Observatories and Research Facilities for EUropean Seismology
Volume 5, no 1 March 2003 Orfeus Newsletter

Fast determination of moment tensors for the recent Molise (southern Italy) seismic sequence

S. Pondrelli2, F. Di Luccio1, E. Fukuyama3, S. Mazza1, M. Olivieri1 and N. A. Pino1

1Istituto Nazionale di Geofisica e Vulcanologia, Via di Vigna Murata 605, Rome, Italy
2Istituto Nazionale di Geofisica e Vulcanologia, c/o Bologna University, Bologna, Italy
3NIED, Tsukuba-shi, Ibaraki-ken, 305, Japan

Introduction - Moment Tensor Solutions
Discussion and Conclusions - Acknowledgements - References


On October 31, 2002, at 10:32 UT, a strong earthquake of Mw=5.7 (Ml=5.4) struck the Molise region in southern Italy. The earthquake caused the collapse of a school building, with 28 casualties and major damages in hundreds of private buildings. No relevant seismic event seems to have occurred in the epicentgal region in the last 1000 years. The CPTI catalogue (CPTI Working Group, 1999) reports only a few large earthquakes in the surrounding areas (red squares in Figure 1):
  • the 1125 earthquake in the immediate vicinity of the present sequence. Little is known about this event, that might have reached Imax VIII-IX;
  • the Apenninic sequence of December 1456, which caused heavy damages and reached MCS intensity Imax XI;
  • the Gargano sequence of the summer of 1627, which also caused damages, assessed as Imax X;
  • the July 1805, Matese earthquake, which produced effects up to Imax X.

Figure 1. Seismicity map of the Molise region (southern Italy). Stars indicate the epicentral location of October 31 and November 1 major events. Instrumental seismicity recorded by the National Central Seismic Network since 1990 (IngNet On Line -INGV Database) is shown with yellow circles for M<4 and red circles for M>4 events. Red squares represent strong historical earthquakes that occurred in the last 1000 years within 100 km from the present sequence (CPTI, CPTI Working Group, 1999; CFTI catalogue, Boschi et al., 1997). Squares are scaled by event dimension: the 1627 event corresponds to Imax X. For more information on the sequence, visit the Molise special web page at INGV.

The recent instrumental seismicity (Figure 1) is mainly distributed in the inner Apennines, west of the present sequence, and in the Gargano promontory to the east. No significant activity was detected in the present epicentral area during last 12 years. Prior to the 2002 sequence, geological and geophysical data did not evidence any active fault in the Molise region. The Database of Potential Sources for Earthquakes larger than 5.5 in Italy reports only the Mattinata fault, an active structure in the Gargano Promontory, with a possible past strike-slip motion. The present sequence is roughly aligned with this structure, but there is no unequivocal evidence of connection between the fault and the 2002 sequence.

Since the first night following the main shock, INGV deployed 30 temporary 3-components stations in the epicentral area, in collaboration with INOGS (Trieste) and Dipteris (University of Genoa) (Figure 2). A description of the survey and first results can be found at the web page Molise special web page. The October 31 event has been followed by a similar magnitude event on November 1 (15:09 UT, Mw=5.7; Ml=5.3) about 7 km farther west, which caused additional damages but no casualties. According to the locations of the permanent INGV seismic network, both events occurred at a depth of about 20 km. At least 400 aftershocks, about 80 of which with Ml>3.0, have been recorded during the temporal experiment, that ended in December 2002. Preliminary locations, kindly provided by INGV National Earthquake Center, predominantly align in a E-W direction (Figure 2).

Figure 2. a) Preliminary locations (blue dots) obtained with a subset of data recorded by the temporary network, constitued by remote (black triangles) and telemetered stations (green triangles). Stars indicate the two major events. b) Three vertical cross-sections: E-W through main shocks (left); N-S through the western November 1 main event (middle); and N-S through the more eastern October 31 main event (right). Locations and figure are courtesy of INGV National Earthquake Center.

Moment Tensor Solutions

Intermediate magnitude events often are destructive in Italy, due to the age and poor quality of many buildings. Fast source parameter determination is thus of prime interest to timely assess the potential damage and to rapidly provide scenarios for the evolution of the seismic sequence.

Recent improvements in near-real time access to broadband data (MEREDIAN Project, van Eck et al., 2002) now allow rapid moment tensor computation. Since the 1997 Umbria-Marche seismic sequence, the Regional Centroid Moment Tensor (RCMT) Project started at INGV in collaboration with the Harvard University (Ekström et al., 1997; Morelli et al. 2000; Pondrelli et al., 2002) with the aim of computing source parameters for large and intermediate magnitude events in the Euro-Mediterranean area. The method, a modification of the CMT technique (Arvidsson and Ekström,1998), inverts for intermediate period surface waves (T>35 s) recorded at regional distances. The regional waveforms allow moment tensor computation for events as small as Mw=4.5, and in favorable circumstances we compute solutions even for Mw=4.0 events.

Figure 3. Broadband stations used for Molise earthquakes moment tensor analyses. Stations directly available with MedNet facilities are in white, stations available through the ORFEUS's MEREDIAN project are shown in yellow.

We pay special attention to the fast computation of moment tensors. Even though it is not automated, we quickly produce preliminary source parameters and publish the solution on the web at the QRCMT site. The data are primarily gathered from the MedNet Network and additional data are collected from stations available on-line from ORFEUS, thought the MEREDIAN Project (van Eck et al., 2002). For both the major events of the Molise sequence reliable Quick RCMT were computed using data from a considerable number of stations (Figure 3) and released within less than one hour. We also obtained QRCMT solution for two Mw= 4.5 aftershocks.

Figure 4. Examples of comparison between data (solid) and synthetics (dotted) computed by using the QRCMT solution, for the October 31, Mw=5.7 event. Source-to-station distances are 1.3 and 8.4 for AQU and GRFO respectively.

The resulting source mechanisms (Figures 4 and 5) display an almost pure strike slip motion with conjugate planes oriented along NS and EW directions, consistent with the aftershocks alignment. The source depths of both main events is fixed to 15 km to solve for the small uncertainties (2-3 km) around this value, obtained during waveform inversions.

Figure 5. Map showing the focal mechanisms obtained for the Molise sequence. Purple dots are relocated aftershocks of the Molise sequence (courtesy of the INGV National Earthquake Center). The red focal mechanisms are the Quick RCMT solution for the last Molise earthquakes, determined at INGV immediately after major events. Black focal mechanisms are solutions obtained by higher frequency (up to 0.05 Hz) waveform inversion (Fukuyama and Dreger, 2000). Both solutions could be computed for four events and are shown side by side. Event data and Mw are reported next to each focal mechanisms of the Molise sequence. Gray focal mechanisms on the background are from the Harvard CMT Catalog (Dziewonski et al., 2001) and from the European-Mediterranean RCMT Catalog (Pondrelli et al., 2002), covering major earthquake activity during the last 25 years in southern Italy. All focal solutions are scaled by the magnitude, but gray focal mechanisms are one third smaller than the others. The yellow line passing through the Gargano Promontory marks the surface evidence of the Mattinata fault ( Database of Potential Sources for Earthquakes larger than 5.5 in Italy).

During this Molise sequence we had the opportunity to test the automatic regional moment tensor determination procedure in use at NIED (Fukuyama and Dreger, 2000; Kubo et al., 2002), based on the moment tensor estimation code written by Dreger and Langston (1995). The algorithm searches for the combination of the fundamental faults and focal depth that provides the best fit to the recorded waveforms. The analysis is performed in the frequency range 0.02-0.05 Hz for M<5.0 and 0.01-0.05 Hz for M>5. In this frequency range, the crustal heterogeneities have only minor influence on the moment tensor inversion results. We used a single laterally homogeneous model (Table 1) representing a reasonable average for the source-to-station paths considered in this study.

Vp (km/sec)
Table 1 - Velocity model used for broadband waveform modeling.

Besides for the two mainshocks of October 31 and November 1, we also computed moment tensor solutions (MT) for several smaller aftershocks, obtaining a reliable source mechanism even for events with magnitude Mw=3.7. We used only MedNet data from the stations AQU, CII, and AIO (Figure 3). Depending on event magnitude and data quality, we used data from 1, 2 or all three stations. In Figure 5 the resulting moment tensors are drawn, while Figure 6 shows a waveform fitting example along with the associated focal solution. The comparison between QRCMT and MT (red and black focal solutions in Figure 5, respectively) for the 4 major events of the sequence shows a very good agreement, in spite of the different methods, frequency ranges and stations used. Our results for the two major shocks are very similar to the solution obtained by the Harvard University and the Swiss Agency using global station distribution. A larger seismic moment (nearly 50%) has been reported by NEIC.

Figure 6. Observed and synthetic waveforms, and the resulting moment tensor solution for the 02/11/02 23:27 aftershock (Mw = 4.3).

Discussion and Conclusions

The Molise (southern Italy) seismic sequence started on October 31, with the occurrence of a Mw=5.7 earthquake in the Apenninic belt, closely followed by a similar magnitude on November 1. At the end of December more than 400 events have been recorded by the permanent and temporary networks. This sequence occurred in a region where, based on historical and instrumental seismicity, no comparable earthquakes occurred in the last 1000 years. Focal depths and mechanisms of the Molise sequence aftershocks differ from the characteristics of the seismicity located within the Apenninic chain in southern Italy, where usually the earthquakes are shallow (within 10 km of depth) normal faulting events (gray focal mechanisms of Figure 5). The 2002 Molise sequence, with hypocentral depths mainly between 10 and 20 km, appears to be confined to the lower parts of the crust. All focal solutions show nearly pure strike-slip mechanisms with tensional axes oriented NE-SW, roughly parallel to those of the normal faulting events in the southern Apennines (Figure 5). With these characteristics, the Molise sequence shows some similarities with the last Apenninic foredeep seismic sequence, occurred on May 5, 1990, near Potenza (Mw=5.8), southern Italy. The entire 1990 sequence was characterized by similarly 20 km-deep, strike-slip events. Both sequences occurred within the foredeep rather than within the Apennines chain.

During the Molise sequence, the aftershocks alignment suggests that the rupture occurred on the EW plane of focal solution, resulting in a dextral fault motion. The only known active structure in the general area with similar kinematics is the Mattinata fault (Figure 5), located in the Gargano Promontory, east of the region hit by the seismic sequence. It is unknown if this structure is continuous with the Molise fault structure beneath the sedimentary terrains that divise the Gargano Promontory and Molise.


Most material shown in the Introduction can also be seen on the Molise events special web page at INGV web site. Part of this work has been produced during the visit of E. Fukuyama (NIED, Japan) at INGV in Rome, during last November. Without the assistance of the MedNet Data Center group these results could not have be obtained. Thanks to Claudio Chiarabba and Massimo Di Bona (CNT-INGV) for providing aftershock locations and temporary deployment information. All maps are produced with GMT (Wessel and Smith, 1993).


  • Arvidsson, R. and G. Ekström, 1998. Global CMT Analysis of Moderate Earthquakes M>4.5 using Intermediate Period Surface Waves, Bull. Seismol. Soc. Am., 88, 1003-1013.
  • Boschi, E., E. Guidoboni, G. Ferrari, G. Valensise and P. Gasperini, 1997. Catalog of Strong Italian Earthquakes from 461 B.C. to 1990, SGA-INGV, Rome Italy.
  • CPTI Working Group, 1999. Catalogo Parametrico dei Terremoti Italiani, ING, GNDT, SGA, SSN, Bologna, 92 pp.
  • Dreger, D.S. and C.A. Langston, 1995. Short course lecture note of IRIS workshop on moment tensor inversion.
  • Dziewonski, A.M., G. Ekström and N.N. Maternovskaya, 2001. Centroid-moment tensor solutions for April-June 2000, Phys. Earth Planet. Inter., 123, 1-14.
  • Ekström, G., A. Morelli, E. Boschi and A.M. Dziewonski, 1998. Moment tensor analysis of the central Italy earthquake sequence of September-October 1997 Geophys. Res. Lett., 25, 1971-1974.
  • Fukuyama, E. and D.S. Dreger, 2000. Performance test of an automated moment tensor determination system for the future "Tokai" earthquake, Earth Planets Space, 52, 383-392.
  • Kubo, A., E. Fukuyama, H. Kawai and K. Nonomura, 2002. NIED seismic moment tensor catalogue for regional earthquakes around Japan: quality test and application, Tectonophys., 356, 23-48.
  • Morelli A., G. Ekström, S. Mazza, S. Pondrelli, E. Boschi and A.M. Dziewonski, 2000. Surface-wave Centroid Moment Tensors in the Mediterranean region: the MEDNET-Harvard project, ORFEUS Electronic Newsletter, Vol.2, 1, 4.
  • Pondrelli, S., A. Morelli, G. Ekström, S. Mazza, E. Boschi and A.M. Dziewonski, 2002. European- Mediterranean Regional Centroid Moment Tensors: 1997-2000, Phys. Earth Planet. Int., 130, 71-101.
  • T. van Eck et al., 2002. Towards a Virtual European Broadband Seismograph Network, ORFEUS Electronic Newsletter, Vol.4, 1, 5
  • Wessel, P. and W.H.F. Smith, 1998. New improved version of the Generic Mapping Tools Released, Eos Trans. AGU, 79, 579.

page 2
Copyright © 2003. Orfeus. All rights reserved.