GEOPHYSICAL RESEARCHLETTERS, VOL. 25, NO.15, PAGES 2861-2864,AUGUST 1, 1998
The 1997 Umbria-Marche, a first
look
at the
main
Italy, earthquake shocks
and
sequence'
aftershocks
A. Amato, R. Azzara, C. Chiarabba, G.B. Cimini, M. Cocco,M. Di Bona, L.
Margheriti,S. Mazza, F. Mele, G. Selvaggi,A. Basili, E. Boschi Istituto Nazionale di Geofisica, Rome, Italy
F. Courboulex, A. Deschamps,S. Gaffet G•osciences Azur, France
G. Bittarelli, L. Chiaraluce, D. Piccinini, M. Ripepe Universita di Camerino, Italy
Abstract. A long sequenceof earthquakes,six with magnitudes between 5 and 6, struck Central Italy starting on September26, 1997, causingseveredamagesand lossof human lives. The seismogenicstructure consistsof a NW-SE elongated fault zone extending for about 40 km. The focal mechanismsof the largest shocksrevealnormal faulting with
NE-SW extensionperpendicularto the trend of the Apennines, consistently with the Quaternary tectonic setting of the internal sector of the belt and with previous earthquakes in adjacent regions. Preliminary data on the main shocks and aftershocks show that extension in this region of the Apenninesis accomplishedby normal faults dipping at low
angle (•40 ø) to the southwest,and confinedin the upper •8 km of the crust. These normal faults might have reactivated thrust planes of the Pliocene compressionaltectonics. The aftershock distribution and the damage patterns also
suggestthat the three main shocksruptured distinct 5 to 15 km-long fault segments,adjacent and slightly offsetfrom
with thousands of earthquakes located by the national net-
work(RSNC) of the Istituto Nazionaledi Geofisica(ING) in the past two decades. Two main sequencesoccurred in the past twenty years at about 30 km from the region of Colflor-
ito, to the northwest(1984 at Gubbio-Perugia,Ms5.3) and to the south (1979 at Norcia, Ms5.8) (Deschampset al., 1984; Haessleret al., 1988). Both these earthquakeshad normal faulting mechanisms(Figure 1), consistentwith the present tectonic setting of the internal sector of the north-
ern Apenninearc (Andersonand Jackson,1987; Frepoli and Amato, 1997;Montoneet al., 1997). Differently from the recentseismicsequences of the region(1979 and 1984),the one which started last September has an unusually high number of strong shocks. Five exceeded ML5, and more than 20
exceededML4 (Figure 2). This paper describesthe distribution of main shocks and aftershocks, as determined from the ING
national
network
and from
a dense local network
installed immediately after the first shocks.
one another.
The
Introduction
On September26, 1997, at 00:33UTC (02:33localtime), central Italy was struck by a strong earthquake, which was followedby a strongershocknine hourslater (09:40 UTC). Both shocks(Mw 5.7 and 6.0) causedseveredamageand lossof human lives in a region betweenUmbria and Marche, and were felt by millions of people in central Italy, including Roma and Firenze. Eleven people died, and unique masterpiecesof Italian architectural and painting art were destroyedor strongly damaged, for example in Assisi and
Foligno(Umbria). Earthquakesof this sizeare commonin the historical catalogue of the Umbria-Marche region, although nonehas been reportedin the epicentralarea of the
1997 shocks(Boschiet al., 1995). The surroundingregions were repeatedely struck by strong earthquakesin the past,
main
shocks
On September3, 1997 (at 22:07 UTC) a Mr4.5 earthquake in the plain of Colfiorito triggered a sequenceof aftershocks which lasted for about three weeks, with a normal decay in the number and magnitudesof shocks(Fig-
ure 3). At 00:33 UTC on Sep. 26, a Mw5.7 shockoccurred at about the same location and at •7 km depth, with a normal faulting mechanism. It causedseveredamage in the small villages located to the south of the epicenter. The
secondstrongshockoccurredat 09:40 UTC (Mw6.0), 3 km WNW of the first. Both earthquakes were located with data from the ING national network, but only for the first one could we use data from four close stations. For this reason, the shock of 09:40 has the largest location uncertainties,
particularly in depth (Table 1). The two earthquakeshad similar focal mechanisms, with NW-trending normal fault-
both to the north (1279,1747,1751)andto the south(1328, ing planes (Ekstrom et al., 1998). Contrary to what was 1703, 1730) (Figure 1). At present,the Umbria-Marche Apennines are characterized by diffuse crustal seismicity,
Copyright1998by theAmericanGeophysical Union.
observed during the first earthquake of the night, the second shock causeddestruction towards the north, suggesting that the two earthquakes ruptured in opposite directions, as also indicated by strong motion recordingsand by the anal-
ysis of regionalbroad-banddata (N.A. Pino, pers. comm.) To test this hypothesis further, we compare the aftershocks located by RSNC in the nine hours following each of the two
Papernumber98GL51842. 0094-8534/98/98GL-51842505.00
2861
2862
AMATO ET AL.- AFTERSHOCKS OF THE 1997 UMBRIA-MARCHE EARTHQUAKE SEQUENCE
Table 1. Hypocentral parametersof the six largestshocks Time UTC
Lat.
97/09/2600:33 97/09/2609:40 97/10/0308:55 97/10/06 23:24 97/10/1211:08 97/10/1415:23
Long.
43001.35 ' 43001.83 ' 43001.97 ' 43ø01.13 ' 42054.62 ' 42ø55.14 '
Depth, km
12053.50 ' 12051.73 ' 12050.25 ' 12050.27 ' 12056.78 ' 12055.56 '
Mo, dynecma Mr a ERH, km
6.6 5.7 6.2 7.1 5.9 6.6
4.0 1024 12.01024 0.9 1024 1.7 1024 0.8 1024 3.4 1024
5.6 5.8 5.0 5.4 5.1 5.5
ERZ, km
Statb
0.9 2.0 0.6 0.5 0.2 0.4
14 (4) 6 (1) 29 (24) 29 (24) 28 (23) 28 (22)
0.6 0.8 0.3 0.3 0.3 0.4
•After Mo after Ekstrom et al. (1998); Mr after Mazza and Casa]e,in preparation bNumberof stationswithin 100 (20) km.
main shocks. Figure 3 showsthat early aftershocksof the first earthquake are concentrated to the east of the main shock epicenter, whereas aftershocks following the second shockare mostly located to the north. This strenghtensthe hypothesisthat the two shocksruptured, in oppositedirections, two adjacent fault segments.Interestingly, the region between the two main rupture zones was affected by the
shocks(Mw between5.2 and 5.7) occurredin the following twenty days (Oct. 3, 7, 12, 14), all with similar focal mechanisms(Ekstrom et al., 1998). The first two were located close to those of Sep. 26, whereas the latter two occurred about
15 km to the southeast.
The total
vated
area thus became
40 km.
foreshocksequenceof Sep. 3 (Figure3). Four more strong Aftershock
about
extent
of the acti-
distribution
A dense temporary network of 26 three- and six-component digital seismicstations (Figure 4) was deployed a few hours after the first shocks by teams of the ING,
44 ø
o
o
ßß 0
O
-•
ß
O
,
the G•osciencesAzur, France (GA), and the Universit• di Camerino(UC) (Figure 4). Severalthousandsof aftershocks were recordedby the temporary network from Sep. 26 to Nov. 3, 1997. The temporal distribution of seismicity(Fig-
Io
- 3OOO
43 ø
-0
1979
I
'
I
'
øt•85 OO•O ,' o 8 %
•
•'
,-.• "
42 ø 12 ø
-5
o
-4
o" •8•s okm 2 •
0 10•
.
o?-a,&i I 'I Figure
--6
1.
Main shocks(stars) and aftershockslocated
4
14
September 1997
24
4
14
October
24
,
3
13
November
by the RSNC from Sep. 26 to Dec. 31, 1997, and focal
mechanismsfrom Ekstrom et al. (1998). Also shownare major historical earhquakesin the last 700 years (squares, Figure 2. Temporal distribution of seismicityfrom Sep.
numbers indicatethe yearsof occurrence afterBoschiet al., 3 through Nov. 3, 1997. Top: cumulative number of earth1995), backgroundseismicityfrom the ING bulletins(1993- quakes with Mr >2.5 and related cumulative seismic mo1997, grey circles), aftershocksand focal mechanismsof the 1979 Norcia and the 1984 Gubbio-Perugiaearthquake.
ment. Bottom: number of earthquakes per hour and mag-
nitude of the largestshocks(Mr _• 4, black stars).
AMATO ET AL.' AFTERSHOCKS
OF THE 1997 UMBRIA-MARCHE
SEQUENCE
2863
shockhypocentersare locatedat the baseof the aftershock zone,at 6-7 km depth. Accordingto the aftershockdistribution at depth,the dip of the faultis •-35øto the SW, slightly lowerthan the oneinferredfrom the focalmechanism(42ø).
Nocera U.
Colfiorito
'970903
431•'
EARTHQUAKE
Discussion
and Conclusion
In the region,the two main NW-trending normal faults active in the Quaternary are located at the easternboundary of the Colfioritobasin,and a few km southwestof it,
respectively (Calamitaand Pizzi, 1994;Celloet al., 1997). Figure4 shows that the ruptureof the Mw6.0 shocknucle-
]oligno km
ated at about 6 km depth and propagated up-dip and to the north, as previouslydiscussed.The rupture probably stoppedvery closeto the surface,as suggested by both the
aftershock andgeodeticdata (Hunstad,et al., 1998). Based on thesesimplegeometricconsiderationsand a comparison with the mappedfaults,it seemslikely that no true surface faultingoccurred.The observedcracksand fault reactiva-
tions,studiedby Cinti et al. (1998),Galli et al. (1998)and
i
i
i
o
o
..
o..
431•'
/Camerino
u.
o
Assisi
ß
43'
o
Figure 3. Top- epicentersof the Sep. 3, 1997 ML4.5 foreshock sequence comparedto thoseof the two shocksof Sep. 26 (stars).Bottom:aftershocks of the firstlargeshock of thenight(blackcircles:26Sep.00:33to 09:39UTC) and of the secondlarge shockof the morning(grey circles:26 Sep. 09:40to 18:40UTC), as determinedby the permanent network. The boxesindicate the approximateextent of the
o
Norcia km
o
5
10
two rupture areas.Focalmechanisms from Ekstromet al., 1998.
I
0 SW
ure2) shows that someof the largeshocks werefollowed by an increasein the number of earthquakes, suggestingthat
E 5
the whole sequenceconsistsof a seriesof differentrupture episodes.However,not all the largeshockswerefollowed
•10
by an evidentincrease in seismicity, asfor instancethe Oct. 3 event (Figure 2 and Table 1), whichoccurrednear an areapreviously rupturedby the largestearthquakes. About 2,550 aftershocks havebeenlocatedwith data of the local network.Figure 4 showsthe 1,747best locatedearthquakes
•
J
,
002
ne
o •.•
15 0
1'0
'
,
Mw6.0
2'0
3O
Distance (km)
(withformallocationerrorslessthanI km),whosedistribu- Figure 4. Epicentraldistribution(top) and SW-NE vertion delineatesa NW-SE trending fault zone,elongatedfor tical section(bottom) of 1,747 selectedaftershockslocated about 40 km, and extendingfor 5 to 15 km in the perpen- by the temporary network (with location errors lessthan 1 dicular direction, The NYV-SE trend is consistent with the km). Earthquakeswithin :k2 km from the profileare plotted
faultplanesolutions of thelargest shocks andwithpreviousin thesection, In themapthedistribution ofthetemporary
earthquakes (Figure1). In a vertical section crossing the (blacktriangles) andof thepermanent network (squares) fault zoneperpendicular to its trend(Figure4) the main areshown.
2864
AMATO ET AL.: AFTERSHOCKS OF THE 1997UMBRIA-MARCHE EARTHQUAKE SEQUENCE
Vittori et al. (1998) could representthe localizedresponse Bally, A. W., L. Burbi, C. Cooper, and R. Ghelardoni, Balanced sectionsand seismicreflection profiles acrossthe central Apenof the shallowest part of the crust to extension occurring at nines, Mem. Soc. Geol. It., 35, 257-310, 1986. depth. Boschi,E., G. Ferrari, P. Gasperini, E. Guidoboni, G. Smriglio, Another interesting issue raised by this earthquake seand G. Valensise, Catalogo dei forti terremoti in Italia dal quence is the role played by normal faults in the active •61 a.C. al 1980, Pubblicazione dell'IstitutoNazionaledi Getectonics of the region. It is a matter of debate whether ofsica, 973 pp., 1995. the regional stress is dominated by crustal extension per- Calamita, F., and A. Pizzi, Recent and active extensional tectonicsin the southernUmbro-MarcheanApennines(centralItaly), pendicular to the Apennines, as suggestedby seismological
and boreholebreakout data (Andersonand Jackson,1987; Haessleret al., 1988; Frepoli and Arnato, 1997; Montone et
al., 1997), or whether the normal faults act as a response to different processes,suchas pull-apart type tectonicsin a
strike-slipstressregime(Cello et al., 1997). Within uncertainties due to the preliminary nature of the data analyzed to date, the following scenario is sug-
gestedfor the Umbria-Marcheearthquakesequence.(a) It is confirmed that a processof NE-SW extension is active in the region. The similarity of the focal mechanismsto those determined for recent earthquakes in adjacent regions
(Figure 1) suggests that this extensionalstressregimeis uniform in this area of the Apennines. (b) The three largest shocks have ruptured distinct, adjacent, sub-parallel normal fault segmentswith a rather constant NW-SE trend. These main segments appear to be laterally offset, particularly the third and southernmost one, for which a rightlateral stepping of about 5 km is suggestedby aftershock
Mem. Soc. Geol. It., •8, 541-548, 1994. Cello, G., S. Mazzoli, E. Tondi, and E. Turco, Active tectonics in the Central Apennines and possibleimplications for seismic hazard analysis in peninsular Italy, Tectonophysics,272, 4368, 1997.
Cinti, F., L. Cucci, F. Marra and P. Montone, Preliminary analysis of the source effects producued by the Umbria-Marche seismicsequenceand possibleseismotectonicimplications(abstract), European GeophysicalSociety Annual Conference, Nice, France, 1998.
Deschamps, A., G. Iannaccone and R. Scarpa, The Umbrian earthquake (Italy) of 19 September 1979, Annales Geophysicae, 2, 1, 29-36, 1984.
Ekstrom, G., A. Morelli, A.M. Dziewonski, E. Boschi, Moment tensor analysis of the Umbria-Marche earthquake sequenceof September-October 1997, Geophys. Res. Lett., in press, 1998. Frepoli, A., and A. Amato, Contemporaneousextension and compressionin the North Apennines from earthquake fault plane solutions, Geophys. J. Int., 129, 368-388, 1997. Galli, P., R. Basili, V. Bosi, F. Galadini, M. Meghraoui, P. Messina, M. Moro and A. Sposato, The central Italy earthquake of Sptember-October 1997: geological effects and seis-
locations. (c) The total extent of the aftershockzone (40 motectonichypotheses(abstract), EuropeanGeophysical Socikin) is larger than the cumulativelength of the individual ety Annual Conference, Nice, France, 1998. rupturesof the largestshocks(25-30 krn), considering their Haessler, H., R. Gaulon, L. Rivera, R. Console, M. Frogneux, seismicmoments. Around the main ruptures, at least three areas with diffuse seismicity are evident, namely the SE edge of the structure, the region north of Sellano, and that between the two main shock epicenters of Sep. 26, where the
foreshocksequence of Sep. 3 occurred.(d) The normalfault planes which ruptured during the largest shocksare those dipping to the southwest. Dip of the distribution is about
•35 ø, in good agreementwith the SW-dippingplane of the focal mechanisms. In view of this, a plausible hypothesis is that the faults ruptured pre-existing thrust planes inherited from the Pliocene compressionaltectonicsof the Apennines
C. Gasparini, L. Martel, G. Patau, M. Siciliano and A. Cis-
ternas, The Perugia (Italy) earthquakeof 29 April 1984: a microearthquake survey, Bull. Seismol. Soc. Am., 78, 6, 19481964, 1988.
Hunstad,I., M. Anzidei, P. Baldi, A. Galvani and A. Pesci(1998) GPS observations of co-seismic displacement of the Umbria-
Marche seismicsequence(abstract), European Geophys.Soc. Annual Conference, Nice, France, 1998. Montone, P., A. Amato, A. Frepoli, M. T. Mariucci, and M. Cesaro, Crustal stressregime in Italy, Annali di Geofsica, XL, 3, 741-757, 1997.
Vittori, E., G. Cello, G. Deiana et al. (1998) Geologicaleffects of the September 26, 1997 earthquakes in central Italy (abstract), EuropeanGeophysicalSocietyAnnual Conference, (Bally et al., 1986),whichare clearlyvisiblein the few seisNice, France, 1998. mic profiles available for the region.
Acknowledgments.
The authorsthank the ING bulletin
staff who helped in field work and analysed the RSNC data, and the ING technical laboratory staff for help in field work. The ML values were kindly provided by P. Casale. Dr. F. Evison and two anonymous referees are thanked for reviewing the manuscript.
A. Amato, Istituto Nazionale di Geofisica, Via di Vigna Mu-
rata, 605, Roma,Italy. (e-mail:
[email protected])
References Anderson, H. J., and J. A. Jackson, Active tectonics of the Adriatic region, Geophys. J. R. Astron. Soc., 91,937-983, 1987.
(ReceivedFebruary9, 1998; revisedMay 6, 1998; acceptedMay 19, 1998.)
