• from Braun et al. (2009)

• from Kagan (2011)

• Fig. 3.4.4 A En-echelon
  Figure 3.4.4A
  
  Photographs of speleoseismite in Soreq Cave
  
  En-echelon style cracks in flowstone
  
  Kagan (2011)
  style cracks in flowstone from Kagan (2011)
• Fig. 3.4.4 B Fractures
  Figure 3.4.4B
  
  Photographs of speleoseismite in Soreq Cave
  
  Fractures on cave ceiling with stalactite growth below them
  
  Kagan (2011)
  on cave ceiling with stalactite growth below them from Kagan (2011)
• Fig. 3.4.4 C Collapsed
  Figure 3.4.4C
  
  Photographs of speleoseismite in Soreq Cave
  
  Collapsed speleothem with post-collapse stalagmite regrowth, suspended between two pillars
  
  Kagan (2011)
  speleothem with post-collapse stalagmite regrowth, suspended between two pillars from Kagan (2011)
• Fig. 3.4.4 D Horizontal
  Figure 3.4.4D
  
  Photographs of speleoseismite in Soreq Cave
  
  Horizontal break in stalagmites
  
  Kagan (2011)
  break in stalagmites from Kagan (2011)
• Fig. 3.4.4 E Collapsed
  Figure 3.4.4E
  
  Photographs of speleoseismite in Soreq Cave
  
  Collapsed ceiling slab, note pre-collapse broken stalactites below ceiling and post-collapse stalagmite regrowth on top of ceiling
  
  Kagan (2011)
  ceiling slab from Kagan (2011)
• Fig. 3.4.4 F Severed
  Figure 3.4.4F
  
  Photographs of speleoseismite in Soreq Cave
  
  Severed stalagmite top with post-damage regrowth (sample SO-32)
  
  Kagan (2011)
  stalagmite top with post-damage regrowth from Kagan (2011)
• Fig. 3.4.4 G Collapsed
  Figure 3.4.4G
  
  Photographs of speleoseismite in Soreq Cave
  
  Collapsed soda straw (macaroni) stalactites embedded in flowstone cave floor
  
  Kagan (2011)
  soda straw (macaroni) stalactites embedded in flowstone cave floor from Kagan (2011)
• Fig. 3.4.4 H Collapsed
  Figure 3.4.4H
  
  Photographs of speleoseismite in Soreq Cave
  
  Collapsed ceiling blocks with post-collapse stalagmites growing below ceiling fracture
  
  Kagan (2011)
  ceiling blocks with post-collapse stalagmites growing below ceiling fracture from Kagan (2011)
• Fig. 3.4.8 H Severed Stalagmite
  Figure 3.4.8
  
  Speleoseismite sample SO-32 [Event B]. A. Hand sample, B-D. Photographs under polarizing microscope. See text for details.
  
  Kagan (2011)
  sample SO-32 [Event B] from Kagan (2011)

• from Kagan (2011:74-76)

Caves create an environment protected from most erosive activity. The calcite and detrital deposits within caves have laminar growth patterns preserving delicate evidence including structural damage from earthquakes. Speleothems can be dated with radiometric methods, making it possible to study the temporal patterns of seismic events [e.g. Kagan et al., 2005; Panno et al., 2009: Plan et al., 2010].

The mechanical relation between earthquakes and the breaking of speleothem structures is not clear. Various aspects of the relation have been investigated in past studies, including investigation of the ground acceleration needed to damage different speleothems, the types of speleothems sensitive to breaking under certain conditions, and predicted modes of failure [e.g. Cadorin et al., 2001; Lacave et al., 2000, 2004; Becker et al., 2006]. Yet speleothems are heterogeneous by nature [Gilli et al., 1999; Lacave et al., 2000] owing to their internal structure, composition, growth rates, and location within a cave. Furthermore, considering site effects and cave depths, shapes and sizes, we are not yet able to precisely predict the effects of earthquakes on speleothems. These considerations make it difficult to evaluate clear intensity values of speleothem damage for intensity scales such as the Environmental Intensity Scale 2007 [ESI07- Reicherter et al., 2009]. Modern observations and detailed investigation immediately following earthquakes need to be carried out for calibration of the past events to quantitative parameters.

Nevertheless, observations of broken speleothems due to modern earthquakes have been documented in caves from around the world [e.g. Gilli et al., 1999; Aydan, 2008; Perez-Lopez et al., 2009]. Dated damaged speleothem samples were reported to have yielded ages of known historical and pre-historic earthquakes in various studies [e.g. Postpischl et al., 1991; Morinaga et al., 1994; Lemeille et al., 1999; Kagan et al., 2005].

Effects of earthquakes on caves and speleothems can come in different forms (Figure 3.3.3). These may include cracks and fissures, severed stalagmites, collapsed and broken speleothems, collapsed ceilings and rockslides, changes in growth axes due to tilting [e.g. Postpischl et al., 1991; Morinaga et al., 1994; Forti, 1998; Gilli et al., 1999; Lemeille et al., 1999]. Another form of earthquake induced damage is the closing or opening of cracks, depending upon their locations in relation to stress fields [Muirwood and King, 1993].



Most paleoseismological and speleoseismological studies seek to date phenomena which occurred almost instantaneously by earthquakes. When a similar age is obtained by a few different speleothem samples from different parts of a cave, it is suggestive that these were not spontaneous separate collapses, but indicative of an earthquake [Kagan et al., 2005; Braun, 2009]. Dating such events can only give an age range due to analytical and geological uncertainties, as in most geological scenarios. These uncertainties can prevent differentiation of closely timed seismic events, but quiescent intervals, as well as periods of clustering, can be identified clearly.

• from Kagan (2011:76-77)

The Soreq and Har-Tuv caves, located 15 km west of Jerusalem (Figure 3.3.1), offer an excellent opportunity for speleoseismic investigation [Kagan et al., 2005]. The two caves have developed under nearly identical geological and climatic conditions. Research in two nearby caves offered the prospect of correlation with most of the 185-ky archive of Kagan et al. [2005]. The Soreq Cave has been studied intensively and shows continuous growth of speleothems for the last 185 ky [Bar-Matthews et al., 2000; Ayalon et al., 2002] and probably the past 350 ky [Bar-Matthews and Ayalon pers. comm.] The Har-Tuv Cave is in an active part of the Har-Tuv Quarry and is intended for destruction. Therefore, sampling in this cave was not limited by considerations for preservation.

Soreq Cave is elongated in the NW-SE direction with an average length of 80 m and an average width of 60 m [Asaf, 1975]. The floor consists of flowstone, stalagmites, fallen speleothems, and locally some mud. There are abundant fractures on the ceiling of the cave, some of them filled with reddish sediments and probably connect the cave to the ground surface. Curtain-type stalactites grow underneath ceiling fractures and form walls that divide the cave into spaces known as rooms, or halls. This cave contains a large amount of fallen cave deposits, of all types and sizes, which provide information on the seismic history of the region. This history is given mainly by dating seismic events, with additional information on local intensity and other physical data pertaining to underground earthquake damage. Mapping showed preferential orientation of collapses and, together with dating clusters, indicated non-random, non-spontaneous collapses. U-series dating of damaged speleothems and of deposits that have grown on them accurately places the causative events into the regional seismic chronological record.

The Har-Tuv Cave was only recently uncovered during quarrying activities. It is part of the same system of karstic caves described above. At the entrance to the cave there is a fault that dips 75/035, showing slickensides orienting 1250, whose age is unknown. It is a small cave, fairly horizontal, about 20 m long and maximum 7 m wide. There are many standing speleothem pillars and some broken ones. There is an abundance of ceiling collapses, mostly covered with stalagmite-stalactite growth. There is also an area covered by a thick layer of flowstone.

The entire Soreq-Har-Tuv Judean Hills archive covers ~185 ky and includes dating of more than 60 speleoseismites. Damages in these caves have been shown to stem from 13-18 earthquakes with a mean recurrence interval of 10–14 ky [Kagan et al., 2005]. This archive shows correlation between numerous events at the Soreq and Har-Tuv caves. However, the Holocene portion of this archive, presented in this paper, shows dissimilar dates of collapses in the two caves (Tables 3.3.1, 3.3.2).

• from Kagan (2011:74-76)

Table 3.3.2



As Table 3.3.2 demonstrates, the Denya cave speleoseismites record two major earthquake events in the Holocene, at ~5ka and ~10.5 ka. The Soreq-Har-Tuv caves record four events: two possible historic earthquakes (younger than the Denya archive and therefore not discussed here further; to be discussed in a future paper), a ~5ka event, and a ~8.6 ka event. In addition, one post-contact sample was dated to 11.6±0.3 ka at Har-Tuv Cave; this poorly constrained collapse event might have occurred prior to this post-contact age.

Three speleoseismite samples from the Har-Tuv Cave cluster to ~5 ka (Table 3.3.2). Considering the error of ages and the sampling process, we suggest that the ages of the three samples may indicate a single event. This event might very well be the same event documented by four fallen speleothems from Denya Cave giving an isochron age of 4.8±0.8 ka.

When comparing this event age (~5 ka) to the results of other paleoseismic studies, complexities arise since this age appears both along the CF, as well as in the JV at all sites discussed here. The shutter ridge sediment along the Yagur section of the CF [Zilberman et al., 2008] indicates a time of increased tectonic movements along the CF at ~5 ka, while an event which occurred at ~5ka displaces the slope seen in the paleoseismic trench on the shores of Lake Kinneret, JV [Katz et al., 2011]. Furthermore, fractured Early Bronze temple walls at Megiddo [Marco et al., 2006] show damage interpreted to be earthquake induced and are dated to ~5 ka. Ferry et al. [2011] also discuss archaeoseismic evidence at ~5 ka (2900±50 BC) from the JV which correlates to this event. Cosmogenic (³⁶Cl) dating of an exposed limestone scarp has inferred a rapid and significant displacement in the lower Galilee around this period (between 4 and 6.5 ka) [Mitchell et al., 2001]. This fault, a part of the E-W Galilee fault system north-west-west of Lake Kinneret, may be related to the regional seismic events indicated in the other sites.

A collapse event recorded in the Har-Tuv Cave speleothems indicates a seismic event along the DSB at ~8.6 ka. As of yet, this is the only location where earthquake evidence has been found for this time interval. However, the extensive and detailed early Holocene paleoseismic archive of the Dead Sea sediments is yet to be investigated [Kagan et al., 2011; Migowski et al., 2004].

The inferred seismic event in the early Holocene (~10.5 ka) seems to have had a significant effect on speleothems in the Denya Cave (five speleothem collapses sampled, see Table 3.3.2). No evidence for a seismic event during this time has yet been found elsewhere along the CF. At the JV Lake Kinneret paleoseismic site, one trench exposed a post-faulting unit that yielded an OSL age range of 10±0.8 ka, while another trench exposed a post-faulting unit with an age range of 9.2±1.9 ka [Katz et al., 2011]. This could be the same event, recorded at both the CF cave and the JV trenches (Figure 3.3.4).

The one sample with only a post-contact age of 11.6±0.3 ka found at Soreq-Har-Tuv is not a well-constrained event and the collapse may have occurred before this post-contact age,/b>. Unfortunately, this is a period when Dead Sea lacustrine seismites are not available at present due to a climatically driven lake level drop [Yechieli et al., 1993; Stein et al., 2010]. Even if this seismite represents a significantly earlier earthquake, then future drill cores in Dead Sea lacustrine sections from this period will not show seismites, as suggested by the results of Kagan et al. [2005], for periods with documented lacustrine Dead Sea sediments. The ~10.5 ka event that caused extensive damage at Denya may have been limited to the JV source, since there is no evidence of an event from on-fault CF studies to date, nor from the DSB. Alternatively, this may be a CF event, for which additional evidence is yet to be uncovered. Different interpretations for potential sources of these events are discussed below using a simplified model that represents the DSB-JV-CF fault system and considering its different paleoseismic proxies.

• from Kagan (2011:87)

We studied two speleoseismic sites, in the Judean Hills and Haifa, each providing individual archives of earthquake shaking in their respective vicinity. Both have been shown to be reliable off-fault cave proxies [Kagan et al., 2005; Braun, 2009]. Complex as dating speleoseismic events may be, each system shows distinct collapse age groupings. Moreover, when compared with one another, and with independent paleoseismic archives from the JV sector, some coupling is suggested for the CF and JV-DSB fault systems. Specifically, an event at ~5 ka is well-recorded at both the CF and Judean Hills caves, as well as in the Lake Kinneret-JV trenches, the CF-Yagur shutter ridge, and at the archaeological site at Megiddo. The study east of the Sea of Galilee reveals surface rupture and sediment deposition at ~5 ka. The study in the Carmel reveals accumulated sediments behind a shutter ridge at about the same age. Probably the DSB, JV, and the CF faults had to slip to account for these pieces of evidence (RC 1- RS A, complete system coupling). The age bracket of the penultimate Denya Cave event at ~10.5 ka is included in the uncertainty range of dated slip evens in the Lake Kinneret trench archive [Katz et al., 2011]. If, as noted above, we assume a complete record from all archives, this could signify a large JV event (Figure 3.3.5-II, Table 3.3.3; RC 9- RS F). Acknowledging that the on-fault record presented by Zilberman et al. [2008]’s is incomplete, RC-9 might indicate coupling between the two branches (RS-B) without coupling to the localized plate boundary. This northern event predates the closest event in the Judean Hills archive by two centuries or more. Uncertainties due to differing dating methods and the lack of a Dead Sea lake seismite archive for that period prevent ruling out a correlation. However, quiescent intervals, which are significant for seismic hazard assessment, can be identified clearly, for example the period between ~10 ka and ~5 ka in the cave in Haifa. Within the framework of the uncertainties of each one of the studies discussed, there are contemporaneous seismic ages that may represent alternatively (a) large earthquakes leaving their mark throughout the region, (b) periods of concurrent seismic activity on the CF and JV and/or DSB, or (c) seismic activity on the JV and/or DSB which triggers activity along the CF or vice versa (Figure 3.3.5-II, Table 3.3.3). Potential for additional data can be realized by more extensive sampling for dating seismites at caves in the region and likewise, from additional on-fault and off-fault paleoseismic studies. This type of multi archive analysis provides a clearer view of regional implications of complex fault activity. Using the model presented here, any further paleoseismic data can be analyzed in the context of the regional setting of this fault system and further enhance our understanding of its workings.

• from Kagan (2011:77)

Kagan (2011:77) noted that the Soreq and Har-Tuv caves show dissimilar dates of collapses in the Holocene (see Kagan, 2011:Tables 3.3.1, 3.3.2).

• from Kagan et al. (2005:261)

The long-term recurrence patterns of past earthquakes are of considerable consequence for hazard assessments, and have implications for earthquake physics. We introduce a rigorously dated record of earthquakes from an extensive number of well-preserved preseismic and postseismic precipitates from caves located off the Dead Sea transform. We dated events directly at the paleoseismic contact by means of a novel correlation method with the oxygen isotope record of the speleothems recovered in one of the caves. Within the 185 k.y. covered, we dated 38 seismite samples. These stem from 13–18 earthquakes with a mean recurrence interval of ~10–14 k.y. We show that the deformational events dated in the study caves complement independent near-fault paleoseismic records by temporal correlation with the earthquakes recorded therein. This opens up a significant new avenue of earthquake research that will provide precise dating and observational constraints on large infrequent earthquakes.

• from Kagan et al. (2005:261)

Records derived from instruments, historical documents, and soft sediments are inherently limited in temporal and spatial span. Regional paleoseismic histories are often incomplete because of lack of surface faulting, scarcity of Quaternary strata, and difficulties in recognizing and locating reliable and dateable earthquake markers (seismites). Cave deposits (speleothems) can undergo various types of damage during earthquakes and offer significant advantages for recovering long histories of earthquakes (Forti, 1998; Gilli, 1999; Gilli et al., 1999; Delaby, 2001) (Fig. 1). A modern-day example is provided by a 1996 M 5.2 earthquake in France that caused the collapse of thin stalactites in a cave 10 km from the epicenter (Gilli et al., 1999). The cave environment is ideal for paleoseismological investigation because earthquake damage is often fossilized by postearthquake calcification and preserved from erosion. The seismic response at different depths below ground varies greatly and may produce amplification or attenuation of as much as a factor of six (Kanai et al., 1966; Bard and Tucker, 1985). Therefore direct magnitude assessment is difficult, although this caveat can ultimately be alleviated by careful comparison with independent paleoseismic records.

The Soreq and Har-Tuv Caves, Israel, are located 40 km due west of the Dead Sea transform, one of the major strike-slip fault systems in the world, active since the Neogene (Garfunkel et al., 1981). The dominant seismogenic element in the area (Fig. 2) is the Dead Sea transform, yet secondary intraplate structures generate microseismicity (Salamon et al., 2003). The location of the source of off-fault paleoseismic evidence is never precisely known, yet the most likely source in our study area is the nearby segment of Dead Sea transform. Although the Jerusalem area, close to the study sites, is not directly on any of the seismic features, it has been affected by many earthquakes. Because this area has been continuously populated and has been a major religious and political center throughout historical times, the record of these earthquakes has been studied in detail (Ambraseys et al., 1994; Amiran et al., 1994; Guidoboni et al., 1994). Paleoseismic studies have provided geological evidence of earthquakes in the Dead Sea transform region (e.g., Marco et al., 1996; Ken-Tor et al., 2001; Amit et al., 2002; Migowski et al., 2004; Begin et al., 2005).

• from Kagan et al. (2005:263-264)

Extreme events are required for noticeable damage in the research caves. Modern examples have shown that the threshold shaking intensity for speleothem damage is likely to exceed VII on the European Medvedev-Sponheuer-Karnik scale, and the magnitude, even for a very close epicenter, exceeds 5 (Gilli et al., 1999). The last strong earthquake in the Dead Sea area was on 11 July 1927, M 6.2, and generated shaking intensities between VI and VII around the research caves (Fig. 2), leaving no evidence in the cave seismites dated thus far. Numerous earthquakes of this magnitude and larger have occurred in the area throughout history (Amiran et al., 1994). No historical earthquakes show in our record, where the youngest event we have found occurred at ca. 5.1–6.3 ka, correlating with archaeological and sedimentary evidence (Fig. 3).

Threshold ground acceleration for breaking speleothems and the attenuation relations provide constraints on earthquake size. A preliminary calculation shows that for flawless, perfectly cylindrical speleothems, the ground acceleration needed to break by flexion is proportional to the diameter divided by the square length, giving a minimum value of 2 m s^-2 for some speleothems in a Belgian cave (Cadorin et al., 2001). The geometry of the speleothems studied here would require higher accelerations in excess of 1 g. Such accelerations would result from an M 7.5–8 event at the Dead Sea transform (using a minimum distance of 60 km to incorporate Arava faulting, attenuation relation of Ambraseys et al. [1996], and amplification factor 6 [Bard and Tucker, 1985]).

Extrapolating the Gutenberg Richter relation from the instrumental window of 100 yr (Salamon et al., 2003) to the recurrence interval on the scale of 10 k.y., apparent from the present study, gives a maximum magnitude of 8.2. Since our record may be at present incomplete, we can consider a shorter mean recurrence interval such as 6 k.y., with a maximum magnitude 7.6 (Klinger et al., 2000). A significantly shorter recurrence would require a less complete (more sparsely sampled) record, which is unlikely since we would then expect to find more than 18 ages in 38 dated collapses. Further dating of collapses in the research caves will test this agreement between the independent considerations of recurrence intervals and ground acceleration.

While the limited number of dated events precludes a rigorous statistical approach, an indication of the sense of the distribution can be derived by the standard deviation of 6 k.y., and a coefficient of variation (standard deviation divided by the mean) of 0.6, which suggests periodic behavior (Kagan and Jackson, 1991; Ben-Zion and Rice, 1995). Even if we grouped two distinct earthquakes to a single event as a result of dating uncertainty, the long-term quiescence between events still seems to show periodicity. Large earthquakes along a single fault have been suggested to show a roughly cyclic behavior (Wesnousky, 1994). When these events are more frequent than expected from Gutenberg Richter relationships, they are coined ‘‘characteristic earthquakes.’’ Infrequent events, whether characteristic or not, may accommodate some of the sizable slip deficits previously noted (Salamon et al., 2003).

Previously, researchers recognized the potential for using cave deposits to reconstruct important paleoseismic events; however, until now, the capability to take advantage of this special medium by precise dating has not been established. We show that dating damage to caves and deposits in them is a practicable paleoseismic method. Cave records from outside the plate-boundary zone can provide very long-term archives of regional earthquakes that are essential for the understanding of the earthquake-generating process. The research caves, rich with datable, damaged speleothems, will permit us to test our interpretation by additional sampling and dating, a project that is currently under way. This method is especially valuable in karstic and tectonically active regions, particularly regions where speleothem growth has been continuous, such as at mid- to low latitudes.