Amotz Agnon

Long-term earthquake clustering: A 50,000-year paleoseismic record in the Dead Sea Graben

The temporal distribution of earthquakes in the Dead Sea Graben is studied through a 50,000-year paleoseismic record recovered in laminated sediments of the Late Pleistocene Lake Lisan (paleo-Dead Sea). The Lisan represents more than 10 times the 4000 years of historical earthquake records. It is the longest and most complete paleoseismic record along the Dead Sea Transform and possibly the longest continuous record on Earth. It includes unique exposures of seismite beds (earthquake-induced structures) associated with slip events on syndepositional faults. The seismites are layers consisting of mixtures of fragmented and pulverized laminae. The places where the seismites abut syndepositional faults are interpreted as evidence for their formation at the sediment-water interface during slip events on these faults. Thicker sediment accumulation above the seismites in the downthrown blocks indicates that a seismite formed at the water-sediment interface on both sides of the fault scarps. Modern analogs and the association with surface ruptures suggest that each seismite formed during a M L ≥5.5 earthquake. The 230 Th- 234 U ages of a columnar section, obtained by thermal ionization mass spectrometry, give a mean recurrence time of ∼1600 years of M L ≥5.5 earthquakes in the Dead Sea Graben. The earthquakes cluster in ∼10,000-year periods separated by quiet periods of similar length. This distribution implies that a long-term behavior of the Dead Sea Transform should be represented by a mean recurrence of at least 20,000 year record. This observation has ramifications for seismic hazard assessment based on shorter records.

Distributed damage, faulting, and friction

We present a formulation for mechanical modeling of geological processes in the seismogenic crust using damage rheology. The seismogenic layer is treated as an elastic medium where distributed damage, modifying the elastic stiffness, evolves as a function of the deformation history. The model damage rheology is based on thermodynamic principles and fundamental observations of rock deformation. The theoretical analysis leads to a kinetic equation for damage evolution having two principal coefficients. The first is a criterion for the transition between strength degradation and recovering (healing), and is related to friction. The second is a rate coefficient of damage evolution which can have different values or functional forms for positive (degradation) and negative (healing) evolution. We constrain these coefficients by fitting model predictions to laboratory data, including coefficient of friction in sawcut setting, intact strength in fracture experiments, first yielding in faulting experiments under three-dimensional strain, onset and evolution of acoustic emission, and dynamic instability. The model damage rheology accounts for many realistic features of three-dimensional deformation fields associated with an earthquake cycle. These include aseismic deformation, gradual strength degradation, development of process zones and branching faults around high-damage areas, strain localization, brittle failure, and state dependent friction. Some properties of the model damage rheology (e.g., cyclic stick-slip behavior with possible accompanying creep) are illustrated with simplified analytical results. The developments of the paper provide an internally consistent framework for simulating long histories of crustal deformation, and studying the coupled evolution of regional earthquakes and faults. This is done in a follow up work.

Late Holocene lake levels of the Dead Sea

This work presents a high-resolution lake-level record of the late Holocene Dead Sea, a hypersaline terminal lake whose drainage basin encompasses both Mediterranean and hyperarid climatic zones. The lake-level curve reflects the regional hydrologic variations in the drainage basin, which in turn represent the Levant paleoclimates. The curve is based on 46 radiocarbon ages of organic remains from well- exposed sedimentary sequences along the Dead Sea shores. These sequences record fluvial and lacustrine depositional environments. The paleolakeshores are marked by shore ridges, coarse-sand units, and aragonite crusts; in the modern Dead Sea, such features indicate the exact elevation of the shore. The late Holocene Dead Sea level fluctuated within the range of 390 to 415 m below sea level (mbsl). For most of the time the lake was below the topographic sill (402 mbsl) separating the northern and southern basins of the Dead Sea and was confined to the deep northern basin. Nevertheless, short-term rises in the late Holocene Dead Sea level caused the flooding of the shallow and flat southern basin. Highstands occurred in the second and first centuries B.C. and the fourth century A.D. during the Roman and early Byzan tine periods, respectively, in the eleventh and twelfth centuries A.D. during the Crusader period, and at the end of the nineteenth century A.D. The rises mark a significant change in the annual rainfall in the region, which likely exceeded the instrumentally measured modern average. The curve also indicates drastic drops that exposed the sedimentary sequences to erosion. The oldest and probably deepest drop in the lake level culminated during the fifteenth and fourteenth centuries B.C. after a retreat from a higher lake stand. The longest lowstand occurred after the Byzantine period and continued at least until the ninth century A.D. This arid period coincided with the invasion of Moslem-Arab tribes into the area during the seventh century A.D. The dramatic fall of the Dead Sea level during the twentieth century is primarily artificial and has been caused by the diversion of runoff water for the drainage basin, but the magnitude is not considered exceptional for the late Holocene. Although the past drops in the lake never exceeded the modern artificial drop rates, they do represent extreme arid conditions that occurred frequently over the past several thousand years.

Prehistoric earthquake deformations near Masada, Dead Sea graben

Earthquake-inducedfluidizations and suspensions of lake sediments, associated with syndepositionalfaults,formapaleoseismicrecordintheDeadSeagraben.Theassociation offluidized beds with surface faulting supports the recognition of mixed layers as reliable earthquake indicators and provides a tool for the study of very long term (>70 kar) seismicity along the Dead Sea transform. The faults compose a fault zone that offsets laminated sediments of the late Pleistocene Lake Lisan. They exhibit displacements of as much as 2 m. Layers of massive mixtures of laminated fragments are interpreted as disturbed beds, each formed by an earthquake. The undisturbed laminated layers between these mixed layers represent the interseismic interval. A typical vertical slip of about 0.5 m per event is separated by several hundred years of quiescence. The fault zone lies within the Dead Sea graben, 2 km east of Masada, where archaeology and historical accounts indicate repeated strong earthquake damage. The distribution of strikes in the fault zone resembles that of the faults exposed in and around the graben, including the seismogenic ones. The excellent exposures over hundreds of metres allow an unprecedented temporal and spatial resolution of slip events on faults.

Earthquake cycle, fault zones, and seismicity patterns in a rheologically layered lithosphere

We study the coupled evolution of earthquakes and faults in a model consisting of a seismogenic upper crust governed by damage rheology over a viscoelastic substrate. The damage rheology has two types of functional coefficients: (1) a “generalized internal friction” separating states associated with material degradation and healing and (2) damage rate coefficients for positive (degradation) and negative (healing) changes. The evolving damage modifies the effective elastic properties of material in the upper crust as a function of the ongoing deformation. This simulates the creation and healing of fault systems in the upper seismogenic zone. In addition to the vertically averaged thin sheet approximation we introduce a Green function for three-dimensional elastic half-space for the instantaneous component of deformation. The formulation accounts in an internally consistent manner for evolving deformation fields, evolving fault structures, aseismic energy release, and spatiotemporal seismicity patterns. These developments allow us to simulate long histories of crustal deformation and to study the simultaneous evolution of regional earthquakes and faults for various model realizations. To focus on basic features of a large strike-slip fault system, we first consider a simplified geometry of the seismogenic crust by prescribing initial conditions consisting of a narrow damage zone in an otherwise damage-free plate. For this configuration, the model generates an earthquake cycle with distinct interseismic, preseismic, coseismic, and postseismic periods. Model evolution during each period is controlled by a subset of physical properties, which may be constrained by geophysical, geodetic, rock mechanics, and seismological data. In the more generic case with a random initial damage distribution, the model generates large crustal faults and subsidiary branches with complex geometries. The simulated statistics depend on the space-time window of the observational domain. The results indicate that long healing timescale, τh, describing systems with relatively long memory, leads to the development of geometrically regular fault systems and the characteristic frequency-size earthquake distribution. Conversely, short τh (relatively short memory) leads to the development of a network of disordered fault systems and the Gutenberg-Richter earthquake statistics. For intermediate values of τh the results exhibit alternating overall switching of response from periods of intense seismic activity and the characteristic earthquake distribution to periods of low seismic activity and Gutenberg-Richter statistics.

Crusader castle torn apart by earthquake at dawn, 20 May 1202

The Crusader castle of Vadum Jacob, an outpost overlooking the Jordan River, was deformed during a destructive earthquake triggered by motion along the Dead Sea Transform. The M >7 earthquake occurred at dawn, 20 May 1202, and offset the castle walls by 1.6 m. This exceptional precision in dating and estimating displacement was achieved by combining accounts from primary historical sources, by excavating the Dead Sea Transform where it bisects the castle, and by dating faulted archaeological strata. The earthquakes of October 1759 and/or January 1837 may account for the remaining 0.5 m out of a total 2.1 m of offset. Our study exploits the potential embodied in interdisciplinary historical-archaeological-geological research and illustrates how detailed histories of seismogenic faults can be reconstructed.

Self-driven mode switching of earthquake activity on a fault system

Theoretical results based on two different modeling approaches indicate that the seismic response of a fault system to steady tectonic loading can exhibit persisting fluctuations in the form of self-driven switching of the response back and forth between two distinct modes of activity. The first mode is associated with clusters of intense seismic activity including the largest possible earthquakes in the system and frequency‐size event statistics compatible with the characteristic earthquake distribution. The second mode is characterized by relatively low moment release consisting only of small and intermediate size earthquakes and frequency‐size event statistics following a truncated power law. The average duration of each activity mode scales with the time interval of a large earthquake cycle in the system. The results are compatible with various long geologic, paleoseismic, and historical records. The mode switching phenomenon may also exist in responses of other systems with many degrees of freedom and nonlinear dynamics.

817-Year-old walls offset sinistrally 2.1 m by the Dead Sea transform, Israel

Abstract Archeological excavations in the Crusader Ateret Fortress near the Jordan River exposed E-W trending walls displaced sinistrally up to 2.1 m by the Dead Sea transform fault. A water duct, probably of Crusader age, is also offset sinistrally across the fault by about 1–2 m, but newer water ducts parallel to the former one show no displacement. The maximum width of the fault zone is about 10 m. Post-Crusader structures show significantly less deformation, and together with the low seismic activity, suggest there has been negligible creep. It is therefore conceivable that in this particular fault segment, stress is occasionally relieved by strong destructive earthquakes associated with surface ruptures. Historical accounts include descriptions of post-Crusader earthquakes in the northern part of Israel in A.D. 1202, 1546, 1759, and 1837. These events caused destruction and casualties over large areas. We conclude that most of the displacement of the Ateret Fortress walls occurred during one of these strong earthquakes, probably that of 1202 A.D., and some additional offset occurred during subsequent events. The associated magnitude is estimated at 6.5–7.1. The Ateret site is extremely valuable for paleoseismic studies in general, and assessment of seismic hazard to nearby population centers in particular, as there is an abundance of well-dated man-made structures and a small number of candidate earthquakes.

The impact of brine-rock interaction during marine evaporite formation on the isotopic Sr record in the oceans: Evidence from Mt. Sedom, Israel

The effect of brine-rock interaction on the composition of strontium in evaporitic basins and its impact on the 87Sr/86Sr ratios in contemporaneous seawater are examined for the Sedom (Dead Sea Rift Valley, or DSR), the Messinian (Mediterranean) and the Louann (Gulf of Mexico) evaporites. For that purpose, mineralogical, chemical and isotopic (Sr, S) analyses were performed on the Sedom Fm. evaporites (halite, anhydrite and dolomite). 87Sr/86Sr ratios are distinctively lower in the Sedom evaporites (dolomites: 0.7082–0.7083; halites: 0.7083–0.7087) than in the contemporaneous late Pliocene seawater (≈0.709). At the same time the sulfur isotope ratios (δ34S ≈ 20‰) are consistent with deposition from late Cenozoic seawater. This duality, together with the variation of strontium isotopes between the dolomites and halites can be explained by modification of the 87Sr/86Sr ratio in the lagoon water by influx of Ca-Chloride brines. The brines were formed by dolomitization of marine carbonates of the DSR Cretaceous wall rocks (where 87Sr/86Sr ∼ 0.7077). Brine-rock interaction can similarly explain the anomalous 87Sr/86Sr ratios in the Messinian and Louann evaporites. It is concluded that this process causes significant changes in the 87Sr/86Sr ratios of evaporitic lagoons. A water and strontium mass balance of the Sedom data is used to show the impact on the strontium oceanic budget. Extrapolation to larger evaporitic basins indicates that the combined global riverine and hydrothermal influx of strontium can be matched by halite or gypsum precipitating lagoon of 2–3.5 × 105 km2. Examples for such evaporitic sites include the Messinian, Louann and Zechstein basins.

Mechanics of oblique spreading and ridge segmentation

Abstract Mid-ocean ridges display a variety of plan view geometries (planforms) that correlate with the tectonic setting. A mechanical analysis is proposed to rationalize the variety of planforms of mid-ocean ridges at various tectonic settings. We model spreading centers as fluid-filled cracks, and find the variation of segment orientation withΔP/ΔS (whereΔP measures a magmatic overpressure within the crack andΔS is the remote ‘tectonic tension’). The analogy suggests that highΔP/ΔS tends to preserve the continuity of an oblique spreading axis, whereas lowΔP/ΔS prompts segmentation. It follows that a planform of the spreading center is an indicator for the forces driving melt injection. The results are in qualitative agreement with principal geological observations. For instance, the concordant, oblique, and continuous Reykjanes Ridge reflects pressurized magma emplacement (ΔP/ΔS> 20), a result compatible with the proximity to the Icelandic hot spot and with lowΔS anticipated in passive margins. Away from major hot spots, the Mid-Atlantic Ridge (MAR) with its passive margins typically has a value ofΔP/ΔS of around unity. The values change widely through space and time, locally becoming negative (amagmatic spreading). This spatial and temporal variability in MAR is consistent with dynamic melt injection. Conversely, in orthogonal-discordant-segmented axes of the Pacific, magma emplacement is dominated by slab pull (highΔS) despite high magmatic activity. The inferred value ofΔP ≈ 0 in the Pacific, stable through space and time, is consistent with passive melt injection.