Taninim Creek Dam

Introduction

The Taninim Creek Dam is located ~850 m inland of the present Mediterranean shore (Marco et al, 2014:1449). Marco et al (2014) excavated trenches behind the Dam on Taninim Creek which showed flame structures which they interpreted to be a direct (liquefaction) or indirect (tsunami deposit induced liquefaction) result of seismic activity.

History of the Dam on Taninim Creek

from Marco et al (2014:1451-1453)

Figures

Fig. 2a - Photo of the Dam

Fig. 2a

View of the dam looking northeastward

Marco et al (2014)

from Marco et al (2014)
Fig. 2b - Closeup - Damaged

Fig. 2b

A closeup view of the damaged section at the northern end of the dam. Building stones are shifted eastward (arrows) as if hit from the west

Marco et al (2014)

section of Dam from Marco et al (2014)

Discussion

The dam on the Taninim Creek (Fig. 2a), which was built by the Byzantines in the fourth century CE as an artificial reservoir, was part of a flushing mechanism to clear silt from the Caesarea harbor, and for operating vertical flour mills as a by-product. The dam that is preserved along 193 m was originally more than 300 m long, 5.5 m wide at the base, and 4.5 m wide at the top. The top of the dam is 7 m above mean sea level.

A shallow fresh-brackish water lake was formed on the eastern side of the dam. Porat (2002) argues that the system was in use at least until the mid-sixth century, based on evidence from the ‘‘Christian Building’’, which received its water from the ‘‘Lower Aqueduct’’, a drinking-water carrier (bypassing the dam system) to Caesarea. After the Arab conquest in 640 CE, all the external water sources of Caesarea ceased to operate, included the studied dam reservoir. The inhabitants instead used water from wells that were dug in almost every house, as was told by the historian Al Mukadasi (cited by Porat 2002). However, a series of Ottoman flour mills powered by water from the dam reservoir indicate much later activity, possibly through the eighteenth century. We therefore argue that the lake persisted until the late Ottoman period.

A tilt of 7° from the vertical, leaning eastward (top moved eastward), most probably because of foundation settling, was fixed already in the dam’s early days. We recognize a late occurrence of damage at the dam’s northern end, where a 10-m section of the western face is destroyed; the upper part of the dam is missing, and the remaining stones are shifted eastward (Fig. 2b). A plausible explanation for the damage is that a large wave hit the dam, most intensely on its northern end. Later, the dam was repaired in the Ottoman period, when a massive support wall was built on the eastern face of the breached section, doubling the thickness of that part of the dam, and a series of 11 flour mills were built on the west side of the dam (Peleg 2002). We interpret the repair as supporting evidence for the operational condition of the dam when it was hit (Fig. 2a). The spillways of the Ottoman flour mills were opened in the dam lower than the watermark of the fourth century CE, possibly because the damaged dam could not hold the lake water at the original level.

Tel Taninim Archaeoseismic Site Webpage

from Taninim Creek Dam Paleoseismic Site Webpage - click link to open new tab

Maps, Aerial Views, Sections, Tables and Photos

Maps and Aerial Views

Fig. 1 - Location Map from

Fig. 1

a general location map showing the Dead Sea Fault (solid black lines)

b A Google Earth image of the Taninim Creek area. Dashed lines delineate the eolianite sandstone (‘‘Kurkar’’) ridge, which is cut by the Taninim Creek between the settlements of Kibbutz Maagan Michael and Jisr el Zarka

Marco et al (2014)

Marco et al (2014)
Fig. 4 - Deduced transport directions

Fig. 4

The transport directions deduced from the flame structures asymmetry (white arrows) and schematic liquefaction intensity gradient (yellow overlay) showing decreasing intensity toward the east

Marco et al (2014)

from Marco et al (2014)
The dam on the Taninim Creek

The dam on the Taninim Creek

click on image to explore this site on a new tab in Google Earth

in Google Earth
The dam on the Taninim Creek

Taninim Creek Dam on govmap.gov.il

click on image to explore this site on a new tab in govmap.gov.il
on govmap.gov.il

Stratigraphic Section for Trench 6

from Marco et al (2014)

Fig. 3

Columnar section of the sediments in a trench east of the dam. The photograph mosaic on the left shows the location of samples. Arrow shows the horizon where flame structures are observed (Fig. 2). A single ¹⁴C age of detrital charcoal from TN8 is 245 ± 30 BP. The earthquakes that were associated with tsunamis according to historical accounts are listed. The temporal constraints imposed on the observed sedimentary section and the correlation with the archeological stratigraphy indicate that the liquefaction and the damage to the dam occurred after the deposition of the laminated lake sediments and accumulation of some 30 cm of clayey soil, and before the construction of the flour mills toward the end of the Ottoman period. The earthquake of November 25, 1759 is the most plausible cause for these features

Marco et al (2014)

Tables

from Marco et al (2014)

Table 1

Summary of injection structure characteristics in studied trenches (Fig. 3 c and d)

Marco et al (2014)

Photos

Fig. 2a - Photo of the Dam

Fig. 2a

View of the dam looking northeastward

Marco et al (2014)

from Marco et al (2014)
Fig. 2b - Closeup - Damaged

Fig. 2b

A closeup view of the damaged section at the northern end of the dam. Building stones are shifted eastward (arrows) as if hit from the west

Marco et al (2014)

section of Dam from Marco et al (2014)
Fig. 2c - Load structures from

Fig. 2c

Load structures were formed by liquefaction of the lower light-colored lacustrine silt that penetrated the overlaying dark clayey soil. We interpret the structure asymmetry and zigzag shapes as result of shear forces exerted on the lake bed sediments by waves sloshing back and forth

Marco et al (2014)

Marco et al (2014)
Fig. 2d - Load structures from

Fig. 2d

Load structures were formed by liquefaction of the lower light-colored lacustrine silt that penetrated the overlaying dark clayey soil. We interpret the structure asymmetry and zigzag shapes as result of shear forces exerted on the lake bed sediments by waves sloshing back and forth

Marco et al (2014)

Marco et al (2014)

Stratigraphy

Table

from Marco et al (2014:1453)
derived from five 1.5 m deep trenches (T1-T5) and one archeological excavation bank (T6) east of the fourth century Byzantine dam in the area that was occupied by the artificial lake.
Marco et al (2014:1453) report that they found exactly the same stratigraphy in all the sections which they divided into 4 units

Unit	Age	Description from Marco et al (2014) with some added comments by JW
1st	Before ~400 CE	Pre-dam - gray-brown clayey soil
2nd	between ~400 CE and ~1400-1700 CE	0.8–1.0-m thick sequence of laminated, whitish, calcareous sand, silt, and some clay Grain-size analysis shows that the lake sediments (unit 2) from the trenches are mostly silt (mode is 3–39 microns) with sand content of up to 11 %. We interpret the lower whitish laminated silt as fresh-to-brackish water deposits of the artificial lake, which was formed behind the dam after it was built in the fourth century The laminated (lacustrine) silt part of this deposit was estimated to have accumulated over 1000-1300 years. Similar laminated sequences are reported from the Dead Sea Basin, where deposition rates of the order of 1 mm/yr are estimated (Schramm et al. 2000; Waldmann et al. 2009). Based on the accumulation of the laminated lake deposits in the studied lake sediments, and assuming that they represent annual cycles typical to the regional climate of dry rainless summers and wet rainy winters, the time equivalent is in the range of 1,000–1,300 years. 2nd unit is enclosed between two clay beds of low permeability, which act as seals that prevent fluid escape making it vulnerable to liquefaction At the boundary between the lake sediments and the topsoil layer, we discovered a series of flame structures that we interpret as injections, where the silty sand is intruded into the dark soil strata (Fig. 2). We documented 30 such structures along a 13-m traverse on the southern wall of trench T-1, 56 structures along a 20-m traverse on eastern wall of trench T-3, and 7 along 3 m on the eastern wall of T5 (Table 1). Many injections have zigzag shapes. The height of the injection structures is about 30 cm on the northwest side of the lake, close to the break in the dam, and it becomes smaller away from the dam toward the southeast. In T5, the injections are only up to 5 cm, and in T6, the easternmost trench, there are no such structures at all and the sediment is undisturbed. The lower parts of most of the injections are strongly asymmetric. We counted 14 injections in T-1 and 42 injections in T-3; the dominant vergence is eastward (12 out of 14 injections in T-1) and southeastward (29 out of 42 injections in T-3), but north- and west-verging structures are also observed. The upper parts of the zigzag injections do no not exhibit uniform vergence (Fig. 2). The vergence and the liquefaction size gradient are summarized in Fig. 4.
3rd	between ~1400-1700 CE and ?	0.5-m darker sequence of light-brown 0–0.15 m silt and clay mixture with abundant fossils and 0.4-m light-brown massive clay-rich soil with carbonate concretions (up to 2 cm in diameter) and shell fragments (0.5 cm) The clay-rich top-soil unit reflects the pedogenic processes, which could start when the lake had dried; hence, the sharp lithological boundary is dated to circa eighteenth century. The abundant Melanopsis shells in the lower brown clay indicate frequent freshwater environment. We found only one dateable piece of detrital organic matter in sample TN8 that yielded a ¹⁴C age 250 ¹⁴C yrs BP, in agreement with the other data (Fig. 3).
4th	?	a dark brown soil within the agricultural zone

Stratigraphic Section for Trench 6

from Marco et al (2014)

Fig. 3

Columnar section of the sediments in a trench east of the dam. The photograph mosaic on the left shows the location of samples. Arrow shows the horizon where flame structures are observed (Fig. 2). A single ¹⁴C age of detrital charcoal from TN8 is 245 ± 30 BP. The earthquakes that were associated with tsunamis according to historical accounts are listed. The temporal constraints imposed on the observed sedimentary section and the correlation with the archeological stratigraphy indicate that the liquefaction and the damage to the dam occurred after the deposition of the laminated lake sediments and accumulation of some 30 cm of clayey soil, and before the construction of the flour mills toward the end of the Ottoman period. The earthquake of November 25, 1759 is the most plausible cause for these features

Marco et al (2014)

Trench Locations

from Marco et al (2014)

Fig. 4

The transport directions deduced from the flame structures asymmetry (white arrows) and schematic liquefaction intensity gradient (yellow overlay) showing decreasing intensity toward the east

Marco et al (2014)

Master Seismic Events Table

Flame Injections Event - ~1500-~1900 CE

Figures

Fig. 4 - Deduced transport directions

Fig. 4

The transport directions deduced from the flame structures asymmetry (white arrows) and schematic liquefaction intensity gradient (yellow overlay) showing decreasing intensity toward the east

Marco et al (2014)

from Marco et al (2014)
Fig. 3 - Columnar Section in

Fig. 3

Columnar section of the sediments in a trench east of the dam. The photograph mosaic on the left shows the location of samples. Arrow shows the horizon where flame structures are observed (Fig. 2). A single 14C age of detrital charcoal from TN8 is 245 ± 30 BP. The earthquakes that were associated with tsunamis according to historical accounts are listed. The temporal constraints imposed on the observed sedimentary section and the correlation with the archeological stratigraphy indicate that the liquefaction and the damage to the dam occurred after the deposition of the laminated lake sediments and accumulation of some 30 cm of clayey soil, and before the construction of the flour mills toward the end of the Ottoman period. The earthquake of November 25, 1759 is the most plausible cause for these features

Marco et al (2014)

Trench E from Marco et al (2014)
Fig. 2c - Load structures from

Fig. 2c

Load structures were formed by liquefaction of the lower light-colored lacustrine silt that penetrated the overlaying dark clayey soil. We interpret the structure asymmetry and zigzag shapes as result of shear forces exerted on the lake bed sediments by waves sloshing back and forth

Marco et al (2014)

Marco et al (2014)
Fig. 2d - Load structures from

Fig. 2d

Load structures were formed by liquefaction of the lower light-colored lacustrine silt that penetrated the overlaying dark clayey soil. We interpret the structure asymmetry and zigzag shapes as result of shear forces exerted on the lake bed sediments by waves sloshing back and forth

Marco et al (2014)

Marco et al (2014)
Table 1 - Summary of injection

Table 1

Summary of injection structure characteristics in studied trenches (Fig. 3 c and d)

Marco et al (2014)

structure characteristics from Marco et al (2014)

Discussion

References

Notes by JW

Marco et al (2014:1449-1450) suggested that the 25 November 1759 CE Baalbek Quake was the most plausible trigger of the sand injections, either directly or via earthquake-induced tsunami. Dating was based on stratigraphic position, archeological context, and historical accounts.

Stratigraphy

Marco et al (2014:1453) derived their stratigraphic succession from five 1.5 m deep trenches (T1-T5) and one archeological excavation bank (T6) east of the fourth century Byzantine dam in the area that was occupied by the artificial lake. Marco et al (2014:1453) report that they found exactly the same stratigraphy in all the sections which they divided into 4 units.

Archaeological Context

Marco et al (2014:1451-1453) argued that the fresh brackish water lake behind the dam persisted into the Ottoman period due to the presence of flour mills on the west side of the dam which would have been powered by water from the dam. This may extend the age of the top of the lacustrine unit into the Ottoman Period, "possibly through the eighteenth century". However, this appears to be in conflict with the time column in their Stratigraphic column (Fig. 3) which dates the top of the lacustrine unit to a little after 1400 CE. The clayey soil above the lacustrine unit, which they stated reflects the pedogenic processes, which could start when the lake had dried did show evidence, based on fossils, that it was inundated by water at least some of the time.

Historical Accounts

Historical accounts specified by Marco et al (2014:1457) were reported in Ambraseys and Barzanagi (1989) that described boats that were swept ashore from the Akko harbor (50 km north of the studied site), and a large wave that was reported from as far south as the Nile Delta.

Flame Injections Event - ~1500-~1900 CE

Effect(s)	Location	Image(s)	Description
Load structures	Taninim Creek Dam Fig. 4 The transport directions deduced from the flame structures asymmetry (white arrows) and schematic liquefaction intensity gradient (yellow overlay) showing decreasing intensity toward the east Marco et al (2014)	Fig. 2c Load structures were formed by liquefaction of the lower light-colored lacustrine silt that penetrated the overlaying dark clayey soil. We interpret the structure asymmetry and zigzag shapes as result of shear forces exerted on the lake bed sediments by waves sloshing back and forth Marco et al (2014) Fig. 2d Load structures were formed by liquefaction of the lower light-colored lacustrine silt that penetrated the overlaying dark clayey soil. We interpret the structure asymmetry and zigzag shapes as result of shear forces exerted on the lake bed sediments by waves sloshing back and forth Marco et al (2014) Table 1 Summary of injection structure characteristics in studied trenches (Fig. 3 c and d) Marco et al (2014)	Brief Description and Interpretation Marco et al (2014:1449) report that flame-shape injections of silty sand that penetrate the overlying clay-rich soil are largest and most frequent within several meters of the point where the dam is badly damaged on the seaward side which they interpreted as a possible result of a large wave. They explained their interpretations as follows: Three features make the sand injections special: their lower extent is commonly asymmetric with dominant southeastward vergence, away from the breach in the dam zigzag shapes characterize the upper parts of many injections the size and frequency of the injections diminish gradually with distance from the dam until they completely disappear some 100 m away from it We suggest that the sand injections can be explained by overpressure that was induced either directly by earthquake shaking or by a tsunami wave that breached the dam, filled the reservoir behind the dam and increased the pressure on the water-saturated silt and sand layers and triggered liquefied sand injections. The movement of water sloshing back and forth in the lake accounts for the zigzag shape of the injections. The similarity to structures that were observed in Thailand after the great 2004 tsunami and other palaeotsunami observations lead us to prefer the tsunami origin of the liquefaction features. Detailed Interpretation Marco et al (2014:1454-1457) discussed their tsunamogenic interpretation in more depth. The sand injections may be interpreted as the result of overpressure in the lacustrine deposits, which could have been triggered either by earthquake shaking or by a sudden increase of overburden (Trifunac 1995). ... Liquefaction and sand injections are often reported in association with earthquakes (Obermeier 2009). Since no evidence for active faulting is found along the coast, we argue that the source of the causative earthquake was remote. The possible sources are the Carmel Fault, about 25 km northeast, or the Dead Sea Fault, about 60 km east of Taninim, both capable of generating M>6 earthquakes (e.g., Arieh 1967; Hamiel et al. 2009; Hofstetter et al. 2007). Compilations of liquefaction–epicentral distance data from Italy and eastern Mediterranean regions (Galli 2000; Papathanassiou et al. 2005) show that an M ~ 6 earthquake on the Carmel Fault or M > 6.5 earthquake at the Dead Sea Fault is capable of triggering liquefaction in our study area. The earthquake history of the Carmel Fault is not well known, but observations on a meticulously built Medieval basilica located on top of the fault, which does not exhibit any earthquake-related damage, indicate that during the last eight centuries there was no significant earthquake along it (Marco et al. 2006). Hence, we conclude that the source was along the Dead Sea Fault. An M > 6.5 on the Jordan Valley segment or an M > 7 on either the northern or the southern segments are plausible. Hence, earthquake-induced liquefaction is a possible cause for the flame structures observed in the study site. The conspicuous, nearly uniform asymmetry of the injection structures here (Fig. 3) is rare in other occurrences of injected sand. Earthquake-related sand injections are commonly symmetrical, upright, exhibiting upward-directed flow features (e.g., Obermeier 2009; Tuttle and Schweig 1995). We therefore hypothesize that the asymmetry may have been caused by shear that was induced by dominant southeastward flow, and the zigzag shapes at the upper parts of many injection structures indicate the sloshing of water back and forth. The flow could trigger shear instability at the upper part of the sediment, also known as Kelvin–Helmholtz Instability (Drazin and Reid 2004). Palaeotsunamis have been reported to trigger sand injection features, e.g., by De-Martini et al. (2003) and Owen and Moretti (2011), and asymmetric flame structures were observed in the deposits of the December 26, 2004 tsunami in Thailand (Matsumoto et al. 2008). The latter describe asymmetric injections of fine sand that was deposited by the tsunami, which intrude the overlying coarse-grained strata, uniformly skewed in the direction of the run-up current. Matsumoto et al. (2008) interpret the sequence as a result of two tsunami waves, where the boundary between the two deposited strata, i.e., the top of the first wave deposits, was simultaneously deformed and truncated by the second run-up current. Another set of asymmetric folds and injections observed in cores from the Aysen Fjord, Chile were associated with mass movement triggered by the April 27, 2007 M_W 6.2 earthquake (Van Daele et al. 2013). In the Lisan Formation (the palaeo Dead Sea sediments), similar zigzag shaped injections are capped by fragmented laminae. The sequence is interpreted as the result of re-suspension of the bottom sediment by the shear with the water during seiche or tsunami events (Alsop and Marco 2012; Wetzler et al. 2010). We suggest that a rapid addition of about 3 m of water (measured from the lake deposits to the top of the dam) during a tsunami surge could increase the overburden on the sandy layer, which is confined at its bottom and top with stiff clayey soil. The surge would have induced a sudden increase of pressure of 0.3 bars, at least three times the confining pressure (assuming soil thickness were similar to that of present, i.e., about 0.5 m and bulk density of about 1.5 g/cc). These conditions are highly likely to trigger liquefaction and the injection of water-saturated sediment into the overlying soil (Trifunac 1995). This scenario is somewhat similar to the case of palaeotsunami deposits that were found in the Kakawis Lake, near the shore of the Vancouver Island, British Columbia (Lopez 2012). We did not find deposits of marine origin, either because of the intensive modern agricultural cultivation of the upper strata that mixed the soil and made the microfossils very rare and hard to find, or because seawater did not go over the dam. An alternative tsunami-related explanation for the liquefaction on the eastern side of the dam could be a pressure gradient developed by increasing water head on the western side of the dam, where the tsunami wave ramped up but did not top the dam entirely (Craig 2004). While this mechanism and the earthquake-induced mechanism may explain why we have not found marine fauna east of the dam, it does not explain the systematic asymmetry of the injection structures. We therefore argue that the observed injections of the sand were triggered by an earthquake, and the shape of the injections combined with the damage at the seaward face of the dam favor the involvement of a tsunami wave. The candidate-triggering events are the earthquakes that postdate the accumulation of the entire sequence of lake deposits and some 20–30 cm of clay-rich soil that was formed at the surface after the lake dried. The historical accounts that associate tsunamis with earthquakes in this region have been reviewed and screened for reliability (Salamon et al. 2011, 2007), but none of the reliably recorded ones stands out as having a significant run-up that could top the 7-m-high dam (possibly 1–2 m less at the breached section). The precise location of the shoreline in the past is unknown, but it may have been different, shore transport along the coast of Israel following the construction of modern ports and marinas (Zviely et al. 2007). In the absence of local active faults, the tsunamis can be induced by remotely triggered submarine slumps (Salamon et al. 2007). We cannot determine independently the precise age of the observed liquefaction; however, the most suitable trigger candidate according to the dam history presented above, and the single ¹⁴C age, is the earthquake of November 25, 1759 (Fig. 3). Reports on this catastrophic earthquake describe boats that were swept ashore from the Akko harbor (50 km north of the studied site), and a large wave that was reported from as far south as the Nile Delta (Ambraseys and Barazangi 1989). The 90-km. long surface rupture of this earthquake was reported by the contemporary French Consul to Beyrouth [JW: it was the French Consul in Saida] after visiting the site in Lebanon [JW: the report was not based on a visit but on second hand information] (Ambraseys and Barazangi 1989) and confirmed by palaeoseismic studies (Daeron et al. 2005; Gomez et al. 2001). Ambraseys and Barazangi (1989) estimate the earthquake magnitude at 7 plus.

Flame Injections Event - ~1500-~1900 CE

Earthquake Archeological

Earthquake Archeological Effects Chart

Click on image to open in a new tab

Rodríguez-Pascua et al (2013)

Effects from Rodríguez-Pascua et al (2013: 221-224)
Environmental Effects (ESI 2007)

Graphic Representation of ESI 2007 Intensity

Click on image to open in a new tab
Synoptic Table of ESI 2007

Synoptic Table of ESI 2007 Intensity Degrees

Accuracy improves in higher degrees, especially VIII–XII.

Click on image to open in a new tab

Michetti et al. (2007)

Intensity Degrees from Michetti et al. (2007)
Environmental Effects vs. Intensity

Diagnostic ranges for environmental effects by intensity

Click on image to open in a new tab

Michetti et al. (2007)

from Michetti et al. (2007)

Effect(s)	Location	Image(s)	Description	Intensity
Load structures	Taninim Creek Dam Fig. 4 The transport directions deduced from the flame structures asymmetry (white arrows) and schematic liquefaction intensity gradient (yellow overlay) showing decreasing intensity toward the east Marco et al (2014)	Fig. 2c Load structures were formed by liquefaction of the lower light-colored lacustrine silt that penetrated the overlaying dark clayey soil. We interpret the structure asymmetry and zigzag shapes as result of shear forces exerted on the lake bed sediments by waves sloshing back and forth Marco et al (2014) Fig. 2d Load structures were formed by liquefaction of the lower light-colored lacustrine silt that penetrated the overlaying dark clayey soil. We interpret the structure asymmetry and zigzag shapes as result of shear forces exerted on the lake bed sediments by waves sloshing back and forth Marco et al (2014) Table 1 Summary of injection structure characteristics in studied trenches (Fig. 3 c and d) Marco et al (2014)	Brief Description and Interpretation Marco et al (2014:1449) report that flame-shape injections of silty sand that penetrate the overlying clay-rich soil are largest and most frequent within several meters of the point where the dam is badly damaged on the seaward side which they interpreted as a possible result of a large wave. They explained their interpretations as follows: Three features make the sand injections special: their lower extent is commonly asymmetric with dominant southeastward vergence, away from the breach in the dam zigzag shapes characterize the upper parts of many injections the size and frequency of the injections diminish gradually with distance from the dam until they completely disappear some 100 m away from it We suggest that the sand injections can be explained by overpressure that was induced either directly by earthquake shaking or by a tsunami wave that breached the dam, filled the reservoir behind the dam and increased the pressure on the water-saturated silt and sand layers and triggered liquefied sand injections. The movement of water sloshing back and forth in the lake accounts for the zigzag shape of the injections. The similarity to structures that were observed in Thailand after the great 2004 tsunami and other palaeotsunami observations lead us to prefer the tsunami origin of the liquefaction features. Detailed Interpretation Marco et al (2014:1454-1457) discussed their tsunamogenic interpretation in more depth. The sand injections may be interpreted as the result of overpressure in the lacustrine deposits, which could have been triggered either by earthquake shaking or by a sudden increase of overburden (Trifunac 1995). ... Liquefaction and sand injections are often reported in association with earthquakes (Obermeier 2009). Since no evidence for active faulting is found along the coast, we argue that the source of the causative earthquake was remote. The possible sources are the Carmel Fault, about 25 km northeast, or the Dead Sea Fault, about 60 km east of Taninim, both capable of generating M>6 earthquakes (e.g., Arieh 1967; Hamiel et al. 2009; Hofstetter et al. 2007). Compilations of liquefaction–epicentral distance data from Italy and eastern Mediterranean regions (Galli 2000; Papathanassiou et al. 2005) show that an M ~ 6 earthquake on the Carmel Fault or M > 6.5 earthquake at the Dead Sea Fault is capable of triggering liquefaction in our study area. The earthquake history of the Carmel Fault is not well known, but observations on a meticulously built Medieval basilica located on top of the fault, which does not exhibit any earthquake-related damage, indicate that during the last eight centuries there was no significant earthquake along it (Marco et al. 2006). Hence, we conclude that the source was along the Dead Sea Fault. An M > 6.5 on the Jordan Valley segment or an M > 7 on either the northern or the southern segments are plausible. Hence, earthquake-induced liquefaction is a possible cause for the flame structures observed in the study site. The conspicuous, nearly uniform asymmetry of the injection structures here (Fig. 3) is rare in other occurrences of injected sand. Earthquake-related sand injections are commonly symmetrical, upright, exhibiting upward-directed flow features (e.g., Obermeier 2009; Tuttle and Schweig 1995). We therefore hypothesize that the asymmetry may have been caused by shear that was induced by dominant southeastward flow, and the zigzag shapes at the upper parts of many injection structures indicate the sloshing of water back and forth. The flow could trigger shear instability at the upper part of the sediment, also known as Kelvin–Helmholtz Instability (Drazin and Reid 2004). Palaeotsunamis have been reported to trigger sand injection features, e.g., by De-Martini et al. (2003) and Owen and Moretti (2011), and asymmetric flame structures were observed in the deposits of the December 26, 2004 tsunami in Thailand (Matsumoto et al. 2008). The latter describe asymmetric injections of fine sand that was deposited by the tsunami, which intrude the overlying coarse-grained strata, uniformly skewed in the direction of the run-up current. Matsumoto et al. (2008) interpret the sequence as a result of two tsunami waves, where the boundary between the two deposited strata, i.e., the top of the first wave deposits, was simultaneously deformed and truncated by the second run-up current. Another set of asymmetric folds and injections observed in cores from the Aysen Fjord, Chile were associated with mass movement triggered by the April 27, 2007 M_W 6.2 earthquake (Van Daele et al. 2013). In the Lisan Formation (the palaeo Dead Sea sediments), similar zigzag shaped injections are capped by fragmented laminae. The sequence is interpreted as the result of re-suspension of the bottom sediment by the shear with the water during seiche or tsunami events (Alsop and Marco 2012; Wetzler et al. 2010). We suggest that a rapid addition of about 3 m of water (measured from the lake deposits to the top of the dam) during a tsunami surge could increase the overburden on the sandy layer, which is confined at its bottom and top with stiff clayey soil. The surge would have induced a sudden increase of pressure of 0.3 bars, at least three times the confining pressure (assuming soil thickness were similar to that of present, i.e., about 0.5 m and bulk density of about 1.5 g/cc). These conditions are highly likely to trigger liquefaction and the injection of water-saturated sediment into the overlying soil (Trifunac 1995). This scenario is somewhat similar to the case of palaeotsunami deposits that were found in the Kakawis Lake, near the shore of the Vancouver Island, British Columbia (Lopez 2012). We did not find deposits of marine origin, either because of the intensive modern agricultural cultivation of the upper strata that mixed the soil and made the microfossils very rare and hard to find, or because seawater did not go over the dam. An alternative tsunami-related explanation for the liquefaction on the eastern side of the dam could be a pressure gradient developed by increasing water head on the western side of the dam, where the tsunami wave ramped up but did not top the dam entirely (Craig 2004). While this mechanism and the earthquake-induced mechanism may explain why we have not found marine fauna east of the dam, it does not explain the systematic asymmetry of the injection structures. We therefore argue that the observed injections of the sand were triggered by an earthquake, and the shape of the injections combined with the damage at the seaward face of the dam favor the involvement of a tsunami wave. The candidate-triggering events are the earthquakes that postdate the accumulation of the entire sequence of lake deposits and some 20–30 cm of clay-rich soil that was formed at the surface after the lake dried. The historical accounts that associate tsunamis with earthquakes in this region have been reviewed and screened for reliability (Salamon et al. 2011, 2007), but none of the reliably recorded ones stands out as having a significant run-up that could top the 7-m-high dam (possibly 1–2 m less at the breached section). The precise location of the shoreline in the past is unknown, but it may have been different, shore transport along the coast of Israel following the construction of modern ports and marinas (Zviely et al. 2007). In the absence of local active faults, the tsunamis can be induced by remotely triggered submarine slumps (Salamon et al. 2007). We cannot determine independently the precise age of the observed liquefaction; however, the most suitable trigger candidate according to the dam history presented above, and the single ¹⁴C age, is the earthquake of November 25, 1759 (Fig. 3). Reports on this catastrophic earthquake describe boats that were swept ashore from the Akko harbor (50 km north of the studied site), and a large wave that was reported from as far south as the Nile Delta (Ambraseys and Barazangi 1989). The 90-km. long surface rupture of this earthquake was reported by the contemporary French Consul to Beyrouth [JW: it was the French Consul in Saida] after visiting the site in Lebanon [JW: the report was not based on a visit but on second hand information] (Ambraseys and Barazangi 1989) and confirmed by palaeoseismic studies (Daeron et al. 2005; Gomez et al. 2001). Ambraseys and Barazangi (1989) estimate the earthquake magnitude at 7 plus.	Marco et al. (2014) write that M > 6.5 earthquake on the Dead Sea Fault, located ~60 km east of the site, would be capable of triggering liquefaction at Tanninim. This suggests a minimum local intensity of VII (calculated using SEISCALC). This value (VII) is consistent with the Earthquake Archeological Effects Chart of Rodríguez-Pascua et al (2013: 221-224)

This evidence requires a minimum Intensity of VII (7).

References

Articles and Books

Ambraseys, N. N., and M.Barazangi (1989), The 1759 Earthquake in the Bekaa Valley: Implications for earthquake hazard assessment in the Eastern Mediterranean Region , J. Geophys. Res., 94(B4), 4007–4013

Marco, S., et al. (2014). Historical sand injections on the Mediterranean shore of Israel: evidence for liquefaction hazard , Natural Hazards 74(3): 1449–1459.

Porat, Y. (2002). The water-supply to Caesarea: a re-assessment, in Amit, D., Patrich, J., and Hirschfeld, Y. (eds), The Aqueducts of Israel, Journal of Roman Archaeology Supplementary Series 46: 104–129.

Websites

Historical sand injections on the Mediterranean shore of Israel: Evidence for liquefaction hazard on Schmuel Marco's webpage at the University of Tel Aviv