Crustal Structure Across the Central Dead Sea Transform and Surrounding Areas: Insights Into Tectonic Processes in Continental Transforms

.

• The Central Dead Sea Transform separates a sediment and crustal structures of a passive continental margin from the Arabian plate interior • Seismic and gravity data show that unlike farther south, shoulder uplift here is absent and Moho step is more gradual • The shape of the Sea of Galilee and recent seismicity is explained by underlapping strike-slip fault stepover activating secondary faults

Supporting Information:
Supporting Information may be found in the online version of this article. 10.1029/2023TC007799 2 of 19 Several seismic refraction surveys of the DST have taken place south of the DST-Carmel Fault junction (Mechie et al., 2009;ten Brink et al., 2006;ten Brink & Flores, 2012;Weber et al., 2004).However, there have not been any seismic refraction surveys across the DST north of that junction (henceforth, the central DST).Sparse refraction and reflection surveys were carried out in 1989 and 1990 along the western portion of the present profile and into the Mediterranean Sea (Ginzburg et al., 1994) but the data from these surveys are no longer available for reprocessing and modeling.We report here the results of a densely instrumented seismic refraction profile across the central DST (Figure 2).We extend the imaged structure laterally and in depth by modeling the free-air gravity anomaly along the profile and extending it ∼70 km on either side.Finally, we carry out static stress modeling to understand the fault kinematics and seismic activity around the Sea of Galilee.
Our results address two topics: (a) the deep crustal structure of the DST and its adjacent areas, and (b) the kinematics of structures within the Sea of Galilee Basin.The geophysical data show that the DST separates between a 7-km thick sediment column overlying a uniform-thickness crust on the east and a ∼10 km thick monoclinal-shaped sedimentary column overlying a progressively thinner crust to the west.The eastern side of the DST is not flexed upward toward the transform in contrast to observations farther south.The Sea of Galilee, previously thought to be a complex basin, can be simulated with static stress models as an asymmetric pull-apart basin formed by a left-lateral slightly divergent stepover of two underlapping strike-slip fault strands.A mismatch between the DST location in the geophysical profile and the static stress model, may hint at temporal changes in the fault geometry.Faults within the Sea of Galilee which were previously suggested to carry part of the relative plate motion, are likely secondary faults.

Background Tectonics of the Dead Sea Transform and the Sea of Galilee Basin
Unlike strike-slip faults above active or fossil subduction zones, the lithosphere of the DST was subjected to relatively minor tectonic and thermal events immediately prior to and during the transform establishment.An ≤80 Ma regional compressional phase oriented at a low angle to the DST has reactivated earlier rift-related normal faults (Syrian Arc fold belt, Figure 1; Sagy et al., 2018).A Tertiary lower lithosphere erosion of the eastern side of the DST under Jordan was suggested from receiver function analysis (Moshen et al., 2006).The lower lithosphere erosion may be the result of asthenospheric flow from the Afar mantle plume (Faccenna et al., 2013).Heat flow in the vicinity of the DST is, however, generally low (45-53 mW/m 3 , Eckstein & Simmons, 1977;Galanis et al., 1986;Shalev et al., 2013) indicating that any deep thermal anomaly has not yet reached the shallow crust.Northern Israel emerged from under relatively deep water during Early Miocene (Gvirtzman et al., 2011, and references therein), but the reason for the emergence is unknown.Volcanic activity is generally not associated with the plate boundary (e.g., Ilani et al., 2001), and Neogene volcanism is found on both sides of the plate boundary in our study area (e.g., Ilani et al., 2001;Wald et al., 2019).
In addition to having a relatively simple lithospheric tectonic and thermal history, the southern and central parts of the DST are also characterized by a relatively stable displacement history.Left-lateral motion between Arabian and the Sinai sub-plate along the southern part of the DST (Israel-Jordan sector) is parallel to the trace of the fault system with minimal compressive or tensile components (e.g., Le Beon et al., 2008;Palano et al., 2013;Sadeh et al., 2012).The relative plate motion is slow (4.6-5.9 mm/yr, Sadeh et al., 2012), but had been continuous since 20-18 Ma (Nuriel et al., 2017), resulting in a cumulative offset of 105-110 km south of the Lebanon restraining bend (Freund et al., 1970;Quennell, 1958).The displacement is confined to one or two parallel or overlapping fault segments (ten Brink et al., 1999).The only exception is the Carmel Fault system which branches to the northwest and terminates at the boundary between the continental margin and the Levant Basin (Figure 1).This branching fault system appears to accommo- 10.1029/2023TC007799 3 of 19 date a small portion of the relative plate motion in a transtensional motion, reducing the relative motion along the DST north of their junction by ∼1-1.5 mm/yr (Sadeh et al., 2012).The Carmel branching fault system may utilize the western part of an Oligocene fault system, the Irbid rift and its southeast extension at the Azraq-Sirhan graben (Figure 1), which cross the DST in a NW-SE orientation (Rosenthal et al., 2019;Segev et al., 2014).
The freshwater lake of the Sea of Galilee (Lake Kinneret) is underlain by a sedimentary basin of poorly determined age and depth (e.g., Ben-Avraham et al., 1996;Eppelbaum et al., 2007;Hurwitz et al., 2002;Reznikov et al., 2004).Geophysical studies indicate that the internal fault structure in the Sea of Galilee is complex and does not conform to a classical pull-apart basin structure.This has led to various suggestions about the nature of displacements along this part of the DST (Barnea Cohen et al., 2022;Ben-Avraham et al., 1996;Gasperini et al., 2020;Hurwitz et al., 2002;Reznikov et al., 2004), discussed in more detail in Section 5.4.

Data Acquisition
We collected a densely instrumented seismic refraction profile across the central DST from the Mediterranean coastal plain near the town of Acre (Akko) in the west to the Golan Heights in the east (Figure 2).The location of the 72-km-long profile maximizes the available length between the Mediterranean Sea and the Syrian border, and minimizes highly populated areas, major traffic routes, secondary faults, and rough topography.The profile is located at the northern end of the Lower Galilee, a region of 100-600 m high basin and ridge topography, and a few kilometers south of the Upper Galilee, an 800-1,200 m high region that continues north into southern Lebanon (Figure 2).It crosses the northern edge of the Sea of Galilee and the Bethsaida Valley.
The seismic sources for the 72-km-long seismic refraction profile were nine shots of 300-400 kg of Emulsion or Anfo explosives detonated at depths of ∼15 m underground.The shots were spaced 6-10 km apart (Figure 2; Table S1 in Supporting Information S1).The seismic waves were recorded by 361 receivers spaced ∼200 m apart.The receivers were vertical Geospace GS-11D 4.5 Hz geophones connected to stand-alone Reftek RT125A (Texan) dataloggers.Prior to deployment, all receiver and source locations and elevations were pre-surveyed and marked with real-time kinematic (RTK) GPS.Final locations were determined by hand-held GPS during the deployment.Data were recorded continuously at a sampling rate of 4 ms from the 9th of April, 2018, 16:00 UTC

Seismic Modeling
First-arrival tomography using the software PROFIT (Koulakov, 2009;Koulakov et al., 2010Koulakov et al., , 2011) ) was used to construct an initial p-wave velocity model (Figure 3).The starting model was a 1-D velocity gradient (Table S2 in Supporting Information S1, Figure 3a) and the model converged after 8 iterations (Table S3 in Supporting Information S1).A checkerboard test shows a good recovery of ±10% variations in velocity structure down to a depth of 5 km (Figure S2 in Supporting Information S1) with a lateral resolution of 9 km.The final tomography model is shown in Figures 3b and 3d and the ray-path distribution are shown in Figure 3c.
The tomographic inversion served as a starting model for a more detailed forward-modeling ray tracing, which incorporated additional observed reflections, head waves, and secondary refraction arrivals (Zelt, 1999).We selected velocity contours of 4.0, 4.5, 5.0, 5.5, and 5.9 km/s to build our initial model (Figure 4) and extended the model laterally to 75 km from the original 70 km and down to 20 km from the original 7 km (Figure S3 in Supporting Information S1).Seismic arrival picks were digitized using the software PASTEUP and the forward model was run interactively using the software MODELING (Fujie, 2008).
The forward model (Figure 5) accentuates the structure around the DST and eliminates the low-velocity region at depths of 4-5 km under shot 3 in Figure 4.The model extends deeper to fit secondary reflections, although the picking error of the arrivals of the bottom two layers (≥6 km depth) is large (Table S4 in Supporting Information S1).The RMS fit (0.064 s) is improved relative to the first arrival tomographic inversion (0.135 s).A summary of the statistics for our 2-D velocity model is shown in Table S4 in Supporting Information S1 and the fit of calculated arrivals to observations is shown in Figure S4 in Supporting Information S1.The forward model achieved better fit of the observations than the first-arrival tomography.

Gravity Modeling
Our 2-D gravity model extends the seismic refraction model both in depth to the Moho and laterally beyond the Mediterranean coastline and the Syrian border (Figure 6).The profile is oriented along Lat.32.88°.The distances along the model match the seismic line with the origin at the coastline.Modeling was carried out with the commercial software package GM-SYS version 4.10.
The gravity data are a compilation of free-air gravity data in the Israel national data set from measurements by the Geophysical Institute of Israel, TGS co. and the Natural Resource Authority of Jordan (blue dots).Additional data are from Rybakov et al. (1997) compilation of gravity data in the Levant (green dots).Rybakov's et al. (1997) data lack information about data sources or data quality and were therefore used only where no other data were available.Topography and bathymetry values were derived from the GEBCO-2020 data set (GEBCO, 2020).The distribution of gravity measurements is variable and generally sparse along the profile (Figure 6a).To include more measurements, we used projected values within ±500 m of the profile (Figure 6b).Misfits between the observed and modeled gravity anomalies may arise not only from the lack of coincidence between elevations along the profile and the elevations of the offline gravity points, but also from 3-D topographic effects in the hilly Galilee region (profile km 15-30) and the eroded gullies at the edge of the Golan Heights (profile km 65-75).
Gravity models are inherently non unique and therefore require external constraints to make them reliable.The following constraints were used: Layer thickness along the seismic profile follows the depth and shape of the forward model layers on land (dashed blue line in Figure 6c), layers extracted from a database of gridded seismic reflection horizons offshore (Gardosh et al., 2008) (dotted blue lines in Figure 6c), the thickness of basalt layer in Dafny et al., 2003, the depth to basement in the Golan Heights from seismic reflection data of Meiler et al. (2011), and mid-crustal and Moho depth estimates in Syria (Brew et al. (2001) and the Mediterranean Sea (Netzeband et al., 2006) (white markers and green dots, respectively in Figure 6d).The gravity model of the depth to Moho under the seismic line is anchored by the estimated Moho depths in Syria (Brew et al., 2001) and under the Levant Basin (Netzeband et al., 2006).In addition, the upper/lower crustal interface follows the depth estimate of Brew et al. (2001) and it is assumed to be horizontal.A horizontal interface only shifts the calculated gravity anomaly by a constant value but does not affect the shape of the anomaly.
Moho depth under our seismic profile is unconstrained by independent data.To pick the model that best fits the observations, we ran sensitivity tests in which the Moho steps vertically between a lower depth of 37 km (Brew et al., 2001) and an upper depth ranging from 31 to 33 km (Figure 7).(A shallower upper step resulted in significantly worse fit).Two step geometries were tested, one in which the Moho started rising under the DST (Figure 7a) and another in which the step is spread at equal distances east and west of the DST (Figure 7b).A model with an upper depth of 32.5 km and a step width of 20 km, extending westward from the DST, was the best fit model (RMS = 4.132 mGal).We also carried out a linear inversion to define depth to Moho under the profile.
The RMS misfit further reduces (4.045 mGal) after 3 iterations, but Moho topography becomes more variable (Black lines in Figures 7a and 7b).Significantly, the slope of the step in the inversion (14°) is similar to the slope of the best fit model (13°).We use below the best-fit model for the sake of keeping the simplest Moho topography.Rock densities are listed in Table 1, and follow the densities used by Bronshtein (2017) to model two profiles located ≤60 km south of our profile.Where available, Bronshtein (2017) used densities from borehole measurements in the Mediterranean Sea, northern Israel, and northern Jordan.Elsewhere, he used previously published We further compared Bronshtein's (2017) shallow densities to lithologies found in two boreholes along the profile.The uppermost 340 section of the Ness-06 borehole (see location in Figure 6a), which is coincident with shot point (SP) 8, is composed of Neogene basalt, paleosol, conglomerate sand and silt with an assumed average density of 2,400 kg/m 3 .The underlying 740 m are Eocene age chalk, marl, and argillaceous limestone with the same assumed density.The bottom 300 m are dolomitic limestones, chert, and chalk of uppermost Mesozoic age with an assumed density of 2,550 kg/m 3 .The top 820 m of Hazon-01 borehole, located between SP 3 and 4 in the Galilee, are Lower Cretaceous marl and sand with limestone and dolomite near the surface.The underlying 200 m section is composed of weathered volcanic rocks and tuff.Both layers are assumed to have an average density of 2,400 kg/m 3 .The bottom 610 m are Jurassic limestone and dolomites with an assumed density of 2,550 kg/m 3 .

Structure of DST Valley
The basin underlying the DST along the profile is ∼4.5 km wide (vertical dotted lines in Figures 5 and 6) and is located under the Bethsaida Valley, a lowland occupying the deltas of the Jordan River and secondary streams flowing from the Golan Heights.The seismic and gravity models do not have the resolution to determine the internal stratigraphy of the basin sediments or the exact locations of its lateral boundaries.The ∼1.5 km thick basin fill is represented by low-velocity and low-density upper layer.The gravity model shows a density contrast extending down 12 km into the crust under the eastern side of the basin (Figure 6c), but only to depth of <5 km under the western side.

The Structure Surrounding the DST Valley
The shallow (<5 km) sedimentary section under the Galilee west of the DST is arched upward with a wavelength of ∼40 km and maximum amplitude of 0.4-0.45km (Figures 5 and 6).The amplitude is measured by comparing the maximum elevations of the velocity layer of ∼5.0-5.3 km/s relative to model-kilometer 5, where ray coverage provides adequate control on the structure (Figure 5a).The uplift amplitude is less than half of that measured by velocity contours west of the Dead Sea Basin to the south (ten Brink & Flores, 2012).The uplift shape is monoclinal.The location of the maximum uplift is under the eastern end of Albian-age outcrops (maximum exhumation in Figure 5b), the oldest rocks along the profile (Bogosh & Sneh, 2014;Sneh et al., 1998; Figure 5b).The velocity and gravity models indicate that the uplifted sedimentary section dips toward the DST starting about 8 km west of its western boundary (Figures 5 and 6c), similar to some transform-perpendicular profiles farther south (Wdowinski & Zilberman, 1997).
To the west, the uplift disappears in the coastal plain west SP 2 (Figure 6 and top Judea Group profile in Figure 5c) and the sedimentary layers continue to descend westward.This descent is likely related to the thinning of the crust.The transition of the pre-Neogene layers from the continental margin to the Levant Basin is steep and the outermost shelf encloses an inner shelf basin east of Foxtrot-1 well (Figure 6c).The steep transition to the Levant Basin and the shelf basin is controlled by the NW continuation of the Carmel Fault offshore (Figure 2) which transects our profile at the shelf edge (Ben-Gai & Ben-Avraham, 1995;Schattner & Ben-Avraham, 2007).The Carmel Fault is characterized by a left-lateral/normal oblique-slip motion (e.g., Sadeh et al., 2012).This sense of motion is responsible for the uplift on the SW side (the Carmel Ridge and its offshore continuation) and subsidence on a confined shelf basin, the Kishon graben, which underlies Haifa Bay (Ben-Gai & Ben-Avraham, 1995).

The Deep Structure
Our interpretation of the deep structure is based on gravity modeling (Figure 6d) because of the limited span of the seismic profile.The model suggests a vertical offset in Moho depth under the DST between depths of 37 and 32.5 km over a lateral distance of 20 km (Figures 6 and 7), and a more gradual change under the continental margin to a depth of 24 km under the Levant Basin.Sensitivity models (Figure 7) indicate that crustal thinning begins under the DST and extends westward while crustal thickness of the eastern side remains constant.

Basement Depth Across the DST
Sedimentary columns of different thickness, velocities, and densities are observed on both sides of the DST (Figures 5c and 6b).Depth to basement east of the DST is ∼7 km.The lowest few kilometers of the sedimentary column there correspond to velocities of 5.6-5.9km/s in the seismic section (Figure 5a) and are interpreted to consist of Paleozoic sediments, similar to Meiler et al. (2011).Approximately 7 km deep basement was also interpreted based on published isopach maps constructed from seismic reflection profiles (Meiler et al., 2011), regional gravity models (Rybakov & Segev, 2004;Segev et al. (2014), and gravity studies in southern Syria (Brew et al., 2001).Top basement in our models dips by about 0.5 km toward the DST within ∼15 km of the plate boundary, which is smaller than the amount of basement dip in Meiler et al. (2011) (1-1.5 km).
The gravity model (Figure 6d) indicates that depth to basement west of the DST is ∼10 km increasing to 11.5 km toward the DST.This depth is larger than the depths proposed by Rybakov and Segev (2004) (7 km) and Segev et al. ( 2014) (8-9 km) based on regional gravity models, and 7 km, proposed by Ginzburg et al. (1994) based on sparse seismic refraction data.Although seismic velocities at depths >6 km are similar across the DST, the corresponding modeled density is ∼2,600 kg/m 3 (Bronshtein, 2017;Rybakov et al., 1999).We therefore interpret these velocities west of the DST as representing Paleozoic and early Mesozoic rocks, partly composed of limestone and dolomites.Deep boreholes in northern Israel and offshore do not penetrate deeper than 6 km and do not reach Paleozoic rocks.More generally, deeply buried carbonate rocks can reach velocities of 6-6.5 km/s and densities of 2,600-2,650 kg/m 3 (e.g., Hamilton, 1978).
The difference in basement depth west and east of the DST is ubiquitous along the entire length of the central and southern DST (Figure 8).The difference reflects the fact that the DST in this region separates the interior Arabian plate from its Triassic-Jurassic continental margin (Steckler & ten Brink, 1986).The crust under the interior Arabian plate retains a constant thickness up to the DST and then thins rapidly westward under the continental margin (Mechie et al., 2013;Segev et al., 2014) allowing for a greater subsidence and sediment accumulation on the west side.Velocities and densities west of the rift reflect the different overall facies of the sedimentary column: limestone and dolomite are deposited on a mostly Mesozoic-Lower Cenozoic carbonate platform west of the DST and variable continental, fluvial and shallow marine deposits of mostly Paleozoic are found east of the DST (Garfunkel & Derin, 1984).
Sediment thickness east of the DST under our profile is larger than along the more southern profiles where basement is at depths of 3-5 km (Figure 8).The higher sediment thickness here may be due to the lack of erosion caused by shoulder uplift and the development of a local Mesozoic syncline on top of ∼2,000 m of Paleozoic sediments (Meiler et al., 2011).

Lack of Uplift of the Eastern Shoulder of the DST
Geological cross-sections south of the study area in Jordan show an upward flexure of the sedimentary layers toward the DST (Bender, 1974;Wdowinski & Zilberman, 1997).Seismic refraction data confirms that the uplift extends to at least the middle crust (Mechie et al., 2009-profile 5 in Figure 8; ten Brink et al., 2006-Profile 6 in Figure 8).The geometry of the eastern shoulder of the DST at the latitude of our cross-section is different.The seismic and gravity structure shows dipping sedimentary layers toward the DST under the Golan Heights.
Neither is an uplifted eastern shoulder of the DST observed north of the Lebanon restraining bend in Syria (Gomez et al., 2006).The reason for the change in the eastern shoulder geometry from south to north is unclear.Crustal thickness under our profile is similar to that farther south (Mechie et al., 2013).Using thermo-mechanical models, Sobolev et al. (2005) proposed that the up-flexed eastern shoulder of the DST to the south is due to Neogene (<10 Ma) thinning of the lithosphere.The thickness of the lithosphere under central Jordan, determined from receiver function analysis, is 80 km thick and is 65 km thick south of 30°N (Moshen et al., 2006).
Although there is no seismically determined lithospheric thickness under our profile, Al Kwatli et al. (2012)

Offset in Moho Depth Across the DST
The westward decrease in crustal thickness starting under the DST is interpreted to be a function of the shape of the Levant margin coupled with the 105-110 km of left-lateral motion along the plate boundary.The Levant continental margin changes its orientation from E-W at the latitude of the Dead Sea Basin (31.2°N) to roughly N-S farther north (e.g., Segev et al., 2014).As a result, the Arabian plate is juxtaposed against a thinner and narrower continental margin due to the left-lateral motion along the DST (Profiles 1-4 in Figure 8).
A broader Moho step is interpreted under our profile than under profiles 2-4 farther south (Figure 8).The preservation of a Moho step throughout the history of the DST (≤18 Ma, Nuriel et al., 2017) is indicative of a relatively high viscosity in the lower crust (ten Brink, 2002), which can resist ductile flow due to horizontal stresses generated by the step.The more gradual step under our profile (13°-14°) than under profiles 2-4 (22°-24°) may reflect lower viscosity relative to the profiles farther south (ten Brink et al., 1990), perhaps due to the extensive Neogene volcanic activity in the area (e.g., Ilani et al., 2001;Segev et al., 2014;Wald et al., 2019).A slightly warmer lower crust in our study area may be reflected by the depth limit of micro-seismic activity along the DST, provided the seismic activity represents crustal brittle deformation.Whereas seismic activity in the central DST reaches depths of 25 km (Braeuer et al., 2012;Hofstetter et al., 2020; Figure 8), micro-seismic activity in our study reaches only 15-20 km and is typically shallower (Haddad et al., 2020; Figure 8).
Detailed investigation of the rheological properties of the lower crust under the DST was conducted farther south at the Dead Sea Basin.There the basin experienced a sudden increase in subsidence rate to 4 km/Myr during the past 1 Myr, accompanied by longitudinal sagging and elongation (ten Brink & Flores, 2012).Petrunin and Sobolev (2006) modeled the rapid subsidence by ductile flow caused by lower crustal heating.However, surface heat flow in the vicinity of the southern DST is too low (45-54 mW/m 2 ; Galanis et al., 1986) to generate ductile flow in the mafic lower crust (ten Brink, 2002).ten Brink and Flores (2012) proposed that the increased subsidence rate and sagging were due to a reduction in crustal shear strength caused by the development of inter-connected mid-crustal ductile shear bands as a result of retrograde metamorphism in the presence of fluids.

The Sea of Galilee as an Asymmetric Pull-Apart Basin
The fault geometry of the DST around the Sea of Galilee had been a subject of debate with several authors suggesting that at least part, if not all the motion is presently accommodated via faults in the western side of the lake and farther north on land.Ben-Avraham et al. (1996) proposed the existence of a western marginal fault (WMF) extending northward to the widest part of the lake (Figure 9b), which may accommodate part of the relative plate motion.Reznikov et al. (2004) extended this fault almost to the northern end of the lake and Hurwitz et al. (2002) extended the WMF north of the Sea of Galilee.Hurwitz et al. (2002) and Reznikov et al. (2004) proposed several diagonal normal faults crossing from SE to NW along the widest part of the lake (Figures 9a and 9b).Gasperini et al. (2020) proposed that the current plate motion in this area is accommodated by a diagonal fault starting from the main fault at the southeast end of the lake, extending to the NW shore of the lake (Figure 9c) and continuing northward to connect with Roum Fault in Lebanon (see Figure 1 for location).Gasperini et al. (2020) further proposed that this fault trace is now the major throughgoing fault accommodating the motion on the DST and bypassing the lake and the main fault north of the lake.However, the offset of walls of Vadum Jacob crusader castle, located on the DST fault strand 12 km north of the lake (Lat.33.03°, Figure 9c) by the 1202 earthquake (Ellenblum et al., 1998), argues against their suggestion, and indicates that the northern strand of the DST continues to be active at the present time.Barnea Cohen et al. ( 2022) proposed NW-oriented system of shallow (<6 km) parallel normal faults extending from the northern end of the DST strand (Figure 9d) which is rooted in a horizontal salt layer.
We propose that much of the above-described complex fault geometry within the Sea of Galilee represents secondary faulting.The primary structure of the basin is suggested to arise from motion on the two main fault strands of the DST, shown in the Geological Survey of Israel map (Sneh & Weinberger, 2014) of major geological structures in Israel (Heavy lines in Figures 9b, 9d and 10).For the sake of modeling, we further simplify the main faults by assuming a single continuous southern strand extending from south of the lake along the eastern side of the Sea of Galilee (Table S5 in Supporting Information S1; red lines in Figure 10).We can simulate the first-order subsidence of the Sea of Galilee by calculating the vertical displacements (black contours in Figure 10) due to left-lateral motion along these two (red) faults strands assuming only a very small normal component on the southern strand (2.5% of horizontal).A left stepover of 2.6 km in the fault system with a gap (underlap) of 9 km between the fault strands creates a subsidence pattern that matches the shape of the Sea of Galilee and Bethsaida Valley.The modeled faults in our model dip 70° to the west and extend between depths of 1-18 km (Table S5 in Supporting Information S1).The westward-dipping fault system generates the asymmetric subsidence.The modeled underlapping gap is a few kilometers longer than marked by Sneh and Weinberger (2014) in order to fit the location of the deepest bathymetry in the Sea of Galilee together with the change from subsidence to uplift along the northwest shore of the lake.
Neither a stress build-up nor slip on the 15-30-km-long Carmel Fault (Achmon & Ben-Avraham, 1997;Sadeh et al., 2012) are expected to affect the subsidence of the Sea of Galilee, because the Carmel Fault is more than 40 km away from the Sea of Galilee (Inset in Figure 10).Both induced static stresses and subsidence from fault slip have been shown to decay within one fault length (Katzman et al., 1995;King et al., 1994).Hence, the stress and slip effects of the Carmel Fault on the subsidence pattern of the Sea of Galilee are expected to be negligible.
The Korazim Plateau (Figures 9 and 10) is a raised block between the Sea of Galilee and the Hula Basin.Although probably uplifted before being covered by Neogene flood basalts, this block shows evidence for Plio-Pleistocene tectonic activity (Heimann & Ron, 1993, and references therein).Rotstein and Bartov (1989) proposed that the block had been partly elevated by a transpressive motion along the DST fault strand north of the lake, whose plane is tilted westward, whereas Heimann and Ron (1993) suggested a compressive component on the Almagor Fault (Figures 9b and 9d).The subsidence model (Figure 10) suggests that at least part of the uplift of the SE corner of the Korazim Plateau and the proposed Kefar Nahum Fault (KNF in Figure 9a) by Hurwitz et al. (2002) could be a by-product of the motion along the DST strands.

Seismic Hazard
If the shape of the Sea of Galilee and Bethsaida Valley are to a first order, the result of slip on the two main strands of the DST, then the faults proposed by previous authors and shown in Figure 8, are likely secondary faults.
We constructed a static stress model to examine which of the previously proposed faults is likely to get closer to failure because of slip on the DST fault strands.We calculated the Coulomb stress change on receiver faults (Figure 12) drawn along simplified traces of some of the published faults shown in Figure 9.The modeled faults are vertical and extend between depths of 1-10 km, except for the fault marked Bar, which extends to 15 km depth, because it runs through Haddad et al. ( 2020) cluster of relocated earthquakes, some of which were relocated at depths of 10-15 km.A positive Coulomb stress change, implying an increasing potential for failure, is concentrated in the northern half of the lake for both normal and left-lateral slip (Figures 12a and 12b).Most of the relocated seismic activity in the area in since 1985 had indeed taken place in the northern half of the lake (Figure 12c; Barnea Cohen et al., 2022;Haddad et al., 2020;Wetzler et al., 2019).The region of increased rupture potential, modeled by the Coulomb stress change, is relatively small (10 × 15 km for left-lateral slip and 12 × 12 km for normal slip) limiting the potential magnitude of secondary earthquakes activated by slip on the main DST fault strands.Only parts of the proposed WMF (Hurwitz et al., 2002) and GAS (Gasperini et al., 2020) faults, which cross the entire Sea of Galilee diagonally, show positive Coulomb stress.The rest of these fault traces show negative stress suggesting that these faults probably do not accommodate through-going ruptures (Figures 12a and 12b).

Temporal Changes in Fault Locations and Slip
A discrepancy in the estimates of the fault geometry is noted between the different methods reported in this paper.
The modeled seismic and gravity profile suggest that the main fault cutting through the sediments and basement is on the east side of the basin (Figure 5).Static stress modeling indicates, however, that the southern strand of the DST does not extend northward to the latitude of the seismic and gravity profile (Figure 10).Dembo et al. (2021) also suggested the presence of underlapping stepover from modeling paleo-magnetic vertical rotations from Pliocene age basalt samples.Their stepover is located 4-6 km south of stepovers suggested by our model and their northern stand extend southward into the lake (orange triangle in Figure 9).There is no evidence, however, for faulting in lines K-02 and K-03 for Dembo's et al. (2021) location of the stepover (Figure 11).
We propose that the discrepancy between the modeled subsidence and Coulomb stress change discussed here (and perhaps Dembo et al., 2021 geometry) may reflect a relatively recent change in fault geometry.The deeper sediment and crustal boundary observed on the seismic and gravity models, may have formed at an earlier stage of the plate boundary activity.Dembo et al. (2021) geometry may also represent an older geometry, given their reliance on the analysis of Pliocene basaltic samples.Several authors have suggested changes in fault geometry in northern Israel within the last 1 Ma (Heimann & Ron, 1993;Hurwitz et al., 1999;Schattner & Weinberger, 2008;Weinberger et al., 2009).Specifically, Hurwitz et al. (1999) documented minor (<1.3%)NNE-SSW extension in Yehudiyya Block (Figure 9), northwest of Bethsaida Valley in a direction similar to the one predicted by our subsidence model (Figure 10) Hurwitz et al. (1999) estimated that the extension started approximately 1 Myr ago and was associated with a major subsidence event farther south in the Sea of Galilee.Changes in fault geometry may possibly respond to minor temporal changes in relative plate motion and the short distance to the pole of rotation (Joffe & Garfunkel, 1987).

Conclusions
Modeling of a seismic refraction and a coincident gravity anomaly profile across the central DST and the continental margin in northern Israel, augmented by static stress modeling in the vicinity of the Sea of Galilee, allows us to draw conclusions about the deformation of this continental transform plate boundary and the surrounding region in northern Israel, a previously less explored area.Similarities and differences between this area and the well-studied area of the DST 150-200 km to the south help elucidate several aspects of this deformation.
The DST separates two sedimentary sections of different thickness, velocity, and density.Sedimentary thickness west of the DST is ∼10 km and the thickness increases approaching the DST.This large thickness is likely the result of a large accommodation space afforded by the underlying thin crust of the continental margin.Sediment thickness east of the DST is ∼7 km and is larger than farther south along the eastern side of the DST in Jordan.These sediments have accumulated during different phases of extension and compression during the Mesozoic and Cenozoic and overlie ∼2,000 m of Paleozoic sediments (Meiler et al., 2011).The sedimentary thickness on the east side also increases toward the DST.
Sedimentary layers under the western side of the DST are arched upward by up to 0.4-0.45km with respect to the coastal plain.The uplift shape is monoclinal and the layers plunge steeply toward the DST.East of the DST, the sedimentary layers also dip toward the DST over a distance of 15 km.The dipping sedimentary section under our profile east of the DST is contrasted with the observed upward flexed sediment and upper crustal uplift along the eastern side of the DST south of Lat.∼32.5°.The reason for the difference in layer geometry east of the DST is unclear.The uplift farther south was previously suggested to arise from thinner and hotter Arabian lithosphere, whereas a normal lithospheric thickness may underly the eastern part of our profile.
The transition from the continental margin to the Levant Basin is shorter in our profile than farther south along the margin, and the outermost shelf encloses an inner shelf basin.The steep transition to the Levant Basin and the inner shelf basin is probably controlled by the NW continuation of the Carmel Fault offshore which transects our profile at the shelf edge.
A gradual step in Moho depth starting from beneath the DST best fits the gravity profile.The Moho step is more gradual than farther south, perhaps because of additional crustal heat by Late Miocene and younger volcanic activity in the area.As in profiles crossing the DST farther south, the Moho east of the DST under our profile is flat, indicating that crustal thinning starts at the DST and that the DST developed along the boundary between the plate interior and its Triassic-Jurassic continental margin.
The shapes of the Sea of Galilee and Bethsaida Valley are modeled here as an asymmetric pull-apart basin formed by a left-lateral stepover of 2.6 km between slightly divergent two strike-slip fault strands dipping 70° to the west.The fault strands are underlapped by 9 km in the northern Sea of Galilee and published seismic reflection data do not show them to be connected.Whether the gap between the fault strands can arrest a future rupture propagation from one strand to the other has not been dynamically modeled.
Several additional faults within the Sea of Galilee have previously been suggested to carry some or all of the relative plate motion.Assuming slip on the main DST strands and using Coulomb stress models, we show that static stress increase is limited to the sections of these faults, that are located within the northern part of the lake Acquisition and processing parameters are given in Hurwitz et al. (2002).Red line corresponds to Reflector Kin-4, tentatively dated at ∼1 Ma (Hurwitz et al., 2002) and reflector TCB, interpreted as Top Cover Basalt (Reznikov et al., 2004) at ∼1.8 Ma (Wald et al., 2019).TWT-Two-way travel time.
motion.The spatial distribution of recent seismicity in the region is indeed centered within the area of predicted increased static stress indicating that these faults are probably secondary to the main DST fault strands.
According to the seismic and gravity models, the DST main fault appears to be on the east side of the basin.Coulomb stress models suggest however, that the southern fault strand currently terminates 8 km south of the latitude of the models.The difference in the assumed location of the fault on the east side of the DST valley between the geophysical and stress models reflects perhaps changes in fault geometry over the life of the DST.2020) earthquake relocation depths at 10-15 km.Alm-Almagor Fault (Hurwitz et al., 2002).Bar-one of the parallel faults of Barnea Cohen et al. (2022).DST-Dead Sea transform fault strands.Gas-Simplified diagonal track of the fault system of Gasperini et al. (2020).KNF-Kfar Nahum Fault (Hurwitz et al., 2002).Rez-one of the parallel faults of Reznikov et al. (2004).WMF-Western Marginal Fault (Hurwitz et al., 2002).Background is similar to Figures 9 and 10 and Reznikov et al. (2004) and are available by request from the Geophysical Institute of Israel national archive.Background topography and bathymetry in maps were created using ArcGIS ® software by Esri from Esri: https://doc.arcgis.com/en/arcgis-online/reference/static-maps.htm.Data sources are Airbus, USGS, NGA, NASA, NOAA, CGIAR, GEBCO, NLS, OS, NMA, Geodatastyrelsen, GSA, SSI, and the GIS User Community.Map image is the intellectual property of Esri and is used herein under license.Copyright © <2020> Esri and its licensors.All rights reserved.GM-SYS (R) Gravity/Magnetic Modeling Software User's Guide Version 4.9.Northwest Geophysical Associates, Inc. 1600 SW Western Boulevard, Suite 200.Corvallis, OR 97333 USA.Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.The following organizations and individuals were instrumental in the data acquisition: IRIS-PASSCAL loaned the Texan receivers and provided a technician to guide the data collection.Steven Harder, University of Texas at El Paso, provided the shot boxes and helped guide data collection in the field.Gideon Tibor and Gideon Amit of the Israel Limnological and Oceanographic Institute organized the deployment of modified Texan receivers in the Sea of Galilee.Lloyd Carothers (IRIS), Nathan Miller (USGS), Guy Lang and other University of Haifa students helped with the field operation.Geophysical Institute of Israel surveyor team and field crew carried out the field work.Field work was funded by the Richard Lounsbery Foundation grant to the University of Haifa and by the Israel Ministry of National Infrastructures, contract 217-17-017 to the University of Haifa.We thank Gou Fujie (JAMSTEC) for sharing his software and guiding us in its use.Amit Segev and Ran Calvo from the Geological Survey of Israel extracted and provided an elevation profile of the top Judea Group along our seismic profile.Reviews by Guy Lang, Nathan Miller, Cenk Yaltirak, Cengiz Zabcı, and anonymous reviewers are gratefully acknowledged.

Figure 1 .
Figure 1.Location map with elements mentioned in the text and regional hill-shaded topography from Esri.

Figure 2 .
Figure 2. Topographic and structural map of the seismic experiment in northern Israel.See Figure 1 for location.Hill-shaded topography is from Esri.Red dots are receivers.Numbered black dots are underground shots.Faults (in yellow, dashed where uncertain) are modified from Sneh and Weinberger (2014).Area below mean-sea level is in blue.BV-Bethsaida Valley.SAF-Sheikh Ali Fault.The two main fault strands of the Dead Sea Transform are marked as Dead Sea Fault.

Figure 3 .
Figure 3. (a) Starting velocity model for first-arrival tomography.See Table S2 in Supporting Information S1 for values.Triangles are shot locations and the upper layers are cut to the topography.Only the top 7 km are shown.(b) Final velocity model after 8 iterations.(c) Ray-path distribution of the first arrivals plotted on the final velocity model.DST-Main fault strand of the Dead Sea Transform.

Figure 4 .
Figure 4. Starting model for forward ray tracing using velocity contours extracted from the first-arrival refraction tomography results.Velocity contours are 4.0, 4.5, 5.0, 5.5, and 5.9 km/s.

Figure 5 .
Figure 5. (a) Forward model using refraction, reflection, and head-wave travel-time picks.Blue lines are reflection interfaces in the model, White numbers are p-wave velocities in km/s.Red triangles are shot locations.Thin gray lines-Ray coverage through layer 3 in the model.(b) Geological (in color, from Sneh et al. (1998)) and fault (black lines, from Sneh and Weinberger (2014)) map centered along the seismic refraction profile.Purple colors on geological map are Neogene basalt.Green and yellow colors are Mesozoic and Early Cenozoic sedimentary rocks, mostly carbonates.Black and red dots are shot and receiver locations, respectively.(c) Zoomed view of the upper part of the model outlined in the dashed green rectangle in (a).Velocity scale is similar to (a).Black line is the elevation of Late Cretaceous top Judea Group along profile X-X' in B, extracted from the regional top Judea map of Segev et al. (2014).DST-Main strand of the Dead Sea Transform.H-Hazon-1 well.N-Ness-6 well.Thin vertical dotted lines are the interpreted edges of the basin under the DST and their projection on the geological map.

Figure 6 .
Figure6.(a) Blue dots and lines are the locations of gravity measurements from the Geophysical Institute of Israel gravity database.Dark green dots are gravity values from a regional data set(Rybakov et al., 1997).Yellow plus signs are locations of control wells discussed in the text.Red line is the location of modeled gravity line in (b) Light green line and red triangles are locations of seismic receivers and shots, respectively.(b) Observed (black dots and calculated (blue line), free air gravity anomaly and the misfit between the two (red line).(c) 2-D gravity model along the red line in A shown to a depth of 15 km.Mediterranean coastline is at zero distance.Blue dashed lines are interfaces from gridded interpretation of offshore multichannel seismic data(Gardosh et al., 2008).Blue dash-dot lines are reflection boundaries from our seismic model in Figure5.Layers densities (in 10 −3 kg/m 3 ) follow Table1.Red triangles are shot locations.DST-Main fault strand of the Dead Sea Transform.SAF-Sheikh Ali Fault.White line is the approximate depth of crystalline basement fromBrew et al. (2001).(d) A full-crustal view of the model in (c) Symbols are similar to (c) Additional symbols are dashed green line, which is the approximate depth to Moho in the Levantine basin(Netzeband et al., 2006), and white lines and square bracket denoting depths to crystalline basement, base upper crust, and Moho fromBrew et al. (2001).

Figure 7 .
Figure 7. Sensitivity tests of the gravity model exploring the width of Moho step between depths of 37 and 32.5 km.(a) Different widths of the Moho step (thin blue lines) all starting under the Dead Sea Transform (DST) overlain on the best-fit gravity model (Figure 6d).Heavy dashed blue line is the model with the lowest root-mean-square (RMS) misfit in (c) Black line-Moho geometry from linear inversion.Layers densities (in 10 −3 kg/m 3 ) follow Table 1.(b) Different widths of the Moho step (thin red lines) centered on the DST with the step widening eastward under the Arabian plate, overlain on the bestfit gravity model (Figure 6d).Heavy dashed red line is the model with the lowest RMS misfit in (c) Black line-Same as in (a) (c) RMS misfit between the calculated and observed gravity for test configuration (a) and (b).Inv.-RMS and step width derived from the inversion.
and Inati et al. (2016) proposed, based on joint modeling of regional gravity, geoid, and heat flow, a 120-km thick lithosphere under the Golan Heights at Lat. 33°N.Based onSobolev et al. (2005) model, a normal thickness lithosphere under our profile is perhaps the reason for the lack of shoulder uplift.As an aside, the Neogene volcanism in the Golan Heights and surrounding areas, does not have to reflect the existence of a thinner lithosphere(Regenauer-Lieb et al., 2015).

Figure 8 .
Figure 8. 1-7-Simplified seismic refraction and gravity profiles crossing the DST and the continental margin and their source references (in parentheses).Horizons traced are the seafloor (blue), basement (green), base upper crust (black), and Moho (red).Brown dots-Earthquakes within ±10 km of each profile projected on the profile.Map shows locations of profiles (solid lines-seismics, dashed lines-gravity) and earthquake epicenters (circles).Earthquake epicenters are from the catalog of the Israel seismic network, accessed 04/25/2022.

Figure 9 .
Figure 9. Various fault interpretations in the Sea of Galilee by (a) Hurwitz et al. (2002), (b) Reznikov et al. (2004), Ben-Avraham et al. (1996), and the Geological Survey of Isreal (GSI) fault map (Sneh & Weinberger, 2014) (c) Gasperini et al. (2020) and (d) Barnea Cohen et al. (2022) and the GSI fault map.WMF and EMF-Western and Eastern Marginal Fault respectively AF-Almagor Fault.KNF-Kfar Nahum Fault.Orange triangles are the interpreted edges of north and south strands of the main DST fault from Dembo et al. (2021).Major and minor faults in the GSI fault map are marked by thick and thin lines respectively.Numbered dashed lines are locations of seismic reflection profiles in Figure 11.White and red dots are shot and receiver locations along the seismic refraction line, respectively.Heavy lines crossing the refraction line denote approximate edges of the basin from the seismic refraction and gravity models.Thin blue line is the Sea of Galilee's coastline at −212 m below sea level.Hillshaded topography is from Esri.

Figure 10 .
Figure 10.Contours of calculated vertical motion (in m) from a half-space elastic model due to a 2-m left-lateral motion on the DST fault strands (red lines).Heavy black lines are mapped DST fault strands from Sneh and Weinberger (2014).Thin dashed lines-Locations of seismic lines K-02-K-05 shown in Figure 11.Dash-dotted line-Approximate boundary of SW-NE extension in Yehudiyya Block (YB)(Hurwitz et al., 1999).KNF-Kfar Nahum Fault.Blue contours are shoreline and bathymetry of the Sea of Galilee.Background is shaded relief topography.See text for model details.

Figure 11 .
Figure 11.Seismic profiles in the northern Sea of Galilee.Profiles are positioned along a common north-south axis.See Figures 9 and 10 for locations.Acquisition and processing parameters are given inHurwitz et al. (2002).Red line corresponds to Reflector Kin-4, tentatively dated at ∼1 Ma(Hurwitz et al., 2002) and reflector TCB, interpreted as Top Cover Basalt(Reznikov et al., 2004) at ∼1.8 Ma(Wald et al., 2019).TWT-Two-way travel time.

Figure 12 .
Figure 12.Coulomb stress change for (a) left-lateral slip and (b) normal slip resolved on fault traces simplified from colored faults in Figure 9. Modeled fault traces are vertical and extend between depths of 1 and 10 km, except for the fault marked Bar, which extends to 15 km depth, following Haddad et al. (2020) earthquake relocation depths at 10-15 km.Alm-Almagor Fault(Hurwitz et al., 2002).Bar-one of the parallel faults of BarneaCohen et al. (2022).DST-Dead Sea transform fault strands.Gas-Simplified diagonal track of the fault system ofGasperini et al. (2020).KNF-Kfar Nahum Fault(Hurwitz et al., 2002).Rez-one of the parallel faults ofReznikov et al. (2004).WMF-Western Marginal Fault(Hurwitz et al., 2002).Background is similar to Figures9 and 10, but map is rotated counterclockwise to facilitate 3-D view of the fault planes.Coulomb stress change is calculated from slip on the main left-lateral strike-slip strands, marked DST, shown by red lines in Figure 10 (c) color-coded polygons encompassing the areas of most of the relocated earthquakes by Wetzler et al. (2019), Haddad et al. (2020), and Barnea Cohen et al. (2022).
Figure 12.Coulomb stress change for (a) left-lateral slip and (b) normal slip resolved on fault traces simplified from colored faults in Figure 9. Modeled fault traces are vertical and extend between depths of 1 and 10 km, except for the fault marked Bar, which extends to 15 km depth, following Haddad et al. (2020) earthquake relocation depths at 10-15 km.Alm-Almagor Fault(Hurwitz et al., 2002).Bar-one of the parallel faults of BarneaCohen et al. (2022).DST-Dead Sea transform fault strands.Gas-Simplified diagonal track of the fault system ofGasperini et al. (2020).KNF-Kfar Nahum Fault(Hurwitz et al., 2002).Rez-one of the parallel faults ofReznikov et al. (2004).WMF-Western Marginal Fault(Hurwitz et al., 2002).Background is similar to Figures9 and 10, but map is rotated counterclockwise to facilitate 3-D view of the fault planes.Coulomb stress change is calculated from slip on the main left-lateral strike-slip strands, marked DST, shown by red lines in Figure 10 (c) color-coded polygons encompassing the areas of most of the relocated earthquakes by Wetzler et al. (2019), Haddad et al. (2020), and Barnea Cohen et al. (2022).

Table 1
Density Values Used in Figure6and Their Corresponding Geologic Interpretation a Gradual increase in density corresponds to increasing seismic velocity with depth.