for the VT

earthquake pairs that are time-related and distant to each other. In

this case, the movement of a volcano sector, for example triggered

by magma batch intrusion, can contemporaneously lead to the failure

of different structures that are even distant to each other. For instance,

taking into account the aforementioned VTs located at Pernicana fault,

related to VTs located ~5 km east of the summit area (Fig. 10b), their re-

lation couldbe due to themovement of the eastern ank of the volcano,

whose portions are bounded by several structures located at relatively

shallow depth in such a region (e.g.Neri et al., 2004).

4. Concluding remarks

The repeating volcano-tectonic earthquakes, taking place during

19992009 at Mt. Etna volcano, were detected and studied. The fol-

lowing points summarize the main conclusions:

i) We found 735 families that amount to 2479 VT earthquakes,representing ~38% of all the analyzed VT earthquakes.

ii) Most of the detected families can be considered burst-type,

and hence their origin cannot be explained by the fairly steady

regional stress eld but rather by temporally variable local

stresses, caused by the volcano, such as magma batch intrusions/

movements and gravitational loading. Such evidence empirically

demonstrates the different behavior of the seismogenic structures

in volcanic areas from those in tectonic areas.

iii) Theve groups of VT families, characterized by the longest re-

peatability over time, were located in the southern, eastern and

north-eastern anks of the volcano. While two of these were as-

sociated with well-known fault systems recognizable at the sur-

face (Timpe and Pernicana Fault Systems), the others are not

related to any known fault visible at the surface.iv) A group of VTs,madeup of 7 families and located inthe Pernicana

area, shows a more tectonic behavior, namely the events in this


8/13/2019 Cannata Et Al., 2013 GR

14/15


group are spread over almost the entire analyzed period. Both

occurrence and slip rates do not remain constant but vary over

time, and such changes are time-related to the occurrence of the

Mt. Etna 20022003 eruption. This is further evidence of the

close relationship between VT earthquake occurrences and erup-

tive activities in volcanic areas.

Acknowledgments

Gruppo Analisi Dati Sismici of Istituto Nazionale di Geosica e

Vulcanologia, Osservatorio EtneoSezione di Catania, is kindly acknowl-

edged forproviding data, P andS phase picking andlocation information.

Placido Montaltois thanked forhis usefulhelp in performing HypoDDlo-

cations. Ferruccio Ferrari and Salvatore Spampinato are acknowledged

for the very fruitful discussion. We are also indebted to the technicians

of the seismological staff for enabling the acquisition of seismic data.

We are grateful to the Associate Editor and to the Reviewers for their

useful suggestions that greatly improved the paper. We thank Stephen

Conway for revising and improving the English text.

References

Acocella, V., Neri, M., 2005. Structural features of an active strike-slip fault on the slidingank of Mt. E tna (Italy). Journal of Structural Geology 27, 343355.

Acocella, V., Behncke, B., Neri, M., D'Amico, S., 2003. Link between major ank slip anderuptions at Mt. Etna (Italy). Geophysical Research Letters 30.http://dx.doi.org/10.1029/2003GL018642.

Allard, P., Behncke, B., D'Amico, S., Neri, M., Gambino, S., 2006. Mount Etna 19932005:anatomy of an evolving eruptive cycle. Earth-Science Reviews 78, 85114.

Alparone, S., Gambino, S., 2003. High precision locations of multiplets on south-easternankof Mt. Etna (Italy): reconstruction of fault plan geometry. Physics Earth PlanetaryInteriors 135, 281289.

Alparone, S., D'Amico, S., Maiolino, V., Ursino, A., 2010. Sismicit all'Etna dal 1989 al2010: evidenze sull'evoluzione spazio-temporale della attivit sismica. ProceedingsGNGTS, Prato, 2628 October 2010.

Alparone, S.,Barberi, G., Bonforte, A., Maiolino, V., Ursino, A., 2011. Evidence of multiplestrain elds beneath the eastern ank of Mt. Etna volcano (Sicily, Italy) deducedfrom seismic and geodetic data during 20032004. Bulletin of Volcanology 73,869885.http://dx.doi.org/10.1007/s00445-011-0456-1.

Alparone, S., Cocina, O., Gambino, S., Mostaccio, A., Spampinato, S., Tuv, T., Ursino, A.,in press Seismological features of the PernicanaProvenzana Fault System(Mt. Etna, Italy) and implications for the dynamics of northeastern ank of the vol-cano. Journal of Volcanology and Geothermal Research.http://dx.doi.org/10.1016/

j.jvolgeores.2012.03.010.Antolik, M., Nadeau, R., Aster, R., McEvilly, T., 1996. Differential analysis of coda Q using

similar microearthquakes in seismic gaps, part 2: application to seismogramsrecorded by the Parkeld high resolution seismic network. Bulletin of the Seismo-logical Society of America 86, 890910.

Aster, R.C., Scott, J., 1993. Comprehensive characterization of waveform similarity in micro-earthquake data set. Bulletin of the Seismological Society of America 83, 13071314.

Azzaro, R., Branca, S., Giammanco, S., Gurrieri, S., Ras, R., Valenza, M., 1998. New evidencefor the form and extent of the Pernicana Fault System (Mt. Etna) from structural andsoilgas surveying. Journal of Volcanology and Geothermal Research 84, 143152.

Azzaro, R., Branca S., Gwinner, K., Coltelli, M., 2012. The volcano-tectonic map of Etnavolcano, 1:100.000 scale: an integrated approach based on a morphotectonic anal-ysis from high-resolution DEM constrained by geologic, active faulting andseismotectonic data. Italian Journal of Geosciences 131, 153170.

Behncke, B., Falsaperla, S., Pecora, E., 2009. Complex magma dynamics at Mount Etnarevealed by seismic, thermal, and volcanological data. Journal of Geophysical Re-search 114, B03211.http://dx.doi.org/10.1029/2008JB005882.

Bonaccorso, A., Bonforte, A., Calvari, S., Del Negro, C., Di Grazia, G., Ganci, G., Neri, M.,Vicari, A., Boschi, E., 2011. The initial phases of the 20082009 Mount Etna eruption:a multidisciplinary approach for hazard assessment. Journal of Geophysical Research116. http://dx.doi.org/10.1029/2010JB007906.

Bonforte, A., Branca, S., Palano, M., 2007. Geometric and kinematic variations along theactive Pernicana fault: Implication for the dynamics of Mount Etna NE ank (Italy).

Journal of Volcanology and Geothermal Research 160, 210222.http://dx.doi.org/10.1016/j.jvolgeores.2006.08.009.

Brancato, A., Gresta, S., 2003. High precision relocation of microearthquakes at Mt. Etna(19911993 eruption onset): a tool for better understanding the volcano seismic-ity. Journal of Volcanology and Geothermal Research 124, 219239.

Burton, M.R., Neri, M., Andronico, D., Branca, S., Caltabiano, T., Calvari, S., Corsaro, R.,Del Carlo, P., Lanzafame, G., Lodato, L., Miraglia, L., Salerno, G., Spampinato, L., 2005.Etna 20042005: an archetype for geodynamically-controlled effusive eruptions.Geophysical Research Letters 32, L09303. http://dx.doi.org/10.1029/2005GL022527.

Buurman, H., West, M.E., 2010. Seismic precursors to volcanic explosions during the

2006 eruption of Augustine Volcano. In: Power, J.A., Coombs, M.L., Freymueller,J.T. (Eds.), The 2006 Eruption of Augustine Volcano, Alaska: U.S. Geological SurveyProfessional Paper, 1769, pp. 4157.

Buurman, H., West,M.E., Thompson, G., 2012. The seismicity of the 2009 Redoubt erup-tion. Journal of Volcanology and Geothermal Research. http://dx.doi.org/10.1016/

j.jvolgeores.2012.04.024.Buurman, H., West, M.E., Roman, D.C., in press. Using repeating volcano-tectonic

earthquakes to track post-eruptive activity in theconduitsystemat RedoubtVolcano,Alaska. Geology. http://dx.doi.org/10.1130/G34089.1.

Cannata, A., Montalto, P., Privitera, E., Russo, G., Gresta, S., 2009. Tracking eruptive phe-nomena by infrasound: May 13, 2008 eruption at Mt. Etna. Geophysical Research

Letters 36, L05304.http://dx.doi.org/10.1029/2008GL036738.Cannata, A., Sciotto, M., Spampinato, L., Spina, L., 2011. Insights into explosive activityat closely-spaced eruptive vents by infrasound signals: example of Mt. Etna 2008 erup-tion. Journal of Volcanology and Geothermal Research 208, 111. http://dx.doi.org/10.1016/j.jvolgeores.2011.09.003.

Cannata, A., Diliberto, S., Alparone, S., Gambino, S., Gresta, S., Liotta, M., Madonia, P.,Milluzzo, V.,Aliotta, M., Montalto, P.,2012. Multiparametricapproach in investigatinghydrothermal systems: the case of study of Vulcano (Aeolian Islands, Italy). Pure andApplied Geophysics. http://dx.doi.org/10.1007/s00024-011-0297-z.

Chen, K.H., Nadeau, R.M., Rau, R.J., 2008. Characteristic repeating microearthquakes onan arc-continentcollision boundary theChihshang faultof eastern Taiwan. Earth andPlanetary Science Letters 276, 262272. http://dx.doi.org/10.1016/j.epsl.2008.09.021.

Chen, K.H., Rau, R.J., Hu, J.C., 2009. Variability of repeating earthquake behavior alongthe Longitudinal Valley fault zone of eastern Taiwan. Journal of Geophysical Re-search 114, B05306.http://dx.doi.org/10.1029/2007JB005518.

Chen, K.H., Burgmann, R., Nadeau, R.M., 2010. Triggering effect of M 45 earthquakeson the earthquake cycle of repeating events at Parkeld, California. Bulletin ofthe Seismological Society of America 100, 522531.

Chen, K.H., Burgmann, R., Nadeau, R.M., 2013. Do earthquakes talk to each other?:

triggering and interaction of repeating sequences at Parkeld. Journal of Geo-physical Research. http://dx.doi.org/10.1029/2012JB009486.

Cociani, L., Bean, C.J., Lyon-Caen, H., Pacchiani, F., Deschamps, A., 2010. Coseismic velocityvariations caused by static stress changes associated with the 2001 Mw = 4.3 AgiosIoanis earthquake in the Gulf of Corinth, Greece. Journal of Geophysical Research115, B07313. http://dx.doi.org/10.1029/2009JB006859.

Ferretti, G., Massa, M., Solarino, S., 2005. An improved method for the recognition ofseismic families: application to the GarfagnanaLunigiana area (Italy). Bulletin ofthe Seismological Society of America 95, 19031915.

Ferrucci, F.,Patan, D., 1993. Seismicactivityaccompanyingthe outbreak of the19911993eruption of Mt. Etna (Italy). Journal of Volcanology and Geothermal Research 57,125135.

Geller, R.J., Mueller, C.S., 1980. Four similar earthquakes in central California. GeophysicalResearch Letters 7, 821824.

Green, D., Neuberg, J., 2006. Waveform classication of volcanic low-frequency earth-quake swarms and its implication at Soufrire Hills Volcano, Monserrat. Journalof Volcanology and Geothermal Research 153, 5163.

Gresta, S., Ghisetti, F., Privitera, E., Bonanno, A., 2005. Coupling of eruptions and earth-quakes at Mt. Etna (Sicily, Italy): a case study from the 1981 and 2001 events. Geo-physical Research Letters 32, L05306.

Gruppo Analisi Dati Sismici, 2012. Catalogo dei terremoti della Sicilia Orientale Calabria Meridionale (19992012). INGV, Catania ( http://www.ct.ingv.it/ufs/analisti/catalogolist.php).

Hirn, A.,Nercessian,A., Sapin, M.,Ferrucci,F., Wittlinger,G., 1991. Seismicheterogeneity ofMt Etna: structure and activity. Geophysical Journal International 105, 139153.

Igarashi, T., Matsuzawa, T., Hasegawa, A., 2003. Repeating earthquakes and interplateaseismic slipin the northeastern Japan subduction zone. Journal of Geophysical Re-search 108, 2249.http://dx.doi.org/10.1029/2002JB001920.

Johnson, J.H.,Prejean,S., Savage, M.K.,Townend,J., 2010. Anisotropy,repeating earthquakes,and seismicity associated with the 2008 eruption of Okmok volcano, Alaska. Journal ofGeophysical Research 115, B00B04. http://dx.doi.org/10.1029/2009JB006991.

Kimura, H., Kasahara, K., Igarashi, T., Hirata, N., 2006. Repeating earthquake activities associ-ated with thePhilippine Seaplate subductionin theKanto district, central Japan: a newplate conguration revealed by interplate aseismic slips. Tectonophysics 417, 101118.

Lahr, J.C., 1999. HYPOELLIPSE: a computer program for determining local earthquakehypocentral parameters, magnitude, and rst-motion pattern. U.S. Geological SurveyOpen-leReport 99023(paper(112pp.) and,online(http://greenwood.cr.usgs.gov/

pub/open-le-reports/ofr-99-0023)).Lengline, O., Marsan, D., 2009. Inferring the coseismic and postseismic stress changes

caused by the 2004 Mw = 6 Parkeld earthquake from variations of recurrencetimes of microearthquakes. Journal of Geophysical Research 114, B10303. http://dx.doi.org/10.1029/2008JB006118.

Li, L., Chen, Q., Niu, F., Su, J., 2011. Deep slip rates along the Longmen Shan fault zoneestimated from repeating microearthquakes. Journal of Geophysical Research116, B09310.http://dx.doi.org/10.1029/2011JB008406.

Matsuzawa, T., Igarashi, T., Hasegawa, A., 2002. Characteristic small earthquake se-quence off Sanriku, northeastern Honshu, Japan. Geophysical Research Letters 29(11), 1543.http://dx.doi.org/10.1029/2001GL014632.

Matsuzawa, T., Uchida, N., Igarashi, T., Okada, T., Hasegawa, A., 2004. Repeating earth-quakes and quasi-static slip on the plate boundary east off northern Honshu, Japan.Earth Planets and Space 56, 803811.

Maurer, H., Deichmann, N., 1995. Microearthquake cluster detection based on wave-form similarities, with an application to the western Swiss Alps. Geophysical JournalInternational 123, 588600.

McNutt, S.R., 2005. Volcanic seismology. Annual Review of Earth and Planetary Sciences32, 461491.

Moran, S.C., Zimbelman, D.R., Malone,S.D., 2000. A model forthe magmatic-hydrothermalsystem at Mount Rainier, Washington, from seismic and geochemical observations.Bulletin of Volcanology 61, 425436.


8/13/2019 Cannata Et Al., 2013 GR

15/15


Nadeau, R.M.,Johnson, L.R.,1998. Seismological studies atParkeldVI: momentreleaserates and estimates of source parameters for small repeating earthquake. Bulletinof the Seismological Society of America 88, 790814.

Nadeau, R.M., McEvilly, T.V., 1999. Fault slip rates at depth from recurrence intervals ofrepeating microearthquakes. Science 285, 718721.

Nadeau, R.M., Foxall, W., McEvilly, T.V., 1995. Clustering and periodic recurrence ofmicroearthquakes on the San Andreas fault at Parkeld, California. Science 267,503507.http://dx.doi.org/10.1126/science.267.5197.503.

Nakahara, H., 2004. Correlation distance of waveforms for closely located events I.Implication of the heterogeneous structure around the source region of the 1995Hyogo-Ken Nanbu, Japan, earthquake (Mw = 6.9). Geophysical Journal Interna-tional 157, 12551268.

Neri, M., Acocella, V., Behncke, B., 2004. The role of the Pernicana Fault System in thespreading of Mt. Etna (Italy) during the 20022003 eruption. Bulletin of Volcanology66, 417430.

Neri, M., Casu, F., Acocella, V., Solaro, G., Pepe, S., Berardino, P., Sansosti, E., Caltabiano,T., Lundgren, P., Lanari, R., 2009. Deformation and eruptions at Mt. Etna (Italy): alesson from 15 years of observations. Geophysical Research Letters 36, L02309.http://dx.doi.org/10.1029/2008GL036151.

Neri, M., Acocella, V., Behncke, B., Giammanco, S., Mazzarini, F., Rust, D., 2011. Structuralanalysis of the eruptive ssures at mount Etna (Italy). Annals of Geophysics. http://dx.doi.org/10.4401/ag-5332.

Palano, M., Aloisi, M., Amore, M., Bonforte, A., Calvagna, F., Cantarero, M., Consoli, O.,Consoli, S., Guglielmino, F., Mattia, M., Puglisi, B., Puglisi, G., 2006. Kinematics andstrain analyses of the eastern segment of the Pernicana fault (Mt. Etna, Italy) derivedfrom geodetic techniques (19972005). Annals of Geophysics 49, 11051117.

Patan, D., Ferrucci, F., Gresta, S., 1993. Leggi di scala e parametri di sorgente per

terremoti all'Etna. 12th Annual Conf. GNGTS, pp. 925928 (in Italian).Patan,D., Privitera, E.,Ferrucci,F., Gresta,S., 1994. Seismicactivity leading tothe 199193

eruption of Mt. Etna and its tectonic implications. Acta Vulcanologica 4, 4756.Patan, D., Cocina, O., Falsaperla,S., Privitera, E., Spampinato, S., 2004. Mt. Etna volcano:

a seismological framework. In: Calvari, S., Bonaccorso, A., Coltelli, M., Del Negro, C.,Falsaperla, S. (Eds.), The Mt. Etna Volcano. AGU, Washington, D.C., pp. 147165.

Peng, Z., Ben-Zion, Y., 2005. Spatio-temporal variations of crustal anisotropy from sim-ilar events in aftershocks of the 1999 M7.4zmit and M7.1 Dzce, Turkey, earth-quake sequences. Geophysical Journal International 160, 10271043.

Peng, Z., Ben-Zion, Y., 2006. Temporal changes of shallow seismic velocity around theKaradere-Duzce Branch of the North Anatolian Fault and strong ground motion.Pure and Applied Geophysics 163, 567600.

Rau, R.J., Chen, K.H., Ching, K.E., 2007. Repeating earthquakes and seismic potentialalong the northern Longitudinal Valley fault of eastern Taiwan. Geophysical Re-search Letters 34, L24301.http://dx.doi.org/10.1029/2007GL031622.

Reasenberg, P., Oppenheimer, D., 1985. FPFIT, FPPLOT and FPPAGE: Fortran computerprograms for calculating and displayng earthquakes fault plane solutions. U.S.Geol. Surv. Open File Rep., 85739.

Richter, C.F., 1958. Elementary Seismology.Freeman and Company, San Francisco (768pp.).Rowe, C.A., Thurber, C.H., White, R.A., 2004. Dome growth behaviour at Soufriere Hills

Volcano, Montserrat, revealed by relocation of volcanic event swarms, 19951996.Journal of Volcanology and Geothermal Research 134, 199221.

Scandone, P., Patacca, E., Radoicic, R., Ryan, W.B.F., Cita, M.B., Rawson, M., Chezar, H.,Miller, E., Mckenzie, J., Rossi, S., 1981. Mesozoic and Cenozoic rocks from Malta Es-carpment (Central Mediterranean). American Association of Petroleum GeologistsBulletin 65, 12991319.

Schaff, D.P., Beroza, G.C., 2004. Coseismic and postseismic velocity changes measuredby repeating earthquakes. Journal of Geophysical Research 109, B10302. http://dx.doi.org/10.1029/2004JB003011.

Schaff, D.P., Richards, P.G., 2011. On nding and using repeating seismic events in and

near China. Journal of Geophysical Research 116, B03309. http://dx.doi.org/10.1029/2010JB007895.Schaff, D.P., Bokelmann, G.H.R., Ellsworth, W.L., Zanzerkia, E., Waldhauser, F., Beroza,

G.C., 2004. Optimizing correlation techniques for improved earthquake location.Bulletin of the Seismological Society of America 94 (2), 705721.

Templeton, D.C., Nadeau, R.M., Brgmann, R., 2008. Behavior of repeating earthquakesequences in Central California and the implications for subsurface fault creep. Bul-letin of the Seismological Society of America 98, 5265.

Thelen, W.,Malone,S., West, M., 2011. Multiplets: their behaviorand utility at dacitic andandesitic volcanic centers. Journal of Geophysical Research 116, B08210. http://dx.doi.org/10.1029/2010JB007924.

Torelli, L., Grasso, M., Mazzoldi, G., Peis, D., 1998. Plio-quaternary tectonic evolutionand structure of the Catania foredeep, the northern Hyblean Plateau and the Ionianshelf (SE Sicily). Tectonophysics 298, 209221.

Tsujiura, M., 1983. Characteristic frequencies for earthquake families and their tectonicimplications: evidence from earthquake swarms in the Kanto District, Japan. Pureand Applied Geophysics 121, 574600.

Uchida, N., Matsuzawa, T., Hasegawa, A., 2003. Interplate quasi-static slip off Sanriku,NE Japan, estimated from repeating earthquakes. Geophysical Research Letters

30.http://dx.doi.org/10.1029/2003GL017452.Uchida, N., Matsuzawa, T., Hirahara,S., Hasegawa,A., 2006. Small repeatingearthquakes and

interplate creep aroundthe 2005 Miyagi-oki earthquake (M = 7.2). Earth Planets andSpace 58, 15771580.

Waldhauser, F., 2001. HypoDD: a computer program to compute double-differenceearthquake locations. U.S. Geol. Surv. Open File Rep., 01113.

Waldhauser, F., Ellsworth, W.L., 2000. A double-differenceearthquake locationalgorithm:Method andapplication to thenorthern Haywardfault,California.Bulletin of theSeis-mological Society of America 90, 13531368.

Waldhauser, F., Ellsworth, W.L., Schaff, D.P., Cole, A., 2004. Streaks, multiplets, andholes: high-resolution spatio-temporal behavior of Parkeld seismicity. GeophysicalResearch Letters 31, L18608. http://dx.doi.org/10.1029/2004GL020649.

Yamashita, Y., Shimizu, H., Goto, K., 2012. Small repeating earthquake activity,interplate quasi-static slip, and interplate coupling in the Hyuga-nada, southwestern

Japan subduction zone. Geophysical Research Letters. http://dx.doi.org/10.1029/2012GL051476.

Zaccarelli, L., Pandol, D., Bianco, F., Saccorotti, G., Bean, C.J., Del Pezzo, E., 2009. Tem-poral changes in seismic wave propagation towards the end of the 2002 Mt Etnaeruption. Geophysical Journal International 178, 17791788. http://dx.doi.org/10.1111/j.1365-246X.2009.04219.x.


Documents

Cannata Et Al., 2013 GR