Rise and fall of a hypothesized seismic gap: source complexity in the 16 December 1857, Southern Italy earthquake (Mw 7.0)

Transcription

1 Rise and fall of a hypothesized seismic gap: source complexity in the 16 December 1857, Southern Italy earthquake (Mw 7.0) Pierfrancesco Burrato* and Gianluca Valensise Istituto Nazionale di Geofisica e Vulcanologia, Rome, Italy *Corresponding author: Pierfrancesco Burrato, Via di Vigna Murata 605, Roma Italy, burrato@ingv.it, Abstract. On 16 December 1857 a strong earthquake (M~7) struck a large portion of the southern Apennines about 150 km southeast of Naples. The earthquake was thoroughly investigated by Irish engineer Robert Mallet, who wrote an extensive report that is still regarded as a landmark in observational Seismology. Due to the concentration of damage in the High Agri Valley, and contrary to Mallet s own findings, for many years the earthquake was referred to as Val d Agri earthquake and believed by most investigators to have ruptured the km normal fault lying beneath this intermontane basin. The magnitude of the earthquake, however, and evidence for earthquake complexity suggest that the true rupture length has been so far underestimated. We contend that the 1857 earthquake ruptured in a cascade fashion two adjacent and relatively well known faults: first the smaller Melandro- Pergola fault, commonly believed to represent a seismic gap between the causative faults of the 1857 and of the 1980 Irpinia (Mw 6.9) earthquakes; then the larger Agri Valley fault proper. Contemporary chronicles reported a time lapse of 2-3 minutes between the two ruptures, thus effectively making them two independent shocks. The rupture must have proceeded unilaterally from the northwestern edge of the Melandro-Pergola fault, where Mallet placed the earthquake epicenter, thus explaining the concentration of damage - and attention by rescue crews and subsequent investigators - in the southeastern portion of the High Agri Valley. Keywords: historical seismicity, Southern Italy, earthquake complexity, seismic gap, 1857 earthquake, Robert Mallet Burrato & Valensise, accepted, October 2007, BSSA. 1

2 1. Introduction On 16 December 1857 a catastrophic earthquake struck southern Italy, killing over 11,000 people (19,000 according to unofficial sources) and causing widespread destruction in the High Agri Valley (hereinafter HAV), Melandro-Pergola Valley (hereinafter MPV) and Vallo di Diano (hereinafter VD) (Figures 1 and 2). The earthquake generated considerable interest among contemporary European scholars, some of whom traveled a long way to survey the earthquake effects. Among them was Robert Mallet, the founder of Seismology as the science that investigates the shaking of the Earth and that he regarded as...the youngest branch of cosmical science... (Mallet, 1848). Soon after the earthquake Mallet sought a grant from the Royal Society of London that would allow him to make a reconnaissance journey through the region struck by the earthquake. He was prepared to devote...a month or five weeks to the enquiry..., and considered that... for this a sum of about One Hundred and Fifty Pounds would... be required... (from Mallet, 1862; page IX of Preface). Eventually Mallet s proposal was accepted, although the expedition turned out substantially more expensive and nearly twice as long as originally envisioned. Upon his return to England in April 1858 Mallet started collecting his experience in a volume (Mallet, 1862) that represents one of the most important contributions to the early development of Seismology and that has been recently been rediscovered and republished (Melville and Muir Wood, 1987; Ferrari, 2004). Mallet reached the HAV from the VD (Figure 2) crossing the southern termination of the Maddalena Mts. and passing by the Magorno Plain, which at that time was a lake (Lake Maorno). He spent a considerable amount of time in the VD inspecting heavy damage suffered by Polla, one of the largest settlements of the area, and a number of smaller villages. Unbeknown to him, he was only at the outskirts of the true meizoseismal area, the large concentration of damage having been caused by extensive and sustained site amplifications at Polla and its nearby villages (Mucciarelli et al., 1999). He later inspected all villages settled on both sides of the HAV, reporting the damage sustained by buildings as well as natural phenomena that accompanied the earthquake (Figure 2). In spite of the earthquake magnitude, similar or even greater than that of the surface breaking, 23 November 1980, Irpinia event (M w 6.9) that occurred a few tens of km further to the northwest (Figure 1), during his visit Mallet did not describe any evidence that could be positively ascribed to surface faulting. Burrato & Valensise, accepted, October 2007, BSSA. 2

3 The earthquake was generated along the well-known Apennines-top seismogenic trend (Pantosti and Valensise, 1988; Galadini et al., 2001; Valensise and Pantosti, 2001a). Although no instrumental evidence exists for such an old earthquake, its location, the characteristics of the modern seismic release in the region (e.g. Cucci et al., 2004; Vannucci and Gasperini, 2004; Moro et al., 2007), the nature of the modern stress field (Montone et al., 2004) and the abundant active tectonics evidence, all suggest that the 1857 event was caused by normal faulting along a NWtrending plane. Overall, the tectonic style of the 1857 earthquake appears to be very similar to that of the well-investigated 1980 earthquake, yet there has been considerable debate over the exact geometry and kinematics of its causative source (e.g. Benedetti et al., 1998; Borraccini et al., 2002; Cello et al., 2003; Galli et al., 2006; Maschio et al., 2005). Modern automatic analyses of intensity data yield an equivalent moment magnitude 7.0 (Gasperini et al., 1999; CPTI Working Group, 2004), making 1857 one of the largest Italian earthquakes of all times and implying a rupture length of roughly 50 km (Wells and Coppersmith, 1994) (see black rectangle in Figure 2). All macroseismic reconstructions, starting with the one elaborated by Robert Mallet, point at a very large meizoseismal area elongated in the NW-SE direction (Figure 3). The earthquake caused extensive damage over an exceptionally large area. Intensity X and larger (MCS scale) were reported over a 900 km 2 region (Figure 2). Most of the damage was suffered by the HAV, although intensity IX and X reports are spread over a region extending from the northern end of the VD to the Sant Arcangelo Basin (Figures 2, 3). The HAV is also the geographic area where Robert Mallet spent most of his time. As a result of these circumstances the 1857 is commonly referred to as the Val d Agri earthquake, in spite of it having generated considerable damage outside the HAV proper. Early geological investigations of the earthquake immediately focused on the 20-km stretch of the HAV extending from Marsico Nuovo to Grumento Nova as the locus of its causative fault (Pantosti and Valensise, 1988; Burrato, 1995; Burrato, 2001, in DISS database v 2.0: Valensise and Pantosti, 2001b), that is also where a faults scarp compatible with the 1857 earthquake has been encountered in paleoseismological investigations (D Addezio et al., 2006). This important geologic, structural and landscape feature, however, is limited in length and can in no way accommodate a seismogenic structure longer than about 25 km. Nearly all subsequent papers on the subject also concentrated on Val d Agri (Benedetti et al., 1998; Borraccini et al., 2002; Cello et al., 2003; Maschio et al., 2005), Burrato & Valensise, accepted, October 2007, BSSA. 3

4 implicitly assuming that the northern and southern terminations of the rupture coincide with the northwestern and southeastern physiographical boundaries of the basin. Subsequent investigations focused on the MPV, a basin located between the northern end of the HAV basin and the southern end of the rupture associated with the 23 November 1980 Irpinia earthquake (Figures 1, 2). The MPV is believed to host a large normal fault similar in style to those underlying Irpinia and Val d Agri (Burrato and Valensise, in DISS database v 2.0: Valensise and Pantosti, 2001b; Cucci et al., 2004; Lucente et al., 2005; Moro et al., 2007). The Melandro-Pergola basin was affected by the 1857 earthquake, but not as much as the Val d Agri proper. On these grounds, it was believed not to have ruptured in 1857, and therefore to be a good candidate for an impending, roughly M 6.5 earthquake (Valensise and Pantosti, 2001b; Montone, 2004; Lucente et al., 2005). A careful examination of 1857 earthquake reports suggests that a significant foreshock occurred about two minutes before the mainshock. According to Branno et al. (1983), this foreshock was a significant (M 6.0 or larger) event that produced damage in an area located to the north of Val d Agri and roughly corresponding with the Melandro-Pergola basin. This area falls within the largest intensity isoseismal as mapped by Mallet in 1858 (Figure 3), and almost entirely within the intensity X area of Boschi et al. (2000) (Figure 2). In this paper we reanalyze the rupture history of the 1857 earthquakes in light of all available evidence, and propose a new rupture scenario that explains both its severity and its damage pattern. 2. The earthquake intensity pattern and focal parameters Robert Mallet investigated in great detail both the distribution of damage induced by the earthquake and its source parameters. Figure 3 contains an excerpt of his Map A, showing the region that suffered the largest damage. Mallet subdivided the region... into four, more or less concentric areas, each marked by a determinate though arbitrary limit, of the seismic effort that acted within it... (Page 252, Part 2 in Mallet (1862); subsequent excerpts from the same source). Figure 3 shows what he referred to as...the meizoseismal area, within which the greater portion of the cities and towns were perfectly prostrated... (Page 253). An examination of the meizoseismal area Burrato & Valensise, accepted, October 2007, BSSA. 4

5 derived from the recent Catalogo dei Forti Terremoti in Italia (Boschi et al., 2000) (Figure 2) shows that the region that suffered the largest damage (intensity X and above) is substantially smaller than the simplified, over 80 km-long oval area envisioned by Mallet (Figure 3). The main discrepancies are seen at the northern end of the VD, where the large damage reported at Polla and neighboring localities is most likely affected by significant site amplifications (Mucciarelli et al., 1999), and in the MPV, where observed intensities appear to be somewhat but consistently smaller than in the HAV. Aside from these differences, both Mallet s and Boschi et al. s intensity distributions are rather symmetric with respect to the HAV and to the fault that is believed to lie beneath it. Although most investigators considered the 1857 a single large earthquake, a detailed examination of the available sources suggests that it was in fact a complex event composed by at least two large shocks. In the framework of a thorough reassessment of the 1857 intensity field, Branno et al. (1983) quoted Leopoldo Del Re, the Director of the Astronomical Observatory of Naples, according to whom... alle ore 10 e minuti 10 di Francia si è sentita una prima scossa di tremuoto della durata di quattro in cinque secondi, la quale è stata dopo due minuti seguita da altra di assai maggiore intensità e della durata di circa venticinque secondi... (... at 10 hours ten minutes French time - we (in Naples, about 150 northwest of the HAV) felt a first shock that lasted for 4-5 seconds, and that was followed two minutes later by a much stronger shock that lasted for about 25 seconds... ). Quite surprisingly, this characteristic of the 1857 earthquake was largely overlooked by Mallet, perhaps because it required the ability to interact with the local population, most of which was not educated and spoke Italian with a broad regional accent (conversely, Mallet did not speak Italian at all). The existence of two separate shocks was confirmed by a number of other sources, though more loosely. Based on all available evidence Branno et al. (1983) attempted a separation of the effects of the first shock from the overall intensity field (Figure 4). They initially singled out localities where the first shock was clearly felt as a distinct earthquake that induced a considerable level of damage (solid circles and solid squares in Figure 4). Their reconstruction shows that the first shock affected the northern portion of Mallet s meizoseismal area, between the villages of Balvano to the northwest and Marsico Nuovo to the southeast. This roughly circular area locates entirely to the northeast of the surface projection of the northeast-dipping fault contained in the DISS database as Melandro-Pergola Source and investigated in detail by Cucci Burrato & Valensise, accepted, October 2007, BSSA. 5

6 et al. (2004). Interestingly, the southeastern end of this portion of the meizoseismal area nearly coincides with the physiographical saddle that separates the Melandro-Pergola basin from the Agri basin and with the gap between the two adjacent seismogenic sources mapped in the DISS database (Figure 4). The earthquake magnitude was estimated by at least three independent groups of investigators using widely different techniques. Table 1 summarizes the assessments available to date, showing that the 1857 was indeed a large earthquake, at least from the point of view of the extent of the area it affected. Recall that the 23 November 1980, Irpinia earthquake is normally assigned a M w 6.9 and that according to the Catalogo Parametrico dei Terremoti Italiani or CPTI (CPTI Working Group, 2004) the 1857 is the sixth largest earthquake in Italian history. The investigation of the exact earthquake location comprises a significant fraction of Mallet s work. Part III of his volume, entitled Deductions and conclusions, contains a number of chapters dedicated to the investigation of the earthquake focal parameters, of wave velocity, of anomalous sounds and precursory shocks. In this paper we are especially concerned with the determination of the epicenter. Mallet carried out 177 observations relating to the wavepath direction at 78 localities and developed a complex methodology to determine the site where the earthquake had presumably originated. In some instances he was able to observe also the angle of emergence of wavepaths, so that he actually determined the location and depth of the earthquake focal cavity, as he called it, not just an epicenter. Mallet found that 16 wavepaths met within a circle having a radius of 500 yards (about 450 m), and 32 within a concentric circle of one mile (about 1850 m), located at the northern end of the meizoseismal area (white star in Figure 3). He then stated that It will be obvious at once that the great mass of the wave-paths radiate from Caggiano, or close about it... (Page 237), and that We have thus ascertained, the point of the surface vertically above the seismic focus, which we find to pass nearly through Caggiano... (Page 247). Given the nature of the method, this has to be intended as the surface projection of the point of rupture nucleation rather than as the centroid of damage. This location falls about 5 km northwest of the region of largest ground shaking as indicated by modern analyses of the intensity pattern (Figure 2). Mallet s determination of the epicenter is made more robust by a consistent determination of the hypocentral depth:... The maximum depth is 8 ½ geographical miles, or 49,359 feet, and the minimum depth is 2 ¾ geographical miles, or 16,705 feet. (Part 2, Page 251). This depth interval Burrato & Valensise, accepted, October 2007, BSSA. 6

7 corresponds to a range between 5.1 and 15.7 km, not far from the depth that releases most of the seismic moment in large Apennines earthquakes. For example, the instrumentally-determined depth interval reported by the DISS database for the 1980 Irpinia earthquake ( Colliano Source in Figure 1) is km. 3. Clues to a complex rupture scenario As mentioned earlier, most investigators including the compilers of the DISS database have so far regarded the 1857 earthquake has having been generated by a km-long fault lying beneath the HAV. Conversely, the km-long fault underlying the Melandro-Pergola basin, located north of the HAV fault along the extensional backbone of the Apennines, has often been regarded as a long-quiescent structure that deserved to be considered with caution in Seismic Hazard Assessment (SHA) analyses (e.g. DISS database v 2.0: Valensise and Pantosti (2001b); Montone (2004); Lucente et al. (2005)). We contend that the 1857 earthquake was a complex event caused by the rupture of both these adjacent, kinematically compatible faults that had experienced a long quiescence prior to the XX century. Our conclusion is based on three independent lines of evidence. 1) Earthquake magnitude and rupture length. Under the assumption that the depth of the seismogenic layer in the region hit by the 1857 earthquake is similar to that characterizing the region of the 1980 Irpinia earthquake, and that coseismic slip in 1857 was also in the same order of magnitude as in 1980 ( m), the HAV and MPV faults are assigned a potential for M w 6.7 an 6.6, respectively, using Hanks and Kanamori (1979) equation for converting M 0 into M w (see Table 2). These geology-based magnitudes must be compared with the estimates obtained for 1857 from intensity data (Table 1). Even rupturing both faults in a single large earthquake results in a magnitude that is at least 0.1 M units smaller than the 7.0 estimate obtained from intensity data alone by the compilers of the CPTI catalogue (CPTI Working Group, 2004) using the automatic technique developed by Gasperini et al. (1999) (Boxer code: available for download from /boxer.html). Aside from the numerous uncertainties that characterize the calculation, the mismatch suggests not only that the Burrato & Valensise, accepted, October 2007, BSSA. 7

8 participation of both the HAV and MPV faults to the 1857 earthquake is a viable scenario, but also that the length, width or coseismic slip of the fault may still be underestimated. The relationships derived by Wells and Coppersmith (1994) for normal faulting suggest that a M earthquake requires rupturing of a 44 to 68 km-long fault (assuming the width and coseismic slip specified in Table 2). There is therefore ample room for a 1857 rupture involving both adjacent segments, even if the magnitude estimated by Branno et al. (1983) is assumed (M 6.84). Notice that the 1980 Irpinia earthquake was definitely a complex event caused by the rupture of a 38 km section of the Apennines seismogenic zone, plus a further 10 km stretch along an antithetic fault (Bernard and Zollo, 1989, Pantosti and Valensise, 1990). The CPTI catalogue (CPTI Working Group, 2004) rates this earthquake M w 6.9 using the same magnitude assessment technique used for ) Evidence for earthquake complexity. The information brought forward by Baratta (1901) and Branno et al. (1983) on the exact timing of the mainshock and the reconstruction of the damage associated with the first shock presented by Branno et al. (1983) (Figure 4) supply compelling evidence for faulting complexity. Earthquake source complexity is a common occurrence in Italian earthquakes, although the time lapse between subsequent shocks may vary substantially. Table 3 summarizes 9 instrumentally documented sequences that a) have occurred in Italy over the past 45 years, b) had M w > 5.5, and c) were positively characterized by source complexity or multiple shocks occurring close in time and space (Figure 5). Our reference catalogue (CPTI Working Group, 2004) lists 22 individual earthquakes (belonging to 19 sequences) with M w > 5.5 that have occurred in Italy and adjacent seas during this time span, suggesting that nearly 50% of the available sample but possibly more - is not made up by isolated shocks. Many other examples are found in the historical record, although most of them are too old to be easily resolved. 3) Earthquake epicentral location. Mallet s epicenter falls within 5 km of the northern end of Melandro-Pergola seismogenic source listed by the DISS database (Figures 2, 3). Even allowing for a substantially greater epicentral uncertainty than that admitted by Mallet, it is remarkable that (a) the presumed rupture nucleation falls near one end of the meizoseismal area, and (b) the location of Mallet s epicenter falls about 30 km from the CPTI catalogue location (Figure 2). Burrato & Valensise, accepted, October 2007, BSSA. 8

9 This evidence suggests that the 1857 may have ruptured in a cascade fashion the MPV fault first, then the larger HAV fault (Figure 6). The rupture would have initiated from the northwestern edge of the MPV fault and propagated unilaterally towards the southeast. According to McGuire et al. (2002), predominantly unilateral rupture propagation is seen in approximately 80% of large shallow earthquakes ruptures, and is therefore a likely rupture mode also for Discussion and conclusions We supplied evidence suggesting that the 1857 earthquake was caused by the cascade-style rupture of two rather large fault segments, together forming a ~40 km-long complex seismogenic source (Figure 6). We found that the geologically-estimated cumulative moment associated with the two faults rupturing in a rapid sequence justifies the extent of the 1857 meizoseismal area, Mallet s presumed location of the rupture initiation and the global seismic moment inferred from routine elaboration of intensity data. If the Melandro-Pergola Valley (MPV) and High Agri Valley (HAV) faults both ruptured in 1857, the MPV fault should no longer be considered a seismic gap, suggesting that a ~100 kmlong section of the Appennines seismogenic backbone has ruptured entirely over the past 150 years (Figure 1). This finding may indeed have significant implications for the local seismic hazard, especially considering a) the lack of historical seismicity in the M range , which suggests that the seismic release is dominated by large characteristic earthquakes, and b) that the typical recurrence interval of large southern Apennines faults is in the order of two millennia (Pantosti et al., 1993; Galadini et al., 2001; DISS Working Group, 2007). The timing of the two 1857 subevents raises an interesting and very general question. The 2-3 minute lapse between shocks puts the 1857 earthquake in between two well-known endmembers of Italian earthquake complexity: the 1980 Irpinia earthquake, that ruptured three discrete faults within about 40 seconds (Table 3), and the catastrophic 1783 earthquake in southern Calabria, that took 48 hours to rupture a ~60 km stretch of the Calabrian Arc extensional zone with three distinct events (Boschi et al., 2000; DISS Working Group, 2007). Similarly to the case of the 1857 earthquake, in both instances the ruptured faults were almost Burrato & Valensise, accepted, October 2007, BSSA. 9

10 perfectly aligned and shared a similar geometry at depth, thus providing optimal conditions for earthquake triggering by Coulomb stress transfer (see Nostro et al., 1997, for 1980 and Jacques et al., 2001, for 1783). Overall, 1857 appears to be more similar to the well-documented case of 1980, a true paradigm of earthquake complexity; but 1980 has always been regarded as one single shock, whereas a 2-3 minutes time lapse is long enough that the two shocks can be considered as separate ones, at least from the seismological point of view. Understanding whether the 1857 earthquake actually shook as a single large earthquake or as two separate smaller events would require a source-time function or information on the 1857 coseismic slip, all of which is unfortunately not available. Nevertheless, the reiteration of similarly large though moderate ground shaking through two similarly large subsequent earthquakes has certainly contributed to increase the level of damage in the meizoseismal area. The strong asymmetry of the earthquake effects with respect to the presumed causative faults larger damage in the southern HAV, smaller in the MPV (Figures 2, 3, 4) can hardly be explained by the relatively limited site effects detected in the HAV (Gallipoli et al., 2003). Rather, it is likely a result of fault directivity, similarly to what is seen in most crustal earthquakes worldwide (e.g., McGuire et al., 2002). This circumstance could be tested by generating a scenario earthquake and superimposing it on the observed intensity field. Unfortunately this characteristic of the 1857 earthquake is partially traded off with the lower density of villages in the MPV and with the fact that Mallet spent most of his time in the investigation of earthquake effects in the HAV and never went to the MPV. Our findings also show that the rendering of large historical earthquakes by conventional parametric catalogues can be very crude and misleading, as also shown by a recent reassessment of the 1456 earthquake, the largest to have ever occurred in peninsular Italy (Fracassi and Valensise, 2007). It is well-known that the location of the instrumental and macroseismic epicenters may differ substantially, but the case of 1857 shows that for a large crustal earthquake (i.e. one generated by a long fault rupture) characterized by unilateral fault rupture the mismatch between the two locations can be 30 km or more (Figure 2). This is generally not an issue for SHA practice if the calculations are based on consistently prepared catalogues (for instance, earthquake locations presented in the CPTI catalogue were purposely obtained from intensity data even if instrumental data were available), but it may become an issue when using mixed historical-instrumental catalogues. The issue is far more serious in Burrato & Valensise, accepted, October 2007, BSSA. 10

11 areas characterized by significant earthquakes, such as most Mediterranean countries. Treating earthquakes as point sources may be an acceptable practice in areas where earthquakes do not normally exceed M 6, such as in central and northern Europe, but is definitely not a satisfactory approximation when dealing with km-long faults. Recall that in conventional SHA practice based on seismogenic zoning, earthquakes are treated as point sources that are allowed to travel through big areas, irrespective of the actual length and geometry of the associated fault. To account at least partially for this physical drawback seismogenic zones must be drawn substantially longer than one fault length. This issue has been recently addressed in the context of the new seismic hazard map of Italy by Meletti et al. (2007), who drew larger seismogenic zones to accommodate longer earthquake ruptures and improve the reliability of earthquake rates. Acknowledgments We wish to thank Graziano Ferrari for suggestions on an earlier version of the manuscript, and an anonymous reviewer. The paper benefited from discussion with components of the DISS Working Group. This work was funded by the project "Assessing the seismogenic potential and the probability of strong earthquakes in Italy" funded by the Italian Civil Defense Department within the agreement with INGV, grant to P. Burrato. Burrato & Valensise, accepted, October 2007, BSSA. 11

12 References Basili, R., G. Valensise, P. Vannoli, P. Burrato, U. Fracassi, S. Mariano, and M. M. Tiberti (2007). The Database of Individual Seismogenic Sources (DISS), version 3: summarizing 20 years of research on Italy s earthquake geology, 2 nd Version submitted to Tectonophysics, April Benedetti, L., P. Tapponier, G. C. P. King, and L. Piccardi (1998). Surface rupture of the 1857 southern Italian earthquake? Terra Nova 10 (4), Bernard, P., and A. Zollo (1989). The Irpinia (Italy) 1980 earthquake: detailed analysis of a complex normal fault, J. Geophys. Res. 94, 1,631-1,648. Borraccini, F., M. De Donatis, D. Di Bucci, and S. Mazzoli (2002). 3D Model of the active extensional fault system of the high Agri River valley, Southern Apennines, Italy, Journal of the Virtual Explorer 6, 1-6. Boschi, E., E. Guidoboni, G. Ferrari, D. Mariotti, G. Valensise, and P. Gasperini (Editors) (2000). Catalogue of strong italian earthquakes from 461 B.C. to 1997 (Version III), Ann. Geophys. 43, with CD-Rom, 259 pp. Bossolasco, M., and C. Eva (1965). Il terremoto del 19 luglio 1963 nel Mar Ligure, Geof. e Meteor. 14, Branno, A., E. Esposito, A. Marturano, S. Porfido, and V. Rinaldis (1983). Studio, su base macrosismica, del terremoto della Basilicata del 16 dicembre 1857, Bollettino della Società dei Naturalisti di Napoli 92, Burrato, P. (1995). Tettonica attiva, sismogenesi e caratteri evolutivi del reticolo idrografico: tre esempi dall Italia meridionale, Unpublished MS thesis, 69 pp and 26 images. Camassi, R., and M. Stucchi (1997). NT4.1, un catalogo parametrico di terremoti di area italiana al di sopra della soglia del danno (published by GNDT, Milano), pp. 95 (also available at: NT/home.html). Cello, G., E. Tondi, L. Micarelli, and L. Mattioni (2003). Active tectonics and earthquake sources in the epicentral area of the 1857 Basilicata earthquake (southern Italy), J. Geodyn., 36, 37 50, doi: /S (03) Chiarabba, C., P. De Gori, L. Chiaraluce, P. Bordoni, M. Cattaneo, M. De Martin, A. Frepoli, A. Michelini, A. Monachesi, M. Moretti, G. P. Augliera, E. D Alema, M. Frapiccini, A. Gassi, S. Marzorati, P. Di Bartolomeo, S. Gentile, A. Govoni, L. Lovisa, M. Romanelli, G. Ferretti, M. Pasta, D. Spallarossa, and E. Zunino (2005). Mainshocks and aftershocks of the 2002 Molise seismic sequence, southern Italy, J. Seism. 9, , doi: /s Chiaraluce, L., A. Amato, M. Cocco, C. Chiarabba, G. Selvaggi, M. Di Bona, D. Piccinini, A. Deschamps, L. Margheriti, F. Courboulex, and M. Ripepe (2004). Complex normal faulting in the Apennines thrust-and-fold belt: the 1997 seismic sequence in Central Italy, Bull. Seism. Soc. Am., 94, ; doi: / CPTI Working Group (2004). Catalogo Parametrico dei Terremoti Italiani, version 2004 (CPTI04). INGV, Milan, available from Cucci, L., S. Pondrelli, A. Frepoli, M. T. Mariucci, and M. Moro (2004). Local pattern of stress field and seismogenic sources in the Pergola-Melandro basin and the Agri valley (Southern Italy), Geophys. J. Int. 156, D Addezio, G., D. B. Karner, P. Burrato, D. Insinga, L. Maschio, L. Ferranti, and P. R. Renne (2006). Faulted Middle Pleistocene tephra layer in the Val d Agri area (Southern Italy), Ann. Geophys. 49 (4-5), 1,029-1,040. DISS Working Group (2007). Database of Individual Seismogenic Sources (DISS), Version 3.0.4: A compilation of potential sources for earthquakes larger than M 5.5 in Italy and surrounding areas, INGV 2005, Ekström, G. (1994). Teleseismic analysis of the 1990 and 1991 earthquakes near Potenza, Ann. Geofis. 37, Ferrari, G. (Editor) (2004). Viaggio nelle aree del terremoto del 16 dicembre L opera di Robert Mallet nel contesto scientifico e ambientale attuale del Vallo di Diano e della Val d Agri. SGA Storia Geofisica Ambiente, 2 Volumes and DVD-ROM, Bologna. Burrato & Valensise, accepted, October 2007, BSSA. 12

13 Finetti, I., F. Giorgetti, H. Haessler, P. Hoang Trong, D. Slejko, and G. Wittlinger (1976). Time-space epicenter and hypocenter distribution and focal mechanism of 1976 Friuli earthquakes, Boll. Geofis. Teor. Appl. 19, Fracassi, U., and G. Valensise (2007). Unveiling the sources of the catastrophic 1456 multiple earthquake: hints to an unexplored tectonic mechanism in southern Italy, Bull. Seism. Soc. Am. 97 (3), , doi: / Galadini, F., C. Meletti, and E. Vittori (2001). Major active faults in Italy: available surficial data, Netherlands J. Geosci. 80, Galanopoulos, A. G. (1961). On magnitude determination by using macroseismic data, Ann. Geofis. 14 (3), Galli, P., V. Bosi, S. Piscitelli, A. Giocoli, and V. Scionti (2006). Late Holocene earthquakes in southern Apennines: paleoseismology of the Caggiano fault, Int. J. Earth Sci. (Geol. Rundsch.), doi: /s Gallipoli, M. R., M. Mucciarelli, D. Albarello, V. Lapenna, M. Schiattarella, and G. Calvano (2003). Hints about site amplification effects comparing macroseismic hazard estimate with microtremor measurements: the Agri Valley (Italy) example, J. Earthq. Eng. 7 (1), 51-72, doi: /s Gasparini, C., G. Iannaccone, and R. Scarpa (1985). Fault plane solutions and seismicity of the Italian peninsula, Tectonophysics 117, Gasperini, P., F. Bernardini, G. Valensise, and E. Boschi (1999). Defining seismogenic sources from historical earthquake felt reports, Bull. Seism. Soc. Am. 89, Guidoboni, E., and G. Ferrrari (1987). Robert Mallet, first modern seismologist, in E. Guidoboni and G. Ferrari (Editors), Mallet s macroseismic survey on the Neapolitan earthquake of 16th december 1857, anastatic reprint of the original volume, ING, Bologna 1987, Hanks, T. C., and H. Kanamori (1979). A moment magnitude scale, J. Geophys. Res. 84, Jacques, E., C. Monaco, P. Tapponnier, L. Tortorici, and T. Winter (2001). Faulting and earthquake triggering during the 1783 Calabria seismic sequence, Geophys. J. Int. 147 (3), , doi: /j x x. Lucente, F.P., N. Piana Agostinetti, M. Moro, G. Selvaggi, and M. Di Bona (2005). Possible fault plane in a seismic gap area of the Southern Apennines (Italy) revealed by receiver function analysis, J. Geophys. Res., 110 (B04307), doi: /2004JB Mallet, R. (1848). Dynamics of earthquakes, Trans. R. Irish Acad. 21, Mallet, R. (1862). The great Neapolitan earthquake of The first principles of observational seismology, Chapman and Hill (Publ.), London. Maschio, L., L. Ferranti, and P. Burrato (2005). Active extension in Val d'agri area, Southern Apennines, Italy: implications for the geometry of the seismogenic belt, Geophys. J. Int. 162 (2), , doi: /j x x. McGuire, J. J., Li Zhao, and T. H. Jordan (2002). Predominance of unilateral rupture for a global catalog of large earthquakes, Bull. Seism. Soc. Am. 92 (8), , doi: / Meletti, C., F. Galadini, G. Valensise, M. Stucchi, R. Basili, S. Barba, G. Vannucci, and E. Boschi (2007). The ZS9 seismic source model for the seismic hazard assessment of the Italian territory, Submitted to Tectonophysics, February Melville, C., and R. Muir Wood (1987). Robert Mallet, first modern seismologist, in E. Guidoboni and G. Ferrari (Editors), Mallet s macroseismic survey on the Neapolitan earthquake of 16th december 1857, anastatic reprint of the original volume, ING, Bologna 1987, Montone, P. (Editor) (2004) Task 1.4: characterization of seismogenic sources in potential gap areas. In: Amato, A., and G. Selvaggi, final report of project Terremoti probabili in Italia tra l'anno 2000 e il 2030: elementi per la definizione di priorità degli interventi di riduzione del rischio sismico, funded by the Italian Civil Protection (available at: Amato_Selvaggi/prodotto_12/TASK1.4.pdf). Burrato & Valensise, accepted, October 2007, BSSA. 13

14 Montone, P., M. T. Mariucci, S. Pondrelli, and A. Amato (2004). An improved stress map for Italy and surrounding regions (Central Mediterranean), J. Geophys. Res. 109, B10410, doi: /2003jb Moro, M., L. Amicucci, F. R. Cinti, F. Doumaz, P. Montone, S. Pierdominici, M. Saroli, S. Stramondo, and B. Di Fiore (2007). Surface evidence of active tectonics along the Pergola-Melandro fault: a critical issue for the seismogenic potential of the southern Apennines, Italy, J. Geodyn. 44, 19 32, doi: /j.jog Mucciarelli, M., G. Valensise, M. R. Gallipoli, and R. Caputo (1999). Reappraisal of a XVI century earthquake combining historical, geological and instrumental information, In: Castelli, V. (Editor), Proc. of the First Workshop of thr WG Historical Seismology of the European Seismological Commission (Macerata, Italy, 1-5 September 1999), Nostro, C., M. Cocco, and M. E. Belardinelli (1997). Static stress changes in extensional regimes: an application to southern Apennines (Italy), Bull. Seism. Soc. Am. 87 (1), Pantosti, D., and G. Valensise (1988). La faglia sud-appenninica: identificazione oggettiva di un lineamento sismogenetico nell'appennino meridionale, Proc. VII Meeting G.N.G.T.S., Rome 1988, Pantosti, D., and G. Valensise (1990). Faulting mechanism and complexity of the November 23, 1980, Campania-Lucania earthquake, inferred from surface observations, J. Geophys. Res. 95, Pantosti, D., D. P. Schwartz, and G. Valensise (1993). Paleoseismology along the 1980 surface rupture of the Irpinia fault: implications for earthquake recurrence in the southern Apennines, Italy, J. Geophys. Res. 98, Valensise, G., and D. Pantosti (2001a). The investigation of potential earthquake sources in peninsular Italy: a review, J. Seismol. 5, Valensise, G., and D. Pantosti (Editors) (2001b). Database of Potential Sources for Earthquakes Larger than M 5.5 in Italy (DISS version 2.0), Ann. Geofis. 44/4, Suppl. 1, , with CD-ROM. Vannucci, G., and P. Gasperini (2004). The new release of the database of Earthquake Mechanisms of the Mediterranean Area (EMMA Version 2), Ann. Geophys. 47(1), Suppl. 1, Wells, D. L., and K. J. Coppersmith (1994). New Empirical Relationships among Magnitude, Rupture Length, Rupture Width, Rupture Area, and Surface Displacement, Bull. Seism. Soc. Am. 84, Westaway, R., and J. Jackson (1987). The earthquake of 1980 November 23 in Campania-Basilicata (southern Italy), Geophys. J. R. astr. Soc. 90, Westaway, R. (1987). The Campania, Southern Italy, earthquakes of 1962 August 21, Geophys. J. R. astr. Soc. 88, Westaway, R., R. Gawthorpe, and M. Tozzi (1989). Seismological and field observations of the 1984 Lazio- Abruzzo earthquakes: implications for the active tectonics of Italy, Geoph. J. R. astr. Soc. 98, Burrato & Valensise, accepted, October 2007, BSSA. 14

15 TABLES Table 1 Summary of available magnitude estimates for the 1857 earthquake Data source I 0 I max M Magnitude type and notes Branno et al., 1983 X NT4.1 (Camassi and Stucchi, 1997) CFTI v. III (Boschi et al., 2000) CPTI04 (CPTI Working Group, 2004) Based on felt area following the method proposed by Galanopoulos (1961) X-XI XI 7.0 M m (macroseismic magnitude obtained from I 0) XI XI 7.0 M e ( equivalent magnitude obtained from all available intensities using the Boxer code: Gasperini et al. (1999). Equated to Mw by regression using intensity data available for instrumentally-recorded earthquakes Table 2 Summary of fault dimensions for the potential causative sources of the 1857 earthquake (from DISS database v. 3.04: DISS Working Group, 2007). Source name/ fault name Fault Length (km) Fault Width (km) Maximum faulting depth (km) Expected M 0 (N/m)* Expected M w** Expected M w*** Melandro- Pergola E Agri Valley E Both faults E * Assumes an average coseismic slip of 1.5 m for similarity with the adjacent causative fault of the 23 November 1980 earthquake. ** Based on M 0 following Hanks and Kanamori (1979). *** Based on fault area following Wells and Coppersmith (1994). Burrato & Valensise, accepted, October 2007, BSSA. 15

16 Table 3 - Instrumental earthquake sequences characterized by similarly large multiple main shocks (see Figure 5). Id Locality Year Description Fault kinematics Reference(s) 1 Irpinia (southern Apennines) 1962 Three large main shocks with Ml 5.7, 6.1, and 6.0 struck on 21 August at 18:09, 18:19, and 18:44, respectively Normal faulting Gasparini et al., 1985; Westaway, Ligurian Sea 1963 Two closely spaced Ms 5.9 shocks one minute apart Thrust faulting Bossolasco and Eva, Belice (western Sicily) 1968 Four shocks of magnitude hit between 15 and 25 January Thrust faulting Gasparini et al., Friuli (northern Italy) Irpinia (southern Apennines) Southern Abruzzo (central-southern Apennines) Potenza (southern Apennines) Umbria-Marche (central Italy) Molise (southern Italy) Three shocks with Ml 6.3, 5.5, 5.3 on 6 and 11 May, followed by four shocks with Ml 5.5, 5.9, 6.1, 6.0 between 11 and 15 September Three separate subevents occurred on 23 November at 0, 20 and 40 s from rupture initiation. Global Mw is 6.9 Two similarly large earthquakes (Mw 5.9) occurred on 7 and 11 May A Ml 5.8 shock on 5 May triggered a sequence that continued for over a year, culminating with a similarly sized event on 26 May 1991 Two Mw 5.7 and 6.0 shocks occurred on 26 September, followed by a further Mw 5.6 shock on 14 October Two Mw 5.7 earthquakes struck on 31 October and 1 November, triggering a swarm-like sequence Thrust faulting Normal faulting Normal faulting Strike-slip faulting Normal faulting Strike-slip faulting Finetti et al., 1976 Westaway and Jackson, 1987; Bernard and Zollo, 1989 Westaway et al., 1989 Ekström, 1994 Chiaraluce et al., 2004 Chiarabba et al., 2005 Burrato & Valensise, accepted, October 2007, BSSA. 16

17 FIGURE CAPTIONS Figure 1. Distribution of Southern Apennines seismicity from the Catalogo dei Forti Terremoti in Italia (Boschi et al., 2000), and surface projection of individual seismogenic sources from DISS database, shown as rectangles (DISS Working Group, 2007; Basili et al., 2007; The line next to the rectangle is the fault cut-off. The stick inside each rectangle indicates the sense of slip. Figure 2. Intensities available for the 16 December 1857 earthquake (MCS scale) from a study reported in the Catalogo dei Forti Terremoti in Italia (Catalogue of Strong Italian Earthquakes: Boschi et al., 2000), plotted over the Melandro-Pergola (to the northwest) and Agri Valley (to the southeaest) seismogenic sources from the DISS database (same as in Figure 1). A white dashed line contours all intensities X and above. The black rectangle is the macroseismic source derived by automatic analysis of intensity data (Gasperini et al., 1999). The stars numbered 1, 2 are the epicenters proposed by Mallet and obtained by automatic analysis (Gasperini et al., 1999; Boschi et al., 2000), respectively. The black line with arrows highlights the route followed by Mallet in the Vallo di Diano and High Agri Valley (see text). Basins: HAV, High Agri Valley; MPV, Melandro-Pergola Valley; SAB, Sant Arcangelo Basin; VD, Vallo di Diano. Localities: CA, Caggiano; GN, Grumento Nova; MN, Marsico Nuovo; MO, Montemurro; PO, Polla. Figure 3. Central portion of Mallet s Map A, showing the meizoseismal region of the 1857 earthquake (outlined with a single dot-and-dash contour), 177 wavepaths inferred for 78 individual localities, the 1-mile radius circle within which 32 wavepaths meet and the proposed epicenter (solid white star filling the circle). See text for further details. Relevant seismogenic sources from DISS database also shown for reference. From Mallet (1862), modified. Figure 4. Intensity VIII contour for the 1857 earthquake and the area (highlighted with diagonal bands) where the first shock was distinctly felt (from Branno et al., 1983, redrawn). Type 1 and 2 localities are those where people were able to distinguish the two shocks (solid circles and solid squares, respectively); Type 3 are those localities where only one large shock was felt; Type 4 are localities not evaluated. Notice that the intensity pattern obtained by Branno et al. is different from that later published by Boschi et al. (2000), shown in Figure 2. Also shown are Quaternary basins (outlined by a hachured pattern), relevant DISS seismogenic sources (same as in previous figures) and Mallet s proposed epicenter (shown by a star: same as in Figure 3). Basins: MPV, Melandro-PergolaValley; SAB, Sant Arcangelo basin; HAV, High Agri Valley; VD, Vallo di Diano Basin. Localities: BA, Balvano; MN, Marsico Nuovo. Burrato & Valensise, accepted, October 2007, BSSA. 17

18 Figure 5. Location of the instrumental earthquake sequences characterized by similarly large complex or multiple mainshocks (solid stars) and Seismogenic Areas from DISS database (DISS Working Group, 2007), color-coded according to faulting mechanism. NF: normal faulting; TF: thrust faulting; SS (RL): right-lateral strike-slip faulting; SS (LL): left-lateral strike-slip faulting. Earthquakes sequences are numbered according to Table 3. Figure 6. Scheme of the proposed cascade-style rupture for the 1857 earthquake. The arrows show the hypotesized rupture directivity that could explain comparatively larger damage in the southeastern portion of the felt area. The stars marked with 1 and 2 are Mallet s epicenter of the first shock (see Figure 3) and our proposed nucleation for the second shock, respectively. Burrato & Valensise, accepted, October 2007, BSSA. 18

19 FIGURES Figure 1 Burrato & Valensise, accepted, October 2007, BSSA. 19

20 Figure 2 Burrato & Valensise, accepted, October 2007, BSSA. 20

21 Figure 3 Burrato & Valensise, accepted, October 2007, BSSA. 21

22 Figure 4 Burrato & Valensise, accepted, October 2007, BSSA. 22

23 Figure 5 Burrato & Valensise, accepted, October 2007, BSSA. 23

24 Figure 6 Burrato & Valensise, accepted, October 2007, BSSA. 24