• Fig. 1 Active faults of Lebanese
  Figure 1
  
  Active faults of Lebanese restraining bend. Map projected upon Shuttle Radar Topography Mission relief and new SHALIMAR EM300 bathymetry. Hitherto unknown thrust system (Mount Lebanon thrust) hugs continental slope base offshore Mount Lebanon. Red box in inset shows location of Lebanese bend along Levant fault system. Red circles are coastal cities that suffered Tsunami effects during the 551 A.D. event. Abbreviations:
  

  • Ar—Arqa
  • Ba—Batroun
  • By—Byblos
  • Ch—Chekka
  • Sa—Sarafand
  • AT—Aakkar thrust
  • TT—Tripoli thrust
  • R-AF— Rankine-Aabdeh fault
  • RF—Roum fault
  • RaF—Rachaya fault
  • SaF—Saida fault

  
  
  click on image to open in a new tab
  
  Elias et al. (2007)
  restraining bend from Elias et al. (2007)
• Fig. 2a Bathymetry and uplifted
  Figure 2A
  
  Left: SHALIMAR bathymetric map of the proximal Lebanese offshore. Bold red lines are "fresh" seafloor seismic breaks. Boxes a and b indicate locations of SAR images in Figure DR4 and Figure 3, respectively.
  
  Right: 2σ calendar ages of ¹⁴C-dated "B1" vermetid samples (red dots), plotted along the coast with elevation above local mean sea level (LMSL) in cm. Most samples predate the A.D. 551 earthquake.
  
  click on image to open in a new tab
  
  Elias et al. (2007)
  vermetid samples along the Lebanese coast from Elias et al. (2007)
• Fig. 2 Structural map of Lebanon
  Figure 2
  
  Structural map of Lebanon and adjacent areas.
  
  Sample locations and their corresponding elevation above sea level are shown on the left-hand side of the figure.
  
  click on image to open in a new tab
  
  Morhange et al. (2006)
  and relative sea-level indicators from Morhange et al. (2006)
• Fig. 4 Sources of the A.D. 551
  Figure 4
  
  Most likely sources of the A.D. 551 earthquake (open star = inferred epicenter, this study) and other large historical earthquakes in Lebanon.
  
  Modified from Daëron et al. (2005).
  
  Colored patches enclose areas where macroseismic intensities > VIII were reported.
  
  Blue shading corresponds to the A.D. 551 earthquake. The intensity VIII isoseismal is from Sieberg (1932).
  
  click on image to open in a new tab
  
  Elias et al. (2007)
  and other major Lebanese earthquakes from Elias et al. (2007)

• Fig. 1 Active faults of Lebanese
  Figure 1
  
  Active faults of Lebanese restraining bend. Map projected upon Shuttle Radar Topography Mission relief and new SHALIMAR EM300 bathymetry. Hitherto unknown thrust system (Mount Lebanon thrust) hugs continental slope base offshore Mount Lebanon. Red box in inset shows location of Lebanese bend along Levant fault system. Red circles are coastal cities that suffered Tsunami effects during the 551 A.D. event. Abbreviations:
  

  • Ar—Arqa
  • Ba—Batroun
  • By—Byblos
  • Ch—Chekka
  • Sa—Sarafand
  • AT—Aakkar thrust
  • TT—Tripoli thrust
  • R-AF— Rankine-Aabdeh fault
  • RF—Roum fault
  • RaF—Rachaya fault
  • SaF—Saida fault

  
  
  click on image to open in a new tab
  
  Elias et al. (2007)
  restraining bend from Elias et al. (2007)
• Fig. 2a Bathymetry and uplifted
  Figure 2A
  
  Left: SHALIMAR bathymetric map of the proximal Lebanese offshore. Bold red lines are "fresh" seafloor seismic breaks. Boxes a and b indicate locations of SAR images in Figure DR4 and Figure 3, respectively.
  
  Right: 2σ calendar ages of ¹⁴C-dated "B1" vermetid samples (red dots), plotted along the coast with elevation above local mean sea level (LMSL) in cm. Most samples predate the A.D. 551 earthquake.
  
  click on image to open in a new tab
  
  Elias et al. (2007)
  vermetid samples along the Lebanese coast from Elias et al. (2007)
• Fig. 2 Structural map of Lebanon
  Figure 2
  
  Structural map of Lebanon and adjacent areas.
  
  Sample locations and their corresponding elevation above sea level are shown on the left-hand side of the figure.
  
  click on image to open in a new tab
  
  Morhange et al. (2006)
  and relative sea-level indicators from Morhange et al. (2006)
• Fig. 4 Sources of the A.D. 551
  Figure 4
  
  Most likely sources of the A.D. 551 earthquake (open star = inferred epicenter, this study) and other large historical earthquakes in Lebanon.
  
  Modified from Daëron et al. (2005).
  
  Colored patches enclose areas where macroseismic intensities > VIII were reported.
  
  Blue shading corresponds to the A.D. 551 earthquake. The intensity VIII isoseismal is from Sieberg (1932).
  
  click on image to open in a new tab
  
  Elias et al. (2007)
  and other major Lebanese earthquakes from Elias et al. (2007)

• Fig. 2c Bench elevations relative
  Figure 2C
  
  Projected total station measurements of bench elevations relative to LMSL (living vermetid surface in the swash zone).
  
  Note significant scatter in elevation measurements on fossil levels, particularly B1 and B4, likely because of sub-levels with rounded edges and irregularities produced by deep microkarst pitting.
  
  click on image to open in a new tab
  
  Elias et al. (2007)
  to local mean sea level from Elias et al. (2007)

• Fig. 2c Bench elevations relative
  Figure 2C
  
  Projected total station measurements of bench elevations relative to LMSL (living vermetid surface in the swash zone).
  
  Note significant scatter in elevation measurements on fossil levels, particularly B1 and B4, likely because of sub-levels with rounded edges and irregularities produced by deep microkarst pitting.
  
  click on image to open in a new tab
  
  Elias et al. (2007)
  to local mean sea level from Elias et al. (2007)

• Fig. 3 Seafloor seismic rupture
  Figure 3
  
  Seafloor seismic rupture. Sidescan sonar image of a "fresh" seismic rupture along the base of the continental slope west of Damour (box "b" in Fig. 2A). Note the sinuous, segmented trace, west-facing main scarp, and smaller parallel scarps on top of the east (hanging) wall and away from the main scarp base on the footwall. Dark shades mark high backscatter. Insonification from the top (west).
  
  click on image to open in a new tab
  
  Elias et al. (2007)
  west of Damour from Elias et al. (2007)

• Fig. 3 Seafloor seismic rupture
  Figure 3
  
  Seafloor seismic rupture. Sidescan sonar image of a "fresh" seismic rupture along the base of the continental slope west of Damour (box "b" in Fig. 2A). Note the sinuous, segmented trace, west-facing main scarp, and smaller parallel scarps on top of the east (hanging) wall and away from the main scarp base on the footwall. Dark shades mark high backscatter. Insonification from the top (west).
  
  click on image to open in a new tab
  
  Elias et al. (2007)
  west of Damour from Elias et al. (2007)

• Fig. 2b Uplifted vermetid benches
  Figure 2B
  
  Uplifted, karst-pitted vermetid benches near Tabarja.
  
  click on image to open in a new tab
  
  Elias et al. (2007)
  near Tabarja from Elias et al. (2007)

• Fig. 2b Uplifted vermetid benches
  Figure 2B
  
  Uplifted, karst-pitted vermetid benches near Tabarja.
  
  click on image to open in a new tab
  
  Elias et al. (2007)
  near Tabarja from Elias et al. (2007)

• from Elias et al. (2007:755)

On 9 July A.D. 551, a large earthquake, followed by a tsunami, destroyed most of the coastal cities of Phoenicia (modern-day Lebanon). Tripoli is reported to have “drowned,” and Berytus (Beirut) did not recover for nearly 1300 yr afterwards. Geophysical data from the Shalimar survey unveil the source of this event, which may have had a moment magnitude (Mw) of 7.5 and was arguably one of the most devastating historical submarine earthquakes in the eastern Mediterranean: rupture of the offshore, hitherto unknown, ~100–150-km-long active, eastdipping Mount Lebanon thrust. Deep-towed sonar swaths along the base of prominent bathymetric escarpments reveal fresh, west-facing seismic scarps that cut the sediment-smoothed seafloor. The Mount Lebanon thrust trace comes closest (~8 km) to the coast between Beirut and Enfeh, where, as 13 ¹⁴C-calibrated ages indicate, a shoreline-fringing vermetid bench suddenly emerged by ~80 cm in the sixth century A.D. At Tabarja, the regular vertical separation (~1 m) of higher fossil benches suggests uplift by three more earthquakes of comparable size since the Holocene sea level reached a maximum ca. 7–6 ka, implying a 1500−1750 yr recurrence time. Unabated thrusting on the Mount Lebanon thrust likely drove the growth of Mount Lebanon since the late Miocene.

• from Elias et al. (2007:755-756)

Mount Lebanon, a ~160-km-long, ~3000-m-high transpressive coastal range emerging from the eastern Mediterranean sea, lies adjacent to the 25°-rightward restraining bend of the active Levant fault system, the left-lateral transform that constitutes the Arabia-Sinai plate boundary (Fig. 1) (e.g., Freund et al., 1970; Daëron et al., 2004). Seaward from Mount Lebanon, the Levantine margin is particularly steep, reaching water depths of ~1500 m only 8 km from shore, adjacent to the deepest part of the Levantine basin (2000 m), which is floored by thickly sedimented (~12 km) oceanic crust of Mesozoic age (Makris et al., 1983). Onland, prominent thrusts that raise and fold Plio-Quaternary marine deposits and continental conglomerates are visible along the range front NE of Chekka (Fig. 1) (Tapponnier et al., 2001; Elias et al., 2001). One such thrust cuts through the city of Tripoli (Fig. DR1 in the GSA Data Repository¹), forming a ~70-m-high cumulative escarpment (Bahsas scarp). Devastating earthquakes have repeatedly shaken Lebanon and adjacent areas during the past 2000 yr (e.g., Ambraseys and Melville, 1988; Guidoboni et al., 1994). Several historical events shook coastal areas more severely than inland regions, implying source locations along the Mediterranean shore or not far to the west. That active faulting takes place offshore is suggested by recent instrumental seismicity maps (Plassard and Kogoj, 1981; Fig. DR2), but little is known of the geometry and mechanism of such faulting. Here, we present new geophysical data obtained during the 2003 SHALIMAR cruise (Briais et al., 2004) that demonstrate the existence of submarine seismogenic thrusts offshore central Lebanon. Along with recently published ages of young, uplifted wave-cut benches on the coastline (Morhange et al., 2006), the seafloor seismic breaks unambiguously define the source of the A.D. 551 earthquake, the largest event to have struck the Levant shore since Roman times. All the active faults responsible for past earthquakes in Lebanon thus now form a self-consistent tectonic framework.

• from Elias et al. (2007:756)

The historical earthquakes that were most strongly felt along the coast of Lebanon belong to two “sequences” separated by 500 yr of apparent quiescence. The so-called seismic crisis of the fourth to fifth century A.D. included five events (local magnitude, M_l, in parentheses): 303–306, Tyre-Saida (~7.1?); 348/349, Beirut (~7?); 450–457, Tripoli; 502, Acre (Akka) (~7?); and 551, Beirut-Tripoli (7.3–7.8?) (Plassard, 1968; Guidoboni et al., 1994). Of the events of the great eleventh to fourteenth century A.D. sequence that ruptured the entire Levant fault system from Aqaba to Antioch, culminating in the A.D. 1202, M_l~7.5 disaster on the Yammouneh fault (e.g., Ellenblum et al., 1998; Daëron et al., 2005), only three caused destruction on the coast, mostly in Tripoli and the Aakkar plain: 1063, Tripoli, Arqa, up to Antakya and Tyre (~7.1?); 1170, northwest Syria (7.5–7.9?), an event known to have primarily ruptured the Myssiaf fault (Meghraoui et al., 2003); and 1339, Tripoli (Plassard and Kogoj, 1981; Ben-Menahem, 1991).

There is no question that the A.D. 551 event was the strongest coastal earthquake, and the only one unambiguously followed by a regional tsunami, as related in detail by many chroniclers (Data Repository item DR3). Historical accounts clearly restrict the maximum earthquake and tsunami devastation to the Phoenician coast from Tripoli to Tyre, even though 100 villages were also destroyed inland.

That this earthquake was tsunamigenic suggests that it ruptured the seafloor, with a fairly large dip-slip component. The location of maximum damage, exclusively along the Phoenician coast (Sieberg, 1932; Guidoboni et al., 1994), implies a local source, ruling out submarine faults far from Lebanon, even along the Cyprus subduction arc. Given the existence of small, instrumentally recorded shocks offshore central Lebanon (Fig. DR2), Plassard and Kogoj (1981) tentatively located the A.D. 551 earthquake epicenter at sea, somewhere between Beirut and Batroun (Fig. 1).

• from Elias et al. (2007:756-757)

During the SHALIMAR survey, we mapped a previously unknown, east-dipping, offshore thrust system with a complex geometry, involving stepping and parallel segments, as observed along most foreland thrusts (Figs. 1 and 2). Major “range-front” ramps reach the seafloor at the base of the continental slope, where it is steepest, between Saida and Tripoli. They are offset by a ~10 km right-step west of Beirut. Farther out into the Levantine Basin, Plio-Quaternary turbidites detached above Messinian evaporites are folded to at least ~30 km from shore, where the seafloor is warped by growing anticlines above shallower, blind thrusts (Briais et al., 2004). Between these anticlines and the slope base, deep canyons incise the Jounieh plateau, which is uplifted by 150–200 m relative to the Levantine abyssal plain (Fig. 2A). The thrust system is limited to the south and north by two oblique lateral ramps: the Saida (SaF) and Rankine-Aabdeh faults (R-AF), respectively (Fig. 1). We interpret this ~160-km-long thrust system (Mount Lebanon thrust), together with the Roum fault and the Tripoli and Aakkar thrusts onland, to be responsible for the Plio-Quaternary growth of Mount Lebanon (Elias et al., 2003).

Using Ifremer’s deep-towed acoustic system (SAR), we surveyed the base of prominent, ~NE-trending, NW-sloping tectonic escarpments transverse to submarine canyons, visible on the SHALIMAR EM300 bathymetry between Enfeh and Damour (Fig. 2). The SAR images, whose pixel resolution is 25 cm, show details of spectacular submarine ruptures and scarps that cut the smoothly sediment-mantled seafloor (Figs. 2 and 3; Fig. DR4). Though segmented, these fairly continuous breaks generally follow the base of the cumulative escarpments. Most of the scarps face west, consistent with east-wall uplift, and have sinuous, stepping traces, as befits dip-slip faulting. They are often paralleled by finer breaks, suggestive of small-scale, hanging-wall collapse or footwall shortening (Fig. 3). Offshore Jounieh, the slope-base break cuts across the Nahr el Kelb submarine channel and slide-blocks within it (Fig. DR4), which suggests a young age. In the south, we could follow two of these relatively fresh, roughly parallel ruptures for 25–30 km along strike, 8–18 km from shore (Fig. 3). In the north, we could only map 5–10-km-long segments of en echelon ruptures that appear to extend over greater distances. Given their geomorphic resemblance to subaerial, seismic dip-slip ruptures, and their position near the foot of cumulative bathymetric escarpments, the seismic origin of such submarine breaks is not in doubt, although assessing whether they result from one or several earthquakes will require further investigation. Similar SAR-imaged breaks along the western stretch of the North Anatolian fault in the sea of Marmara were recently confirmed, by close-range imaging with Ifremer’s remotely operated vehicle Victor camera, to be due to one historical earthquake (1912 event) (Armijo et al., 2005).

The SHALIMAR data showed no evidence of submarine landslides except for small-scale slump scars and rockslides on or at the base of steep slopes south of Damour and near Batroun. It is thus possible to rule out the occurrence of a large local submarine landslide as potential sources of historical tsunamis along the Lebanese coast. [JW: not so fat my friend]

• from Elias et al. (2007:757)

The discovery of active offshore thrusts sheds new light on the Quaternary geomorphology of coastal Lebanon. The stairs of abandoned, marine-cut terraces between Tripoli and Beirut (Sanlaville, 1977), whose heights in central Lebanon (up to 500 m near Tabarja) are unmatched either north or south, are most simply interpreted to reflect long-term Quaternary uplift resulting from the thrust-driven rise of Mount Lebanon (Elias et al., 2003). As documented in other regions of active tectonic uplift (e.g., Lajoie, 1986), such high terrace levels likely record ancient marine highstands coeval with warm interglacials/-stadials (Sanlaville, 1977).

The presence of smaller-scale indicators of recent shoreline uplift, such as elevated marine notches, beach deposits, and wave-cut, shoreline-fringing bioconstructional platforms (“trottoirs marins”) (Data Repository item DR5), chiefly also between Tripoli and Beirut, can be accounted for as well. Specifically, two benches (“double trottoirs”), including the present (living) one, are visible only between Jounieh and Enfeh and on the islands offshore Tripoli (Sanlaville, 1977). Moreover, north of Jounieh and Batroun, a few stretches of multiple benches are observed. At Petite Fontaine beach near Tabarja, for instance, our total station measurements (Figs. 2B and 2C) document four distinct emerged bench levels (B1, B2, B3, B4), at average elevations of, respectively, 80 ± 30, 175 ± 15, 290 ± 30, and 380 ± 40 cm above the local mean sea level (LMSL).

Given the dearth of preserved material, few of the higher fossil benches have been dated, but a sizable set of radiocarbon ages exists on the lowest one (Morhange et al., 2006), which corresponds to B1 at Tabarja (Table DR6). Between Enfeh and Beirut, 13 radiocarbon ages out of 15 argue for a relatively sudden “end” of the stable sea level recorded by the lowest “B1” bench during the sixth century A.D. at the latest (Fig. 2) (Morhange et al., 2006). Though slightly younger, one more Tabarja and two Palmier Island sample ages are not inconsistent, at the 2σ level, with a rapid, relative sea-level change at that time. Perhaps the death of emerged vermetids was delayed in the island because of the larger swash zone in the more open sea.

In map view, the active and recently ruptured traces of the submarine Mount Lebanon thrust ramps come closest to the coastline precisely where it is endowed with double trottoirs (Fig. 2). Hence, there is little doubt that the sudden sixth century A.D. emersion of “B1” in central Lebanon, by ~80 cm, was of tectonic origin, namely the coseismic shoreline uplift due to the A.D. 551 earthquake, whose macro-seismic epicenter was in all likelihood located offshore Tabarja (Fig. DR7). Higher fossil benches likely reflect similarly sudden uplifts caused by more ancient earthquakes, and it comes as no surprise that the clearest and highest raised,marine-cut Quaternary terraces are found near Tabarja.

• from Elias et al. (2007:757-758)

The source of the A.D. 551 earthquake and the processes responsible for shoreline uplift in Lebanon have long been debated. Sanlaville (1977), who first questioned whether the raised trottoirs might be related, in part, to the A.D. 551 earthquake, finally chose to interpret this relative sea-level change as a “brief, positive eustatic movement” (“Tabarjan”). Pirazzoli et al. (1991) later attributed prominent shoreline uplift around much of the eastern Mediterranean to a regional “seismic crisis,” the “Early Byzantine Tectonic Paroxysm” (EBTP), including the great A.D. 365 subduction earthquake offshore Crete (Guidoboni et al., 1994). Darawcheh et al. (2000) speculated that the A.D. 551 event ruptured a postulated extension of the Roum fault at sea, which the Shalimar survey has now demonstrated not to exist. More recently, on the basis of onland tectonic and geomorphic evidence, we proposed that the inferred Tripoli-Roum thrust—now mapped offshore as the Mount Lebanon thrust—was the most likely source of the A.D. 551 and 1063 earthquakes (Elias et al., 2001; 2003). Morhange et al. (2006) discuss the latter hypothesis in the light of their new vermetid ages, but fail to conclude on the relative roles of the EBTP, Tripoli-Roum thrust, and Yammouneh fault.

The SHALIMAR data essentially settle the issue, even though direct dating of the seafloor scarps will provide the ultimate proof. The submarine Mount Lebanon thrust range-front ramps are ideally positioned. To raise the Tabarja trottoirs 80 ± 30 cm above the LMSL, simple dislocation modeling in an elastic half-space (Okada, 1985) requires 1.5–3 m of seismic slip on these ramps, assuming they dip ~45º eastward in the upper 20 km of the crust (Data Repository item DR8). Such slip amounts are consistent with the estimated magnitude of the A.D. 551 earthquake, and sufficient to account for the tsunami observed. Historical evidence combined with the extent of vermetid death in the sixth century A.D. implies a rupture length of at least ~100 km, and possibly up to 150 km if the Rankine-Aabdeh lateral ramp was involved (Figs. 1 and 4), as suggested by two ages on Palmier Island (Table DR6). For such rupture lengths on thrust faults, empirical scaling laws predict an Mw of ~7.4–7.6 (Wells and Coppersmith, 1994), consistent with macroseismic estimates. Because strike-slip motion on the Yammouneh fault has been shown to produce only small local uplift (less than ~1 m in ~10,000 yr; Daëron et al., 2005), the inference that events on this fault might raise shorelines north of Beirut (Morhange et al., 2006) can be safely ruled out. The coastal ¹⁴C vermetid ages confirm that the great A.D. 1202 earthquake, for instance, produced no uplift along the Lebanese shoreline. That benches offshore Tripoli are older than the seventh century A.D. in fact excludes the possibility that any of the earthquakes of the eleventh to fourteenth century A.D. sequence, including the A.D. 1063 event, ruptured the offshore Mount Lebanon thrust system. Hence, the destruction of Tripoli and Arqa by the latter earthquake may have been caused by slip on the Aakkar and/or Tripoli thrusts (Fig. 4).

The existence of multiple uplifted benches along the central Lebanese coast, and the similar vertical separation of the four benches we measured at Tabarja (~95 ± 20 cm on average), imply recurrent shoreline uplift by A.D. 551-type events since the early Holocene, when the sea reached its latest, highest level, likely ca. 7–6 ka (Lambeck and Purcell, 2005). The return time of A.D. 551-size events must in any case be at least ~1500 yr. The seismic behavior of the Mount Lebanon thrust might thus be characterized by ~1500–1750-yr-long quiescence periods (four?) separating earthquakes clustered in a few centuries, as in the fourth to sixth century A.D. sequence. If so, the 1837 and 1956 events on the Roum fault and the 1918 event offshore Byblos (Daëron et al., 2005; Plassard and Kogoj, 1981) might be “forerunners” of worse to come. In keeping with this interpretation, one might expect the Mount Lebanon thrust to slip at an average rate of ~1 to ~2 mm/yr, the central Lebanese coast to rise above LMSL by 0.5 ± 0.1 mm/yr at most, and hence the highest raised marine terrace above Tabarja to be ~1 m.y. old, all plausible order-of-magnitude values. If the age of the latest sea-level highstand were only ca. 4 ka, as possibly consistent with archaeological evidence on the Israeli coast (Sivan et al., 2001), then the above rates might be 30% faster and the return time of A.D. 551-size events 30% less (~1 k.y.), which would make a repeat of this type of event long overdue. Clearly, dating of the higher vermetid benches and of cumulative seismic deformation on the seafloor are essential to test such inferences, as well as the apparent evidence for local, characteristic slip.

• from Morhange et al. (2006:99)

Twenty-nine ¹⁴C dates of precise biological sea-level indicators from the Lebanese coast show evidence for two significant regional crustal uplift episodes during the past 6000 years. We elucidate: (1) an upper shoreline at ca. +120 to +140 cm, which lasted from ca. 6000 to 3000 BP; and (2) a lower shoreline at +80 ± 40 cm, developed between the fifth century BC and the sixth century AD. These movements are associated with: (1) two major seismic crises along the Yammuneh fault and the Roum-Tripoli Thrust (RTT); and (2) subsequent seismic events on a series of second-order ENE trending dextral transpressive faults. Vertical movements affected north Lebanon, whilst the coasts of south Lebanon generally underwent crustal downlift. This is in contrast with relative stability in northern Israel, suggesting an area of stationary tectonic conditions west of the Dead Sea–Rosh Hanikra/ Ras Nakoura fault. The main ¹⁴C age cluster, corresponding to the second uplift event, may have resulted from fault movements during the “Early Byzantine Tectonic Paroxysm” (EBTP), between ca. 1750 and 2000 BP. Relative sea level stabilised to present level around 1000 BP.

• from Morhange et al. (2006:99-100)

The most active tectonic structure along the eastern shore of the Mediterranean is the north–south trending left-lateral Levant fault system, the plate boundary between Arabia and Africa. In Lebanon, it forms a 160 km restraining bend responsible for the uplift of Mount Lebanon (Daëron et al., 2004). The study of relative sea-level mobility can provide quantitative constraints for crustal deformation along the Lebanese coast Dead Sea Transform Fault and more particularly the role of the Roum-Tripoli Thrust (RTT). The coast of Lebanon is characterized by well-developed uplifted subtidal bioconstructions e.g. Dendropoma petraeum. Our new data allow us to: (1) test the Early Byzantine Tectonic Paroxysm (EBTP) hypothesis (Pirazzoli, 1986a,b; Pirazzoli et al., 1996); and (2) interpret relative sea-level changes in terms of crustal deformation notably the roles of the Roum, Yammuneh and Serghaya faults in connection with the RTT.

Over the past three decades, studies of marks left by former sea levels have shown that numerous vertical displacements took place in the Eastern Mediterranean between ca. 1750 and 2000 BP. After calibration, the period concerned ranges from the mid-4th century to mid-6th century AD. Field results from the islands of Crete and Antikythira, and data compilation from other areas, suggested that these movements could be correlated (Pirazzoli, 1986a,b). This tectonic episode was called the EBTP. Later, these vertical displacements were further documented along the coasts of Hatay (Pirazzoli et al., 1991), Syria (Dalongeville et al., 1993), the Ionian Islands (Pirazzoli et al., 1994a), and the Gulf of Corinth (Pirazzoli et al., 1994b). Further evidence for the EBTP has been reported from the Gulf of Corinth (Stiros and Pirazzoli, 1998), possibly Samos Island (Stiros et al., 2000), and Tripolitana (Di Vita, 1990).

Elevated fossil benches along the Lebanese coast, mainly between Tripoli and Beirut, were first reported 40 years ago and ascribed to historical tectonic movements (Fevret and Sanlaville, 1965, 1966). These features were reinterpreted as elevated eustatic shorelines, lasting from the 2nd century BC to the 2nd or 3rd centuries AD, and called the “Tabarjan” (Sanlaville, 1970, 1977). This interpretation was challenged by Pirazzoli (1976) and interpreted as possible evidence for the EBTP along the Lebanese coast.

• from Morhange et al. (2006:100-102)

Lebanon coincides with a restraining bend of the left-lateral Dead Sea fault system that marks the transform boundary between the African and Arabian plates (Fig. 1). The fault system, which stretches from the Gulf of Aqaba in the south up to the East Anatolian fault in the north, has propagated across the Levantine regions since the early Pliocene (Garfunkel, 1981). The restraining bend of the transform fault system, which controls the first-order topography of the mountainous areas of Lebanon and Syria, is not confined to a simple fault trace (Fig. 2). It is a fan-shaped fault system resulting from the split of the Dead Sea fault north of the Hula pull-apart in northern Israel (Zilberman et al., 2000), into three main splay faults that are, from west to east, the Roum, Yammuneh and Serghaya faults (Walley, 1988, 1998). Recent research suggests that the RTT might correspond to an offshore thrust system linking the Roum fault and the Tripoli thrust (Elias et al., 2003; Daëron et al., 2005).

Strong and repeated historical earthquakes occurred in the Levantine regions (Kallner-Amiran, 1950; Plassard and Kogoj, 1981; Guidoboni et al., 1994; Mart and Perecman, 1996) but none of these events has been linked to any kinematics nor fault segments in Lebanon. Recent palaeoseismological investigations indicate a high level of activity on the Yammuneh and Serghaya faults throughout the Holocene with repeated M = 7 events and a high level of seismic hazard along the Yammuneh fault that cuts across the entire restraining bend (Gomez et al., 2003; Daëron et al., 2005). Preliminary cosmogenic dating of offset fans along the Yammuneh fault indicates an integrated slip rate of 5–10 mm/yr over the past 8000 years (Daëron et al., 2004), whilst the late Holocene slip rate along the Missyaf segment (i.e. the northern extension of the fault in Syria) is estimated to be around 7 mm/yr (Meghraoui et al., 2003). This palaeoseismic estimate appears, however, to be quite high with respect to the integrated late Pleistocene slip rate of 4 mm/yr deduced from offset geomorphic features along the transform south of the Lebanese splay (Klinger et al., 2000). Despite its seismogenic character, the seismic behavior and Holocene slip rate of the Roum fault are not constrained. Yet, based on geomorphic arguments, the fault is considered to be a particularly active segment of the Levantine restraining bend (Butler et al., 1998; Khair, 2001).

Most of the late Holocene palaeo-shorelines exposed north of Beirut are found on rocky shorelines of a crustal panel, the Mount Lebanon block, bounded to the south by the Roum fault, to the east by the Yammuneh fault and to the north and north-west by the RTT. The most prominent structural feature of this panel is the monocline that runs almost parallel to the coast and the series of WSW–ENE to E–W trending faults showing a dominant finite right-lateral component of slip (Dubertret, 1955; Dubertret, 1975; Beydoun, 1977; Walley, 1998; Fig. 2). Palaeo-shorelines found south of Beirut belong to a slightly lower-elevation panel, called hereafter the Tyre–Saida block, bounded to the east by the Roum fault and to the south by the Rosh Hanikra–Ras Nakoura fault (Fig. 2). The internal fan-shaped fault pattern of this panel is characterised by dominantly NW trending dextral faults to the south and dominantly WNW trending sinistral faults to the north slip (Ron et al., 1984; Ron and Eyal, 1985).

• from Morhange et al. (2006:102)

Fossil biological and geomorphological indicators have been used to identify and estimate former sea-level positions (see Laborel and Laborel-Deguen, 1994 for detailed discussion). The limit between the midlittoral and subtidal zones is marked by a sudden increase in species diversity, which corresponds to biological sea level. This benchmark, which incorporates tidal variations and wave exposure, appears as a line between the lower cupulae of midlittoral limpet erosion and the subtidal bioconstructions (e.g., a vermetid rim). Bioconstructions, on the outer edge of the abrasion platform, are built by the close association of two species: the vermetid gastropod D. petraeum and the coralline alga Neogoniolithon notarisii (Lipkin and Safriel, 1971; Safriel, 1975; Tzur and Safriel, 1978; Barash and Zenziper, 1985). Thus, the upper surface of Dendropoma rims precisely marks the upper limit of the subtidal zone (Laborel and Laborel-Deguen, 1994). They are accurate markers of mean sea level. When colonies of elevated in situ Dendropoma rims were found capping the rock, their flat upper surface was used as a fossil sea-level indicator (accuracy of ± 10 cm to ± 20 cm). The lower limit of the bioconstructions is much less constant relative to wave exposure. The present vertical range of the living species varies from 10 cm in sheltered areas to 40 cm, sometimes more, at exposed sites. The accuracy and precision of the upper limit of this indicator are therefore appropriate for estimating the elevation of uplifted remains.

Marine calcareous alga develop in the infralittoral zone and their upper extension therefore indicates a minimum biological sea-level position. Tidal notches are often good morphological sea-level indicators, with an accuracy up to ±10 cm (Pirazzoli, 1986a,b). Spring tidal range is less than 45 cm.

Samples were cut using a diamond saw and the section checked under stereomicroscope. The vermetid tubes comprised aragonite, with biologically altered parts being chiseled away. The material was ¹⁴C dated at the Laboratoire de Géochronologie de Lyon, and the Poznan and Groningen AMS laboratories. Radiocarbon dates from marine specimens have been conventionally corrected using ¹³C measurements and later calibrated (Stuiver et al., 1998a,b). Calibrated dates are quoted to 2σ.

• from Morhange et al. (2006:102-109)

The sites investigated are presented from north to south. Radiocarbon data are given in Table 1 and sample locations are indicated on Fig. 2.

4.1. Ile du Phare (Ramkine island, Tripoli)

On the southwestern cliffs of the island, two elevated notches undercutting the Miocene substratum have been measured at ca. +120 cm and +140 cm above present biological mean sea level (Fig. 3A). At the same location, the outer part of a calcareous algal crust and associated Dendropoma construction collected at +100 cm (sample LIB 2000-23) yielded an age of 2630 ± 35 BP. On the wall of an indentation, marks of a double organic cornice were measured at ca. +100 cm and +140 cm respectively (Fig. 3B), but the upper outcrop could not be reached for sampling. The two cornices clearly indicate, with an uncertainty range of ca. ±20 cm, the position of two former biological sea levels.

On the northern coast of the island, a double elevated bench exists at +90 cm and +120 cm, respectively. A Dendropoma sample (LIB 2000-22), collected at ca. +110 cm from the base of the notch delimiting the upper bench, yielded a much older age: 5975 ± 40 BP.

4.2. Ile du Palmier (Tripoli)

A date of 1880 ± 50 BP (MC-146) was previously reported for a sample of vermetids bordering an elevated bench at ca. +60 cm, and another date of 3490 ± 80 BP (MC-145) for a crust of unidentified vermetids, sampled at +220 cm from a rock fissure on the northern shore of the island (Sanlaville et al., 1997).

On the northern shore of the island, well-preserved colonies of Dendropoma are found at ca. +80 cm (Fig. 4). Sample (LIB 2000-21) was dated 1810 ± 35 BP.

4.3. Hannouch

Near the promontory of Ras Chekka, a Dendropoma sample (LIB 2000-24) collected from an elevated bench at ca. +35 ± 15 cm yielded an age of 2195 ± 30 BP. A sample of Vermetus gigas collected nearby at +40 ± 15 cm (LIB 2001-11) was dated 1930 ± 25 BP.

4.4. Ras Koubba–Salaata harbour

A network of micro-creeks has cut into the Miocene limestone formations north of Ras Koubba. The creek floor, submerged by shallow seawater, is bound by a slightly emerged, irregular fringing bench. A vermetid sample (LIB 2000-20), collected in growth position at ca. +20 ± 15 cm on the bench, yielded an age of 2075 ± 35 BP. Another sample of Dendropoma, from the bench at 50 ± 10 cm (LIB 2001-18), was dated 2615 ± 25 BP.

The jetties of Salaata's new harbour have modified the surrounding area's exposure to wave action, causing a recent local drop of biological sea level by around 30 cm relative to the wave exposure existing prior to harbour construction. In the harbour, two stepped bioconstructions are visible at ca. +30 cm (or present level factoring in the wave exposure). Dendropoma sample (LIB 2000-19) collected from the upper rim yielded an age of 2585 ± 35 BP. Two other Dendropoma samples (LIB 2000-18 and LIB 2000- 18A), from the lower rim were dated 2750 ± 35 BP and 2600 ± 30 BP. These data demonstrate that at exposed sites with a favourable bedrock morphology, Dendropoma may simultaneously colonise the swash zone, 60 cm apart.

4.5. Ras Madfoun

South of the Madfoun fault (Fig. 2), erosion benches ending in notch profiles at ca. +100 cm, cut the Turonian limestones. A sample of juvenile Dendropoma (LIB 2000-9) collected at ca. +100 cm from the base of a mushroom-rock notch yielded an age of 2485 ± 35 BP. Other Dendropoma (LIB 2000-10), developed at ca. +110 cm on the outer rim of a bench, gave an age of 2340 ± 30 BP. A different sample from this same colony (LIB 2000-10A) yielded an age of 2410 ± 45 BP.

Further south of the Madfoun fault (ca. 1 km from sample LIB 2000-10), biological remains show lower elevations. A Dendropoma sample at 80 ± 10 cm (LIB 2001-27) has been AMS dated to 1995 ± 25 BP.

4.6. Fidar creek

North of the southern Fidar creek (Fig. 5), a limestone bench is capped on its outer edge by a Dendropoma rim and partially covered by beach-rock. A Dendropoma sample (F1) at 50 ± 10 cm, collected from the rim, was dated 3020 ± 35 BP. This "old" date results from the death of the molluscs by sediment overlapping, prior to any relative sea-level change.

4.7. Bouar–Tabarja area

An elevated shoreline is seen almost continuously from south of Nahr Ibrahim to the northern border of Jounié Bay. The evidence mainly comprises a bench a few decimetres high, cut into the Turonian limestone, and often ending landward with a palaeo-tidal notch. This elevated bench surface is currently dissected by weathering and marine erosion. We were, however, able to observe and sample marine organisms in growth position, attached to the bench surface or at the notch base. It is important to note that, generally, this elevated platform is twice to three times more developed than the present one (Sanlaville, 1977). The platform is narrow along the sheltered micro-creeks and larger in more exposed sectors.

South of the Nahr Ibrahim estuary, a vermetid sample (LIB 2001-10) collected at the base of the +60 cm midlittoral notch was dated 2065 ± 40 BP. North of Bouar creek, a small man-made basin has been cut into the limestone. It possibly corresponds to an ancient funerary chamber later used as a fish tank, with a channel opening to the sea. The channel leads to a bench at ca. +80 cm. Barnacle shells, collected in living position from a channel wall (LIB 2000-11 at 80 cm), were dated 1160 ± 30 BP.

In a small creek north of Bouar, a marine algal crust partially composed of N. notarisii, lies at the base of a midlittoral notch (ca. +80 cm). The sample (LIB 2000-17) yielded an age of 3195 ± 35 BP. Near Safra, at La Petite Fontaine beach (Fig. 6), a Dendropoma sample (LIB 2000-3) collected at ca. +60 cm, yielded an age of 1960 ± 35 BP. Also in Safra, north of the Rabiya Marine Royal, another Dendropoma sample (ca. +80 cm) from the base of an elevated midlittoral notch (LIB 2000-8), was dated 1975 ± 45 BP.

In the same geomorphic context, dates for the so- called "Tabarjan" shoreline are derived from two vermetid samples collected near Tabarja (MC-63 at 2035 ± 13 BP and MC-64 at 1960 ± 40 BP) (Sanlaville et al., 1997). In the same area, a Dendropoma sample (LIB 2000-7), collected at ca. +60 cm from the surface of the elevated bench, yielded an age of 1970 ± 35 BP, consistent with previous dates. A second Dendropoma sample (LIB 2000-4) collected at +80 cm, just a few metres from the previous one, yielded a younger age (1585 ± 35 BP), whilst an algal crust (LIB 2000-6) collected at +50 cm gave a slightly older date (2220 ± 35 BP) (Fig. 7). A few metres to the south, a well-preserved Dendropoma sample (LIB 2001-8) yielded an age of 1805 ± 30 BP at +120 cm.

4.8. Saida area

South of Beirut, elevated shoreline markers are more scattered and their elevation is generally lower. On Zire island, just offshore from Saida, the bottom of an ancient quarry (Fig. 8) is sealed by a beach-rock at ca. +50 cm containing quarried blocks mixed with well-preserved shells of Pirenella conica. The radiocarbon age of these shells (LIB 2001-27) constrains the submergence of the quarry to 2210 ± 50 BP. The margins of the quarry also show an uplifted notch at +50 cm concomitant with this drowning phase. Close examination of beach-rock specimens in thin sections shows a sediment dominated by sand-sized pieces of branched coralline algae, mixed with debris of molluscs and siliceous particles (Fig. 8B and C). The rock is of grain-stone texture, changing laterally into wackestone. The intergranular pores are partially to totally infilled by two distinct generations of cements. The earlier precipitated cement fabric was pore-rounding, with calcite granular fringes. It is clearly of meteoric origin. The later cementation stage has yielded a structureless, micrite matrix that has embedded marine detritus and occupied the residual pore spaces. It is obvious that this sediment has experienced at least two different diagenetic environments, meteoric followed by marine (Coudray and Montaggioni, 1986). The depositional and diagenetic history of the beach could have been as follows: 1) deposition of sand-grained sediments along the shore; 2) relative sea-level drop promoting the precipitation of meteoric cements within the sand pores; 3) relative sea-level rise, accompanied by the precipitation of marine micrites that have occluded residual porosity; 4) final emergence of the beach-rock deposit. Zire island has therefore undergone a minor ‘yo-yo’ movement, with just the onset of the downward tendency being dated.

Fifteen kilometres south of Saida, near Ras Qantara, a sample of Vermetus triqueter (LIB 2000-1), collected in growth position on a midlittoral platform at ca. +50 cm, yielded an age of 2230 ± 35 BP.

4.9. Ras Abou Zeid

Between Minet Abou Zebel and Ras Minet Abou Zeid (Aadloun sector), an elevated midlittoral notch at +70 cm above the present abrasion platform indicates a former sea-level position (Fig. 9). A V. triqueter sample (LIB 2000-2) at this same elevation was dated 2525 ± 35 BP near marks of Lithophaga holes. This outcrop is sealed by beach-rock, and its age therefore corresponds to the death of the vermetids by sand accumulation before any relative sea-level change took place.

4.10. Khaizerane area

South of Hotel Mounes, the Khaizerane coastal area is characterised by patchy Dendropoma bioconstructions overlying a beach-rock at +40 cm. These bioconstructions (LIB 2002-19) yielded a radiocarbon age of 1095 ± 30 BP. This young age may be correlated with the elevated barnacles dated at Bouar (see Discussion below).

4.11. Tyre area

Our underwater geoarchaeological surveys in Tyre uncovered a submerged quarry floor 2 m below present sea level. The quarry belongs to the so-called "southern harbour" basin. On the northern side of Tyre, the top of the submerged (Roman?) mole, which would have sheltered the ancient northern harbour, is also found 2.5 m below present sea level (Poidebard, 1939; Marriner et al., 2005; Marriner et al., 2006). A living Dendropoma rim (Lib 2002-20) was sampled on the outer edge of an abrasion platform at −5 ± 5 cm, in order to date the base of the present bioconstruction. Constrained to 450 ± 50 BP, it marks the onset of sea-level stability in the Tyre area.

It is critical to discuss two main issues: (1) the timing of relative sea-level changes since 6000 BP, and (2) the spatial distribution of the former sea-level markers with respect to the tectonic pattern.

• from Morhange et al. (2006:112-)

The Lebanese coast provides evidence for two main elevated Holocene sea levels: (1) an upper shoreline at ca. +1.2 to +1.4 m, which lasted from ca. 6000 to 3000 BP. On the nearby coasts of Syria and Turkey, this upper shoreline suggests the occurrence of seismotectonic displacement(s) around 3000 BP, decreasing in amplitude from Turkey to southern Lebanon; (2), a lower shoreline at +0.8 ± 0.4 m, developed between 2700 BP and the 6th century AD, at the time of the EBTP. These uplifted shoreline remains result from the activation of the Yammuneh and the RTT, as well as slip along transverse faults. The Rosh Hanikra/Ras Nakoura fault marks the southern boundary of the Levantine vertical displacements, with no evidence for coastal uplift being reported from Israel during the Holocene.

Elias et al. (2007) examined uplifted benches on the Lebanese coast between Sarafand and Tripolis, including areas near Tabarja (~20 km NE of Beirut). They estimated that ~80 cm of uplift occurred on the lowest bench (B1) in the 6th century CE. Deep-towed sonar data collected offshore showed fresh west-facing fault scarps cutting across the otherwise smooth seafloor, which they associated with seismic activity along the newly identified offshore Mount Lebanon Thrust Fault system. They concluded that ~100–150 km of this fault ruptured in 551 CE, generating an earthquake with a moment magnitude (M_w) of ~7.5.

Previous researchers had suggested that elevated fossil benches along the Lebanese coast (mostly between Beirut and Tripolis) were formed by past earthquake activity. Morhange et al. (2006) used radiocarbon dating of fossil vermetids on the tops of these benches to determine when they were last in the subtidal zone (approximately mean sea level). By integrating these dates with seismic evidence, Elias et al. (2007) concluded that the well-documented 551 Beirut earthquake caused 80 ± 30 cm of uplift of the lowest bench (B1) during that seismic event.

Elias et al (2007) discovered an ~160 km. long offshore thrust system in the process of collecting and analyzing their geophysical data. They termed this new thrust fault system the the Mount Lebanon Thrust.

The deep towed sonar data showed fresh west facing seismogenic fault scarps on the smooth ocean floor which allowed them to map and characterize this thrust system. In commenting on the fault scarps, they stated that

given their geomorphic resemblance to sub-aerial, seismic dip-slip ruptures, and their position near the foot of cumulative bathymetric escarpments, the seismic origin of such submarine breaks is not in doubt, although assessing whether they result from one or several earthquakes will require further investigation.

They later added that "direct dating of the sea-floor scarps will provide the ultimate proof". They noted that their survey "showed no evidence of submarine landslides except for small-scale slump scars and rockslides on or at the base of steep slopes south of Damour and near Batroun" concluding that it was "possible to rule out the occurrence of a large local submarine landslide as potential sources of historical tsunamis [e.g. in 551 CE] along the Lebanese coast." This last conclusion does not seem justified given the limited amount of side scan sonar data they collected. Further, the textual sources for the 551 CE Beirut Quake spoke of an initial ebbing of the sea which suggests that the tsunami was caused by a large submarine landslide. Nonetheless, the geographic coincidence of the stretch of coast exhibiting uplifted benches with the observed areal extent of submarine fault scarps appears to confirm that the Mount Lebanon Thrust fault was the source of large ancient earthquakes and tsunamis along the coast.

Elias et al (2007) estimated a moment magnitude (M_w) of ~7.4-7.6 for the 551 CE Beirut Quake and offered the following discussion:

To raise the Tabarja trottoirs [benches] 80 ± 30 cm above the LMSL [Local Mean Sea Level], simple dislocation modeling in an elastic half-space (Okada, 1985) requires 1.5-3 m of seismic slip on these ramps, assuming they dip -45° eastward in the upper 20 km of the crust (Data Repository item DR8). Such slip amounts are consistent with the estimated magnitude of the A.D. 551 earthquake, and sufficient to account for the tsunami observed. Historical evidence combined with the extent of vermetid death in the sixth century A.D. implies a rupture length of at least -100 km, and possibly up to 150 km if the Rankine-Aabdeh lateral ramp was involved (Figs. 1 and 4), as suggested by two ages on Palmier Island (Table DR6). For such rupture lengths on thrust faults, empirical scaling laws predict an Mw of ~7.4-7.6 (Wells and Coppersmith, 1994), consistent with macroseismic estimates. Because strike-slip motion on the Yammouneh fault has been shown to produce only small local uplift (less than ~1 m in ~10,000 yr; Daeron et al., 2005), the inference that events on this fault might raise shorelines north of Beirut (Morhange et al., 2006) can be safely ruled out. The coastal ¹⁴C vermetid ages confirm that the great A.D. 1202 earthquake, for instance, produced no uplift along the Lebanese shoreline. That benches offshore Tripoli are older than the seventh century A.D. in fact excludes the possibility that any of the earthquakes of the eleventh to fourteenth century A.D. sequence, including the A.D. 1063 event, ruptured the offshore Mount Lebanon thrust system. Hence, the destruction of Tripoli and Arqa by the latter earthquake may have been caused by slip on the Aakkar and/or Tripoli thrusts (Fig. 4

Figure 4

Most likely sources of A.D. 551 (open star—inferred epicenter, this study) and other large historical earthquakes in Lebanon (modifi ed from Daëron et al., 2005). Colored patches enclose areas where macroseismic intensities >VIII were reported. Blue color corresponds to A.D. 551. (VIII isoseismal from Sieberg, 1932).

Elias et al (2007)

).