Active oblique extension in the central Apennines (Italy): evidence from the Fucino region

Transcription

1 Geophys. J. Int. (1999) 139, Active oblique extension in the central Apennines (Italy): evidence from the Fucino region Luigi Piccardi,1 Yves Gaudemer,2 Paul Tapponnier2 and Mario Boccaletti3 1 CNR, Centro di Studio di Geologia dell Appennino e Catene Perimediterranee, Firenze, Italy 2 Institut de Physique du Globe de Paris, L aboratoire de T ectonique, Paris, France. tappon@ipgp.jussieu.fr 3 Università degli Studi di Firenze, Dipartimento di Scienze della T erra, Firenze, Italy Accepted 1999 July 7. Received 1999 July 7; in original form 1997 July 7 SUMMARY The Fucino Basin is a flat Quaternary depression within the central Apennines. It is surrounded by active normal faults with small oblique-slip components. Surface faulting was observed along the east side of the basin during the M =7.0 Avezzano earthquake s of In order to understand the kinematics of recent strain better and to assess the seismic hazard of this region, we carried out a detailed geomorphic and structural study of the faults around the basin. Fault scarp heights were accurately measured with total station profiles, and vertical slip rates were estimated assuming slope offsets to postdate the end of periglacial abrasion (14±4 kyr BP). To the north, the most prominent fault, the Magnola fault, appears to have an average postglacial throw rate of 0.7±0.3 mm yr 1 and, together with the Velino fault, to be capable of generating earthquakes of maximum magnitudes of with recurrence intervals of yr. East of the basin, the Serrone, Parasano and Ventrino faults form a right-stepping horsetail of the Giovenco fault. Current vertical slip rates on the three former faults appear to be between 0.5 and 1.4, 0.5 and 1.0, and 0.3 and 1.1 mm yr 1, respectively. Infrequent maximum-magnitude earthquakes on them may also exceed 7, particularly in the case of coupled rupture, with loosely constrained recurrence intervals (up to several thousand years). The right-lateral slip components implied by the most recent slickensides and by geomorphic offsets on the NW SE-trending normal faults of the area suggest that the blocks they bound rotate counterclockwise, consistent with oblique left slip on the NNW SSE-trending Giovenco and Ovindoli faults to the east. Sinistral shear parallel to the latter faults, the maximum relief across the Magnola fault, and the postglacial slope offsets measured suggest that the extension rate across the central Apennines might be of the order of 6±3mmyr 1 in a N20 ±10 E direction, more northerly than hitherto inferred. Key words: central Apennines, extension, normal faulting, oblique slip, seismic hazard. INTRODUCTION The Apennines are one of the most seismically active areas in the Mediterranean. Throughout history, this range has been rocked by many destructive earthquakes with intensities of up to XI and magnitudes of up to M=7 (Baratta 1901; CNR 1985a,b; Armijo et al. 1985; Guidoboni 1989; Boschi et al. 1997). Most of the earthquakes for which a mechanism is known have ruptured normal faults (McKenzie 1972; Westaway & Jackson 1984; Tapponnier et al. 1986; Pantosti & Valensise 1990). The present-day tectonic regime is thus markedly different from that which created the mountain range in Mio- Pliocene times. Coseismic surface faulting has only been rarely described (Oddone 1915; Westaway & Jackson 1984; Benedetti 1999 RAS et al. 1998). Hence, the correspondence between active faults and earthquakes is poorly known, even for relatively large events. Knowledge of the faults associated with earthquakes, however, is fundamental for evaluating the rates and styles of present-day deformation and for assessing seismic hazard. Here we document evidence of postglacial movements on several large faults of the central Apennines. Our study essentially concerns the part of the Abruzzi that surrounds the Fucino Basin, where the 1915 Avezzano earthquake occurred, and is based on the combined use of satellite images, aerial photographs, topographic maps and field measurements. The geometry, structure and morphology of fault scarps are mapped at different scales. We estimate slip directions and rates, attempt to place bounds on plausible earthquake magnitudes, and 499

2 500 L. Piccardi et al. compare them with trench results and historical records. Our the range is km (Bally et al. 1986; Nicolich 1989), while measurements are then used to discuss the direction and rate lithospheric thickness is up to 90 km (Müller & Panza 1986). of present-day extension across the whole range at 42 N, The Apennines formed as a thrust belt with NE shortening 13.5 E. and NE-vergent thrusting during Mio-Pliocene times, while the Adria plate subducted beneath the Tyrrhenian active margin. The overall structure is thus characterized by complex SUMMARY OF THE STRUCTURE AND stacking of various thrust slices, with thick units of Mesozoic GEOLOGICAL HISTORY OF THE carbonates (Apenninic platform) and Neogene flysch emplaced APENNINES onto the Apulian platform (Parotto & Praturlon 1981; Bally The Apennines are a long, narrow mountain range, oriented et al. 1986; Malinverno & Ryan 1986; Patacca & Scandone approximately NW SE, that forms the backbone of the Italian 1989). peninsula (Fig. 1). The highest peak is the Gran Sasso s Corno Most of the recorded seismicity is localized along the highest Grande (2912 m asl), about midway along the range, with part of the range (CNR 1985a; Cucci et al. 1996), and occurs most other summits between 1500 and 2000 m asl. The average at crustal depth (Gasparini et al. 1985; Bernard & Zollo 1989). altitude is less than 1 km. In a broad sense, the Apenninic belt Fault plane solutions south of 43 N show predominant normal belongs to the Alpine orogenic system, and is a late result of faulting with T-axes in the NE quadrant (Fig. 1) (Deschamps the collision between fragments of the African and European & King 1984; Westaway & Jackson 1984, 1987; Gasparini plates (Tyrrhenia and Adria) (Boccaletti et al. 1971; Boccaletti et al. 1985; Anderson & Jackson 1987; Westaway et al. 1989; & Guazzone 1972; Tapponnier 1977). Crustal thickness under Dziewonski et al. 1985, 1987, 1988, 1989). Only regions such Figure 1. Seismotectonic map of the central-southern Apennines. Fault traces are from satellite images, aerial photographs and fieldwork. Topography and bathymetry are from Bigi et al. (1992), epicentres from CNR (1985b), and fault plane solutions from Dziewonski et al. (1985, 1987, 1988, 1989).

3 Active oblique extension in the Apennines 501 as the Po Plain and the north-central Adriatic foreland are layer dated at #0.54 Myr (39Ar/40Ar) was found at a depth of still undergoing crustal shortening. The limit between areas in #100 m in a borehole in the southwest part of the Bacinetto extension and areas in compression remains unclear, and the (Follieri et al. 1991), implying a sedimentation rate of central Apennines appear to be a transition area between an #0.2 mm yr 1 in the corresponding time span. The basin is extensional internal domain to the SW and a compressional surrounded by long-recognized Quaternary normal faults external domain to the NE (Boccaletti et al. 1982; Westaway (e.g. Bosi 1975) (Fig. 2b). The topographic relief is greater 1990, 1992; Patacca & Scandone 1989). across the northeastern and northern sides (up to #1880 m The plate tectonic evolution of the central and western at Mt Velino) than across the other two sides of the basin Mediterranean is not as clearly defined as that of the eastern (Figs 3a, b and c), suggesting greater Quaternary uplift on Mediterranean and Aegean. Different models have been proposed faults along those sides, as noted by many authors (Beneo 1939; to explain extension in the internal Apenninic zones Raffy 1970, 1982; Giraudi 1989b; Galadini & Messina 1994; in relation to the formation of the Tyrrhenian Sea and the Salvi 1995). Such asymmetry is confirmed by the facts that migration of the outer thrust belt towards the Adriatic foreland. Plio-Pleistocene sediments are most widespread in the east, Most involve westward subduction of Adria, with eastward the basin infill increases to a maximum thickness of 1300 m at retreat of the slab (e.g. Royden et al. 1987; Doglioni et al. its NE border (Bosi et al. 1995; Di Filippo et al. 1997), and 1994). Difficulties arise from the fact that such subduction the palaeolake floor slopes eastwards by less than 2. According has ceased in the south and that the Apennines are a secondary to Bosi et al. (1995), the formation of the basin started in the belt within the realm of Africa Europe convergence. The Pliocene with normal movement on the NE SE-trending faults deformation history of the region thus probably includes that bound the basin to the east. According to Galadini & components of intraplate deformation (Westaway 1990, 1992). Messina (1994), some of these faults have had a left-lateral Many authors (e.g. Lavecchia 1988; Calamita et al. 1994) slip component since the upper Pleistocene. consider the present-day normal faults as pre-existing thrusts Currently, the fault system chiefly responsible for subsidence reactivated through a mechanism of tectonic inversion. continues to be along the northeast border of the basin Our study of the extensional, active tectonics around the (Blumetti et al. 1993; Galadini et al. 1993), with at least four Fucino Basin in the Abruzzi was undertaken to improve under- parallel faults (Serrone, Parasano, Muricci and Ventrino faults; standing of the most recent deformation regime. Following Figs 2 and 3) splaying from a single fault along the Giovenco the work of Bosi (1975), Galadini et al. (1993), Piccardi and Sangro valleys to the southeast. To the north, three of (1993, 1995), Vittori et al. (1995), and Pantosti et al. (1996), the four faults appear to merge again near Celano, where the we present a map of the main active faults based on geomorphic Tre Monti fault system, whose strike is almost perpendicular observations, a systematic analysis of their geometric and (Fig. 2b), truncates them. Farther north, recent deformation is kinematic relationships, and measurements placing constraints taken up by the Magnola Velino Fault, and by the Ovindoli on their slip vectors and rates. Fault, which splays into the Piano di Pezza and Campo Felice faults north of the Magnola mountains. The Magnola and Velino faults (Figs 2b and 3a and b) display the most spectacular ACTIVE FAULTING AROUND THE FUCINO topographic and geomorphic signatures. They limit the steep BASIN southwestern flank of the Magnola Velino Mountains, joining Tectonic setting of the Fucino Basin at an angle of 140 north of Magliano (Figs 2b, 3a and b and 4b). The Fucino is the most conspicuous intramontane Quaternary The four faults on the eastern side of the Fucino Basin basin of the Apennines south and west of the highest summits separate a flight of three well-developed steps (Figs 2b and 3c) of the Abruzzi (Figs 1 and 2a and b). The basin is roughly covered by fluvio-lacustrine terraces. The highest terraces rhomb-shaped, with long and short sides oriented NW SE ( m asl) are of Late Pliocene Lower Pleistocene and E W, respectively. It extends from Magliano in the age, whereas those standing at m asl are of middle northwest to Gioia dei Marsi in the southeast (Fig. 2b). It is Pleistocene age (Giraudi 1989b; Blumetti et al. 1993). The floored by a very flat plain, approximately 30 km wide and lowest terrace ( m asl) is of Late Pleistocene Holocene 650 m asl, whose lowest part, the Bacinetto, stretches parallel age (Giraudi 1989a). This suggests basinward migration of to the eastern edge of the basin (Fig. 2b). This plain is faulting (Galadini & Messina 1994), but, as discussed below, completely landlocked within mountains that culminate above at least three of the faults appear to be still active. The terraces 1500 m and expose Meso-Cenozoic carbonate shelf strata are deeply incised by small rivers (Fig. 2a), but these rivers and Miocene terrigenous flysch (e.g. Bigi et al. 1992). After remain perched, implying that uplift along the normal faults deglaciation, the plain was flooded by the Fucino Lake, the is faster than stream incision. The four faults may root at third largest lake in Italy before it was drained by the Romans depth on an older shallow west-dipping thrust, as proposed in (#50 AD), and again in 1875 (Giraudi 1989b). Until #1 Myr large-scale reconstructions of Apenninic tectonics elsewhere ago, the lake had a much greater extent ( Grande Fucino, (e.g. Calamita et al. 1994). The chronostratigraphy of the Valensise & D Addenzio 1995). According to Giraudi (1986b), continental series that fills the basin has been interpreted in this bigger lake received water from the palaeo Liri river at various ways (e.g. Raffy 1970, 1982; Giraudi 1988; Blumetti Capistrello (Fig. 2b). Drainage inversion, due to uplift and tilt et al. 1993; Bosi et al. 1993, 1995; Giraudi 1995). In Figs 4(b) of the NE part of the river course, occurred prior to the upper and 12(b), we use elements of those interpretations that are Pleistocene. supported by our field mapping. The Fucino depression is filled by fluvio-lacustrine sediments Other normal faults cut the sediments of the Fucino Plain from Late Pliocene Early Pleistocene to Holocene in age (Giraudi 1986a; Galadini et al. 1993; Galadini & Galli 1996; (Mostardini & Merlini 1986; Bosi et al. 1993). A volcanic Michetti et al. 1996) but their geomorphic expression has been

4 502 L. Piccardi et al. Figure 2. (a) Thematic Mapper scene (13/10/88) of the Fucino Basin and surrounding area. Boxes show locations of Figs 4(a) and 12(a). ( b) Seismotectonic map of the Fucino Basin and surrounding area. Topography from Istituto Geografico Militare maps at 1/ scale, sheets No 145, 146, 151 and 152 and event surface ruptures are from Oddone (1915). Fault traces within the Fucino plain are from Giraudi (1986a) and Galadini & Messina (1994). Instrumental seismicity ( ) is from the PDE catalogue. Boxes show locations of Figs 4(b) and 12( b). Dashed lines show positions of topographic profiles of Figs 3(a) (c).

5 Active oblique extension in the Apennines 503 Figure 3. Topographic profiles across the Velino Fault (a), the Magnola Fault ( b) and the Serrone, Parasano and Ventrino faults (c) (see Fig. 2b for location). Topography is from Istituto Geografico Militare maps 145 II NE, 146 III SE and 146 II SO at 1/ scale. Geology is from 1/ geological map, Avezzano and Sulmona sheets. erased by deposition and cultivation. The principal one is the tunnel, was located. Only minor scarps, affecting Würmian Trasacco Fault, which offsets a Roman channel by cm deposits, mark the E W-trending southern limit of the basin (Galadini & Galli 1996). South of Avezzano, the west side of at the base of the limestone slope, particularly near Trasacco. the basin is bounded by a small normal fault at the foot of Mt Salviano, marked by a small step at the bottom of the slope. The geomorphic signature of this fault is subdued, with Historical seismicity of the Fucino area only degraded scarps, locally, at the bedrock colluvium con- The Fucino Basin was devastated by the Avezzano earthquake tact, and only one clear stretch, near Luco (Galadini & Messina of 1915 January 15. The main shock (I=XI MCS, M =7.0; s 1994). It is on that side of the basin that the original karstic Margottini & Screpanti 1988) (Fig. 2b) destroyed most of the drainage, used by the Romans to construct the aqueduct villages around the Fucino Plain, causing more than

6 504 L. Piccardi et al. (a) (b) Figure 4. (a) Detail of multispectral SPOT scene KJ (01/10/86). (b) Morphotectonic map of area shown in Fig. 4(a). Topography and geology as in Fig. 3.

7 Active oblique extension in the Apennines 505 (a) (b) Figure 12. (a) Detail of multispectral SPOT scene KJ (01/10/1986). ( b) Morphotectonic map of the area shown in (a). Topography and geology as in Fig. 3.

8 506 L. Piccardi et al. deaths. Surface faulting occurred along the east side of the topographic maps (Figs 2a and b and 4a and b). The SSWbasin, with open cracks m wide, and average vertical dipping fault (Fig. 3) consists of two principal segments with throws of cm with maxima of 3 m (Oddone 1915). different trends, #10 and #5 km long, striking N110E and From local witness reports and fieldwork, Serva et al. (1986) N130E in the west and east, respectively. The two segments confirmed rupture along the NW SE-trending, SW-dipping meet at an angle of #50 in a complex corner between sites 7 Serrone and Parasano normal faults (Figs 2b and 12b). A and 9 (Fig. 4b). At that corner, faulting occurs along several comparison of two levelling surveys made in 1872 and 1917 is parallel scarps, including the northwestern segment, which consistent with an average coseismic downthrow of the plain appear to branch from the southeastern fault segment in a of about 50 cm (Ward & Valensise 1989). The epicentre, horsetail manner (Fig. 4b), implying oblique sinistral slip on initially located macroseismically at Avezzano (Oddone 1915), this latter segment. In the middle part of the fault, the mountain was later relocated near Gioia dei Marsi (CNR 1985a), and front shows a maximum relief (Magnola summit, 2220 m asl) a recent reassessment of the macroseismic field (Galadini relative to the Fucino Plain, of m (Figs 3b and 4b). et al. 1995) confirmed both locations to be distinct centres of Towards the northwest, the Magnola fault joins the Velino maximum damage (XI MCS). The focal mechanism of the fault at the bend north of Magliano. Towards the southeast, 1915 earthquake first computed by Gasparini et al. (1985) is the fault terminates just NW of Celano (Fig. 4b). in contradiction with that obtained later by Basili & Valensise Ongoing slip on the fault is attested by all geomorphic (1991) (Fig. 2b), although both are consistent with oblique and geological evidence (Figs 4 7). The shape of mountainextension. front profiles (Fig. 3b), whether downslope (profile 1) or down Palaeoseismological studies establish the occurrence of incised stream valleys (profile 2), is convex upwards. The earthquakes similar to that of 1915 in the Fucino Basin during action of erosional processes alone would have produced the Late Pleistocene and Holocene. Giraudi (1989a) found at stable, concave profiles (e.g. Bull 1977; Wallace 1978). The least four large earthquakes in the area since yr BP, at mountain front displays well-defined faceted spurs (Figs 3b, least one in the post-roman period. A detailed review of the 4a and b and 5a and b), characteristic of sustained normal palaeo- and historical seismicity of the area, based on historical, throw (e.g. Bull 1977; Wallace 1978; Armijo et al. 1986; Briais geological and trenching evidence, is given by Galadini et al. et al. 1990). The triangular facets are planar, with slopes >30, (1995). On the Serrone fault, at Casali d Aschi, these authors and maximum heights of the order of 500 m. Their bases meet identified at least four slip events after 2783±213 yr BP. On along the fault trace (Figs 5a and b) without any embayment. the Parasano Fault, near Collarmele, they recognized at least A previous generation of facets, with gentler slopes, is visible three events after ±650 yr BP, with long recurrence above 1800 m. Valleys perpendicular to the fault trace, between times (>1000 yr). From logging of trenches dug across the the facets, narrow just upstream from the fault, forming the 1915 fault scarp ( voragine ) of the Serrone fault at S. Benedetto wine glass pattern typical of active normal faulting. Such dei Marsi, Brunamonte et al. (1991) and Michetti et al. (1996) valleys, including the glacial valley northwest of Forme (Figs 4b demonstrated the occurrence, prior to 1915, of at least two and 5a and b), are in a hanging position relative to the debris events similar to the M#7 Avezzano earthquake, the first slope below the fault. Farther down, Late Würmian (13 18 ka, between the 6th and 9th centuries AD, and the second between Giraudi 1989a, 1995) slope colluvium and coalescent debris the 10th century AD and 1349 AD. This implies a recurrence fans are faulted. The fault trace is marked by an almost contime of about 700 yr. Similar studies across the Piano di Pezza tinuous cumulative scarp, with a slope steeper than that of the fault (D Addenzio et al. 1995; Pantosti et al. 1996) indicate a facets (Figs 5a, b and c and 6b). Along the base of that scarp, long recurrence time (3000 to 1000 yr) for this fault, with the last event between 1019 and the 13th century AD. an even steeper scarplet exists in places, exposing exhumed, lighter-coloured limestones ( nastro di faglia, Bosi 1975). In Afar, such light-coloured scarplets, which systematically follow POSTGLACIAL MOVEMENT ON THE the boundary between basaltic bedrock and debris wedge FAULTS SURROUNDING THE FUCINO along the most active normal faults, are unambiguously related BASIN to instrumental earthquakes (e.g. Tapponnier et al. 1990). In Greece and Crete, similar light-coloured scarplets in lime- Here, we document quantitatively, from surface measurements, stones mark faults activated by historical earthquakes (Armijo postglacial movement on the principal normal faults surround- et al. 1991) or by repeated postglacial events (Armijo et al. ing the Fucino Basin, with emphasis on those that show the 1992; see also Stewart & Hancock 1991 and Yeats et al. 1997, greatest cumulative offset. We describe the border faults north pp ). At places along the Magnola fault, such of the Fucino (Magnola and Velino faults) separately from nastri expose the slickensided fault plane that dips under those of the east margin, south of the Tre Monti fault zone. the Würmian conglomerates (Figs 5b, c and e). As in other mountain environments where active normal faulting occurs Northern faults (e.g. Armijo et al. 1986, 1991, 1992), the most plausible way to account for the steeper cumulative scarp at the base of Magnola Fault the facets is to invoke smoothing of the mountain slope by periglacial surficial mass transport (e.g. Peltzer et al. 1988) Structurally and morphologically, this fault is the real northern during the last glacial maximum followed by steady growth of boundary of the Fucino depression (Figs 2a and b). The main the cumulative scarp after deglaciation, as such transport ceases fault trace runs along the base of the Magnola range front for to erase increments of throw along the fault trace. Such a slope #15 km (Figs 2a and b and 4a and b). It is particularly sharp evolution process has been inferred in the Apennines by on Landsat and Spot satellite images, air photographs and Westaway et al. (1989) and Blumetti et al. (1993). In keeping

9 Active oblique extension in the Apennines 507 (a) (b) Figure 5. (a) NNE facing view of Magnola Mountain. Note the prominent triangular facets and wineglass valleys. Note also the upward convexity of the mountain front slope. Black triangles show the location of the Magnola Fault. (b) Close-up of cumulative scarp and scarplet (nastro) along the Magnola Fault near Forme (site 5). (c) Close-up of slickensided scarplet at the base of the cumulative scarp shown in (b). (c) slickenside measurements confirm these authors results, and indicate dominant dip-slip, with minor either right- or left- lateral components (Fig. 7). Most of the slickensides we measured lie on the main fault plane, at the debris/bedrock contact, and thus result from Pleistocene motion, but it is difficult to tell whether they reflect postglacial to present slip, or exhumed traces of deeper, hence older, albeit Quaternary, throw. The average slip vectors deduced from such kinematic indicators are reported for each site along the fault (Figs 4b with the Late Würmian age of colluvial wedge deposits in the hanging wall (Giraudi 1993), we thus infer the cumulative scarp along the Magnola fault to have accrued in the last 18 kyr at most, and interpret the scarplet visible at the base of that scarp to represent, as in most other regions, the trace of the last or of the last few earthquakes to have ruptured the fault. Slickensides on the scarplet (Fig. 5e) have been described along the southern segment of the Magnola fault (e.g. Vittori et al. 1991; Blumetti et al. 1993). With few exceptions, our

10 508 L. Piccardi et al.

11 Active oblique extension in the Apennines 509 Figure 6. (Continued.) and 7). Evidence of the superposition of distinct slickenside generations was found at site 8 (Figs 4b and 7). In general, the movement recorded by slickensides has a small sinistral component. Along the fault, there are short kinks of the fault plane (Fig. 4b, site 3, and Fig. 7, inset), across which the slip vector keeps a constant attitude despite changes of up to 90 in faultstrike. Between sites 4 and 5 (Fig. 4b), there are at least three Figure 6. (a) Orthophoto map (original scale 1/10 000, section n , Forme) showing locations of topographic profiles levelled across the Magnola Fault at site 5. (b) Projection of transverse profiles M2, M4 5, M7 9 on a vertical plane striking N42 E, parallel to average profile azimuths. Bold numbers are vertical throw values. Regional slopes and steepest scarplet slopes are also indicated. Ratios of scarp-to-scarplet heights are indicated in the upper-left corner. (c) Projection of fault trace profile M10 on a vertical plane striking N30 E, perpendicular to local fault orientation. (d) Projection of all profiles on a vertical N120 E plane parallel to the fault trace. The difference in vertical offsets east and west of the stream is interpreted to result from the dextral offset of surfaces sloping parallel to the local strike of the Magnola Fault (inset). (e) Projection of river bed profile M9 on a N120 E plane dipping 25 S (see text). Bounds of right-lateral offsets are discussed in the text.

12 510 L. Piccardi et al. Figure 7. Lower-hemisphere stereographic plots of slickenside measurements (fault plane and striae) along the Magnola Fault. Numbers refer to sites shown in Fig. 4(b). Thick symbols show average trend (Dm) and dip (Im) of striae at each site. distinct scarplets in the slope deposits. In the junction area positioned here at the debris wedge/limestone bedrock contact. (sites 7 and 9, Fig. 4b), there are two parallel, cumulative The last source of uncertainty is the fault dip, with shallow scarps, as indicated by debris fans fed from above the upper dips yielding greater offsets. Here, however, the fault surface is scarp but deposited mostly below the lower scarp. At site 9 exhumed or easily reconstructed from the fault-trace profile (Fig. 4b), a well-preserved, 2 3 m high nastro lies at the base (Peltzer et al. 1988) (Fig. 6c). Hence, the primary source of of a #5 m high cumulative scarp. The slope debris below the uncertainty at site 5 arises from the regional slopes fit. scarp is also faulted, with an additional vertical throw of Our profiles (Figs 6a e) are long enough to constrain the #2 m. In contrast with slickensides, geomorphic indicators, slopes of the colluvial wedge below the fault well. Such slopes such as the offsets of small streams that incise both the footwall are similar, steeper (27 29 ) west of the stream than along and the hangingwall, consistently show a slight right-lateral and east of it (18 22 ) (Fig. 6b). The irregular relief of the component of movement (Fig. 5c). eroded footwall makes it more difficult to estimate slopes Detailed topographic profiles were measured at site 5 above the scarp, and only three profiles (M4, M7 and M8) are (Fig. 4b), across and along the fault (Figs 6a, b and d), long enough. All the profiles, however, yield similar footwall transverse ones to determine the vertical offset of the mountain slope values, #10 steeper than those obtained on the hanging- slope, and longitudinal ones, north and south of the fault, to wall. Such slopes are greater than 28, with those away from determine horizontal displacements of transverse markers. The the stream incision reaching (Fig. 6b). The fault dip approach is discussed in detail by Peltzer et al. (1988), Avouac was determined in three ways. On each profile, the local dip et al. (1993), and Gaudemer et al. (1995). We used a Wild was taken to be that of the steepest portion of the exhumed T2000 electronic theodolite coupled with a Wild DI20 infrared limestone scarp. Such dips are steeper east of the stream (up distancemeter, and digital measurements were recorded on a to 58.6 on M5) than west of it (as little as 44.3 on M7), but Wild GRE3 field terminal. The offset measurements are subject average #50. A second determination was made by projecting to several sources of error (Gaudemer et al. 1995). Instrumental the fault-trace profile (M10) onto a vertical plane striking errors, which are very small (a few millimetres), are neglected. N30 E, orthogonal to the N120 E fault strike (Fig. 6c). Again The vertical offset is the elevation difference between the intersection the dip is steeper in the east (up to 56.8#57 S), where the of the fault plane and the regional slopes below and fault trace stands at lower elevation, than in the west above the fault scarp (Fig. 6b). Since such slopes are not (40.6#41 ). It has an average value of 48.9#49 S (Fig. 6c). perfectly planar surfaces, an error arises from fitting a straight Finally, the dip of the slickensided scarplet measured directly line through the levelling points. Following Gaudemer et al. with a compass near the stream is 55 (Fig. 7, site 5). Hence, (1995), We estimate that this induces here an uncertainty of the dip of the fault at the site is fairly well constrained to be #±0.5 m on the vertical offset. A second source of error may 53±4 SSW. arise from locating the fault plane scarp slope intersection The vertical offsets determined on the six transverse profiles (Gaudemer et al. 1995), but this surface trace is accurately vary along strike. They increase from profile M4 (7.8±0.5 m)

13 Active oblique extension in the Apennines 511 on the left bank of the stream to profiles M2 (14.7±0.5 m), The dextral offset of the stream at site 5 may also be and M7 (17±0.5 m) on the right bank (Fig. 6b). Profiles estimated directly (Fig. 6e). In Fig. 6(e), profile M9 is projected M5 and M8, at the east and west sides of the site, show vertical onto a plane striking N120 E, parallel to the fault, and dipping offsets of 10.1 and 13.9 m, respectively, #12 m on average. The 25 to the south, a slope intermediate between the regional stream channel itself (M9) is vertically offset by 14.5±0.5 m. slopes below and above the fault (#18 and #31, respectively). Although this latter value might be biased by differential Although the dog-leg offset of the projection is clearly dextral, incision above the fault and deposition below, it is close to the the short length of the profile makes it difficult to constrain offset west of the stream. Note that the stream profile knick- this offset accurately (Gaudemer et al. 1995). If one assumes point has receded by only m from the scarp top. On segments AB and GH to be representative of the original average, the cumulative vertical offset is about 6 m less east channel, then the offset is the distance between point C and F of the stream (#9 m) than west of it (#15 m). We take the on the straight line that coincides with the fault trace (Fig. 6e). values of 14.7 and 7.8 m deduced from profiles M2 and M4, In that case, the offset might be up to 15.9 m, which represents respectively, which exhibit the smoothest, best-constrained an upper bound. A lower bound is obtained by measuring the slopes above and below the scarp, to be most representative distance between points D and E, projections of B and G on of the vertical cumulative throw on either side of the stream the fault, respectively. In this case, the offset is only 3.7±1.0 m, incision. closer to the 5.1 m deduced from the vertical offset difference. The throw change upon crossing the stream might result The actual dextral offset is thus probably of the order of 5 m. from different factors. It might be due to the scatter of We conclude that, at site 5, the N120 E-striking, 53 S measurements along profiles that span as much as 120 m of dipping Magnola Fault has had a total of #11 m of movement, the fault length. It might reflect non-uniform tectonic slip, with about 10 and 5 m of vertical throw and dextral slip, erosion or deposition along the fault. More plausibly, as implied respectively, since glacial mountain slope smoothing ceased by the existence of lateral stream offsets along the fault, it 14±4 kyr ago (Bard et al. 1990). Its average postglacial reflects apparent vertical offsets due to a horizontal component vertical slip rate is thus 0.7±0.3 mm yr 1. of slip (Fig. 6d). Because the west side of the stream incision Due to the rather loose point spacing, the upper limit of the has a slope a#39.8 towards the ESE, a right-lateral offset h steepest scarplet (nastro) at the base of the cumulative scarp would add a vertical apparent offset h tan a to the real vertical (s, Fig. 6b) is poorly constrained by our profiles, but that offset v (Fig. 6d, inset) (Gaudemer et al. 1995). Concurrently, scarplet height can be at most #1/3 and as little as 1/7 of the the WNW slope, b#27.2, of the east side of the incision cumulative vertical offset. More detailed measurements are would locally reduce the vertical offset by h tan b (Fig. 6d, needed to assess the exact size of the nastro, and whether it inset). Hence, measured offsets on the west side of the stream is a composite surface made of exhumed ribbons of different can be greater than east of it if slip on the fault is dextral in ages with different degrees of alteration, in keeping with the addition to normal: idea that it corresponds to the surface scarp of one or a few v =v+h tan a, individual earthquakes. If, where smallest (profile M8), that a scarplet resulted from only two events with about 1 m of v =v h tan b. b throw, then as many as 14 earthquakes (M#6.7, Tables 1 Eliminating v from the two equations leads to and 2) would have occurred in the last 14±4 kyr, with a mean h=(v v )/(tan a+tan b). a b recurrence time of 1000±400 yr. If the scarplet were the result of only one event (M#7) with as much as 3 m of throw With v =14.7 m (M2) and v =7.8 m (M4), the dextral offset (profile M2), then only five earthquakes with recurrence times a b is h=5.1 m. About 5 m of cumulative right-lateral slip would on average three times larger (3000±1000 yr) might have be consistent with the dog-leg offset of the stream visible when ruptured the Magnola Fault during that same time span. Given looking at the mountain front from a distance. With 5.1 m of scaling laws (Scholz 1990; Well & Coppersmith 1994), events dextral offset, the actual vertical offset v for the segment of the this large should simultaneously rupture the northern segment fault between M2 and M4 would be 10.4 m, within 10 per cent of the Velino Fault (Table 2). Clearly, such recurrence times of the average value deduced from the profiles furthest from and magnitudes require confirmation by trenching, but they the stream. represent plausible upper and lower bounds. In particular, the Table 1. Slip rates and seismic behaviour of active faults in the Fucino Basin area. Fault Closest event Holocene throw rate Max. scarplet height Recurrence interval Max. M w Facet age dd/mm/yy (mm yr 1) (m) (kyr) (Myr) Magnola ± ± Velino 24/02/ Serrone 13/(1/ ± Parasano (13/01/15) Ventrino 0.7±

14 512 L. Piccardi et al. Table 2. Determination of moment and magnitude of plausible earthquakes on active faults in the Fucino basin area. W is the thickness of the seismogenic layer, L is the fault length, d is the fault dip, Du is the average coseismic slip, A is the fault area, M o is the seismic moment given by M o =madu, where m= is the shear modulus and M w is the moment magnitude given by Kanamori & Anderson (1975). Fault W (km) L (km) d ( ) Du (m) A (km2) M o (1019 Nm) M w Magnola Serrone Parasano Ventrino lower bound inferred here from the morphology is compatible (Fig. 4b). The cumulative fault scarp is clear, steeper than with that derived from trenches on the Serrone Fault at San the facets (Fig. 8b) although more degraded than that of the Benedetto dei Marsi. There are no reports of surface faulting Magnola Fault, and only along short segments is a recent on the Magnola Fault during the 1915 earthquake, but this scarplet with slickensides exhumed. Slickensides indicate normal might reflect a lack of observation, as attention was drawn to slip with a minor left-lateral component (Fig. 9). The uniform the spectacular fissures ( voragine ) in the populated Fucino shape of the scarp along strike implies a maximum throw of Plain. the order of 10 m. The 1904 February 24 earthquake (M =5; k CNR 1985a), whose epicentre was located north of the village Velino Fault of Magliano, at the base of Mt Velino (Fig. 2b), may have ruptured the Velino Fault at depth. The western front of the range formed by the Velino and The SE segment extends south of Mt Velino, where the Duchessa Mountains, north of Magliano, is bounded by a height of the calcareous range front decreases abruptly south west-dipping, normal fault striking NW SE (Figs 2a and b of the intersection with the Magnola Fault. Its cumulative and 4a and b). This fault is at least 22 km long and consists scarp is about 6 m high across the Würmian piedmont deposits of two main right-stepping fault segments, about 12 and 10 km (Fig. 4b, site 12). Current activity of this segment is clear from long, NW and SE of Mt Velino, respectively. As for the the deep entrenchment of the two rivers NW of site 12 (Fig. 4b) Magnola Fault, geological and morphological observations and from the deposition by these rivers of three inset terraces suggest ongoing movement, with sustained uplift of the base within the Würmian alluvial surface that slopes gently from the of the mountain front, at a rate faster than that of smoothing Magnola Fault to the Fucino Plain. Immediately after the 1915 by erosion. earthquake, surface cracks were described on this segment, SE The NW segment of the fault trace (Fig. 4b) runs along of Magliano dei Marsi (Oddone 1915), implying involvement the base of prominent triangular facets (Fig. 2) and follows the of the Velino Fault in the event. mountain piedmont slope-break (Fig. 3a). Given the attitude of the bedrock limestones and the geometry of the drainage network, no process other than continuing truncation of the T re Monti Fault base of the mountain spurs by active slip on the fault can The southern front of the Tre Monti range limits the Fucino account for the merging of the adjacent facets at their lower Plain to the north (Figs 2a and b and 4a and b). The range apices. The best-defined facets lie near the middle of the fault, front has a youthful morphology and is bounded by a normal where the mountain front has its maximum local relief (up to fault zone, composed of several right-stepping segments, 1500 m at Mt Velino, 2486 m). The heights of the facets are of 3 6 km long, trending ENE WSW (Figs 4b and 10). The main the order of 400 m, with slopes of about 30. The facet heights fault scarps generally follow the tectonic contact between decrease regularly towards the north, while to the south the bedrock limestones and Würmian slope debris. These debris mountain front bends abruptly at the Velino and Magnola deposits are faulted in several steps with scarps parallel to the fault junction (Fig. 4b). The faceted spurs are separated by main fault attesting to slip in the last yr. distinct wine glass valleys (Figs 4b and 8a). Some of the still In places the scarps show exhumed nastri with slickensides. active alluvial fans on the piedmont are dissected, and fanhead Consistent kinematic indicators indicate a complex evolution entrenchment is present near the fault trace. The fault trace is with superposition of striae in many places. Slickensides with linear, and corresponds to the bedrock/debris slope contact. a right-lateral slip component always overprint those with a Other minor fault scarps affect the Würmian piedmont sediments left-lateral component. The same relationship holds on more

15 Active oblique extension in the Apennines 513 (a) Figure 8. (a) E-facing view of the Velino Fault. Note the prominent triangular facets and steep cumulative scarp along the base of facets to the left. (b) N-facing close-up view of Velino postglacial cumulative fault scarp. Note light-coloured Würmian conglomerates in foreground right and recent vegetation-free surface break along right-hand side of the cumulative scarp. (b) weathered fault planes with broader, deeper grooves. Sharp local changes in fault strike (up to 90 ) do not alter the direction of slip much (Fig. 11a, inset). Although complex, the recent kinematics of the Tre Monti Fault thus resemble those of the Magnola Fault. In a quarry located in one step of the fault zone (Fig. 4b, site 16) faulting in the limestones splays on several planes that offset the overlying Pleistocene deposits along small, distinct surface scarps. On the main fault plane, from top to bottom, the pitch of striae and grooves rotates from 80 E (normal left lateral) to at least 45 W (right lateral normal) (Fig. 11a). Just west of Celano, active uplift on the fault is attested by the deep footwall entrenchment, into Würmian sediments, of two streams flowing down from Ovindoli and the Magnola Mountains (Fig. 4). Even deeper entrenchment of the larger river east of Celano (Fig. 4b, between sites 17 and 18), resulting in a narrow canyon carved into the limestones (Fig. 4a), testifies to strong differential uplift past the NE corner of the basin. The right-stepping Tre Monti Fault zone thus continues northeastwards, truncating the SE slope of Mt Sirente (Fig. 2b). Eastern faults At least three of the main faults east of the basin, the Mt Serrone, Mt Parasano and Mt Ventrino faults, show evidence of recent surface faulting. The three faults, and the

16 514 L. Piccardi et al. Figure 9. Plot of slickenside measurements along the Velino Fault. Symbols as in Fig. 7. Serrone Fault The Serrone Fault bounds the Fucino Basin to the northeast (Figs 2b, 12b and 13a and b). Its two main segments overlap near Gioia dei Marsi (Figs 12b and 13b). The southern segment, #6 km long, extends from Sperone to Gioia dei Marsi, and the northern segment, #15 km long, continues past S. Benedetto dei Marsi. The 1915 earthquake is reported to have caused extensive surface fissuring for 13 km along this ranges they bound, are arranged in a left-lateral horsetail relative to the Sangro-Giovenco Fault to the east (Figs 2a and b and 12a and b). Each is composed of two main right- stepping segments, connected by complex overlaps, where strain is distributed on several minor scarps. Streams flowing down from the NE across the three steps separated by the faults (Fig. 3e) incise deeper into the footwall of each fault, attesting to ongoing relative uplift (Figs 2a and 12a and b). Westwards, the three faults bend counterclockwise before being truncated by the Tre Monti Fault zone. None of them connects with the Magnola, Velino and Ovindoli faults. fault, with narrow collapse grabens 1 2 m deep and #10 m long (Oddone 1915; Serva et al. 1986). The southern segment s trace is continuous, with slightly left-stepping subsegments (Fig. 13a). The cumulative scarp (Fig. 13a), about 8 m high, follows the debris slope/limestone bedrock contact. Much of that scarp was in existence before the 1915 earthquake, which reportedly added a whitish, #50 cm high scarplet (nastro) along its base. Hence, from a geomorphic point of view, this scarp is identical to those found along the northern faults. Kinematic analysis indicates superposition of normal dextral over older, normal sinistral slickensides (Figs 12b and 14, site 31). The fault surface is not planar, but shows kinks and corrugations (Fig. 14, inset). The overlap area, north of Gioia dei Marsi, developed as a step-down, with a high-angle, NE-striking fault splay veering away from the main southern segment (Figs 12b and 13a). Concurrently, the northern segment veers counterclockwise to about E W to accommodate subsidence of this step-down (Fig. 12b, site 30, Fig. 14, site 30k, and Fig. 13b). Within the overlap there are three principal #NNW-striking scarps, one at the mountain front, one at the basin edge and one in between (Figs 12b and 13b, arrowed). The latter has the highest relief, and is continuous with the northern segment. The 1915 voragine extended along this scarp. The two small villages located just above are called Le Grippe and Le Crette, dialectal forms of the cracks (Serva et al. 1986), a possible indication of a prior surface faulting event similar to that of Along the northern segment, near site 30 (Fig. 12b), a quarried section of the fault shows cumulative development of a small graben, sealed by thick slope colluvium (Figs 13c and e). About 10 m downslope, this colluvium is in turn downfaulted by #6 m (Figs 13d and f ). On a local scale, faulting thus appears to have migrated basinwards. Discrete changes in the thickness of colluvial deposits across the small graben faults, and the upward sealing of these faults by such deposits, imply that at least four events led to the development of that Figure 10. N-facing view of the Tre Monti fault. Note right-stepping, postglacial cumulative scarps and nastri and recent incision on footwalls.

17 Active oblique extension in the Apennines 515 Figure 11. (a) Plots of slickenside measurements along the Tre Monti Fault. Symbols as in Fig. 7. ( b) Plots of slickenside measurements along the Ovindoli Fault. Symbols as in Fig. 7. graben (Fig. 13e). The incremental vertical throw during each to the northern Serrone Fault segment at S. Veneziano and event was cm, comparable with the average coseismic S. Benedetto dei Marsi (Fig. 15), close to the 1915 rupture. throw on the voragine in 1915 (Oddone 1915; Serva et al. The profile at S. Veneziano (Fig. 15c), about 1 km NW of the 1986; Ward & Valensise 1989). Hence, the events that produced site in Fig. 13(c) (f ), crosses two cumulative scarps, the first the graben may have had comparable magnitudes (Schwartz & at the foot of the mountain front, and the second within the Coppersmith 1984), and may correspond to ancient, characteristic colluvial slope, 150 m SW and 40 m below the first (Fig. 15c). earthquakes on the northern Serrone Fault segment. The first scarp offsets the topographic surface by 8.1±0.5 m. Faulting becomes more localized approaching Venere The age of the colluvium is between 18 and 7±0.5 kyr (Giraudi (Fig. 12b), north of which the northern Serrone segment con- 1989a). Therefore, the throw rate on this first branch of the tinues as a single scarp. The mountain front above that scarp fault is at least of 0.45 mm yr 1. The second scarp is similar is faceted (Fig. 13b), with a few wine glass valleys. Active to that at San Benedetto dei Marsi (Figs 15a and b), but alluvial fans and slope debris at the base of the mountain connects two surfaces with steeper slopes (5 6 ). The offset is slope are faulted (Fig. 15c). To estimate recent slip rates, 11.2±2.9 m, close to those measured on the San Benedetto topographic profiles (Figs 15a c) were levelled transverse profiles (S4 7, Figs 15a and b). Given the colluvium age,

18 516 L. Piccardi et al. (a) Figure 13. (a) NE-facing view of Mt Serrone. Note the clear cumulative scarp and 1915 scarplet (nastro) at mid-slope. Note the high-angle fault splay veering upslope where the main cumulative scarp loses height. ( b) N-facing view of the Mt Serrone fault overlap. Large and small white triangles show major and minor scarps, respectively. (c), (d) NW-facing view of postglacial colluvium offset by small normal faults. (e), (f ) Interpretative sketches of faulting events (numbers increase with diminishing age). The dashed box shows the location of small grabens with antithetic faults in (c) and (e). Younger colluvium post-dating events in (e) is offset by events in (f ). (b) the throw rate on that second branch of the fault is Since this free face is located #200 m NE of the 1915 voragine, 0.9±0.45 mm yr 1. The profiles near San Benedetto (S4 7, at an elevation close to the 19th century highstand of the lake Figs 15a and b) show that the nearly horizontal surfaces on (Giraudi 1989b), it probably corresponds to the ancient shoreline either side of the cumulative scarp are offset by 8 12 m over cliff. The fact that the lake level has varied during the a distance of several hundred metres. The scarp slope increases Holocene (Giraudi 1989b) makes it difficult to determine the upwards, to become almost vertical just below the top surface throw rate on that segment of the fault. Assuming the surface (Figs 15a and b), except for profile S5, which was levelled east of the scarp to be that of the Giovenco River delta would along a road. The steepest part of the scarp, about 2 m high, imply that it is kyr old (Giraudi 1989a). Southeast of was interpreted by Giraudi (1989b) as a receding free face. the fault, 3 m deep trenches (Michetti et al. 1996) have yielded

19 Active oblique extension in the Apennines 517 (c) (e) (d) (f) Figure 13. (Continued.) ages younger than 6600 yr, suggesting that the kyr old with a small right-lateral component (Fig. 19). Other minor deposits, if present, should lie deeper. The minimum vertical scarps, 1 3 m high, form a left-stepping, right-lateral array at offset of the delta surface would thus be 10+3=13 m, and the base of the Würmian colluvial wedge (Fig. 17a). Small fans minimum vertical slip rate, #0.7 mm yr 1. lie mostly below those scarps (Fig. 17a). The splaying of the fault into two scarp zones may be related to slip-partitioning Parasano Fault in the colluvial wedge (e.g. Armijo et al. 1986). The surface at the foot of the Parasano Mountain front, about 1000 m asl, is Like the Serrone Fault, this normal fault (Figs 2a and b and the geomorphic equivalent of the uplifted Late Pliocene Lower 12a and b), which is composed of two principal segments, is Pleistocene terraces that stand at comparable elevation to the reported to have ruptured during the 1915 earthquake, with north (e.g. Mt Ventrino, Blumetti et al. 1993; Piccardi 1995) a break at least #10 km long, and small collapse grabens (Fig. 12b). along the northern segment (Oddone 1915; Serva et al. 1986; Several profiles were levelled transverse to the main cumulative Galadini et al. 1993, 1995). According to Blumetti et al. (1993), scarp (Figs 17a c), two of them (P1 and P4) to the base of the fault started moving after the Middle Pleistocene. Its two the colluvium, across the entire fault zone. The topographic right-stepping NW-striking segments are connected by an slopes above and below the uppermost fault are similar, #30. overlap area, between Mt Parasano and Pescina, where strain The slope of the steepest scarplet, m high, is between 55 release is distributed on minor NNW-striking, left-stepping and 58. On three profiles (P2, P3 and P4), a second topographic scarps (Figs 12b, 16a and 17a). The southern segment, about step appears to have receded upslope (Fig. 17c). There 8 km long, lies high up in the relief, along the SW slope is no fault at the base of this step, whose unclear origin adds of Mt Parasano. The northern segment, about 13 km long, to the uncertainty on the profile offsets. When measured using stretches from Pescina to Celano at a much lower elevation. the distant slopes, the vertical offset decreases from 12.7±0.5 m Along the southern segment, the fault plane separates in the SE (profile P4) to ±0.5 m in the NW (profile P1). Mesozoic limestones from reddish Würmian conglomerates It decreases from only 8.6 m in the SE (profile P4) to 5.6 m in (Figs 17a and b). A well-preserved slickensided scarplet runs the NW (profile P1) if the near-scarp slopes are taken. The along much of the fault, at the base of a cumulative scarp northwestward decrease is consistent with the existence of a (Figs 17b and c). The slickensides indicate dip-slip motion second fault near the base of the colluvium on P1, which takes

20 518 L. Piccardi et al. Figure 14. Plots of slickenside measurements along the Serrone Fault. Numbers refer to sites shown in Fig. 12(b). Symbols as in Fig. 7. Measurements made at kinks and corrugations of the main fault plane (see text) are specified. up a fraction (4.2±0.5 m) of the movement. Adding this fraction to the m measured on the main scarp yields a total of 10.9±1.0 to 9.8±1.0 m, close to the m measured on P4. Thus, m of vertical slip appears to have occurred along the southern segment of the Parasano fault since late-glacial smoothing of its 30 slope by surficial mass transport ceased after #18 kyr BP. This yields a vertical slip rate of at least 0.48 mm yr 1. If smoothing had continued until 13 kyr BP, and if the corresponding offset were 12.7 m, this rate might be as much as #1mmyr 1 (Fig. 17c). Although the steepest, whitish scarplet (Fig. 16b) at the base of the cumulative scarp probably results from the most recent earthquake(s), it is unclear whether that fault segment ruptured in 1915, and whether only a fraction or all of the scarplet formed then. Because the appearance of a#3 m high scarplet only a few kilometres from Pescina, 80 yr ago, should have attracted attention, we suspect it is in part older. Like that of the cumulative scarp, however, the scarplet height tends to decrease northwestwards (Fig. 18c). The fact that this height is about one half that of the cumulative scarp might be taken to suggest that only two large earthquakes ruptured this segment of the Parasano fault in the last yr. If so, the recurrence intervals might range between 5500 and yr, but the corresponding events would be implausibly large. Northwest of the Parasano scarp, the principal fault array of the overlap is composed of three small left-stepping segments. Slip vectors near the overlap are consistent with those found on the southern segment (Figs 12b and 18). Northwest of the overlap, the fault continues as a single scarp along the western edge of the Collarmele terrace, cutting Plio-Quaternary sediments. At Pescina a large crack is reported to have opened in 1915 where the Giovenco river crosses the fault. It is said to have swallowed the river for several days (Oddone 1915). The name Pescina itself, a dialectal form of little pool, refers to

21 Active oblique extension in the Apennines 519 Figure 15. (a) and (b) Projection on a N50 E vertical plane, roughly perpendicular to local fault strike, of profiles S4 7 across the Serrone Fault near S. Benedetto dei Marsi (see Fig. 12b for location) v 1 and v 2 are the lower and upper bounds, respectively, of vertical offset due to uncertainty on the hangingwall regional slope. Vertical exaggeration is 10. (c) Projection on a N20.5 E vertical plane, roughly perpendicular to local fault strike, of profile S1 across the Serrone Fault at San Veneziano near Ortucchio (see Fig. 12b for location). v 1 and v 2 as in (b). a pond still visible today on the footwall upstream from the fault. NW of Pescina (Figs 2b and 12b), the scarp becomes less clear because it cuts less resistant gravel terraces, but as it veers back to a more NNW SSE direction south of Celano, it resumes a sharp geomorphic signature, with canyons incised into the footwall, and active debris fans aligned along its base. Ventrino Fault The Ventrino Fault runs roughly parallel to and #5 km northeast of the northern segment of the Parasano Fault (Fig. 12b). Although shorter than the Parasano and Serrone faults, it also is composed of two main right-stepping segments, each #5km long and striking WNW ESE, that overlap through a zone of more diffuse strain, with several minor fault scarps. The range front has its maximum relief (450 m) at Mt Ventrino (1507 m asl). Both segments lie at an elevation of 1200±100 m, in general m below the crests of Mt Ventrino and Capo di Moro. Like that of most of the other faults described above, the Ventrino Fault trace follows a late-glacial colluvium/mesozoic limestone contact (Giraudi 1993) and bears clear geomorphic traces of ongoing movement. The perched flat surface SW of the fault stands at a uniform elevation of #1100 m. The fault geometry, morphology and kinematics were described in detail by Piccardi (1995).

22 520 L. Piccardi et al. (b) (c) Figure 16. (a) Mosaic of aerial photographs (Nos 2526 and 2528) showing the Parasano Fault zone. Box shows location of Fig. 17(a). ( b) NWfacing view of the exhumed Parasano Fault plane under Late Würmian colluvium. (c) NE-facing view of the Parasano Fault cumulative scarp and basal whitish scarplet (nastro). We levelled seven topographic profiles across the SW slope of except in the stream channel profile (V5). The gentler 16.4 Mt Ventrino (Figs 19a d), four (V1, V3, V5, V6) perpendicular slope below the fault in this profile is due to deposition in the to the fault, and three (V2, V4, V7) parallel to it. The slopes stream s fan (Fig. 19d). The values of the cumulative vertical above and below the fault scarp are comparable (23 28 ), offset show no increase or decrease along strike. The almost

23 Active oblique extension in the Apennines 521 Figure 17. (a) Simplified map of the Parasano Fault zone. The locations of slickenside measurement sites and profiles are indicated. ( b) Projection on vertical planes perpendicular to local fault strike of profiles P1 5 (see location in a). No vertical exaggeration. (c) Detail of ( b) showing cumulative vertical throws and scarplet heights and dips across the fault. Ratios of scarp-to-scarplet heights are indicated in the upper-left corner. continuous slickensided scarplet is at most #3 m high. The maximum slope-offset, in V6, is 13.8 m (Fig. 19d). V1 yields an offset of 7.1±0.5 m or 9.0±0.5 m, depending on the slope value below the fault. V3 yields an offset of 8.1±0.5 m, close to the mean value (8.05 m) in profile V1. The values obtained with profile V5 (5.2 m and 16.6 m) are very different because the slopes below and above the fault differ due to footwall incision and hangingwall deposition, and because the upstream slope, which is difficult to assess, ranges between 16 and 24. The mean offset value (10.9 m), however, lies between those of the other profiles. Taking the mean and standard deviation of the six values found, the vertical offset of the cumulative scarp amounts to 10±4.3 m along that segment of the fault. If this offset were due to cumulative throw on the fault since at most 18 kyr BP and at least 10 kyr BP, as elsewhere, the average vertical slip rate would be 0.7±0.4 mm yr 1, consistent with that inferred, from less accurate measurements, by Piccardi (1995). Assuming discrete scarp height increments of #3 m during large earthquakes would imply recurrence times between 4000

24 522 L. Piccardi et al. Figure 17. (Continued.) In recent years, much work has been devoted to the assessment of seismic hazard in peninsular Italy, with emphasis on areas struck by large recorded earthquakes (Giraudi 1989a; Ward & Valensise 1989; Pantosti & Valensise 1990; Blumetti et al. 1993; Galadini & Messina 1994; Galadini & Galli 1996; Michetti et al. 1996). This has led to significant insights into the palaeo- seismological behaviour of several active faults. In the Fucino and surrounding mountains, trenching across strands of the 1915 surface breaks, the Trasacco Fault, and the Ovindoli and Piano di Pezza Faults (Michetti et al. 1996; Galadini & Galli 1996; Pantosti et al. 1996) has revealed the existence of several ancient earthquakes. With the exception of the latter work, and 6000 yr. However, the small length of the fault (Scholz 1990; Wells & Coppersmith 1994) suggests smaller individual slip events, hence shorter recurrence times (Table 1). There was no report of surface faulting on this fault during the 1915 earthquake, but, as elsewhere, this might be due to a lack of observation in a scarcely populated area. Slickensides on the exposed fault plane show a dextral-slip component (Piccardi 1995). This is consistent with the #7m dextral offsets of creeks flowing downslope on the 1/ orthophotomap (Fig. 19a). Projections of the stream profile V5 (Fig. 19d), analysed as at site 5 on the Magnola Fault, yield dextral offset values between 7.8 and 15 m, which would imply a postglacial dextral slip rate of 0.7±0.3 mm yr 1, as previously inferred by Piccardi (1995). SUMMARY AND DISCUSSION

25 Active oblique extension in the Apennines 523 Figure 18. Slickenside measurements along the Parasano Fault. Numbers refer to sites shown in Fig. 17(a). Symbols as in Fig. 7. Measurements made at kinks and corrugations of the main fault plane (see text) are specified. however, relatively little effort was directed towards quantitative in 1915, but the cumulative scarps and light-coloured scarplets study of the geomorphic signature of active faults, whether in that consistently, and rather uniformly, offset the mountain the Fucino area or in the rest of the Apennines. While the traces slopes along the other faults indicate that they too must slip of Quaternary faults have long been described (Demangeot and break the surface in the event of fairly large, if infrequent, 1965; Raffy 1970; Bosi 1975; Serva et al. 1986; Dufaure et al. earthquakes, and have done so repeatedly since the last glacial 1989; Blumetti et al. 1993; Galadini et al. 1993; Piccardi 1993), maximum. and analyses of slickensided surfaces have been performed In terms of its contribution to creating youthful relief, the to unravel the recent evolution of tectonic stress along the Magnola Fault, together with its NW continuation, the Velino mountain belt (Calamita et al. 1982; Vittori et al. 1991; Fault, is the most prominent of all. The cumulative elevation Hippolyte et al. 1994; Cello et al. 1997), studies addressing difference across the Magnola Velino system is the greatest recent stress have rarely been coupled with those aimed at of all, and this fault system exhibits the highest truncated assessing current strain. triangular spurs of any faults near the basin. Not only does Here, we performed a systematic study of recent motion this fault lie at the foot of the highest range, but the upward on the most prominent faults north and east of the Fucino convexity of the range front, and the steep cumulative scarp Basin by combining constraints derived from geometry, slip and scarplet as its base are characteristic of world-class, indicators (Figs 11a, 15, 18 and 20), and short- and long-term active normal faults. The throw rate we obtain is of the order offsets of the ground surface. Whilst much less accurate than of 0.7±0.3 mm yr 1. It is possible, however, that the profiles trench-logging to decipher a succession of dated palaeo-events made at site 5 (Figs 4b and 6a c), being not quite long enough on individual fault segments, this approach yields a more upslope, have led to an underestimation of the postglacial slope exhaustive view of the Late Quaternary regional deformation offset and hence the corresponding rate. At 0.7 mm yr 1, the than trenching, which is always limited, in practice, to a small fault would have created most of the 1500 m high range front number of favourable sites. The results summarized below thus in about 2 Myr, and the steepest facets in about 700 kyr. Using complement those obtained by other workers in the area. Kanamori & Anderson s (1975) moment magnitude relationship [M =(2/3) log M 6.01, with M = A Du, w o o and A=LW/sin d; Table 2], and based on the plausible fault Slip rates, event sizes and recurrence times lengths and scarplet heights listed in Table 3, the Magnola The geomorphic signature of the faults shown in Fig. 21(a) Fault might be expected to produce earthquakes of magnirequires that they all be active (e.g. Wallace 1977; Armijo et al. tude M = with recurrence times of the order of w 1986, 1991, 1992; Tapponnier et al. 1990; Gaudemer et al yr (Tables 1 and 2). Simultaneous slip on the Velino 1995). Only the Serrone Fault, the northern segment of the and Magnola faults, with as much as 3 m of vertical throw, Parasano Fault and the southern segment of the Velino Fault would produce the largest possible earthquake in the region (Oddone 1915) are documented to have ruptured the surface (M #7.3, Table 2). w

26 524 L. Piccardi et al. (b) (a) (c) (d) Figure 19. (a) Orthophoto map (original scale 1/10 000, section No , Monte Ventrino) of the Mt Ventrino area showing locations of topographic profiles V1 7. ( b) Map view of topographic profiles V1 7. Hatched area is the cumulative fault scarp; shaded area is the nastro. Letters and numbers refer to various estimates of dextral offset of the stream derived as in Fig. 6. (c) Projection of profiles V2, V4, V5 and V7 on a vertical N135 E plane, parallel to fault strike, showing right-lateral offset of a stream channel (profile V5). (d) Projection of profiles V1, V3, V5 and V6 on a vertical N45 E plane perpendicular to fault strike. Numbers indicate cumulative vertical offsets, regional slopes and fault dip. The Serrone Fault, between Gioia and San Benedetto mountain strand of the fault are insufficient to constrain slip dei Marsi, also exhibits fast slip rates, between 0.5 and per event, but if the scarplet at the base of the cumulative 1.4 mm yr 1. The topography of the Serrone range is less scarp were the trace of just one event, then such an event spectacular than that of the Magnola Mountain, but this is might reach a magnitude of 7 or more and would have a compensated by the greater depth of the basin, since up to particularly long recurrence time, between 1000 and yr 1300 m of sediments have accumulated in the Bacinetto (Tables 1 and 2). The sums of both the short- and long-term (Di Filippo et al. 1997). Taking the vertical throw of the 1915 rates derived here across the Parasano and Serrone faults are earthquake to have been m (e.g. Ward & Valensise at least twice the minimum rate, 0.4 mm yr 1, inferred by 1989; Michetti et al. 1996; Fig. 13e) would imply recurrence Michetti et al. (1996) across the same faults. times for comparable M #7 earthquakes of between 370 In order to have vertically offset, by 10 m, the postglacial slope w and 2200 yr using our extreme vertical slip rates of 0.45 and of Mt Ventrino, the Ventrino Fault ought to have a vertical 1.35 mm yr 1. The fact that the recurrence times documented slip rate of the order of 0.7±0.3 mm yr 1 (Tables 1 and 4). If, in trenches at San Benedetto (Michetti et al. 1996) are closer as on the Parasano Fault, an incremental throw of 3 m per event, to yr would imply that the slip rate on that fault is comparable to the height of the freshest scarplet, is assumed, in the high range of those we infer here from surface offsets this fault might also produce large (M #6.8) earthquakes w and extant ages. with long recurrence times ( yr) (Tables 1 and 2), Our measurements on the Parasano Fault are consistent but scaling laws (e.g. Wells & Coppersmith 1994) and the small with a throw rate of mm yr 1. The data on the length of its segments (Table 3) make this very unlikely.

27 Active oblique extension in the Apennines 525 Figure 20. Summary of slickenside measurements on faults around the Fucino Basin. Our data on the Tre Monti and Velino faults are insufficient to discuss their seismic behaviour, but the fact that their cumulative scarps closely resemble those along the other faults suggests that such behaviour might be within the bounds found for these other faults. The range of values of dip-slip rates, slip per event, event magnitudes and recurrence times estimated in Tables 1 and 2 for the main active faults around the Fucino Basin are comparable to those found by Pantosti et al. (1996) for the Ovindoli Piano di Pezza Fault, further north (Fig. 2). Like this latter fault, the Magnola, Velino and Ventrino faults appear to have been seismically quiet throughout recorded history; however, our results suggest that one should not underestimate their capacity to generate large earthquakes. Regional kinematics While it has long been accepted that all the major faults north and east of the Fucino Basin contribute to absorb tectonic extension, and have done so for at least several hundreds of

28 526 L. Piccardi et al. Figure 21. (a) Synthesis of tectonic and kinematic observations on faults surrounding the Fucino Basin. Large bold arrows are the preferred direction of regional extension, implying a sinistral component of motion between Adria and Tyrrhenia along the central Apennines. (b) Kinematic model of regional extension with a left-lateral shear component, consistent with sinistral horsetail and CCW block rotations west of the Sangro-Giovenco and Ovindoli faults and with the T-axis of Basili & Valensise s (1991) fault-plane solution for the 1915 earthquake. All faults have normal components of throw. Single, full-headed white arrows indicate the preferred horizontal projection of fault slip vectors. Antiparallel arrows indicate the sense of lateral slip on faults with most oblique slip. Bottom-left inset shows how an oblique sinistral component of slip on faults striking in a direction 45 more northerly than horsetail faults can induce a dextral component of slip on the latter. Top-right inset shows the preferred direction of regional extension, equally compatible with the opposite lateral components of slip on the Giovenco and Tre Monti faults. Table 3. Morphological and structural characteristics of active faults in the Fucino Basin area. Fault Av. strike Total length Max. relief Facet height Av. scarp height Max. scarplet height (E from N in ) (km) (m) (m) (m) (m) Magnola Velino (12+10) Tre Monti Serrone Parasano Ventrino ± thousands of years, consistent with our analysis of the long-term morphology (Tables 1 and 4), several aspects of the regional kinematics have remained unclear, and hence a subject of debate, from both a qualitative and quantitative point of view. The overall direction of extension appears to be at fairly high angle to the trend of the mountain belt, but data from various sources allow for different interpretations. Most authors infer the direction of extension to be roughly perpendicular to