1 1 2 3 Concurrent eruptions 4 at Etna, Stromboli, and Vulcano 5 6 Andrea Billi (1) , Emily E. Brodsky (2) , and Renato Funiciello (1) 7 8 (1) Dipartimento di Scienze Geologiche, Università “Roma Tre”, Rome, Italy 9 (2) Department of Earth Sciences, University of California, Santa Cruz, CA, USA 10 11 12 13 14 15 16 Correspondence: 17 Andrea Billi 18 19 Dipartimento di Scienze Geologiche, 20 Università “Roma Tre” 21 Largo S. L. Murialdo, 1 22 00146, Rome, Italy 23 Tel: +39 0654888016 24 Fax: +39 0654888201 25 Email: [email protected]26 27 28
45
Embed
Concurrent eruptions at Etna, Stromboli, and Vulcanobrodsky/reprints/Billietal_paper.pdf · 2007. 4. 25. · 51 between Etna and the Aeolian volcanoes, whereas Mercalli [12] supposed
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
1
2
3
Concurrent eruptions 4
at Etna, Stromboli, and Vulcano 5
6
Andrea Billi (1), Emily E. Brodsky (2), and Renato Funiciello (1) 7
8 (1) Dipartimento di Scienze Geologiche, Università “Roma Tre”, Rome, Italy 9 (2) Department of Earth Sciences, University of California, Santa Cruz, CA, USA 10 11
12 13
14
15
16
Correspondence: 17
Andrea Billi 18 19
Dipartimento di Scienze Geologiche, 20 Università “Roma Tre” 21 Largo S. L. Murialdo, 1 22
Anecdotes of concurrent eruptions at the Southern Italian volcanoes have persisted for more than 30
2000 years. Here we perform a statistical analysis of the eruption catalog to find that concurrent 31
eruptions among Etna, Stromboli, and Vulcano are much more frequent than would be expected by 32
chance. Based on the frequency of eruptions at each individual volcano, we calculate the expected 33
occurrence rate of concurrent eruptions and compare it to the observed rate. We find that pairs of 34
simultaneous eruptions occur as much as about 13 times more often than expected and triple 35
eruptions occur as much as 26 times more often. All of these concurrencies are statistically 36
significant at the 99% level. 37
38
39
Keywords: Etna, Stromboli, Vulcano, eruption, and volcano 40
3
1. Introduction 41
Interactions between large earthquakes and volcanic systems have excited a great deal of interest 42
in recent years [1-8]. Eruptions triggering other eruptions have been speculated about since at least 43
the 1st Century BC (e.g. Diodorus Siculus in Bibliotheca Historica), but recent systematic studies 44
have been rare [9, 10]. 45
Southern Italy’s abundant volcanoes (Fig. 1) and long historical records make it ideal for the 46
study of concurrent eruptions. A link between Mt Etna and the Aeolian volcanoes was suggested by 47
Diodorus Siculus (in Bibliotheca Historica, 1st Century BC) and Julius Solinus (in Collectanea 48
Rerum Memorabilium, 4th Century AD) wrote that there should have been subterranean conduits 49
that linked and fed these volcanoes. de Dolomieu [11] hypothesized direct feeding relationships 50
between Etna and the Aeolian volcanoes, whereas Mercalli [12] supposed an indirect physical (i.e. 51
dynamic) link between these volcanoes. 52
Statistical evidence for a link between Etna, Stromboli, and Vulcano (Fig. 1) is addressed in this 53
paper. Our aim is to determine whether concurrent activity at these three volcanoes occurs at 54
significantly higher frequencies than expected based on the background rate. We will approach this 55
problem in three different ways. We will first do a statistical study of the cataloged eruptions using 56
a bootstrap probability method. The statistical method has the advantage of being quantitative, but 57
the disadvantage of providing little insight into the plethora of behavior that can fall under the 58
heading of eruption. In order to provide more qualitative information, we will go on to describe in 59
detail the activity during a few anecdotal instances of simultaneous eruptions. We finally discuss a 60
negative observation that rules out the role of large earthquakes in producing the concurrency. 61
62
2. Volcanic setting and activity 63
Mt Etna and the Aeolian Islands (Fig. 1) are located along the active destructive plate boundary 64
between Africa and Eurasia in the central Mediterranean region. In this tectonic framework, the 65
Aeolian volcanoes form an alkaline island arc whose activity started about 1.3 Myr ago and is still 66
4
active on at least three of the seven major islands, namely Lipari (last eruption in 729), Vulcano 67
(last eruptions in 1888-1892), and Stromboli. Hydrothermal activity offshore Panarea has occurred 68
in historical times and in recent years [13]. Stromboli is characterized by a persistent volcanic 69
activity consisting in continuous small explosions occurring at approximately regular intervals of a 70
few minutes (i.e. strombolian activity); however, individual events, such as lava effusions and 71
paroxysmal eruptions, are recorded at Stromboli during recent and historical times [14, 15]. The 72
volcanic activity of Vulcano has been characterized by powerful and impressive explosions (i.e. 73
vulcanian eruptions) alternated with less frequent effusive events [15, 16]. 74
Volcanism in the Etnean area started about 0.5 Myr ago with tholeiitic magmas. The modern 75
volcano formed by a succession of volcanic edifices consisting in alternating pyroclastic and 76
effusive rocks [17]. The volcanic activity of Etna is, in fact, mostly effusive; however, explosive 77
activity has often occurred at the summit craters [18]. Lateral effusive and mildly explosive 78
eruptions have also been frequent at Mt Etna [19]. The source of Etna magmatism is possibly 79
connected with a mantle plume that Montelli et alii [20] imaged above 1000 km using seismic 80
tomography. 81
82
3. Concurrent eruptions 83
In this paper, we investigate the frequency of simultaneous eruptions. As eruptions are often 84
ongoing sequences with poorly defined start dates, we do not study the start of the eruptions. The 85
problem studied here is whether or not multiple ongoing eruptions overlap more often than expected 86
by chance. 87
We can construct a null hypothesis by using the observed record to calculate the expected rate of 88
concurrent eruptions if the volcanoes do not interact. If the probability of volcano A erupting in any 89
given month pA and the probability of volcano B erupting is pB and the two volcanoes are unlinked, 90
statistically independent systems, then the probability of a concurrent eruption pAB is the product 91
pApB. In this paper, we evaluate the null hypothesis that eruptions at each of the volcanoes are 92
5
statistically independent by calculating pApB from time-randomized catalogs and comparing it to the 93
observed frequency of concurrent eruptions. 94
Before evaluating the probabilities, we first limit the catalogs to provide uniform completeness 95
over the study interval. The catalogs are expected to be complete for the largest, most easily 96
observable eruptions, but progressively more incomplete at the smaller sizes. Volcanic eruptions 97
generally follow a power law distribution in sizes with many more small eruptions than large ones 98
as quantified by the Volcano Explosivity Index, VEI [21]. Significant depressions at small VEI 99
relative to the trend at large VEI suggest that there are missing eruptions in the catalog. This method 100
is similar to the standard comparison with the Gutenberg-Richter distribution used to determine the 101
magnitude at which seismic catalogs are complete [22]. Fig. 2 shows that the volcanic catalog used 102
for this region [15] is only complete for VEI ≥ 2. This rule is equally valid for eruptions in the 0-103
1993, 1300-1500, 1500-1993, 1800-1993, and 1800-1900 periods of time (Fig. 2). For this reason, 104
we only use eruptions with VEI ≥ 2 for the analysis. The same threshold has been used previously 105
for global studies of eruption triggering [9]. 106
We use the resulting catalog to estimate the probability of eruption per month and per year of 107
each individual volcano (i.e. Etna, Stromboli, and Vulcano) assuming statistical independence. The 108
simplest estimate of the probability of eruption is the number of time periods with eruptions divided 109
by the total number of time periods studied. The sampling bias of the limited catalog will 110
undoubtedly lead to errors in this estimate. These errors are evaluated using the standard bootstrap 111
method [23], which resamples the data set randomly in order to empirically capture probability 112
distribution (Table 1). Reported error bars are 1 standard deviation on 5000 bootstrap trials. The 113
advantage of bootstrapping is that no probability distribution needs to be assumed. The observed 114
data set is taken as a sampling of the true distribution and, therefore, the distribution of the 115
probability of occurrence is inferred from multiple samplings of the observations [23]. 116
The next step is to multiply the individual probabilities to predict the probability of a concurrent 117
eruption if the systems are independent. This null hypothesis is evaluated by comparing the 118
6
observed frequency of concurrent eruptions to the predicted probabilities (Table 1). In the table, the 119
expected probability is the product of the independently inferred probabilities, e.g. pApB, and the 120
observed probability is the actual fraction of time periods containing concurrent eruptions. Every 121
pair of volcanoes erupts more frequently than would be expected based on the individual rates. 122
Stromboli and Vulcano are more closely correlated to each other than either is to the more distant 123
Mt Etna (Table 1). The most convincing correlation comes from the occurrence rate of triple 124
eruptions, this rate being more than 20 times as often than expected from individual eruptions rates 125
(Table 1). We again used bootstrapping to calculate the significance of the result. We time-126
randomized each volcano catalog separately and then used a bootstrap method to determine the 127
probability distribution of concurrent eruptions in the time-randomized catalogs. We then compared 128
the observed rate of concurrency to the derived empirical probability distribution function and 129
found that the observation would occur by chance less than 0.5% of the time in all cases. Therefore, 130
the null hypothesis can be rejected at the 99.5% confidence level. 131
We limit our analysis to time periods of a month and longer because the exact dates are often 132
poorly constrained in the historical catalog. Nonetheless, it is interesting to investigate whether the 133
eruptions often occur in a particular order. During the 1700-1993 period, for which the eruption 134
dates are more reliable than those of the previous centuries [15], the order of eruption is often 135
Vulcano, Etna, and then Stromboli (Table 2). If only two of the three erupt, the order is often 136
preserved. 137
In order to check whether the statistical analysis is sensitive to the time window used to identify 138
concurrent eruptions, we repeated the analysis for year-long windows (Table 1b). Results from both 139
windows are consistent, which suggests that the result of statistically significant concurrent 140
eruptions is independent of the time scale used. Shorter time windows are not investigated because 141
the start of an eruptive episode often lasts weeks and thus concurrency on a time scale much less 142
than a month is poorly defined. 143
144
7
4. Examples of concurrent eruptions 145
The statistical evidence suggests a correlation between Southern Italian volcanoes; however 146
volcano catalogs are notoriously subjective. Eruptive activity is gradational and activity during a 147
given month may mean a variety of different phenomenon. To fill out this story, we therefore 148
provide narratives of four of the most notable examples of concurrent activity. Two of these 149
episodes fall outside our statistical study period and three involve observations from other Aeolian 150
volcanoes, but we include these stories anyhow to provide examples of the type of phenomena 151
expected. 152
153
4.1. Concurrent activity during 2002-2003 154
The 2002 eruption of Etna started the night between 26 and 27 October 2002 from the summit 155
craters, concurrently with the formation of ground-surface fractures in this area. From the summit 156
craters, fractures propagated toward the south and toward the northeast. Newly generated vents and 157
cones developed along these fractures. For about two months since the onset of the eruption, 158
voluminous emissions of ash occurred. Volcanic ashes fell in the nearby regions of Sicily and 159
Calabria. After about 95 days of almost continuous eruption, on 29 January 2003, Etna ceased most 160
of its effusive and explosive activity. The 2002-2003 eruption of Etna was highly explosive, 161
possibly one of the most explosive of the last 350 years (VEI ≥ 3 [24]). 162
Early in the morning of 3 November 2002, local fishermen observed spots of anomalous 163
seawater boiling and high mortality of fishes offshore Panarea, and perceived smell of sulfurs. In 164
particular, five major sites of gas emissions were found around the islets of Lisca Bianca, Bottaro, 165
and Lisca Nera. At these sites, ascent of gas bubbles with diameter up to 1 m on the sea surface 166
signaled the presence of submarine gas-emissions. With the exception of one site close to Lisca 167
Bianca, these sites were previously unknown as gas-emitting spots. Gas emissions originated from 168
NW- and NE-striking fractures in volcanic rocks lying at depths between 8 and 30 m from the sea 169
surface [25]. Gas emissions offshore Panarea gradually ceased during the first months of 2003 [26]. 170
8
On 28 December 2002, after 17 years of mild although continuous volcanic activity, an intense 171
eruption began at Stromboli. The eruption was preceded by seismic swarms and by an increase in 172
volcanic tremors since 3 November 2002, concurrently with the onset of gas emissions offshore 173
Panarea. Since 28 December 2002, lava flowed from the Stromboli summit crater into the sea along 174
the steep northwest flank. Renewed lava flow occurred since 30 December 2002. The eruption 175
mostly ceased by 5 April 2003, when a powerful explosion occurred at the summit crater of 176
Stromboli [27]. 177
178
4.2. Concurrent activity during 1886-1890 179
From 10 January to 31 March 1886, strombolian activity occurred at Vulcano. On 22 January 180
1886, volcanic explosions occurred at Stromboli. On 18 May 1886, an eruption began at the summit 181
craters of Mt Etna [28]. This eruption was highly explosive (VEI = 3). 182
During January, March, and November 1887, volcanic explosions occurred at Stromboli. On 3 183
August 1888, an eruption began at Vulcano. The eruptive activity of Vulcano ceased in March 184
1990. The 1888-1890 eruption of Vulcano was highly explosive (VEI = 3). On 23 October 1888, 185
volcanic explosions and lava fountains occurred at Stromboli. On 29 November 1888, offshore 186
Vulcano, local fishermen observed boiling seawater and associated fish mortality, and perceived 187
smell of sulfurs [28-30]. 188
189
4.3. Concurrent activity during 1865 190
On 30 January 1865, an explosive (VEI ≥ 2) eruption occurred at Etna [31, 32]. At the onset of 191
the eruption, NE-striking fractures propagated across the summit crater of Mt Etna. Several vents 192
and cones formed along these fractures. The effusive and explosive activity at Mt Etna lasted for 193
about 90 days. Voluminous emissions of ash, which fell for about two months in the nearby regions 194
of Sicily and Calabria, accompanied the eruption from the beginning. During the 1865 eruption of 195
9
Etna, emissions of sulfurous gases occurred offshore Panarea. From 26 January to 2 February 1865, 196
volcanic explosions occurred at Stromboli [30]. 197
198
4.4. Concurrent activity during 126 BC 199
Paulus Orosius, a Latin historian lived in the 5th Century, in his Historiarum Adversus Paganos 200
wrote: “Under the consulate of M. Aemilius Laepidus and L. Orestes, Mt Etna was violently shaken 201
by a powerful tremble and poured out waves of fire globes. The day after, the Lipari Islands (i.e. 202
also known as Aeolian Islands) and nearby the sea reached such a high boiling point that rocks were 203
burnt and broken up. Ships, axes, and wax were carbonized and melted, dead fishes were burnt on 204
the sea surface, and several men, except those who were able to escape, were stifled and their inner 205
organs burnt by breathing”. Strabo, a Greek geographer, and Julius Obsequens, a Latin historian, 206
had reported the same event in documents dated back to the 1st Century BC (Geographia) and to the 207
4th Century (De Prodigiis), respectively. The events reported by the above-cited authors refer to the 208
explosive eruption of Mt Etna started in June 126 BC and to the contemporaneous emissions of gas 209
occurred in the offshore area between Vulcano and Panarea [15]. 210
211
5. Large earthquakes as triggers? 212
An advantage of studying this region is that there exists an extensive pre-instrumental seismic 213
record. The historical chronicles [15, 33] report six large earthquakes (epicentral MCS intensity ≥ 214
IX-X and magnitude possibly ≥ 6.5) in the eastern Sicily (Fig. 1) during historical times. (1) The 4 215
February 1169 earthquake, which hit the eastern Sicily and western Calabria. The epicentral MCS 216
intensity for this event is XI. An eruption from the southern flank of Mt Etna is doubtfully dated 217
back to February 1169. (2) The 10 December 1542 earthquake, which hit the southeastern Sicily 218
with epicentral MCS intensity of IX-X. No concurrent volcanic and seismic phenomena are 219
documented for the Etnean and Aeolian areas. (3) The 11 January 1693 earthquake, which hit the 220
southeastern Sicily with epicentral MCS intensity of XI. No concurrent volcanic and seismic 221
10
phenomena are reported for the Etnean and Aeolian areas except for a doubtful eruption at Mt Etna 222
dated back to 9 January 1693. (4) The 5 March 1823 earthquake, which hit the Patti Gulf along the 223
northeastern coast of Sicily. The maximum MCS intensity recorded is X. No concurrent volcanic 224
and seismic phenomena are reported for the Etnean and Aeolian areas. (5) The 28 December 1908 225
earthquake, which hit the Messina Straits and the surrounding region. The maximum MCS intensity 226
recorded is XI. No concurrent volcanic and seismic phenomena are reported for the Etnean and 227
Aeolian areas. (6) The 15 October 1911 earthquake, which hit the southeastern flank of Mt Etna. 228
The NNW-SSE-elongated epicentral area was characterized by a maximum MCS intensity of X. 229
The earthquake was preceded by an eruption occurred between 10 and 22 September 1911 from the 230
northeastern flank of Mt Etna. This eruption was contemporaneous with an intense seismic activity 231
at Stromboli. 232
Notably, there is no evidence – anecdotal or otherwise – that either local or remote large 233
earthquakes triggered eruptions in the region (see the NEIC Catalog available on-line at 234
http://neic.usgs.gov/neis/epic/ and the INGV Catalog available on-line at http://www.ingv.it/). 235
236
6. Discussion 237
Pairs of simultaneous eruptions occur between about 6 and 13 times more often than expected 238
from the separate rates of each system when analyzed per month, and between about 4 and 6 times 239
more often when analyzed per year (Table 1). Triple eruptions occur about 21 times more often than 240
expected if the systems were independent, when analyzed per month and 26 times more often when 241
analyzed per year. Although the exact value of the increased occurrence depends on the timescale, 242
in all cases the results are statistically significant at the 99.5% level, indicating that concurrency is a 243
robust observation independent of the time scale used (i.e. months and years). 244
The statistics make it clear that there is a significant correlation between volcanoes in the 245
recorded catalog. The key issue in interpreting the results is determining how accurately the catalog 246
reflects the true state of the eruptions. We have partially addressed this issue by limiting the study to 247
11
mid-sized explosive eruptions (VEI ≥ 2), which are unlikely to be missed in a densely populated 248
region and appear to be complete according to a test of the size distribution (Fig. 2). Another 249
important issue is the treatment of ongoing eruptions. Here we exclusively study the overlap time of 250
eruptions rather than trying to pinpoint the eruption start. We also do not attempt to distinguish 251
between eruptive styles or locations within a given volcanic system. This leaves the eruptive 252
triggers poorly constrained, but makes the statistics possible for the historical data. 253
A final, and potentially fatal, problem is reporting bias. After an eruption, observers are more 254
likely to make note of activity on other area volcanoes. This is less likely a problem late in recent 255
years and the VEI distribution suggests that at least most large eruptions (VEI ≥ 2) are recorded. 256
However, it is possible that the concurrence statistics are simply reflecting a trend in the reporting. 257
In that case, the apparent concurrency can be used as a tool to highlight catalog errors. However, 258
given the anecdotal evidence, we think that it is more likely that the signal is real. The hypothesis of 259
reporting bias is untestable in retrospect, but can be easily tested in the future as eruptions continue 260
to develop. We look forward to our predictions of a high incidence of concurrent eruptions being 261
confirmed (or refuted) with the passage of time. 262
The above-discussed statistical and anecdotal evidence indicate links in the activity of Etna, 263
Stromboli, and Vulcano, but the data do not allow an unequivocal identification of a plausible 264
triggering mechanism. Some constraints on the triggering mechanisms can, however, be derived 265
from our results. Previous studies on earthquake-eruption and eruption-eruption triggering 266
mechanisms [10] established that eruptions and volcanic unrest could be triggered by earthquakes 267
through static stresses (i.e. the fault slip induces a permanent deformation in the crust) or dynamic 268
stresses (i.e. transient stresses induced by the passage of seismic waves). The same reasoning can be 269
applied to eruption-eruption triggering [34]. Either the static stress generated by the movement of 270
magma or the shaking generated by the eruptions themselves can trigger adjacent volcanoes. The 271
lack of eruption triggering from the larger earthquakes in the region argues that the eruption-related 272
earthquakes themselves are not the main agent of coupling. More prolonged dynamic stresses 273
12
generated through eruptive processes would be a more viable candidate [34]. We speculate that 274
since Vulcano is often the most explosive of the volcanoes, it likely generates the strongest seismic 275
waves and hence usually begins the sequence. Of course, whether or not a given input successfully 276
triggers an eruption depends on the state of the volcano. We do not expect to observe perfect 277
correlations or a deterministic sequence of eruptions. 278
Alternatively, the concurrent eruptions may be responding to a common, external stimulus. The 279
“superswarm” of earthquakes and eruptions could be a result of either: (1) a dynamic stress from 280
strong, remote earthquakes, or (2) a tectonic stress associated with the N-S convergence between 281
Africa and Eurasia. Although the possibility of remotely triggered eruptions cannot be generally 282
ruled out, this mechanism did not probably activate the concurrent eruptions discussed in this paper 283
because no remote strong earthquakes are known to be chronologically near (i.e. within hours or 284
days) to these phenomena. Both seismological and geodetic evidences show that a N-S tectonic 285
compression is active in the northeastern Sicily [35-37]. Such stress may be periodically relieved by 286
a large earthquake on one segment of the regional fault network [38] or, more often, by 287
superswarms consisting in eruptions and low-magnitude earthquakes where the triggering threshold 288
is low (i.e. in the active volcanic districts). The model of a tectonic stress stimulating occasional 289
“superswarms” in the volcanic districts of Southern Italy revives Mercalli’s [12] original idea of 290
“dynamic relationships” between Etna and the Aeolian volcanoes, and of a “dependence of these 291
volcanoes on the same fracture system”. Mercalli, in fact, claimed that the nature of the 292
relationships between these volcanoes should not have been magmatological [11] but physical and 293
external to them. 294
295
7. Concluding remarks 296
The analysis of data presented in this paper demonstrates the statistical significance of temporal 297
correlations between reported volcanic phenomena at Etna, Stromboli, and Vulcano. The statistical 298
evidence here provided for coupled phenomena compels further research to better constrain the 299
13
temporal pattern of such phenomena and to identify their causes. The identification of volcanic 300
precursors is fundamental to mitigate the relative hazard in densely populated areas such as the 301
Aeolian Islands and Mt Etna. An integrated monitoring system, including measurements of 302
deformation, seismicity, and gas flux, may significantly help in revealing the causes for coupled 303
volcanic phenomena at Etna and Aeolian Islands in the next years. 304
The apparent causal relationships between seismic and volcanic phenomena in regions 305
characterized by low-magnitude earthquakes are still mostly unexplained. One of the models (i.e. 306
“superswarms”) proposed in this paper for explaining the concurrent phenomena at Etna, Stromboli, 307
and Vulcano may be a suitable one for other regions where volcanism and seismicity are nearly 308
concurrent but the earthquakes are low in magnitude. 309
The importance of the statistical analyses presented in this paper consists in the fact that, 310
whereas earthquakes and eruptions triggered by large earthquakes are increasingly being well 311
documented [1, 7], except a few studies [6, 9, 10], quantitative analyses of concurrent eruptions at 312
adjacent volcanoes are poorly documented. Because of the long historical records, Southern Italy is 313
the only place in the world where interactions between active volcanoes can be analyzed over about 314
the last two millennia. 315
316
317
Acknowledgments. We thank F. Barberi and D. Peacock for suggestions on an early version of 318
the manuscript, C. Cimarelli for Fig. 1, L. Da Riva for helping in Latin translations. Financial 319
support to AB comes from a GNDT project coordinated by L. Beranzoli with the help of P. Favali 320
and C. Faccenna. C. Jaupart and two anonymous reviewers are thanked for their comments. 321
14
References 322
[1] Hill, D.P., Reasenberg, P.A., Michael, A., Arabaz, W., Beroza, G.C., Brune, J.N., Brumbaugh, 323 D., Davis, S., DePolo, D., Ellsworth, W.L., Gomberg, J., Harmsen, S., House, L., Jackson, S.M., 324 Johnston, M., Jones, L., Keller, R., Malone, S., Nava, S., Pechmann, J.C., Sanford, A., Simpson, 325 R.W., Smith, R.S., Stark, M., Stickney, M., Walter, S., Zollweg, J., 1993. Seismicity in the 326 western United States remotely triggered by the M 7.4 Landers, California, earthquake of June 327 28, 1992. Science 260, 1617-1623. 328
from rising bubbles as a mechanism for remotely triggered seismicity. Nature 371, 408-410. 331 332 [3] Gomberg, J., Davis., S., 1996. Stress/strain changes and triggered seismicity at The Geysers, 333
California. Journal of Geophysical Research 101, 733-749. 334 335 [4] Brodsky, E.E., Sturtevant, B., Kanamori, H., 1998. Earthquakes, volcanoes, and rectified 336
diffusion. Journal of Geophysical Research 103, 23827-23838. 337 338 [5] Nostro, C., Stein, R.S., Cocco, M., Belardinelli, M.E., Marzocchi, W., 1998. Two-way coupling 339
between Vesuvius eruptions and southern Apennine earthquakes, Italy, by elastic stress transfer. 340 Journal of Geophysical Research 103, 24487, 10.1029/98JB00902. 341
342 [6] Miklius, A., Cervelli, P., 2003. Interaction between Kilauea and Mauna Loa. Nature 421, 229. 343 344 [7] Stein, R.S., 2003. Earthquake conversations. Scientific American 288, 72-79. 345 346 [8] McNutt, S.R., 2005. Volcanic seismology. Annual Review of Earth and Planetary Sciences 33, 347
461-491. 348 349 [9] Linde, A.T., Sacks, I.S., 1998. Triggering of volcanic eruptions. Nature 395, 888-890. 350 351 [10] Manga, M., Brodsky, E.E., 2006. Seismic triggering of eruptions in the far field: volcanoes and 352
geysers. Annual Reviews of Earth and Planetary Sciences 34, 263-291. 353 354 [11] de Dolomieu, D.G., 1783. Voyage aux Ĩles de Lipari fait en 1781. Cuchet, Paris. 355 356 [12] Mercalli, G., 1879. Contribuzioni alla geologia delle Isole Lipari. Atti della Società Italiana di 357
Scienze Naturali 22, 369-380. 358 359 [13] Gabbianelli, G., Romagnoli, C., Rossi, P.L., Calanchi, N., 1993. Marine geology of the 360
363 [14] Rittmann, A., 1931. Der Ausbruch des Stromboli am 11 September 1930. Zeitschrift für 364
Vulkanologie 14, 47-77. 365 366 [15] Simkin, T., Siebert, L., 1994. Volcanoes of the World. Geoscience Press, Tucson. 367 368 [16] Frazzetta, G., Gillot, P.Y., La Volpe, L., Sheridan, M.F., 1984. Volcanic hazards at Fossa of 369
Vulcano: data from the last 6000 years. Bulletin of Volcanology 47, 105-124. 370 371
15
[17] Gillot, P.Y., Kieffer, G., Romano, R., 1994. The evolution of Mount Etna in the light of 372 potassium-argon dating. Acta Vulcanologica 5, 81-87. 373
374 [18] Allard, P., Burton, M., Murè, F., 2005. Spectroscopic evidence for a lava fountain driven by 375
previously accumulated magmatic gas. Nature 433, 407-410. 376 377 [19] Rittmann, A., 1964. Vulkanismus und Tektonik des Ätna. Geol. Rundsch. 53, 788-800. 378 379 [20] Montelli, R., Nolet, G., Dahlen, F.A., Masters, G., Engdahl, E.R., Hung, S.-H., 2004. Finite-380
frequency tomography reveals a variety of plumes in the mantle. Science 303, 338-343. 381 382 [21] Newhall, C.G., Self, S., 1982. The volcanic explosivity index (VEI): an estimate of explosive 383
magnitude for historical volcanism. Journal of Geophysical Research 87, 1231-1238. 384 385 [22] Wiemer, S., Wyss, M., 2000. Minimum magnitude of completeness in earthquake catalogs: 386
examples from Alaska, the Western United States, and Japan. Bulletin of the Seismological 387 Society of America 90, 859-869. 388
389 [23] Efron, B., 1979. Bootstrap methods: another look at the jackknife. Annals of Statistics 7, 1-26. 390 391 [24] Patanè, D., De Gori, P., Chiarabba, C., Bonaccorso, A., 2003. Magma ascent and the 392
pressurization of Mount Etna’s volcanic system. Science 299, 2061-2063. 393 394 [25] Esposito, A., Giordano, G., Anzidei, M., 2006. The 2002-2003 submarine gas eruption at 395
Panarea volcano (Aeolian Islands, Italy): volcanology of the seafloor and implications for the 396 hazard scenario. Marine Geology 227, 119-134. 397
398 [26] Caliro, S., Caracausi, A., Chiodini, G., Ditta, M., Italiano F., Longo, M., Minopoli, C., Nuccio, 399
P.M., Paonita, A., Rizzo, A., 2004. Evidence of a recent input of magmatic gases into the 400 quiescent volcanic edifice of Panarea, Aeolian Islands, Italy. Geophysical Research Letters 31, 401 L07619, doi:10.1029/2003GL019359. 402
2005. Chronology and complex volcanic processes during the 2002-2003 flank eruption at 405 Stromboli volcano (Italy) reconstructed from direct observations and surveys with a handheld 406 thermal camera. Journal of Geophysical Research 110, B02201, doi:10.1029/2004JB003129. 407
408 [28] Silvestri, O., 1893. L'eruzione dell'Etna del 1886. Atti dell'Accademia Gioenia di Scienze 409
Naturali in Catania 69, 1-37. 410 411 [29] Silvestri, O., 1890. Etna, Sicilia ed isole vulcaniche adiacenti sotto il punto di vista dei 412
fenomeni eruttivi e geodinamici avvenuti durante l’anno 1889. Atti dell’Accademia Gioenia di 413 Scienze Naturali in Catania 66, 222-249. 414
415 [30] Mercalli, G., 1888. L’Isola Vulcano e lo Stromboli dal 1886 al 1888. Atti della Società Italiana 416
di Scienze Naturali 31, 1-15. 417 418 [31] Fouqué, F.A., 1865. Sur l’eruption de l’Etna du 1 fevrier 1865. Comptes Rendus de 419
l’Academie des Sciences 61, 210-211. 420 421
16
[32] Silvestri, O., 1867. I fenomeni vulcanici presentati dall’Etna nel 1863-64-65-66. Atti 422 dell’Accademia Gioenia di Scienze Naturali in Catania 23, 55-317. 423
424 [33] Boschi, E., Guidoboni, E., Ferrari, G., Valensise, G., Gasperini, P., (Eds.) 1997. Catalogo dei 425
forti terremoti in Italia dal 461 a.C. al 1990, 2. ING-SGA, Bologna. 426 427 [34] Kanamori, H., Givens, J.W., 1982. Analysis of long period seismic waves excited by the May 428
18, eruption of Mount St. Helens: a terrestrial monopole? Journal of Geophysical Research 87, 429 5422-5432. 430
inferences from seismic-strain tensor computations at Mt Etna Volcano, Italy. Bulletin of 433 Volcanology 62, 318-330. 434
435 [36] D’Agostino, N., Selvaggi, G., 2004. Crustal motion along the Eurasia-Nubia plate boundary in 436
the Calabrian Arc and Sicily and active extension in the Messina Straits from GPS 437 measurements. J. Geophys. Res. 109, B11402, doi:10.1029/2004JB002998. 438
439 [37] Hollenstein, C., Kahle, H.G., Geiger, A., Jenny, S., Goes, S., Giardini, D., 2003. New GPS 440
constraints on the Africa-Eurasia plate boundary zone in southern Italy. Geophys. Res. Lett. 30, 441 1935, doi:10.1029/2003GL017554. 442
443 [38] Billi, A., Barberi, G., Faccenna, C., Neri, G., Pepe, F., Sulli, A., 2006. Tectonics and seismicity 444
of the Tindari Fault System, southern Italy: crustal deformations at the transition between 445 ongoing contractional and extensional domains located above the edge of a subducting slab. 446 Tectonics 25, TC2006, doi:10.129/2004TC001763. 447
Table 1. Observed and expected compound probabilities for the occurrence of (a) months and (b) years including eruptions at two or three volcanoes among Etna, Stromboli, and Vulcano during 0-1993 AD. The relative data are from [15] and are listed in Tables SM1 and SM2 in the Supplementary Material. (a) volcanoes compound probability (%) observed expected Etna-Stromboli 0.19 0.027 ± 0.0027 Etna-Vulcano 0.28 0.045 ± 0.0036 Stromboli-Vulcano 0.042 0.0033 ± 0.00042 Etna-Stromboli-Vulcano 0.0042 0.00020 ± 0.000026 (b) volcanoes compound probability (%) observed expected Etna-Stromboli 1.6 0.25 ± 0.037 Etna-Vulcano 0.50 0.14 ± 0.027 Stromboli-Vulcano 0.25 0.042 ± 0.0093 Etna-Stromboli-Vulcano 0.10 0.0038 ± 0.00086
Table 2. Time order of erupting volcanoes in the occasion of duple and triple concurrent eruptions between 1700 and 1993. Data are from [15]. type of concurrent eruptions
erupting volcanoes
no. of times in which the volcano erupted before the others in the
Table SM1. Eruption (with VEI ≥ 2) occurrences by month between 0 and 1993 AD at Etna, Stromboli, and Vulcano. VEI = Volcanic Explosivity Index [21]; 0 = no occurrence; 1 = occurrence. Data are from the Simkin and Siebert’s Catalog [15].
Table SM2. Eruption (with VEI ≥ 2) occurrences by year between 0 and 1993 AD at Etna, Stromboli, and Vulcano. VEI = Volcanic Explosivity Index [21]; 0 = no occurrence; 1 = occurrence. Data are from the Simkin and Siebert’s Catalog [15].