Journal of Seismology 5: 449–474, 2001. © 2001 Kluwer Academic Publishers. Printed in the Netherlands.

449

Late Pleistocene and Holocene slip rate of the Northern Wadi Araba fault, Dead Sea Transform, Jordan Tina M. Niemi1, Hongwei Zhang1 , Mohammad Atallah2 & J. Bruce J. Harrison3 1 Geosciences Department, University

of Missouri – Kansas City, Kansas City, MO 64110, U.S.A.; 2 Department of Earth and Environmental Sciences, Yarmouk University, Irbid, Jordan; 3 Department of Earth and Environmental Science, New Mexico Institute of Mining and Technology, Socorro, NM 87801, U.S.A.

Received 30 November 1999; accepted in revised form 28 October 2000

Key words: alluvial fans, Dead Sea transform, earthquake recurrence, Holocene, Jordan, late Pleistocene, slip rate, Wadi Araba fault Abstract The Wadi Araba Valley is a morphotectonic depression along part of the Dead Sea Transform (DST) plate boundary that separates the Arabian plate on the east from the Sinai subplate on the west. The Wadi Araba fault (WAF) is one of the main strike-slip faults between the Gulf of Aqaba and the E-W trending Khunayzira (Amatzayahu) fault that bounds the southern end of the Dead Sea. Just south of the Dead Sea, the WAF cuts across several generations of alluvial fans that formed on tributaries to the Wadi Dahal after the regression of Late Pleistocene Lake Lisan ca. 15 ka. Geomorphic and stratigraphic evidence of active faulting, including left-laterally offset stream channels and alluvial-fan surfaces, yielded fault slip-rate data for the northern segment of WAF. Typical cumulative displacements of 54 m, 39 m, and 22.5 m of stream channels and alluvial-fan surfaces across the fault were measured from detailed geologic and topographic mapping. The 54 m offset of the oldest alluvial-fan surface (Q f 1 ) occurred after the final lowering of Lake Lisan (16–15 ka) and before 11 ka yielding a slip-rate range of 3.4 mm/yr to 4.9 mm/yr. Based on radiocarbon ages of charcoal and landsnail shell samples from the buried Qf 2 alluvial-fan deposits exposed in trenches excavated across the fault, the 39 m and 22.5 m offsets occurred after 9 ka and 5.8 ka, respectively. These data yield a slip-rate range between 3.9 mm/yr and 6.0 mm/yr. The small variability in these slip-rate estimates for different time periods suggests that the northern Wadi Araba fault has maintained a relatively constant slip rate in the past 15 ka. We calculate an average slip rate of 4.7 ± 1.3 mm/yr since 15 ka based on the three separate displacements and age estimates. Five separate offsets of 3 m were measured from gully bends and the offset of small fault-scarp alluvial fans. These displacement data suggest a coseismic slip of 3 m in the last earthquake, or a cumulative slip of 3 m in the past few earthquakes. A maximum slip of 3 m correspond to a Mw 7 earthquake that ruptures about 49 km of fault length. Using an average slip rate of 4.7 ± 1.3 mm/yr together with a 3-m slip-per-event suggests a maximum earthquake recurrence interval of this fault segment of 500 to 885 years.

Introduction The Dead Sea Transform (DST) fault system is the plate boundary fault between the Arabian plate on the east and the Sinai subplate on the western (Figure 1A). The transform fault zone strikes N15◦E and links the Taurus-Zagros Collision Zone in the north with the Red Sea spreading center in the south. According to Courtillot et al. (1987), the extension direction in the

Red Sea is almost perpendicular to its axis, while leftlateral, strike-slip motion is occurring on the DST. In addition, a counterclockwise rotation of 5.5◦ to 6◦ of Arabia with respect to Africa is estimated (Quennell, 1959; Freund et al., 1970; Joffe and Garfunkel, 1987). Cumulative post-Eocene displacement along the DST is approximately 107 km (e.g., Quennell, 1959; Freund, 1965; Freund et al., 1970; Garfunkel et al., 1981; Garfunkel and Ben-Avraham, 1996). Palinspastic re-

450

Figure 1. (A) The Wadi Araba Valley lies along a portion of the Dead Sea Transform fault (DST) which is the plate boundary that separates the Sinai plate on the west from the Arabia plate to the east. (B) Generalized geologic map of Wadi Araba Valley showing location of selected major active faults and juxtaposition of different bedrock lithologies across the valley. The study area, marked by a box, is located along the Northern Wadi Araba fault near the Khunayzira fault escarpment that bounds the southern Dead Sea Ghor. Map modified after Bender (1974).

451 Table 1. Geologic estimates of lateral slip rates on the Dead Sea transform fault system Fault segment

Evidence

Amount of offset

Age of datum

Slip rate

Authors

DST fault system

offset Miocene rocks

40–45 km

3.5–6 mm/yr 4–6 mm/yr 9–15 mm/yr 7–10 mm/yr

Freund et al., 1968 Freund et al., 1970 Steinitz et al., 1978 Garfunkel et al., 1981

1.5–3.5 mm/yr (seismic slip rate) 5–10 mm/yr

Garfunkel et al., 1981 Garfunkel et al., 1981

Yammuneh fault South-Central Jordan fault Central Wadi Araba fault DST fault (Hula area) Southern Wadi Araba fault Arava fault

Northern Wadi Araba fault Northern Wadi Araba fault Northern Wadi Araba fault

offset Homs basalts

8 km

offset Litani river Methodo offset channels

5 km 100–150 m

7–12 Ma 7–10 Ma 3.1–3.7 Ma PliocenePleistocene (4–5 Ma) Last 1,000– 1,500 years MiocenePliocene 1–2 Ma post-Lisan

offset alluvial fans

3 km

0.3–0.6 Ma

5–10 mm/yr

offset of E-W trending walls in the Crusader Ateret Fortress offset alluvial fans

2.1 mm

817 years

∼ 2.5 mm/yr

150 m

20–23 ka

7.5 mm/yr

offsets in drainage basins and alluvial fans slumps in Lisan deposits offset of the late Pleistocene Dahal alluvial fan offset gullies and fan surfaces on tributaries to Wadi Dahal

15 km

Late Pliocene or early Pleistocene

3–7.5 mm/yr

500 m

77–140 ka

54 m 39 m 22.5 m

16–11 ka 9–6.5 ka 5.8 ka

construction of offset geological formations and structural features show that lateral displacement on the DST occurred in two steps: 62 km of offset by early Miocene, and 45 km of offset since the late Miocene or Pliocene to the present (Quennell, 1959; Freund, 1965; Freund et al., 1970). The seismic, historical, and archaeological records of the DST region show that past strong earthquakes (M = 5.5) have caused widespread destruction and casualties (e.g. Ben-Menahem, 1979, 1991; Ben-Menahem and Aboodi, 1981; Poirier and Taher, 1980; Russel, 1985; Amiran et al., 1994; Ambraseys et al., 1994; Ambraseys and Jackson, 1998; Guidoboni, 1994). However, the recent

5–10 mm/yr 10 mm/yr

6.4 mm/yr (seismic slip rate) 2–6 mm/yr preferred rate of 4 mm/yr 3.4–4.9 mm/yr 4.3–6.0 mm/yr 3.9 mm/yr avg. 4.7±1.3 mm/yr

Garfunkel et al., 1981 Garfunkel et al., 1981 Marco et al., 1997 Zak and Freund, 1966 Ginat et al., 1998 El-Isa and Mustafa, 1986 Klinger et al., 2000a Niemi et al., this paper

seismic activity has been relatively low, especially in the southern segment of the DST where the Wadi Araba fault is the main strike-slip fault (Shapira, 1981). This may indicate that the recurrence of very strong earthquakes generated on the Wadi Araba fault is very long, or the paucity of earthquakes may be due to the aseismic creep along the Wadi Araba fault. Determination of earthquake recurrence intervals is a major factor of earthquake-hazard assessments, and fault slip rates may be the single most important parameter for understanding earthquake hazards along a fault (Allen, 1986). Fault slip rates on the DST proposed by previous studies are variable (Table 1). Because there

452 are few well-constrained slip rate estimates for the Wadi Araba fault, large uncertainties remain about the earthquake-hazard assessment of this segment of the DST. This research focuses on documenting both geomorphic and stratigraphic evidence of recent faulting on the northern segment of the Wadi Araba fault, south of the Dead Sea, Jordan in order to assess seismic hazard along this fault (Figure 1B). We define the northern Wadi Araba fault (NWAF) as the DST fault segment north of the structural high in the Araba Valley at ArRisha (Figure 1B). At our research site immediately south of the junction between the Khunayzira (Amatzyahu) fault and the Wadi Araba fault, the northern Wadi Araba fault has clear geomorphic expression of active tectonics that includes offset stream channels, offset alluvial-fan surfaces, shutter ridges, pressure ridges, and sag depressions. The main purpose of this study is to determine the amount of displacement of alluvial-fan surfaces and stream channels and estimate the age of the offsets in order to calculate fault slip rates and infer their relationship to earthquake recurrence.

Previous studies Few publications relate directly to the neotectonic and active tectonic behavior of the DST in the Wadi Araba Valley. Several small-scale geologic maps show the surface trace of the Wadi Araba fault. Blake (1939) produced the first geological map that includes the study area at a 1:1,000,000 scale. The Wadi Araba fault was mapped as part of the Jordanian Natural Resources Authority 1:50,000-scale mapping project (e.g., Ibrahim, 1991, 1993; Rashsan, 1988; Tarawneh, 1992); the location of the Wadi Araba fault is mostly inferred or is based on a photolineaments. Other researchers including Shaw (1947), Quennell (1956), Wetzel and Morton (1959), Zak and Freund (1966), Garfunkel (1970), Garfunkel et al. (1981), Bowman (1995), and Klinger (1999) mapped faults that deform Quaternary and Late Pliocene age sediments in the Wadi Araba Valley. Geological studies defining surface rupture segments, recurrence of ground-rupturing earthquakes, and Holocene slip along the active fault in Wadi Araba are generally lacking. Prior to the present study, subsurface investigations of active faulting in Wadi Araba Valley (Arava Valley) were limited to the study of normal faults near Elat, Israel. Paleoseismic studies at two sites in Is-

rael have provided data on the activity of the Southern Wadi Araba fault, which is called the Evrona fault system. These sites are the Shehoret fan site and the Evrona playa site. Gerson et al. (1993) excavated twelve trenches across fault scarps and other lineaments at the Shehoret site, which is located 7 km north of Elat. These studies identified several faulting events separated by periods of quiescence using geomorphic evidence. Infrared-stimulated luminescence dates on buried colluvial wedge deposits indicate that four surface faulting events with vertical displacements of 1.5 m occurred between 35 ka and 14 ka (Amit et al., 1995, 1996; Porat et al., 1996). Smaller, more frequent fault displacements in the Holocene deposits in the Shehoret trench stratigraphy led Leonard et al. (1998) to suggest a change in seismic activity about 14 ka, and according to Amit et al. (1995), a shift in activity toward the center of the DST. Based on fault-scarp degradation modeling, Enzel et al. (1996) estimated that at least one large (M>6.5) earthquake occurred on the normal faults of the Evrona fault system in past 2000 years, and the recurrence interval of large earthquakes is 1200–2000 years in the Holocene. Trenches have also been excavated across the fault in the Evrona playa, a narrow basin traversed diagonally by the Evrona fault system (Amit et al., 1999). Based on seismic-reflection data, the fault is interpreted as a 1-km-wide zone of en echelon fault segments that form a flower structure (Shtivelman et al., 1998). Trenched normal faults at this site (Amit et al., 1999) revealed at least six surface faulting events since 14 ka. The most recent at this site occurred in the past 1000 yrs and produced 1 m of uplift. Marco and Agnon (1995) estimated an average recurrence interval of 1400 years for strong earthquakes that caused fluidization of varved sediments by counting the interval between the seismites recorded in a section of Lisan formation located along the southwest margin of the Dead Sea in Israel. The paleoseismological record in the Lisan formation is more than 10 times longer than the historical and archaeological record. At the same research site as discussed in this paper, Bowman (1995) concluded that only dip-slip motion has occurred on the faults based on the examination of oblique air photos. Bowman noted that the fault cuts two generations of alluvial fans, offsets some gullies, and forms a 1.5-km-long scarp. He interpreted no morphologic evidence for lateral slip, but this conclusion is not supported by the present or previous studies (Zhang, 1998; Klinger, 1999; Klinger et al., 2000a).

453 Galli (1999) recorded at least 2.5 m of right-lateral offset of ruins of a Roman (?) aqueduct in Wadi Khunayzira to the north of the study site. The aqueduct, he thought, was the same age as offset ruins of Qasr Tilah to the south of the study site. He suggested the movement occurred in the past 15–16 centuries. Galli (1999) also noted the left-lateral displacement of small gullies in the study area of this paper. However, no fault offsets were determined based on these observations. Offset archaeological ruins at Qasr Tilah to the south of the study site were studied by Klinger (1999), Klinger et al. (2000b), Niemi (2000), and Niemi and Atallah (2000). The northwest corner of a water reservoir at Qasr Tilah lies across the Wadi Araba fault and is offset a minimum of 1.8 ± 0.2 m (Niemi and Atallah, 2000) or 2.2 ± 0.5 (Klinger et al., 2000b). The average of two radiocarbon ages on charcoal within the wall plaster and mortar yields a construction date for the reservoir of 1370 ± 24 yr B.P., which corresponds to a two-sigma, tree-ring calibrated, calendar date of 641–687 A.D. (Niemi and Atallah, 2000). The reservoir at the Qasr Tilah site was probably ruptured in one or more of the large historical earthquakes recorded in this region in A.D. 1068, 1202, 1212, 1293, and 1458 (Ambraseys et al., 1994). Zhang (1998) and Klinger (1999) both conducted detailed studies at this site, including mapping the alluvial fans and measuring offsets of displaced alluvial-fan surfaces and stream channels. It is noteworthy that their independent work yielded similar results on the alluvial-fan separations and fault-offset determinations. Klinger (1999) did not report fault slip rates using these data because of the absence of age control of these offset deposits. We discuss the details of our research in the following section. Study area The study area is located along the northeasternmost part of the Wadi Araba Valley, which is a 170-kmlong valley located between the Gulf of Aqaba (Elat) to the south and the Dead Sea to the north (Figure 1). The valley is bounded by normal faults that have dropped the valley floor and uplifted the flanking mountain plateaus. The width of the valley floor ranges between 10–25 km and is mainly covered by alluvium, playa deposits, and sand dunes of Miocene, Pliocene, Pleistocene and Holocene age. The study site is a 4 km by 2 km area that lies in the transitional zone between the Dahal mountains to

the east and the Wadi Araba Valley floor to the west – an ideal environment for the development of alluvial fans (Figure 2). Alluvium, lacustrine deposits, and lake terraces ranging in age from Middle Pleistocene to the Recent are found along the mountain front. The Wadi Araba Valley floor is an area of badland topography, which is formed on the easily eroded, Upper Pleistocene marls and clastic sediment of the Lisan Formation. Base-level lowering of the Wadi Araba drainage has been controlled by changes in the level of the Dead Sea and the Late Pleistocene precursor, Lake Lisan. The Dahal mountainous area is mainly the bare bedrock region east of the Wadi Araba fault. The area’s high relief ranges from 170 m below the mean sea level (MSL) to 408 m above MSL. The bedrock has been strongly dissected by headward erosion of the major drainages. Two streams cross the NWAF near the study site (Figure 2 and 3). The Wadi An Nakhbar drainage basin to the north covers an area of 7 km2 while the Wadi Tilah drainage basin to the south covers an area of 18 km2. Both drainages originate in the interior of the Dahal mountains and have wide, long, deeply incised, stable courses. When Wadi An Nakhbar intersects with the NWAF, it turns to north and follows the fault trace before joining Wadi Khunayzira (Figure 2). Wadi Tilah also flows to the Dead Sea. It incises the Lisan Formation and joins Wadi Dahal through the Wadi Araba floor, then crosses the Khunayzira fault scarp. The watershed area of the study site is only 2 km2 (Figures 2 and 3). All channels and gullies in the study site drain to the west and are tributaries to Wadi Dahal. These ephemeral streams are small and short; most drainages are less than 500 m long from the head to where the channels cross the NWAF. Because of the stream’s short length, the sediments deposited by these streams are recycled Quaternary sediments or eroded Upper Cretaceous bedrock. At the study site, bedrock clasts in the fan deposits are predominantly Upper Cretaceous limestone, gypsum, and chert, and Upper Cambrian and Lower Cretaceous sandstone. Methodology Aerial photograph interpretation: A Quaternary geologic map of the study area was constructed based on interpretation of three 1978, black-and-white, 1:10,000-scale aerial photographs obtained from the Royal Jordanian Geographic Society (Figures 2 and 3), and further modified after field mapping. Due to

Figure 2. Scanned image of the 1:10,000-scale, black-and-white aerial photograph of the study area south of the Khunayzira fault escarpment that bounds the Dead Sea Ghor. The arrows indicate the location of the linear trace of the Wadi Araba fault in the study area. Note the sinistral offset alluvial fan surfaces and deflected gullies above the region marked by an ‘O’, the sag depression (marked by ‘S’), and several well-developed, pressure ridges to the north. East of the fault are several prominent lake terraces that have been eroded in the region of deflected gullies. Alluvial fans are built out over the Upper Pleistocene lacustrine deposits of the Lisan formation visible in the foreground of the image. See Figure 3 for a map of the surficial deposits.

454

Figure 3. (A) Quaternary map of the study area based on air photo interpretations and field mapping. Sinistral motion along the Wadi Araba fault has offset alluvial and lacustrine deposits along this portion of the Dahal Mountains. Lake terrace elevations are approximations and shown as meters below sea level. The map shows the location of three paleoseismic trenches, marked as T1, T2, and T3, pressure ridge exposure (PR), and cross section A-A’. (B) Inset shows the region of detailed, 1:250-scale topographic survey and geologic mapping of offset alluvial fan surfaces and gullies. The star marks the location of a subsurface pit excavated into the abandoned g4 gully deposits.

455

456 the extremely arid conditions in the study area, the evidence for active faulting is clear. Fault-generated landforms are rarely degraded by runoff or eroded and are not masked by vegetation cover. Topographic survey: To acquire precise displacement data, a detailed 1:250-scale topographic map with 30-cm contour intervals was constructed using approximately 5,000 topographic data points collected with a Total Station. A version of this map with 40 cm contours is shown in Figure 4. Field mapping: Field observations showed that both geomorphic and stratigraphic characteristics along and near the fault zone are more complex than those interpreted from aerial photographs. Numerous small fluvial terraces and offset gullies, which are not visible on the 1:10,000-scale air photos, are present on the older alluvial-fan surfaces. Lake terraces and shoreline features located along the mountain front were divided into four mappable units. Deep incision by ephemeral wadis in the study area provided exposures of stratigraphic units, which were measured and described (Zhang, 1998). The field context of the stratigraphic units provided a relative chronology for reconstruction of the depositional history of the study area. Detailed field mapping (Figures 3B and 4) focused on documenting geomorphic and stratigraphic evidence of horizontal displacements by matching left-laterally offset alluvial-fan units and stream channels that cross the fault nearly perpendicularly. Trenching: Subsurface stratigraphy in three faultperpendicular trenches and a pre-existing exposure in a gravel borrow pit focused on documenting evidence for past earthquakes and provided a stratigraphic framework for correlating displaced subsurface stratigraphic units with offset geomorphic features. Several soil pits on geomorphic surfaces and abandoned channel fill help establish the timing of cumulative coseismic slip on the fault. Furthermore, charcoal and landsnail shells (Sphincterochila fimbriata?) from sedimentary deposits in each of the trenches and in several soil pits provide radiocarbon ages for these deposits. All radiocarbon ages have been corrected for isotope fractionation and are calibrated using the treering calibration program, Calib reversion 4.2, by M. Stuiver and P.J. Reimer (Dataset reference, Stuiver et al., 1998). Results are rounded to the nearest decade (Table 3).

Late Quaternary geomorphology The modern topography and morphological features of the study area are controlled by both tectonic processes and lake-level fluctuations. Tectonically, the Dead Sea area attained its recent structural shape during the Late Pliocene to Early Pleistocene (Garfunkel et al., 1981). Active tectonics continue to modify the geomorphology as is evident by fault scarps, offset surficial features, and other deformation. Most surficial features are less than ∼15 ka in age and are related to the lowering of Lake Lisan and Dead Sea fluctuations that occurred during the Late Pleistocene to Recent (Bowman, 1997). Thus, the last retreat of Lake Lisan plays an important role in shaping the landforms. Quaternary sediments were subdivided into two groups based on their geomorphic characteristics: (1) lake terraces sediments, and (2) alluvial-fan and fluvial deposits. We mapped four lake terraces according to their different elevations – QL1 , QL2 , QL3 , and QL4 (Figure 3). The terraces consist of well-defined treads and risers. The treads are flat, gently west-sloping surfaces covered with a gravel of well-rounded cobbles and boulders. Desert pavement and Reg soil are well developed, and the lake terrace surfaces appear dark gray and smooth on air photos. The lake terrace surfaces (QL ) are cut into both pre-Lisan and Lisan deposits (Figure 5), and are covered by a veneer of alluvium (Qof over the L1 and L2 surface, and QLf over the QL3 and QL4 terraces). The surface of QL3 contains a series of well-preserved, parallel, very narrow, closed-spaced abandoned beach ridges deposited by the last retreat of the Lake Lisan. The elevation of this terrace is approximately 160 m below MSL, which is above the often cited highest stage of Lake Lisan at 180 m below MSL (Neev and Emery, 1967, Begin et al., 1974). Our data on the elevation of this Lisan beach ridge may support the conclusion that the highest stage of Lake Lisan was ∼150 m below MSL (Bowman and Gross, 1992), or that the higher elevation is due to the post-Lisan tectonic uplift (Clark, 1988). The youngest and lowest lake terrace, QL4 , is composed of Lisan lacustrine and fan-delta deposits, which is cut-and-filled or overlapped by the alluvial fan and fluvial deposits. Alluvial-fan surfaces are separated into three age groups (from oldest to youngest, Qf 1 , Qf 2 , and Qf 3 ) based on their relative elevation and surficial characteristics, such as tonal color, erosional and depositional features, and the degree of weather-

457

Figure 4. Detailed topographic map of the study area showing areas of 1:250-scale field mapping and the location of Trenches 1 and 2. Map was generated using 5000 survey data points collected with a Total Station. Field data were contoured using a digital terrain model program in Autocad. Contours are spaced every 40 cm with bold contours are every 2 m. Map was constructed using a relative elevation datum. Actual elevation range in diagram is approximate –175 to –150 mbsl.

ing, smoothness, and amount of vegetation on the alluvial-fan surfaces. The three alluvial fans units lack beach features as noted above, but have the characteristic cone-shaped morphology typically developed by braided fluvial systems in arid climates. This morphology provides a useful field trait for distinguishing post-Lisan features from Lisan and pre-Lisan surfaces (Bowman, 1997). Younger alluvial fans are incised into or deposited on older fan deposits, and are lighter in color compared to older fans. The typical braided drainage pattern is better developed on younger fan surfaces. The two older fans (Qf 1 and Qf 2 ) were cut by active faults and have no vegetation on their surfaces, whereas the youngest fans (Qf 3 ) have abundant vegetation in channel courses, and active deposition and erosion has largely obscured evidence of active faulting. A comparison of the geomorphic and stratigraphic units of this study to previous studies is given in Table 2.

Stratigraphy A cross section showing the stratigraphic relations of geologic units (Figure 5) drawn perpendicular to the trend of the Wadi Araba fault was developed from exposed sections (Zhang, 1998) in incised wadis and from Trench 2. Upper Cretaceous limestone is overlain by Pleistocene conglomerates that trend N10◦ E, 42◦ SE. These deposits have been beveled by two lake terraces, QL2 and QL1 (not shown) and blanketed by an older alluvium (Qof ). The lake terraces probably formed during the Lake Samra cycle, which filled the Dead Sea basin around 350–63 ka (Kaufman et al., 1992). Two lower lake terraces QL3 and QL4 developed during the Late Pleistocene when Lake Lisan covered the area. Outcrops of the Lisan formation (QLL ) are predominately oolitic calcarenites. The uppermost layer of the Lisan formation is an alluvial fan unit (QLf ). Additional stratigraphic detail of the study area are gleaned from a bulldozer exposure of a pres-

458

Figure 5. Cross section showing the stratigraphic relationship of geologic units in the study area. This section drawn perpendicular to the trend of the Wadi Araba fault was developed from measured sections (Zhang, 1998) of exposures in incised wadis and from Trench 2. Upper Cretaceous limestone bedrock (K) is overlain by older deformed Pleistocene? conglomerates (Qg) and (Qc). An erosional unconformity marks the base of lake terraces QL2 and QL1 developed on the older units and lake terraces QL3 and QL4 developed over Late Pleistocene Lisan formation (QLL ). Lake terraces are covered with an alluvium veneer (Qof and QLf ). Post-Lake Lisan sedimentary units include three phases of alluvial-fan deposition (Qf 1 , Qf 2 , (not shown) and Qf 3 ). Motion on the active Wadi Araba fault formed the shutter ridge and sag depression. See Figures 6 and 7 for details of these features. The location of the line of section is given on the map in Figure 3.

sure ridge (Figure 6) and three paleoseismic trenches (T1, T2, and T3) excavated perpendicular to the fault trend (Figures 7, 8, and 9). The structure of a pressure ridge is exposed in a borrow pit that was excavated during construction of the highway in the early 1970s (Figure 6). At this site, located just north of Trench 2 (PR on Figure 3), a 5-m-thick section exposes a well-developed, positive flower structure (Figure 6). The core of the flower structure is the limestone bedrock, probably of the Upper Cretaceous Na’ur formation. We divided the bedrock in two units: a highly fractured limestone (K1 ), and a limestone breccia (K2 ). In fault contact with the bedrock is complexly deformed blocks of the Upper Pleistocene Lisan formation (QLL ). At least three different facies of Lisan formation are present in the outcrop: 1) a deep-water facies composed of thinly laminated marl and gypsum, 2) a beach facies composed of calcarenite and oolitic sandstone with interbeds of well-rounded pebbles and cobbles, and 3) a fan-delta facies composed of graded beds of coarse sand and gravel. The bedrock and deformed Lisan section are capped by an erosional unconformity marked by a layer of rounded cobbles; we interpret this contact to be the last regressional lag marking the final retreat

of Lake Lisan. The lag deposits are overlain by a 3-mthick sequence of alluvial sheet-wash sediments that represent the youngest deposits of the Lisan formation (our map unit, QLf ). A similar series of relatively thin alluvial sheets, or fan veneers, which never attained the classic fan shape was documented for latest Lisan deposition along the western margin of the Dead Sea (Bowman, 1988). On the west side of the ridge, the top-Lisan alluvial-fan deposits are in fault contact with the channel gravels and interbedded sandy silt of the oldest post-Lisan alluvial-fan deposits (Qf1 ). The structure of a pressure ridge was also exposed in Trench 2 (Figure 7). Here, two fault-bounded pressure ridges are cored by ripple, dune, and planarbedded oolitic sand interbedded with clay (QLL), and the topmost alluvial-fan deposits (QLf ) of the Lisan formation. Fault A juxtaposes QLf deposits against Qf 1 fan sediments. This western ridge acts as a shutter ridge. Alluvium filled the depression that formed as the shutter ridge disrupted flow of several small gullies (Figure 4). The alluvium, which we correlate to Qf 3 sediments, was deposited on the older fan gravels of Qf 1 . We divide the alluvial fill into five mappable units. Unit 1 is a silt interbedded with sandy gravel. Char-

459 Table 2. Correlaltion of Late Quaternary stratigraphic units in the Wadi Araba Valley Northern Wadi Araba Valley Niemi et al., Bowman, this paper 1995

Qf3 Youngest alluvial fan

Active alluvial fan

<5.8–present

Qf2 Post-Lisan Intermediate alluvial fan 9–6.5 ka Qf1 Oldest alluvial fan 16–15 ka to 13–11 ka QLL and QLf Lisan formation 16–15 ka (Age of the end of lake Lisan)

Inactive alluvial fan Lisan formation

Klinger, 1999

Q3c 4400–2010 ka (preferred 3400 ka)

Q3b 8800–5200 ka (preferred 7000 –6200 ka) Q3a 15–12 ka

Lisan formation

coal from campfires stratigraphcally in the middle of the layer provide excellent age control; unit 1 silt was deposited around 3.1–3.2 ka. Overlying unit 1 are channel gravels of unit 2 and the silt layer of unit 3. Another in situ burn in unit 3 yielded an age of 1.2 ka. Very recent gravel deposits of unit 4 mark the location of active drainages across the depression. Unit 5 is a mixture of colluvium derived from the adjacent slope and alluvium. Three radiocarbon samples (T2-S5, T2S6, T2-S7) from landsnail shells yielded older ages out of stratigraphic order indicating either redeposition of the shells, or contamination of the carbon in the shells by older carbonate. Recent displacements on faults B and C have deformed the Qf 3 fill in Trench 2. The fault in Trench 1 forms a 9-m-wide flower structure (Figure 8). The Lisan formation (QLL ) on the east side of fault consists of interbedded oolitic sand, well-rounded cobbles, and alluvial gravel, and is capped by an unknown thickness of QLf alluvial-fan deposits. The QLL /QLf contact is exposed in outcrop in an incised wadi north of the trench site (Figure 10). The Qf 2 alluvial-fan forms the surface deposits east of the fault. The contact between the QLf and the

Southern Wadi Araba Valley Nahal Shehoret Wadi Yutim Fan, Aqaba (Enzel et al., 1996) (Niemi et al., in prep.) (Amit et al., 1995) (Niemi and Smith, 1999) Qa4+5 Late Holocene and active surface Qa3 > 3 ka Middle Holocene Qa2 > 7 ka Early to Middle Holocene

Qf4 Historical fluvial, aeolian and beach sediment Qf3 2–3 ka Qf2 and Qf1 < 6 ka Chalcolithic

Qa1 ∼ 60 ka Late Pleistocene

Qf 2 fan deposits is unclear. On the west side of the fault, we have divided the sequence into three alluvialfan deposits. The boundary between the lower, old Qf 1 (or possibly QLf ) deposits and the overlying Qf 2 sediments is inferred to follow a sharp sedimentary contact. This boundary also marks a change in sediment age (See Table 3 for radiocarbon data). Radiocarbon data from shell samples from Qf 1 yielded ages of 16.5–13 ka. Samples from the overlying Qf 2 sediments have radiocarbon ages of 8.5 ka from charcoal, and 9.8 ka from shell. Once again, though positioned in stratigraphically younger sediments, the shell age is older and may have been contaminated by older carbon. The surface west of the fault developed by recent Qf 3 alluvial sediments. Channels of the Qf 3 deposits have eroded the underlying Qf 2 sediments, but a portion of the Qf 2 soil has been preserved and buried. Radiocarbon analyses of charcoal in the soil indicate that the Qf2 surface soil at this location was not buried until after 4.7 ka. Trench 3 was excavated across the fault within an active channel on the Qf 3 surface (Figure 3). The exposed deposits included cobble to boulder debris-flow

460

Figure 6. Log of a pressure ridge exposed in a gravel excavation along strands of the Wadi Araba fault (See Figure 3 for location). A well-developed, positive flower structure is exposed in the outcrop. A prominent erosional unconformity that beveled the Cretaceous bedrock, fault breccia, and Upper Pleistocene Lisan formation is marked by a layer of rounded cobbles. This erosional horizon was produced by the final regression of Lake Lisan around 15 ka. The lag deposits are overlain by a 3-m-thick sequence of alluvial sheet-wash sediments of the topmost Lisan formation (QLf ). The deformed erosional unconformity and QLf fan deposits are in fault contact with younger post-Lisan alluvial-fan sediments (Qf 1 ).

layers with cobble and gravel channel deposits (Figure 9). Three stratigraphic units were identified in this trench based on buried soil horizons. At the base of the trench, an old alluvial-fan surface was exposed, which may correlate to the Qf 1 surface, or might be a stable surface within the Qf 2 fan deposits. Above this lower soil, alluvial-fan sediments are capped by a second buried soil that we interpret as top of the Qf 2 fan. Radiocarbon ages on three landsnail shell samples above the soil yielded ages of 3.6 and 5.8 ka. These dates provide the maximum age of the onset of deposition for the youngest, Qf 3 alluvial fan that overlies the Qf 2 soil.

Age of stratigraphic units The Lisan formation (QLL ) in the study area consists of oolitic sandstone that is fairly unique for the formation. Similar oolitic deposits have only been described from one other location, the Nahal Amazyahu section in Israel, along the southwestern margin of the Lake Lisan outcrop belt (Druckman et al., 1987). We do not have direct age control on the Lake Lisan deposits at our study site, however, Druckman et al. (1987) reported a radiocarbon age of 14,600 ± 240 yr B.P. on organic plant remains collected from below the oolitic sand at the top of the Amazyahu section (elevation –190 msl). They interpreted this age to be the

461

Figure 7. Map of northern wall of Trench 2 showing multiple fault strands that bound a pressure ridge and a shutter ridge. The central zone is a depression filled with alluvium ponded by the shutter ridge on the west. Faults are identified by rotated and aligned clasts, fracture fills, fault gouge zones, and stratigraphic offsets. Locations of radiocarbon samples from the south wall are shown at their approximate stratigraphic level. See Table 3 for further information on radiocarbon data. See Figures 3, and 11 for location of trench.

time of the maximum areal extent of the Lake Lisan. Furthermore, Druckman et al. (1987) interpreted the formation of the calcite cement to indicate humid climatic conditions between 14–13 ka. We used the tree-ring calibration program of Stuiver et al. (1998) to convert this age to 15,284 ± 1002 cal yr B.P. (two sigma, Method B), and have assigned the median age of this range (15 ka) to the final lacustrine phase of the Lisan formation in our study area. Begin et al. (1985) developed a curve of levels of Lake Lisan primarily from the elevation and radiocarbon ages of organic carbon from algal stromatolites. This curve shows three Lake Lisan lowstands, with a minimum elevation of –370 m, at approximately 35– 28 ka, 24–22 ka, and 17 ka. Lake Lisan apparently continued a cycle of rises and falls until it reached a maximum elevation of –180 m (or perhaps –150 mbsl, Bowman and Gross, 1992) by 15 ka. The Lisan Formation has also been dated using the 230 Th/234 U dating methods (Kaufman, 1971; Kaufman et al., 1992, Yechieli et al., 1993, Bartov, 1999, Klinger et al., 2000a; Schramn et al., 2000); these dates support the conclusion that deposition of the Lisan formation terminated around 20–15 ka. In the study area, the final retreat of Lake Lisan is marked by beach ridges that developed on the QL3 and

QL4 lake terraces. A layer of well-rounded cobbles covers the terrace surface. Fluvial sheet gravels and alluvial-fan deposits overlie the regressional lag and are designated as the topmost Lisan sediment (QLf ). These sheet gravels were deposited prior to full entrenchment of the major stream courses that deposited the Qf 1 , Qf 2 and Qf 3 alluvial-fan sequences. We postulate that the age of deposition of QLf is penecontemporaneous with the final Lake Lisan highstand and subsequent regression that commenced at about 15 ka. The retreat of Lake Lisan lowered the base level of drainages flowing toward the Dead Sea and caused incision. The oldest post-Lisan alluvial fan (Qf 1 ) crosscuts, overlies, and is in fault contact with the Lake Lisan sediments. The maximum age of Qf 1 is the age of the topmost Lisan formation (15 ka). If the sediment at the base of Trench 1 west side of the fault (Figure 8) is a Qf 1 deposit, then our radiocarbon ages from three landsnail shells (13 ka, 15 ka, and 16.5 ka, see Table 3) also provide a maximum age for the onset of Qf 1 deposition. However, alluvial-fan sediments are difficult to differentiate based on their sedimentary structure or soil formation (which ceases with burial and differs from the surface soil of the same fan unit). Thus, these deposits could also correlate to the QLf deposits.

Figure 8. Map of northern wall of Trench 1 showing a 9-m-wide flower structure. Pulverized rock in a fault gouge indicates repeated motion along two main strands of the fault. Extensional fractures extend several meters east of the main deformation zone. Faults are identified by rotated and aligned clasts, fracture fills, fault gouge zones, and stratigraphic offsets. Locations of radiocarbon samples from the south wall are shown at their approximate stratigraphic level. See Table 3 for further information on radiocarbon data. See Figures 3 and 4 for location of trench.

462

463

Table 3. Radiocarbon data Unit

Sample No. Lab No.

Sample type

Qf3 Unit 5 Qf3 Unit 3 Qf3 Unit 3 Qf3 Unit 3 Qf3 Unit 2 Qf3 Unit 1 Qf3 Unit 1

T2-S7 Beta-118977 T2-C6 Beta-118982 T2-S5 Beta-118975 T2-S6 Beta–118976 T2-S1 Beta-118974 T2-C4 Beta-118983 T2-C1 Beta-118981

Shell

Qf3

T3-S10 Beta-118979 T3-S5 Beta-118978 T3-S11 Beta-118980

Shell

Qf3 Qf3

Qf3

T1-C5 Beta-103462

Qf2

T1 (OSL-9) Beta-118973 T1-C1 Beta-103461 g4 pit (OSL-8) Beta-118972

Qf2 Qf2

Qf1 Qf1 Qf1

T1-S3 Beta-103456 T1-S4 Beta-103457 T1-S5 Beta-103458

Measured C14 age yr B.P.

C13/C12 ratio


Qf3 deposits in Trench 2 3820 ± 50 –2.8

Corrected C14 age yr B.P.

Calibrated age (2σ ) cal yr B.P.∗

4180 ± 50

4700 ± 130

Charcoal

1250 ± 50

–23.4

1270 ± 50

1230 ± 110

Shell

4980 ± 40

–3.4

5340 ± 40

6100 ± 110

Shell

2770 ± 50

0.0

3180 ± 50

3400 ± 80

Shell

2890 ± 50

–0.1

3300 ± 50

3540 ± 100

Charcoal

2910 ± 50

–23.3

2930 ± 50

3080 ± 130

Charcoal

3050 ± 50

–26.0

3030 ± 50

3250 ± 110

Qf3 deposits in Trench 3 2940 ± 50 +0.1

3360 ± 50

3580 ± 120

Shell

4660 ± 50

0.0

5070 ± 50

5810 ± 110

Shell

4690 ± 40

+0.2

5110 ± 40

5840 ± 90

Qf3 deposits in Trench 1 Organic 4010 ± 60 –16.1 4150 ± 60 sediment Qf2 deposits in Trench 1 and g4 soil pit Shell 8420 ± 50 +0.6 8840 ± 50 Organic sediment Shell

Shell

4680 ± 150

9840 ± 130

7640 ± 60

–16.8

7780 ± 60

8530 ± 130

7680 ± 50

–0.9

8080 ± 50

9030 ± 100

10970 ± 50

13,010 ± 150

Qf1 deposits in Trench 1 10580 ± 50 –1.2

Shell

12130 ± 50

–0.1

12540 ± 50

14,880 ± 630

Shell

13400 ± 70

+0.1

13820 ± 70

16,600 ± 490

∗ Rounded to the nearest decade. Calibrated using radiocarbon calibration program revision 4.2 by M. Stuiver and P.J. Reimer. Data set reference Stuiver et al. (1988).

464

Figure 9. Map of northern wall of Trench 3 showing multiple fault strands. Faults are identified by rotated and aligned clasts, fracture fills, fault gouge zones, and stratigraphic offsets. Locations of radiocarbon samples from the south wall are shown at their approximate stratigraphic level. See Table 3 for further information on radiocarbon data. See Figures 3, 4, and 10 for location of trench.

We assume that Qf 1 deposits aggraded during a wet climatic phase. In the study area Qf 1 deposits are extensive and cover a larger area than the Qf 2 and Qf 3 fan surfaces. According to Horowitz (1992), the Late Würm is marked by an explosion in epipaleolithic culture and an arboreal pollen peak at 13–11 ka. These data indicate a brief return of wet climatic conditions, which may be the local manifestation of the Younger Dryas event. Sedimentologic data from a 34.5 m borehole located in the Zeelim alluvial fan along the western margin of the Dead Sea indicate a shift to wetter climatic conditions at 11 ka (Yechieli et al., 1993). The basal Lisan aragonite, U-Th dated to 21 ka, is overlain by a clay layer radiocarbon dated to 11 ka. The hiatus in deposition (or erosion) between 21 and 11 ka brackets the age of retreat of the Lake Lisan level. We conclude that the Qf 1 fan deposits in our study area probably accumulated between 13–11 ka. However, in light of our radiocarbon results, we use a maximum age of 16–15 ka for Qf 1 deposition. Inset into the Qf 1 alluvial-fan surface is the deposits of the intermediate-age Qf 2 alluvial fan. We have three radiocarbon ages on carbon from the Qf 2 deposits. Radiocarbon dates from Trench 1 on organic sediment near the base of the layer yielded an age of 8.5 ka, and on a landsnail shell in a stratigraphically higher portion in the trench wall gave an age of 9.8 ka. The third sample from Qf 2 deposits was collec-

ted from the fluvial terrace in an abandoned channel (g4) on the Qf 1 alluvial fan (Figures 3 and 10). A radiocarbon age determination on a landsnail from the beheaded g4 gully yielded an age of 9 ka. An early Holocene wet phase commencing at ca. 9 ka and associated with the Pre-Pottery Neolithic cultures and ending ca. 6.5 ka is documented in several studies. Palynological studies of late Pleistocene/Holocene prehistoric sites in the Lower Jordan Valley (Leroi-Gourhan and Darmon, 1993) and of archaeological sites in Israel (Horowitz, 1992) suggest wetter-then-present climatic conditions after 10–9 ka. Frumkin (1997) reports relatively high levels of the Dead Sea during the early Holocene. In the Zeelim borehole (Yechieli et al., 1993), the 11-ka-clay is overlain by a 6.5-m-thick halite layer that is capped by a clastic section. Radiocarbon dating of wood overlying the halite indicates that the deposition marking arid conditions ceased around 8.5 ka. Given these data, we believe that the Qf 2 alluvial-fan sediments in our study area were deposited during a wet phase between 9–8.5 ka and 6.5 ka. The Qf 3 young alluvial-fan unit in the study area is exposed in all three trenches, and 11 radiocarbon ages from both charcoal and landsnail shells indicate that the maximum age for onset of Qf 2 alluvial-fan aggradation is 5.8 ka (Table 3). This age correlates with wetter climatic conditions in the region that are

Figure 10. Detailed geologic map showing offset gullies developed on the Qf 1 alluvial fan surface. Piercing points used to calculate slip are marked as solid dots on the map. We use uppercase ‘G’ for upstream gullies east of the fault and lowercase ‘g’ for downstream channels west of the fault trace. Likewise, projected piercing points are uppercase letters east of the fault and lowercase letters west of the fault. The gullies are numbered from north to south. See Figures 3 and 4 for location of map.

465

466 inferred from evidence of high Dead Sea lake levels, expansion of Chalcolithic culture, and other proxy records described in Frumkin (1997). The Qf 3 fan is currently active. Fault traces The northern Wadi Araba fault (NWAF) is distinct both on aerial photographs and in the field investigation. In the study area, fault traces are defined by: (1) abrupt changes in drainage courses, (2) straight low scarps in late Pleistocene and Holocene sediments, and (3) the linear alignment of pressure ridges and sag ponds. Evidence of tectonic offsets along the NWAF is well preserved in the study area. The larger streams (Wadi An Nukhbar and Wadi Tilah) and several tributaries with stronger erosional ability and steep gradients have destroyed the scarp along the active fault. However, most of the drainages in the study area that are nearly perpendicular to the NWAF, are laterally offset along the well-defined linear traces of the NWAF. Their small drainage area and short development history offer excellent possibilities to reconstruct the offset history of these channels. Where Wadi Tilah enters the Wadi Araba Valley, the NWAF forms a bedrock scarp at the Dahal mountain front (Figure 2); the straight fault trace separates late Quaternary alluvial-fan sediments from bedrock. To the north, the fault bifurcates: one fault branch follows the straight mountain front and shows no sign of recent activity, whereas the other branch trends more northerly, cuts through the alluvial fans, and is marked by very clear fault scarps (Figure 3). Fault scarps are easily degraded and modified by mass wasting, erosion and sedimentation, especially in young unconsolidated sediments. The well-preserved fault scarps at the study site are due to two factors: (1) the young faulting, and (2) the extremely arid climatic conditions. Fault-scarp heights on the Qf 1 and Qf 2 fan surfaces in the study area vary from 1–2.5 m. Height variations are largely the product of horizontal displacement and the juxtaposition of pre-existing topographic features. Fault-scarp heights do not depend on the age of the faulted deposits. On the Qf 1 alluvialfan surface, the scarp height is only 1–1.5 m, whereas near trench 1, the scarp height on the younger Qf 2 alluvial-fan surface is 2–2.5 m. Displacements on this segment of the NWAF do have a small vertical component of slip, however,

strike-slip motion is dominant. Based on our study, the Qf 1 alluvial-fan surface is lateral offset 54 m compared to vertical displacements of 1–1.5 m, which indicates a ratio between vertical and horizontal displacement of 0.02–0.03. The strike-slip motion on the NWAF has formed features such as shutter ridges where it traverses undulating terrain. Streams blocked by shutter ridges are forced to flow along the fault trace, and then follow in the previous channel or form a new channel. This process produces sharp bends (or ‘dog-leg’ bends, Wallace, 1970) in stream courses, which provides excellent evidence of horizontal fault offsets. The formation of sharp bends is largely dependent on fluvial erosional processes. Optimum conditions are where the erosion is strong enough to maintain a channel, but not too strong to greatly modify or destroy the channel. The active fault trace in alluvial-fan deposits as mapped on the detailed 1:250-scale field topographic map is not a single, continuous straight line (Figures 10 and 11); at this scale, the fault is composed of discontinuous, en echelon traces that generally step toward the right. This geometry creates localized zones of compression that produces pressure ridges. Elsewhere, local extension at left steps between strands form small pull-apart sag depressions. Pressure ridges are very prominent, especially in the northern part of the study area (Figures 2 and 3). These pressure ridges have been transported along the fault trend to form shutter ridges and disrupt drainages. Commonly, the fault zone is characterized by two parallel fault strands that splay from a single trace in the south into two traces in the north. A third, quaternary fault trace is present in preLisan sediments. The fault scarp trends N-S and subparallel to the modern fault trace and separates the QL2 and QL3 lake terraces. This scarp was modified later into a lake terrace riser during the drop of Lake Lisan. The arrangement and activity of these fault strands may indicate a basinward migration of fault activity as mentioned by Gardosh et al. (1990) who studied the western margin of the Dead Sea. Alternatively, various fault splays may be active at different times. Fault offsets, age of offsets, and slip rates Lateral-slip (Table 4) is determined by matching geomorphic features of post-Lisan age across the fault, such as offset alluvial-fan surfaces, offset channels, and beheaded gullies. These offset geomorphic features allow multiple interpretations of displacement

467

Figure 11. Detailed geologic map showing offset gullies in the vicinity of Trench 2. Piercing points used to calculate slip are marked as solid dots on the map. We use uppercase ‘G’ for upstream gullies east of the fault and lowercase ‘g’ for downstream channels west of the fault trace. Likewise, projected piercing points are uppercase letters east of the fault and lowercase letters west of the fault. The gullies are numbered from north to south. See Figures 3 and 4 for location of map.

events. Lisan-age and pre-Lisan deposits do not offer good piercing points for measuring fault offsets because the lake terraces and Lisan deposits on the east side of the fault zone can not be traced to the west of the fault zone. Offset fluvial channels provide the best piercing points because drainages flow nearly perpendicular to the fault trace. The channel offset data are summarized on Table 4 and are shown in Figures 10 and 11. In the following discussion, we use uppercase ‘G’ for upstream gullies east of the fault zone and lowercase ‘g’ for downstream channels west of the fault trace. Likewise, projected piercing points are uppercase letters east of the fault and lowercase letters west of the fault. The gullies are numbered from north to south on each map sheet.

Excellent examples of offset channels are found on the offset Qf 1 alluvial-fan surface (Figure 10), where three large gullies (named G1, G4 and G6) were matched to downstream gullies, and three large fault offsets (54 m, 39 m and 22.5 m) were measured. Since the gullies tend to bend and lengthen along the fault with successive faulting events, projecting the straight, downstream axis of the gully to the fault as a piercing point provides the best estimate of these offsets. Our best estimate of the 54 m offset was determined by matching three piercing points (Figure 10): 1) point A1 of gully G1 with point a1 of gully g2, 2) projection point A2 of G4 with projection point a2 of the beheaded gully g5, and 3) point A3 of G6 with point a3 of gully g8. Gully g8 is a large channel developed on Qf 2 alluvial-fan surface. It is more likely that the course of this channel was once a margin of

468 Table 4. Estimates of laterally offset features along the northern Wadi Araba fault, Jordan. Location of offset features described below are shown in Figures 3, 10 and 11 Offsets 1

2

3

Evidence

Best estimate

54 m

Maximum

66 m

Minimum

40 m

Best estimate

39 m

Minimum

31 m 33 m

Best estimate

22.5 m

Maximum

25 m

Minimum

23 m 20 m 21 m

offset Qf1 alluvial-fan surface from A1 to a1. offset gully G4 from A2 to a2 of gully g5. offset gully G6 from A3 to a3 of gully g8. offset Qf1 alluvial-fan surface from m to n using projecting point. of fluvial terrace margin to the fault as piercing point. offset gully G4 from c to d of gully g5. offset gully G6 from e to f of gully g7. offset Qf1 alluvial-fan surface from A1 to b1. offset gully G4 from A2 to b2 of gully g4. offset gully G6 from A3 to b3. offset from c of gully G4 to o of gully g4. offset from e of gully G6 to b3. offset gully G4 from A2 to c2 of gully g3. offset gully G6 from A3 to c3 of gully g6. offset from A1 to c1. offset gully G11 from A11 to a11 of g11. offset gully G11 from A22 to a22 of g22. offset from A1 to d1. offset Qf 2 alluvial-fan surface from A33 to a33.

4

6m

small fault-scarp alluvial fans. offset Qf 1 alluvial-fan surface from A1 to k. offset gully G4 from A2 to L. bend of gully G3.

5

3m

small fault-scarp alluvial fans. offset of gully G5. bend in gully g33. bend of gully G4 from A2 to c? bend in gully G6 from A3 to e?

Qf 1 alluvial fan based on the observation that gully G6 is the boundary between Holocene Qf 1 sediment and Late Pleistocene Lisan fluvial and lacustrine sediments (QLf and QLL in Figure 10). Fluvial terraces (Qt 1) developed in gully g2 indicate that it connects with the larger upstream gully G1, not from small gullies of G2 and G3. Projecting the terrace edge to the fault and matching the intersection ‘m’ with ‘n’ gives the maximum offset value of 66 m. A minimum offset estimate of 40 m is measured from the intersection of the thalweg and the fault (which is the midpoint in the dog-leg bend). These are between gully G4 (point ‘c’) and g5 (point ‘d’), and between gully G6 (point ‘e’) and gully g8 (point ‘f’). These offsets fall within a range of 54 ± 12 m.

The 54 m offset is no older than 16–15 ka, based on the assumption that the lowering of Lake Lisan began around 16–15 ka as discussed earlier. During the lake’s regression, a series of alluvial fans and strandlines developed on older lacustrine sediments at the foot of the eastern escarpment. Alluvial fans that formed after the last retreat of Lake Lisan lack beach features. We postulate that the Qf 1 alluvial fan aggraded during a return to humid climatic conditions during the Younger Dryas time ca. 13–11 ka (Horowitz, 1992, Niemi, 1997). Thus, the maximum age of the Qf 1 is 16–15 ka, whereas its minimum age is 13–11 ka. Using the maximum age, a minimum slip rate for the 54 m offset is 3.5 ± 0.1 mm/yr. Using our preferred age estimate for the Qf 1 fan deposition of 13–11 ka

469 Table 5. Slip rates for the Wadi Araba fault determined in this study Cumulative offset (best estimate)

Time interval (Stratigraphic and geomorphic evidence and climatic interpretation)

Slip rate (SR)

Recurrence interval (RI) based on 3 m individual offset

54 m

16–15 ka (maximum age) (End of Lake Lisan, begin QLf to Qf 1 deposition) 13–11 ka (minimum age) (End of Qf 1 deposition, dry period starts)

minimum SR 3.5 ± 0.1 mm/yr

maximum RI 860 ± 25 yr

preferred SR 4.5 ± 0.4 mm/yr

preferred RI 680 ± 60 yr

9 ka (maximum age) (Begin Qf 2 deposition, wet period starts. Three new gullies with straight channels form on the Qf 1 surface) 6.5 ka (minimum age) (End of Qf 2 deposition, dry period starts)

minimum SR 4.3 mm/yr

maximum RI 700 yr

maximum SR 6 mm/yr

minimum RI 500 yr

5.8 ka (maximum age) (Begin Qf 3 deposition, wet period starts. Second generation of new gullies with straight channels form on the Qf 1 surface)

minimum SR 3.9 ± 0.5 mm/yr

maximum RI 770 yr

39 m

22.5 ± 2.5 m

and a 54 m offset, we calculate a preferred slip rate of 4.5 ± 0.4 mm/yr (Table 5). A 39 m offset was determined by matching channel offsets on the Qf 1 alluvial-fan surface from three piercing points (Figure 10): 1) points A1 to b1 of gully g1, 2) point A2 of gully G4 with b2 of gully g4, and 3) point A3 of gully G6 with point b3 of gully g7. Gully g7 is one of the three beheaded channels on the offset Qf 2 alluvial-fan surface. The downstream gully g4 is large and has a fluvial terrace developed in its channel. Gully g4 is displaced from the larger, upstream gully G4. The drainage of g4 was subsequently captured by the upstream G5 gully. A minimum offset of 31 m can be measured from point ‘c’ of gully G4 to point ‘o’ of gully g4. A 33 m offset is measured from point ‘e’ of gully G6 to point b3 of gully g7. We estimate that the 39 m offset started no earlier than 9 ka based on dating of a landsnail shell from the sediment fill in gully g4, which yielded a 2σ age of 9030 ± 100 cal yr B.P. (Figure 10, Table 3). Two additional radiocarbon dates from the Qf 2 alluvium from Trench 1 (Figure 8) yield 2σ ages of 8530 ± 150 cal yr B.P. on charcoal and 9840 ± 130 cal yr B.P. on shell. We believe that the charcoal age is more reliable. We use the maximum age of 9 ka and 39 m of

offset to calculate a minimum slip rate of 4.3 mm/yr. A minimum age for the Qf 2 deposit is based on a shift to more arid conditions as recorded in the Dead Sea climatic proxy records (Frumkin, 1997). This minimum age of 6.5 ka yields a maximum slip rate for the 39 m offset of 6 mm/yr. The best estimate of 22.5 m offset is from the present courses of gullies G1, G4 and G6, which gullies are obviously bent along the fault zone due to cumulative offset. The displacement is measured by matching point A2 of gully G4 with point c2 of gully g3 and point A3 of gully G6 with point c3 of gully g6. The distance from point A1 to point c1 is 25 m, and from point A1 to point d1 is 20 m. Similar offsets are present along the fault zone to the north (Figure 11). Near Trench 2, two gullies, G11 and G22 were left-lateral offset to gullies g11 and g22. Measuring the piercing point from A11 to a11 gives an offset of 25 m, and from A22 to a22 gives an offset of 23 m. The offset of Qf 2 alluvial-fan surface from A33 to a33 is 21 m. For this group of offsets, the maximum is 25 m, the minimum is 20 m, and the best estimate value is 22.5 m. The offset falls in the range of 22.5 ± 2.5 m. A maximum age of the 22.5 m offset of 5.8 ka is based on the subsurface stratigraphic relations mapped

470 in Trench 1 and 3 (Figures 8 and 9). In Trench 3, snail shells stratigraphically above the buried soil of the Qf 2 surface yield a 2σ age of 5810 ± 110 and 5840 ± 90 cal yr B.P. The Qf 2 buried soil in Trench 1 contained charcoal dating to 4680 ± 150 cal yr B.P. The maximum age of the beginning of deposition of the Qf 3 alluvial fan surface is therefore 5.8 ka. We calculated a minimum average slip rate of the 3.9 mm/yr using a best estimate of the offset of 22.5 m and a maximum age of 5.8 ka. Given the range of offsets between 20 and 25 m, we calculated an average minimum slip rate of 3.9 ± 0.5 mm/yr. In addition to the three large offsets (54 m, 39 m, and 22.5 m), we measured other smaller offsets. Both 6 m and 3 m offsets are common along the fault trace (Table 4). A 6 m offset is measured from three piercing points: 1) point A1 to K where Qf 1 alluvial-fan surface was displaced, 2) the bend of gully G3, and 3) point A2 to L of offset gully G4. A 3 m offset is indicated by the offset of gully G5 along the fault, and perhaps the bend of gully G4 from A2 to ‘c’ and gully G6 from A3 to ‘e’ (Figure 10). Small alluvial fans developed on the fault scarps also appear to be offset by 3 m (Figure 10). In the northern map area (Figure 11), the bend in gully g33 is approximately 3 m. These offset data may be evidence of 3 m of coseismic slip in the last earthquakes or a cumulative slip of 3 m in the past few earthquakes. To better illustrate the offset history along this part of the fault, we constructed a paleogeographic map for the three large offset channels on the Qf 1 alluvial-fan surface (Figure 12). Around 15 ka following the final retreat of Lake Lisan, the oldest alluvial fan (Qf 1 ) was deposited on the Lisan lacustrine deposits and older sediments at the foot of the eastern escarpment. Three large gullies G1, G4, and G6 formed on the active Qf 1 surface (Figure 12a). We assume that initially, these gullies flowed straight across the fault. Slip on the fault progressively offsets these gullies left laterally. By around 9 ka, 15 m of displacement had occurred along the fault (Figure 12b). At this time, the oldest alluvial fan (Qf 1 ) became inactive, and a new alluvial fan (Qf 2 ) started to form. Three downstream channels, gullies g2, g5, and g8, were abandoned, and three new gullies, g1, g4, and g7, formed. This simultaneous shift of three gullies and development of straight channels was probably driven by a shift to wetter climatic conditions. By about 5.8 ka, a total cumulative displacement of 31.5 m had occurred (Figure 12c). Approximately 16.5 meters of this offset was recorded by the offset Qf 1 alluvial-fan margin. Sometime after

5.8 ka, the Qf 2 alluvial fan also was abandoned, and the Qf 3 alluvial fan began to develop. This resulted in the abandonment of gullies g1, g4, and g7, and the formation of another three new gullies g3, g6, and a subsidiary of g1. This hydrologic shift was also probably driven by a shift to a wetter climate. Since 5.8 ka there has been 22.5 m of offset of the Qf 2 alluvial fan surface. Both 6 m and 3 m offsets are present as bends in current gully courses. Collectively, a cumulative offset of 54 m has occurred over the past 15 ka (Figure 12d). Discussion The slip rate on a fault is generally related to the recurrence rate of surface-rupturing earthquakes on the fault (Wallace, 1970; Schwartz and Coppersmith, 1984). Faults with high slip rates generally generate large magnitude earthquakes with short recurrence intervals, while low slip rates will produce earthquakes with long recurrence intervals. If both fault segmentation and characteristic earthquake models are true, then fault slip rates can be used to estimate the earthquake recurrence time on individual fault segments. Therefore, determining fault slip rates is one of the most important parameter for understanding earthquake hazards (Allen, 1986; McCalpin, 1997; Yeats et al., 1997). Based on our studies, we have identified three prominent offsets and estimated their ages to yield long-term average fault slip rates for different time intervals (Table 5). We estimate that the slip rate falls within the range of 3.4–6 mm/yr, and our preferred average estimate is 4.7 ± 1.3 mm/yr. Comparison between the slip rates for each of the different displacements and time intervals demonstrates that the Wadi Araba fault has moved at a relatively constant slip rate since 15 ka. The similarity of minimum and maximum slip-rate estimates suggests that strain accumulation across the Wadi Araba fault is fairly uniform. Therefore, the average recurrence interval deduced from this slip rate may yield realistic representation of the typical earthquake recurrence for the fault. The average slip rate from this study of 4.7 ± 1.3 mm/yr is similar to most of the long-term geologic slip rate estimates (Table 1). Previous estimates for the DST of 7–10 mm/yr are based on an accumulative displacement of 105 km since the Middle Miocene (Table 1). Our rate is consistent with the slip rate range of 3.5–6 mm/yr that was determined by Freund et al. (1968, 1970) from a 40–45 km offset of Miocene-age

471

Figure 12. Paleogeographic map illustrating the offset history of gullies developed on the Qf1 alluvial fan. We use uppercase ‘G’ for upstream gullies east of the fault and lowercase ‘g’ for downstream channels west of the fault trace. Likewise, projected piercing points are uppercase letters east of the fault and lowercase letters west of the fault. a) Three prominent gullies flow across the fault (G1-g2, G4-g5, and G6-g8) ca 13–11 ka. b) Around 9 ka, after 15 m of cumulative slip, gullies G1, G4 and G6 abandon their downstream channels that lie west of the fault. New downstream gullies g1, g4 and g7 form. c) Around 5.8 ka after 16.5 m of cumulative slip, gullies G1, G4, and G6 once again abandon their downstream channels and develop new straight courses (g1 at point c1, g3, and g6). d) Map of the present configuration of gullies.

472 rocks (7–12 Ma), and with the work of Klinger (1999), who estimated a rate of 2–6 mm/yr from a 500 m offset of the Late Pleistocene (77–140 ka) Dahal alluvial fan. When using slip rates to calculate a recurrence interval, the displacement-per-event must be known. The offsets of 54 m, 39 m, and 22.5 m used to calculate the average slip rate are cumulative displacements caused by multiple large earthquakes. Because the exact number of earthquakes that produced these offsets is unknown, the average recurrence interval can not be determined from these data. Individual offsets-per-earthquake on Wadi Araba fault in the study area is not well constrained, but our five 3-m offset measurements suggest that this might be the maximum amount of slip produced in a single earthquake. Using an average slip rate of 4.7 mm/yr together with a 3-m slip-per-event implies a recurrence interval of 500 to 885 years (Table 5). If we apply the equations derived from global data that relate earthquake magnitude to rupture length and maximum displacement (Wells and Coppersmith, 1994), a 3-m slip event corresponds to a Mw 7 earthquake along a 49-km-long fault rupture. Klinger et al. (2000b) also measured channel offsets in this study area that cluster about 0.3 m and 2.7 m. These data suggest that the 3 m offset may be cumulative slip of two earthquakes rather than from a single event. Five large historical earthquakes are recorded for the region between the Dead Sea and the Gulf of Aqaba; (Ambraseys et al., 1994; Ben Menahem, 1991); these include the earthquakes of A.D. 1068, 1202, 1212, 1293, and 1458. The earthquakes of A.D. 1068 and 1212 probably had their epicenters in the southern Araba Valley where ground-rupture has been documented in the Evrona playa (Amit et al., 1999) and Aqaba (Mansoor and Niemi, 1999). Ground rupture along the Dahal tributaries reported in this study may have occurred in the earthquakes of A.D. 1202, 1293, and/or 1458. Conclusions About 25 km south of the Dead Sea, the Wadi Araba fault has a clear geomorphic expression of active tectonic motion that includes offset stream channels, beheaded streams, offset alluvial-fan surfaces, shutter ridges, pressure ridges, and sag depressions. This geomorphic evidence, combined with stratigraphic evidence from three trenches excavated across the Wadi Araba fault, permit the reconstruction of the history of fault offsets, slip rate, and probable earthquake recurrence intervals.

This study focused on defining the stratigraphic framework of Late Quaternary sediments along the northern Wadi Araba fault south of the Dead Sea escarpment where tributaries to the Wadi Dahal are offset by recent faulting. Detailed mapping along the northern Wadi Araba fault reveals systematic displacements of 54 m, 39 m, and 22.5 m; these values are derived from matching stable upstream gullies east of the active fault to three generations of downstream gullies west of the fault. From our mapping, we identified three generations of alluvial fans and fluvial terraces that post-date the highest stand of Lake Lisan (∼15 ka). The maximum age of the Qf 1 alluvial fan surface is no earlier than the regression of Lake Lisan ca. 16–15 ka. The minimum age of the Qf 1 deposits is inferred to correlate with the Younger Dryas at 13–11 ka. Using the Qf 1 offset of 54 m and an age range of 16–11 ka, we calculate a range of slip rates between 3.4 and 4.9 mm/yr. Using the 14 C ages of samples collected from Trench 1 and the soil pit excavated in g4 gully terrace, we estimate that the Qf 2 alluvial fan deposition began around 9 ka and corresponds with the development of a new generation of gullies on the Qf 1 surface. These gullies are offset 39 m. Qf 2 fan deposition is inferred to have ceased around 6.5 ka when the region experienced extreme aridity. Using an offset of 39 m and an age range of 9–6.5 ka, we calculate a minimum slip rate of 4.3 mm/yr and a maximum slip rate of 6 mm/yr. Approximately 22.5 ± 2.5 m of slip have occurred on gullies eroded since 5.8 ka thus yielding a slip rate of 3.9 ± 0.5 mm/yr. Comparison between these slip rates on piercing points ranging in age from 16 ka to 5.8 ka indicates that motion on the Wadi Araba fault has been a nearly constant, average slip rate of about 4.7 ± 1.3 mm/yr. Five separate offset measurements of 3 m from bends in gullies and offset of small fault-scarp alluviual fans were mapped. Archaeological evidence suggests that the 3-m displacement may be the cumulative coseismic slip from more than one earthquake. If we assume a 3-m slip-per-event value for this segment, then the calculated recurrence of similar-magnitude earthquakes using a rate of 4.7 ± 1.3 mm/yr is 500 to 885 years. A maximum slip of 3 m corresponds to a Mw 7 earthquake that ruptures about 49 km of fault length. However, to resolve whether the 3-m offset is a single or multiple-event displacement requires additional detailed investigations.

473 Acknowledgments We are grateful to those who assisted this study in the field, including J. Christian Gierke, Andrew M. Smith II, Maher Jarrar, Nasser Mansoor, Martha (Missy) Eppes, and Eileen Niemi. We deeply appreciate the generosity of the Jordan Valley Authority for permitting us to use their resthouse facilities in Safi. We thank Nasser Mansoor for drafting figures 5–7 and J. Christian Gierke for using a digital terrane modeling program to generate figure 9. This manuscript was greatly improved by the critical reviews of Anthony J. Crone and two anonymous reviewers. Support for this research was provided by a University of Missouri Research Board grant and a UMKC Faculty Research grant to T.M. Niemi. References Allen, C.R., 1986, Seismological and paleoseismological techniques of research in active tectonics, In: Wallace, R.E. (ed.), Active Tectonics: Studies in Geophysics, Washington, D.C., Natl. Acad. Press, pp. 148–154. Ambraseys, N., Melville, C. and Adams, R., 1994, The Seismicity of Egypt, Arabia, and the Red Sea, Cambridge Univ. Press, New York, 181 pp. Ambraseys, N. and Jackson, J.A., 1998, Faulting associated with historical and recent earthquakes in the Eastern Mediterranean region, Geophys. J. Int. 133, 390–406. Amiran, D.H.K., Arieh, E. and Turcotte, T., 1994, Earthquakes in Israel and adjacent areas – macroseismic observations since 100 B.C.E., Israel Expl. J. 44(3-4), 260–305. Amit, R., Harrison, J.B.J. and Enzel, Y., 1995, Use of soils and colluvial deposits in analyzing tectonic events – The southern Arava rift, Israel, Geomorph. 12, 91–107. Amit, R., Harrison, J.B.J., Enzel, Y. and Porat, N., 1996, Soils as a tool for estimating ages of Quaternary fault scarps in a hyperarid environment – the southern Arava Valley, the Dead Sea rift, Israel, Catena 28, 21–45. Amit, R., Zilberman, E., Porat, N. and Enzel, Y., 1999, Relief inversion in the Avrona playa as evidence of large-magnitude historical earthquakes, southern Arava Valley, Dead Sea rift, Quat. Res. 52, 76–91. Bartov, Y., 1999, The Geology of the Lisan Formation in Massada Plain and the Lisan Peninsula, The Hebrew University of Jerusalem, Master thesis (in Hebrew with English abstract), 63 pp. Begin, Z.B., Ehrlich, A. and Nathan, Y., 1974, Lake Lisan. The Pleistocene precursor of the Dead Sea, Geol. Surv. Isr. Bull. 63, 30 pp. Begin, Z.B., Broecker, W., Buchbinder, B., Druckman, Y., Kaufman, A., Magaritz, M. and Neev, D., 1985, Dead Sea and Lake Lisan levels in the last 30,000 years, A preliminary report, Geol. Surv. Isr. Rep. GSI/29/85, 18 pp. Bender, F., 1974, Geological map of Jordan, 1:100,000: Geological Survey of the Federal Republic of Germany, Hannover. Ben-Menahem, A., 1979, Earthquake catalog for the Middle East (92 B. C. – 1980 A.D.), Bull. Geof. Teor. Appl. 84, 245–310. Ben-Menahem, A. and Aboodi, E., 1981, Micro and macroseismicity of the Dead Sea rift and offcoast eastern Mediterranean, Tectonophysics 80, 199–233.

Ben-Menahem, A., 1991, Four thousand years of seismicity along the Dead Sea Rift, J. Geophy. Res. 96(B12), 20,195–20,216. Blake, G.S., 1939, Geological Map of Palestine: Jerusalem, Survey of Palestine, scale 1:250,000. Bowman, D., 1988, The declining but non-rejuvenating base level – the Lisan Lake, the Dead Sea area, Israel, Earth Surf. Proc. and Landforms 13, 239–249. Bowman, D. and Gross, T., 1992, The highest stand of Lake Lisan: –150 meter below MSL, Isr. J. Earth Sci. 41, 233–237. Bowman, D., 1995, Active surface ruptures on the northern Araba fault, the Dead Sea rift, Isr. J. Earth Sci. 44, 51–59. Bowman, D., 1997, Geomorphology of the Dead Sea western margin, In: Niemi, T.M., Ben-Avraham, Z. and Gat, J.R. (eds), The Dead Sea – The lake and its setting, Oxford Monographs on Geology and Geophysics 36, pp. 217–225. Clark, G.A., 1988, Some thoughts on the southern extent of the Lisan Lake as seen from the Jordan side, Bull. Am. School of Oriental Res. 272, 42–43. Courtillot, V., Armijo, R. and Tapponnier, P., 1987, Kinematics of the Sinai triple junction and a two phase model of Arabia-Africa rifting, Geol. Soc. Spec. Pub. 28, 559–573. Druckman, Y., Magaritz, M. and Sneh, A., 1987, The shrinking of Lake Lisan, as reflected by the diagenesis of its marginal oolitic deposits, Isr. J. Earth Sci. 36, 101–106. El-Isa, Z.H. and Mustafa, H., 1986, Earthquake deformations in the Lisan deposits and seismotectonic implications, Geophys. J. Royal Astron. Soc. 86, 413–424. Enzel, Y., Amit, R., Zilberman, E., Harrison, J.B.J. and Porat, N., 1996, Estimating the ages of fault scarps in the Arava, Israel, Tectonophysics 253, 305–317. Freund, R., 1965, A model of the structural development of Israel and adjacent areas since Upper Cretaceous times, Geol. Mag. 102, 189–205. Freund, R., Zak, I. and Garfunkel, Z., 1968, Age and rate of the sinistral movement along the Dead Sea rift, Nature 220, 253– 255. Freund, R., Garfunkel, Z., Zak, I., Goldberg, M., Derin, B. and Weissbrod, T., 1970, The shear along the Dead Sea rift, Philos. Trans. R. Soc. London, Ser. A. 267, 107–130. Frumkin, A., 1997, The Holocene history of Dead Sea levels, In: Niemi, T.M., Ben-Avraham, Z. and Gat, J.R. (eds), The Dead Sea – The Lake and its Setting, Oxford Monographs on Geology and Geophysics No. 36, pp. 237–248. Gardosh, M., Reches, Z. and Garfunkel, Z., 1990, Holocene tectonic deformation along the western margins of the Dead Sea, Tectonophysics 180, 123–137. Garfunkel, Z. and Ben-Avraham, Z., 1996, The structure of the Dead Sea basin, Tectonophysics 266, 155–176. Garfunkel, Z., Zak, I. and Freund, R., 1981, Active faulting in the Dead Sea rift, Tectonophysics 80, 1–26. Garfunkel, Z., 1970, The Tectonics of the Western Margins of the Southern Arava, Ph.D. Thesis, Hebrew University, Jerusalem, 204 pp. (In Hebrew, English abstract) Gerson, R., Grossman, S., Amit, R. and Greenbaum N., 1993, Indicators of faulting events and period of quiescence in desert alluvial fans, Earth Surf. Proc. and Landforms 18, 181–202. Ginat, H., Enzel, Y. and Avni, Y., 1998, Translocated PlioPleistocene drainage systems along the Arava fault of the Dead Sea transform, Tectonophysics 284, 151–160. Guidoboni, E. (ed.), 1994, Catalogue of Ancient Earthquake in the Mediterranean Area up to the 10th Century, Istituto Nazionale di Geofisica, Rome, Italy. Horowitz, A., 1992, Palynology of Arid Lands, Amsterdam, Elsevier, 546 pp.

474 Ibrahim, K., 1991, The Geology of the Wadi Rahma, Bulletin 15, Geological Mapping Division, Geology Directorate, Natural Resources Authority, Jordan, 122 pp. Ibrahim, K., 1993, The Geology of the Wadi Gharandal Area, Bulletin 24, Geological Mapping Division, Geology Directorate, Natural Resources Authority, Jordan, 77 pp. Joffe, S. and Garfunkel, Z., 1987, Plate kinematics of the circum Red Sea – a re-evaluation, Tectonophysics 141, 5–22. Kaufman, A., 1971, U-series dating of Dead Sea basin carbonates, Geochimica et Cosmochimica Acta 35, 1269–1281. Kaufman, A., Yechieli, Y. and Gardosh, M., 1992, Reevaluation of the lake-sediment chronology in the Dead Sea basin, Israel, based on new 230 Th/U dates, Quat. Res. 38, 292–304. Klinger Y., 1999, Seismotectonics of the Levantine fault zone, Universite Louis Pasteur – Strasbourg I., Ph.D. dissertation (in French and English), 238 pp. Klinger Y., Avouac, J.P., Abou Karaki, N., Dorbath, L., Bourles, D. and Reyss, J.L., 2000a, Slip rate on the Dead Sea transform in northern Araba Valley (Jordan), Geophys. J. Int. 142(3), 755– 768. Klinger Y., Avouac, J.P., Dorbath, L. and Abou Karaki, N., 2000b, Seismic behaviour of the Dead Sea fault along Araba Valley, Jordan, Geophys. J. Int. 142(3), 769–782. Leonard, G., Steinberg, D.M. and Rabinowitz, N., 1998, An indication of time-dependent seismic behavior – An assessment of paleoseismic evidence from the Arava fault, Israel, Bull. Seism. Soc. Am. 88(3), 767–776. Leroi-Gourhan, A. and Darmon, F., 1987, Analyses palynologiques de sites archeologiques de Pleistocene final dans la Vallee du Jourdain, Isr. J. of Earth Sci. 36, 65–72. Mansoor, N. and Niemi, T.M., 1999, Active oblique-slip faults across the Wadi Yutim fan complex at the head of the Gulf of Aqaba, Dead Sea Transform, Jordan, Geol. Soc. Am. Absts. with Prog. 31(7), A-377. Marco, S., Agnon, A., Ellenblum, R., Eidelman, A., Basson, U. and Boas, A., 1997, 817-year-old walls offset sinistrally 2.1 m by the Dead Sea Transform, Israel, J. Geodyn. 24(1-4), 11–20. Marco, S. and Agnon, A., 1995, Prehistoric earthquake deformations near Masada, Dead Sea graben, Geology 23(8), 695–698. McCalpin, J.P., 1996, Paleoseismology, Academic Press, San Diego, 588 pp. Neev, D. and Emery, K.O., 1967, The Dead Sea, depositional processes and environments of evaporites, Geol. Surv. Isr. Bull. 41, 147 pp. Neev, D. and Hall, J.K., 1977, Climatic fluctuations during the Holocene as reflected by the Dead Sea levels, In: Greer, D.C. (ed.), Terminal Lakes, Int. Conf. on Desertic Terminal Lakes, Ogden Utah, 1977, Proc., pp. 53–60. Niemi, T.M., 2000, Wadi Araba earthquake project, In: Egan, V., Bikai, P.M. and Zamora, K. (eds), ‘Archaeology in Jordan’, Am. J. Archaeology 104(3), pp. 567–568. Niemi, T.M. and Atallah, M., 2000, Offset of the Early Islamic ruins of Qasr Tilah along the Wadi Araba fault, Dead Sea transform, Jordan (abst), Geol. Soc. Am. Absts. with Prog. 32(7), 443. Niemi, T.M. and Smith, A.M., II, 1999, Initial results of the southeastern Wadi Araba, Jordan geoarchaeological study: Implications for shifts in late Quaternary aridity, Geoarchaeology 14(8), 791–820. Niemi, T.M., 1997, Fluctuations of Late Pleistocene Lake Lisan in the Dead Sea Rift, In: Niemi, T.M., Ben-Avraham, Z. and Gat, J.R. (eds), The Dead Sea – The Lake and its Setting, Oxford Monographs on Geology and Geophysics No. 36, N.Y., Oxford University Press, pp. 226–236.

Poirier, J.P. and Taher, M., 1980, Historical seismicity in the Near and Middle East, north Africa and Spain from Arabic documents (VIIth – XVIII century), Bull. Seism. Soc. Am. 70, 2185–2201. Porat, N., Wintle, A.G., Amit, R. and Enzel, Y., 1996, Late Quaternary earthquake chronology from luminescence dating of colluvial and alluvial deposits of the Arava Valley, Israel, Quat. Res. 46, 107–117. Quennell, A.M., 1956, The Geological Map of the Kingdom of Trans-Jordan, scale 1: 250,000. Quennell, A.M., 1959, Tectonics of the Dead Sea rift, Int. Geol. Congr., 20th , Mexico. Assoc. Serv. Geol. Afr., pp. 385–403. Rashdan, M., 1988, The Regional Geology of the Aqaba-Wadi Araba Area, Bulletin 7, Geological Mapping Division, Geology Directorate, Natural Resources Authority, Jordan, 87 pp. Russel, K., 1985, The earthquake chronology of Palestine and northwest Arabia from the 2nd through the mid-8th century A.D., Bull. Am. School Oriental Res. 260, 37–59. Schramn, A., Stein, M. and Goldstein, S.L., 2000, Calibration of the 14 C time scale to >40 ka by 234 U – 230 Th dating of Lake Lisan sediments (last glacial Dead Sea), Earth and Planet. Sci. Lett. 175, 27–40. Schwartz, D.P. and Coppersmith, K.J., 1984, Fault behavior and characteristic earthquakes – Examples from the Wasatch and San Andreas faults, J. Geophys. Res. 89, 5681–5698. Shapira, A., 1981, Assessment of the potential earthquake risk in Israel and adjacent areas, Isr. J. Earth Sci. 30, 135–141. Shaw, S.H., 1947, Geological Map of Southern Palestine with Explanatory Notes, Govt. Printer, Jerusalem, 42 pp., scale 1:250,000. Shtivelman, V., Frieslander, U., Zilberman, E. and Amit, R., 1998, Mapping shallow faults at the Evrona playa site using high-resolution reflection methods, Geophysics 63, 1257–1264. Steinitz, G., Bartov, Y. and Hunziker, J.C., 1978, K-Ar age determination of some Miocene-Pliocene basalts in Israel – their significance to the tectonics of the rift valley, Geol. Mag. 115(5), 329–340. Stuiver, M., Reimer, P.J., Bard, E., Beck, J.W., Burr, G.S., Hughen, K.A., Kromer, B., McCormac, G.V.D., Plicht, J. and Spurk, M., 1998, INTCAL 98 radiocarbon age calibration, 24,000 – 0 cal BP, Radiocarbon 40, 1040–1083. Tarawneh, B., 1992, The Geology of the Fifa Area, Bulletin 20, Geological Mapping Division, Geology Directorate, Natural Resources Authority, Jordan, 44 pp. Wallace, R.E., 1970, Earthquake recurrence intervals on the San Andreas fault, California, Geol. Soc. Am. Bull. 81, 2875–2890. Wells, D.L. and Coppersmith, K.J., 1994, Empirical relationships among magnitude, rupture length, rupture area, and surface displacement, Bull. Seism. Soc. Am. 84, 974–1002. Wetzel, R. and Morton, D.M., 1959, Contribution a la geologie de la Transjordan, Noteset Mem. Moyen Orient. 7, 95–191. Yeats, R.S., Sieh, K. and Allen C.R., 1997, The Geology of Earthquakes, Oxford University Press, New York, 568 pp. Yechieli, Y., Magaritz, M., Levy, Y., Weber, U., Kafri, U., Woelfli, W. and Bonani, G., 1993, Late Quaternary geological history of the Dead Sea area, Israel, Quat. Res. 39, 59–67. Zak, I. and Freund, R., 1966, Recent strike-slip movements along the Dead Sea rift, Isr. J. Earth Sci. 15, 33–37. Zhang, H., 1998, Late Pleistocene and Holocene Slip Rate of the Northern Wadi Araba Fault, Dead Sea Transform, Jordan, University of Missouri – Kansas City, Master’s thesis, 128 pp.