Transcript Diapositiva 1 - Earth

Cities On Volcanoes 4
23-27 January 2006
Mount Etna eruptions and damaging earthquakes: new insights
from recent broad band data by Giuliano Milana, Giuliana Mele, and Antonio Rovelli
Between 26 and 29 October 2002, the Mt. Etna area was struck by a seismic swarm that counted some
hundred events some of them well felt by the population. A first group of epicenters is located in the E-NE
part of the volcano, reaching a maximum local magnitude of 4.6 on October 27 at 02:50 UT. A second group,
occurring since October 28, is located in the SE flank close to the town of Zafferana Etnea and culminated
with the ML 4.4 earthquake of October 29 at 10:02 UT. The seismic swarm was concomitant with the
opening of a system of fractures in the eastern part of the volcano and the reactivation of the eruptive
fracture system originated by the 2001 eruption.
The most damaging event was the ML 4.4 earthquake of October 29, which occurred in a densely urbanized
territory and attained intensity VIII on the European Macroseismic Scale (EMS). In spite of its small local
magnitude, the shock caused significant damage to many buildings including reinforced concrete ones in
an area that extends for about 4 km in a NNW-SSE direction and is centered around the villages of S.
Venerina and Guardia. A long system of surface fractures originated along the damaged area. Other similar
and even stronger events of the 2002 swarm in the northern zone did not cause diffuse damage since they
occurred in a not densely urbanized area; however, they were highly destructive for the few sparse buildings
close to the epicenters.
The sequence was well recorded by the Broad Band Station of Antillo (Mednet Network), the most energetic
events also triggered the Strong Motion Stations of Bronte (BRNT) and Catania (CATA). The latest part of
the sequence (Dec. 2003 – Jan. 2004) was recorded by the local micronetwork installed in Santa Venerina
(SVN). Two of these stations were also operating in Santa Venerina from Nov. 2003 to Aug. 2004.
Moreover, data from a broad band station installed in Catania are available from Mar. 2001 to Dec. 2001.
INGV
Red circles: Events recorded by
strong motion Instruments
Blue diamonds: Aftershock
recorded by local seismic
stations.
Bongiardo
8
Felicetto
8
Scura
8
S. Venerina
7-8
Guardia
7-8
Catania
5
Messina
4
EMS-98 Macroseismic Intensities for
the 2002 Oct. 29 10:02 4.6 event.
Map of the study area.
Damage on reinforced concrete buildings produced by 2002 Oct. 27 02:52 (left) and
2002 Oct. 29 10:02 (right).
Strong motion data enhance an anomalous low-frequency content
• The activation pattern of strong motion stations reflects the temporal and spatial evolution of the swarm.
Sample acceleration and velocity waveforms for BRNT and CATA stations.
Date
Time
ML
Distance
PGA SP
PGA Obs.
PGV SP
PGV Obs.
2002/10/27
01:28
4.3
19
28.01
4
0.91
0.7
2002/10/27
01:58
4.6
12.5
49.89
14
1.86
2.3
2002/10/27
02:50
4.6
17
38.27
9
1.39
2.4
2002/10/27
02:50
4.6
27
24.89
4
0.89
1.6
2002/10/29
10:02
4.4
18
31.58
15
1.07
4.5
2002/10/29
16:39
4.1
16
28.4
6
0.87
0.9
• The accelerometer of BRNT was triggered by three of the events of October 27 located in the N-NE area of Mount
Etna.
• The ML 4.6 event of October 27 at 02:50 UT is the only one that triggered both BRNT and CATA.
• The accelerometer of CATA was also triggered by two of the events of October 29 occurred in the SE flank of Mount
Etna, very close to Santa Venerina area.
• Waveforms indicate a 20-s long strong motion phase characterized by predominance of low frequencies.
• Horizontal PGA and PGV values deviate from the statistical expectations of the Sabetta and Pugliese regression
(SP87). Even if the SP87 did not include earthquakes of volcanic origin we believe that this comparison still is useful
to enhance the inadequacy, for volcanic events, of the commonly adopted ground motion scaling laws. This is
especially important for the hazard assessment in the Mt. Etna area.
• Recorded PGAs are always smaller than predictions, in contrast recorded PGVs are always larger, and the
discrepancy tends to increase for the strongest events.
• The low-frequency amplitudes of the strongest volcanic events are larger than those of tectonic events at the same
magnitude and distance whereas the opposite occurs for the high-frequency band, implying a source scaling
completely different between tectonic and volcanic events.
• Acceleration spectra of CATA show a significant depletion in the high-frequency amplitude and a large spectral bump
in the frequency band 0.1 – 1 Hz.
• BRNT station has the same tendency, although the exaggeration of low-frequency amplitudes has a smaller extent.
• The difference in the spectral bump between BRNT and CATA for the same event is not due to a site effect at CATA
since the station does not show amplification at low frequency in the spectra of the local tectonic events.
• The evident deviation from the conventional Brune spectrum is therefore the origin of the anomalously small
accelerations and large velocities of volcanic events.
Comparison between observed source and Brune model horizontal
spectra for Etna events (a-e) and for a southern Sicily tectonic event (f).
Comparison between observed and predicted strong motion parameters.
Weak motion data confirm the predominant low-frequency radiation
• Between 2002 Dec. 3 and 2003 Jan. 25 five local seismic stations were operating in S. Venerina (SVN)
• Eleven good S/N ratio events were recorded with magnitude between 2.2 and 3.7.
• Events were located in the N-NE and S-SE sectors of Mount Etna.
• The majority of the records shows a low-frequency content regardless of epicentral distance.
• Some high-frequency events were also recorded mainly at the end of the recording period.
Comparison between records and Fourier spectra for two weak motion data.
The Brune model spectrum for the high frequency event is also showed.
Fourier spectra for weak motion data. Epicentral distance are: a) 3-5 km, b) 13-16 km.
Does the high-to-low frequency amplitude ratio relate to volcanic activity?
Seismic records are low- and high-pass filtered with a 1 Hz Butterworth filter, PGV are then evaluated on filtered signals (PGVHP, PGVLP) along with their ratio. It is possible to find some temporal variations of the high-to-low
frequency amplitude ratio. This procedure is applied to: i) 2001 data recorded at Catania, ii) 2002-2003 data recorded at Antillo and Santa Venerina, iii) 2003-2004 data recorded at Santa Venerina.
100
100
enlarged in the next plot
100
100
Catania Station
10
10
10
10
1
1
1
1
2001
0.1
PGVHP / PGVLP
0.01
30-Mar
27-Apr
25-May
22-Jun
2002-2003
0.1
0.1
20-Jul
17-Aug
14-Sep
12-Oct
9-Nov
7-Dec
PGVHP / PGVLP
Fractures Opening
Catania Station
0.01
12-Jul
13-Jul
14-Jul
15-Jul
16-Jul
17-Jul
18-Jul
0.01
25-Oct
2003-2004
0.1
PGVHP / PGVLP
Antillo Station
S. Venerina Station
S. Venerina station
0.01
8-Nov
22-Nov
6-Dec
20-Dec
3-Jan
17-Jan
3-Jan
31-Jan
28-Feb
27-Mar
24-Apr
22-May
19-Jun
17-Jul
The large low-frequency content is responsible for high damage
• 5% damped pseudovelocity response spectra for the strongest of the October 2002 earthquakes compared to statistical expectations of the Sabetta and Pugliese regression (SP96) show again a significant (up to a factor of 3)
depletion in the observed high-frequency range.
• In contrast, at low frequency, response spectra computed from the local recordings are fit by spectral ordinates statistically corresponding to ML 5.5 to 6 of tectonic earthquakes in Italy. In this frequency range the effects of volcanic
earthquakes on buildings could be as large as those experienced in tectonic regions of Italy during much stronger (up to ML 6) earthquakes.
• The low-frequency content brings to high values for ground displacement, also the damage of the 2002 events were consistent with high values of displacement.
• The frequency band 0.3 to 1.5 Hz gives the largest contribution to the pseudovelocity response spectra of volcanic events of Mt.Etna.
• For a given earthquake, it is possible to define for each recording site a measure of ground shaking evaluating the Housner intensity in 0.3-1.5 Hz frequency band and to determine the magnitude that, in the SP96 regression, would
attain the same spectral amplitude.
CONCLUSIONS
•The Housner spectral intensity can be
evaluated almost in real time as an
indicator of potential damage
produced by volcanic events even
when their magnitude is as low as 4.
Istituto Nazionale di Geofisica e Vulcanologia,
Via di Vigna Murata 605,
00143, Roma, ITALIA
[email protected]
•The volcanic event can cause local
damage comparable to the tectonic
events associated to the main
seismogenetic structures of Eastern
Sicily and must be taken into account
in hazard evaluation.