Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

The modest seismicity of the northern Red Sea rift

The modest seismicity of the northern Red Sea rift Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 Geophys. J. Int. (2018) 214, 1507–1523 doi: 10.1093/gji/ggy176 Advance Access publication 2018 May 05 GJI Marine geosciences and applied geophysics 1 2 Neil C. Mitchell and Ian C. F. Stewart School of Earth and Environmental Sciences, University of Manchester, Williamson Building, Oxford Road, Manchester M13 9PL, United Kingdom. E-mail: neil.mitchell@manchester.ac.uk Stewart Geophysical Consultants Pty. Ltd, Adelaide, South Australia, Australia Accepted 2018 May 4. Received 2018 April 29; in original form 2017 November 29 SUMMARY Inferring tectonic movements from earthquakes (seismotectonics) relies on earthquakes faith- fully recording tectonic motions. In the northern half of the Red Sea, however, events of magnitude 5.0 and above are almost entirely absent from global catalogues, even though GPS −1 and other plate motion data suggest that the basin is actively rifting at ∼10 mm yr . Seismic moments computed here from event magnitudes contributed to the International Seismology Centre (ISC) suggest that the moment release rate is more than an order of magnitude smaller than for the southern Red Sea and for the Southwest Indian Ridge (SWIR), which is spreading at a comparable rate to the central Red Sea and is more remote from recording stations. A smaller moment release rate in the northern Red Sea might be anticipated from its smaller spreading rate, but seismic coupling coefficients, which account for spreading rate variations, are also one order of magnitude smaller than for the other two areas. We explore potential explanations for this apparently reduced seismicity. The northern Red Sea is almost continu- ously covered with thick evaporites and overlying Plio-Pleistocene sediments. These deposits may have reduced the thickness of the seismogenic layer, for example, by elevating lithosphere temperatures by a thermal blanketing effect or by leading to excess pore fluid pressures that reduce effective stress. The presence of subdued seismicity here implies that tectonic move- ments can in places be poorly recorded by earthquake data and requires that alternative data be sought when investigating the active tectonics of sedimented rifts in particular. Key words: Heat flow; Continental neotectonics; statistical seismology; mid-ocean ridge processes; paleoseismology; seismicity and tectonics. northern Red Sea lithosphere is more cold and rigid than it is in the INTRODUCTION southern Red Sea closer to the Afar mantle plume and that rigidity Earthquakes provide valuable information on the distribution, mag- has prevented the northern Red Sea proceeding to a sea floor spread- nitude and orientation of strain in the lithosphere (Scholz 2002), but ing stage. However, as we show below, the apparently suppressed the tectonic information is rendered incomplete by varied efficiency seismicity could instead be interpreted as indicating a locally thin of seismic release (seismic coupling) and by the possibility of some brittle layer and hence a weak lithosphere. Studies of this region’s faults slipping stably without generating major earthquakes (e.g. seismicity may therefore contribute to a more general geodynamic Steinbrugge et al. 1960). Characterizing the reduced seismicity in understanding of how continental rifting proceeds by identifying regions such as the northern Red Sea and investigating possible causes of lithospheric weakening. causes may ultimately provide general insight into how regions of The low seismicity in the Zagros Mountains of Iran illustrates suppressed seismicity arise and help us to consider how or whether why the Red Sea may be more generally important. Jackson & tectonic information can be reliably extracted from seismological McKenzie (1988) estimated that less than 15 per cent of plate- data in such regions. Knowledge of the conditions in such areas may tectonic shortening is accommodated co-seismically in an area also help to assess whether regions presently lacking seismicity are where the Hormuz evaporites reached 1–4 km in original depo- persistently aseismic or whether they are merely gaps with seismic sitional thickness (Jahani et al. 2007). Although the seismic strain potential and are hence more hazardous. estimate has since been revised to 30 per cent, centroid depths de- The Red Sea is considered to be one of the best examples of a rived from body wave inversion in the Southern Fold Belt occur rifted continental shield actively proceeding to the sea floor spread- within the lower part of the strata overlying the salt and not in ing stage of ocean basin formation (Buck et al. 1988). Relevant the basement underlying the sedimentary sequence (Nissen et al. to that issue, Cochran & Karner (2007) have suggested that the 2011; Allen et al. 2013; Nissen et al. 2014). Because the tectonic The Author(s) 2018. Published by Oxford University Press on behalf of The Royal Astronomical Society. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 1508 N.C. Mitchell and I.C.F. Stewart plates are converging under the Zagros, the low basement seismic- (2016) study, instead following the Solomon et al. (1988) method ity could potentially be caused by folding of the basement rather for reasons given below, so our values are not directly comparable than faulting (Nissen et al. 2011). In contrast, as the northern Red with those of the former study. Nevertheless, the results are useful Sea is extending, basement folding cannot explain the suppressed for comparing rates between the areas we have characterized. The seismicity there, so it forces us to seek mechanisms that lead to coupling is found to be at least one order of magnitude smaller in the widespread aseismic slip on faults. northern Red Sea than it is in the southern Red Sea and Southwest While small magnitude events have been recognized to occur in Indian Ridge, and coincides with a decrease in plate spreading rate −1 the northern Red Sea (Daggett et al. 1986;Gharib et al. 1997), to below 10 mm yr . This difference contrasts with global trends, as a few authors have remarked previously on the low frequency of seismic coupling typically increases toward slow spreading rates at magnitude M > 4.0 earthquakes compared with other active rifts divergent plate boundaries (Rundquist & Sobolev 2002; Frohlich & (Fairhead & Girdler 1970; El-Isa & Al Shanti 1989;Al-Amri 1995; Wetzel 2007), as would be expected from axial lithosphere increas- Al-Almadi et al. 2014;Zahran et al. 2016). In one earlier study, this ingly thickened by conductive cooling. This extremely low coupling was attributed to a poor distribution of seismic stations in the region in the northern Red Sea casts doubt on the ability of earthquakes to (Fairhead & Girdler 1970). record tectonic motions in sedimented rifts unless corroborated by Al-Amri (1995) speculated that the apparently limited seismic- independent data, such as from space geodesy. ity could be due to deformation occurring outside the Red Sea rift. However, although some outcrops have been interpreted as suggest- ing intraplate deformation within Arabia (e.g. Schettino et al. 2016), DATA SETS AND METHODS there is little seismicity outside the rift (Fig. 1). GPS measurements have been made around the Red Sea since 1992 (Mahmoud et al. Seismic data 2005; McClusky et al. 2003, 2010; ArRajehl et al. 2010; Reilinger et al. 2015). Reilinger et al. (2015) described the most recent eval- Event data uation of continuous and survey mode measurements carried out at the sites located in Fig. 1. Their site velocities within Arabia show Earthquakes of all magnitudes reported for the Red Sea region oc- no deformation of the plate that is resolvable within their 1σ uncer- curring from 1964 January to 2016 July were obtained from the ISC −1 tainties of 0.4–1.0 mm yr (which are small compared with the Red event catalogue including events not reviewed by the ISC (On-line −1 Sea spreading velocities of 10–16 mm yr , Chu and Gordon 1998). Bulletin, http://www.isc.ac.uk, Internat. Seismol. Cent., Thatcham, There have been fewer measurements made on the western side of United Kingdom, 2016; Willemann & Storchak 2001). The spatial the Red Sea, though measurements on the western side of the Gulf distribution of events of magnitude (typically m ) 4.0 and above is of Suez in Mahmoud et al. (2005) show no significant movement shown in Fig. 1, along with events for which seismic moments were of those sites relative to the rest of Nubia. Reilinger et al. (2015) available from the Global Centroid-Moment-Tensor database (CMT, found that GPS-measured rates of movement across the northern http://www.globalcmt.org/; Dziewonski & Woodhouse 1983). Fo- Red Sea were within 10 per cent of geologic rates, whereas their cal mechanisms of the CMT database (not shown) suggest normal measurements in the southern Red Sea were within 6 per cent of faulting events within the Red Sea, except around and south of 17 geologic rates. The lack of seismicity within the northern Red Sea N, where strike-slip events are associated with the boundaries of the is therefore not caused by the deformation occurring significantly Danakil tectonic plate (Chu & Gordon 1998) [a strike-slip mech- ◦ ◦ outside the sea. anism reported for a 1967 event at 19.79 N, 38.82 E(Fairhead This study has been made possible by the submission of lo- &Girdler 1970) has not since been repeated at that latitude]. For cal magnitude data from the Saudi Geological Survey (SGS) to comparison, we have obtained similar event and CMT data for the the International Seismology Centre (ISC). From around 2004, the Southwest Indian Ridge (Supporting Information Fig. S1). mainly university-run stations were supplemented with 13 broad- One event (1967 March 28) had an extreme m = 6.7. Although band stations (Endo et al. 2007) and the Saudi Arabian network has such a magnitude is physically possible [the seismic moment cal- culated using the procedure below of 4.9 × 10 Nm could occur since been extended to more than 200 stations distributed across the kingdom. The utility of the data to studies of the Red Sea on a fault ∼100 km long (Scholz 1994), which is within the spatial rift is somewhat compromised, as the sea lies outside the network, scale of segmentation of the Red Sea free air gravity field (Mitchell making epicentre locations uncertain by amounts that vary with lo- &Park 2014; Mitchell 2015)], it appears anomalous compared with cation and magnitude (Stewart 2007). The velocity model used for the other seismicity data. Information on this event originally sub- locations has varied over time and is more appropriate for shield mitted to the ISC by the US Geological Survey (USGS) suggests lithosphere. Attenuation-distance relationships were also developed that only one seismometer recording was used in estimating this for the Arabian shield so magnitudes may have a bias that is difficult magnitude and the USGS no longer lists the event on their web- to quantify, though potentially of up to a few tenths of a magnitude site (www.usgs.gov). The seismic coupling calculations described unit. Nevertheless, more than 1000 events have been contributed by below were therefore carried out without this event. the SGS to the ISC catalogue and represent a significant resource, According to Bergman & Solomon (1984), deformation about which we examine here. mid-ocean ridges generating observable seismicity is concentrated In the following, we describe our calculations of seismic moments in oceanic lithosphere up to 15 m.y. old. The red and blue polygons from magnitudes in the ISC catalogue. We then estimate rates of in Fig. 1 outline crust that would be 15 m.y. or younger in the moment release. This allows us to estimate a minimal seismic cou- Red Sea if it were all oceanic and formed with the spreading rate pling (the proportion of plate tectonic movement occurring during and direction of the central Red Sea according to Chu & Gordon earthquakes) for the first time for the Red Sea, following similar (1998). Although the nature of the crust in the Red Sea has been calculations in earlier studies of mid-ocean ridges (Solomon et al. controversial and not necessarily all oceanic (Bonatti 1985; Cochran 1988;Cowie et al. 1993; Olive & Escart´ ın 2016). Our calculations 2005; Cochran & Karner 2007; Ligi et al. 2011; Ligi et al. 2012; are somewhat different from those used in the Olive & Escart´ ın Augustin et al. 2014; Mitchell & Park 2014), we find these polygons Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 Northern Red Sea aseismicity 1509 Figure 1. Seismicity of the Red Sea region. Red-filled circles scaled by magnitude (key) represent event locations obtained from the International Seismological Centre for the period 1964–2016 (only M > 4.0 shown). Black-filled squares locate seismic recording stations according to the ISC. Black-filled diamonds with white outlines are GPS stations of Reilinger et al. (2015). Location of the Danakil triple junction (TJ) is based on data in Chu & Gordon (1998). Light green fill in areas seaward of coastlines represent various indications of exposed basement rocks from sonar data (see main text) and igneous and metamorphic rocks reported in PetDB (www.geomapapp.org) and by Altherr et al. (1988). Background hill-shaded topography in this and Figs 6 and 8 is from the Becker et al. (2009) compilation. Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 1510 N.C. Mitchell and I.C.F. Stewart usefully encompass most seismicity within the rift. The polygon thresholds of data sets from the different regions and reporting or- edges also run close to the GPS stations shown to be moving with ganizations, and it allows a simple adjustment for incompleteness Arabia and Nubia (Fig. 1; Reilinger et al. 2015). explained below. The along-rift variations in magnitudes and moments are shown The grey dashed line in Fig. 5(a) is a least squares regression over in Fig. 2(a), in which ISC events and CMTs within the polygons the data range shown and is reproduced in Fig. 5(b) for comparison. ◦ ◦ ◦ ◦ were projected onto a line from 14 N, 42.5 Eto28 N, 34 E(zero The b-value implied by this line is 0.89. Similar plots were made distance corresponds to a point southeast of the Red Sea). Latitudes for the Red Sea over the full period and a restricted period of 2005– given along the top were computed for points along the Red Sea 2016 as shown. Dashed regression lines were similarly computed axis and the vertical grey bar locates the Danakil Triple Junction. A over the ranges of the data shown. Other features of Fig. 5 are similar graph was generated for the Southwest Indian Ridge (Fig. explained below. 2b) by projecting events within the 15 Ma polygon for the ridge ◦ ◦ ◦ ◦ from 25 S, 70 Eto41 S, 42 E [zero distance corresponds with the Rodrigues or Indian Ocean Triple Junction (Mitchell 1991)]. Seismic moments ISC event occurrences within the Red Sea polygons are shown by Converting magnitudes to seismic moments is complicated by the distance and date in Fig. 3, colour-coded by magnitude to highlight diversity of magnitude types. In their study of seismic coupling on varied reporting of the small magnitude events to the ISC. Fig. 4 the Mid-Atlantic Ridge, Olive & Escart´ ın (2016) used a relation similarly reveals the varied reporting of events for the northern Red of Das et al. (2011) to estimate moment magnitudes (M ) from Sea polygon by magnitude range over time. The increase with time w in the 2466 reported 1.0 < M < 3.0 events is mostly due to the body-wave magnitudes (m ), attempting to correct the tendency for improvements in the Saudi Arabian networks, with responsibility magnitudes to be underestimated from body waves at large mag- passing from the Saudi National Seismic Network (764 events) to nitude. However, under-reporting of small M in the ISC database the SGS (1497 events) in 2007 (Endo et al. 2007). Other authors likely biases such relations (Gasperini et al. 2013). We chose to use included the King Saud University, National Research Institute of the following relations to estimate M , which are from Lolli et al. Astronomy and Geophysics (NRIAG), Helwan, Cairo, Egypt, the (2014) and account for European Mediterranean moment-tensor Geophysical Institute of Israel and Institute for Petroleum Research data extending to smaller magnitude than the global data sets. Only and Geophysics, Israel. The majority of magnitudes are listed as M = 3.5 and above were adjusted. Other magnitude types were local (M ), although some other types were also reported in this considered sufficient to represent M for the purpose of our study L w range (52 M ,62 M ,2 M ). Of the 56 events of M ≥ 4.0, the and were left unchanged. b d m majority (42) were M , with lesser M and M The author of those b L d. M = exp (2.133 + 0.063M ) − 6.205, (2a) w s events was mainly the ISC (38 events). Events of 3.0 < M < 4.0 have varied types and originating institutions. M = exp (0.719 + 0.212m ) − 0.737. (2b) w b M was then converted to seismic moment (M in Nm) using the w 0 Magnitude-frequency distributions Hanks & Kanamori (1979) relation: Magnitude-frequency incremental distributions of the seismicity in log M = 1.5 (M + 6.033) . (3) 10 o w each polygon are shown in Fig. 5. Event frequencies (every 0.2 magnitude interval) were computed for all magnitude types and A similar equation to eq. (3) was found by Bakun (1984)us- normalized by measurement period and polygon area as shown. For ing a wide range of local magnitudes from southern California the Southwest Indian Ridge, a distribution was computed for data (3 < M < 6). The equation is also similar to the recommendations covering only the eastern end of the ridge, where major transform of Bormann and Dewey (2014), differing only by the constant 6.033, faults are absent (Sauter & Cannat 2010) because this may be the which their equation suggests would be 6.0667 in their scheme. A best analogue for the central and southern Red Sea where multibeam 6.0667 constant would change the derived moments by a uniform and gravity data suggest a lack of large transform faults (Augustin factor of 1.12. Hence, the seismic coupling values derived below et al. 2014; Mitchell & Park 2014; Augustin et al. 2016). The north may be underestimated by 12 per cent, although this bias has no and central Red Sea also lack strike-slip focal mechanisms in the effect on our comparisons of moment release between areas. CMT catalogue. ◦ ◦ The estimated seismic moments were summed within 0.1 X 0.1 Many earthquake cumulative distributions obey a law of the form geographical cells to produce the map in Fig. 6. Note that the greater (Gutenberg & Richter 1954): reporting of small events by the SGS in the northern Red Sea gives the impression of a broad zone of low activity (blue-green colours Log (N /t ) = a − bM, (1) in the figure) that is absent in the southern Red Sea, but its absence where N/t is the cumulative frequency of events of magnitude M in the latter is more a result of underrecording of small events (the (frequency of events of that magnitude and larger). Equation (eq. 1) scale in Fig. 6 is logarithmic—notice that the higher cumulative represents an exponential distribution, so (N/t) can be represented by moments in red and orange form a comparably narrow band around a cumulative probability density function P(m > M) that is propor- the spreading axis in the north as in the south). − β M tional to e ,where β= 2.3b (e.g. Ogata and Katsura 1993). Con- With event locations within the two polygons projected along- sequently, the (non-cumulative) probability density function p(M), rift as described earlier, the seismic moments of those events were which is obtained by differentiating P(m > M), is also proportional summed within 100 km along-axis increments to produce the graphs − β M to e . Therefore, if event data follow equation (eq. 1), their graph in Fig. 2(a) (connected turquoise circles). That is, the summed mo- of the logarithm of incremental frequencies against magnitude will ments in Fig. 2(a) represent total 1964–2016 moment release every also have a straight line with gradient −b. We prefer to use an incre- 100 km along the rift and within the polygons (Fig. 1). A similar mental distribution here as it more clearly highlights the detection calculation was carried out for the SWIR (Fig. 2b). Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 Northern Red Sea aseismicity 1511 Figure 2. ISC earthquake magnitudes (red symbols) are plotted versus along-rift distance for the Red Sea and Southwest Indian Ridge. Purple-filled stars with blue outlines represent seismic moments derived from the Global Centroid-Moment-Tensor database. Only data within the rift polygons marked in Figs 1 and Supporting Information Fig. S1 are shown. Distances of lines of latitudes and position of Zabargad Fracture Zone (ZFZ) in (a) were computed along the rift axis. Grey line in (a) is the approximate limit of the Danakil tectonic plate and triple junction according to Chu & Gordon (1998). Grey bands in (b) locate major fracture zones. Light blue-filled circles represent cumulative moment every 100-km along-rift derived from the magnitude data (see text for details). Seismic coupling Our approach has instead been to compute average seismic mo- ment release rates for the Red Sea simply from the ISC magnitude- The seismic coupling coefficient is defined as the ratio: derived moments and describe the coupling coefficient derived from them as ‘apparent’ to highlight that some unrecorded small- χ = , (4) ˙ magnitude seismicity is not accounted for. Nevertheless, the con- tribution of small magnitude events from the northern local seis- ˙ ˙ where M is the seismic moment release rate and M is the mo- 0 0 mometer arrays, but not in the south, suggests that the southern ment rate calculated from plate motions (Scholz 2002). The in- Red Sea and SWIR moment release rates are more greatly under- completeness of the seismic record causes problems in applying estimated (i.e. there is a greater downward bias from unreported eq. (4). Cowie et al. (1993) used the observation that total seismic small events in the south). This implies that the northern Red Sea moment release rate is typically about twice the moment released deficit of moment release compared with the south suggested by the by the largest earthquake in a region (Wesnousky et al. 1983)to averages in Fig. 2(a) is a useful minimum. We return to this issue adjust their estimate derived from the CMT catalogue. Olive & Es- again later. cart´ ın (2016) used a method that relies on the power-law slope and We have returned to the method used by Solomon et al. (1988)to range of the Guttenburg–Richter magnitude-frequency relationship estimate χ for slow-spreading ridges by assuming extension occurs to correct for incompleteness (Frohlich 2007). Given the difficul- ◦ on inward-dipping normal faults of average ∼45 dip (Huang et al. ties in characterizing the slope of the relationship here (Fig. 5)and 1986). Although low-angle faults have been found at the seabed only weak knowledge of maximum magnitude, we considered this over slow-spreading ridges (Cann et al. 1997), none has so far been unreliable in the Red Sea. Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 1512 N.C. Mitchell and I.C.F. Stewart Figure 3. Distribution of ISC events with along-rift distance by year colour-coded by magnitude for (a) northern and (b) southern Red Sea polygons (Fig. 1). Figure 4. Number of events reported for the northern Red Sea polygon by year and magnitude range, revealing increasing reporting of events below magnitude 4.0 with time but likely increasingly irregular reporting to the ISC. found in multibeam data from the Red Sea (Mitchell et al. 2010; was derived for situations in which all plate spreading is ac- Augustin et al. 2014), and in practice such faults are expected to commodated by slip on normal faults, ignoring aseismic slip evolve by slipping at steeper angles within the crust (Escart´ ın et al. and dyke intrusion (which both contribute to seismic coupling 2008). Using the relation between total moment of a shear zone deficits). and total mean slip on faults (Brune 1968) and rift fault geometry, To estimate v as a function of along-rift distance, we regressed the Solomon et al. (1988) showed that spreading full rates from Chu & Gordon (1998,fig. 9)asshown by the dashed line in Fig. 7(a). Combined with μ = 30 GPa, L = 10 m M = 4μLvh, (5) and h = 3000 m [from the half depth of mid-ocean ridge seismicity where μ is the shear modulus (usually assumed to be 30 GPa), as explained by Solomon et al. (1988)], we estimated M and thus L is length of deforming zone being characterized, v is seafloor the coupling coefficients χ shown with black symbols in Fig. 7(b). spreading half-rate and h is the seismogenic layer thickness. Eq. (5) Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 Northern Red Sea aseismicity 1513 A χ -value computed similarly for the easterly SWIR without major transform faults (red bar on left in Fig. 7b) is similar to that of the southern Red Sea. Heat flow data The conductivity and temperature gradient data in Fig. 8 [from the Hasterock (2013) compilation] were mainly derived from heat flow probe measurements, supplemented with a few borehole (mainly Deep Sea Drilling Project) data. Makris et al. (1991) noted the great variability between heat flow values, which is comparable with that around mid-ocean ridges and interpreted by them as caused by ad- vection of fluids within the sediments or crustal rocks. However, heat flow data are derived from measurements of sediment temper- ature gradient and conductivity, so heat flow variability could also arise from variability of the conductivity measurements. Outside the deeps, sediment profiler data typically reveal a simple draping sequence of reflections (Mitchell et al. 2015; Mitchell et al. 2017), suggesting that the sediment has a spatially uniform or gradually varying composition, rather than one with many abrupt changes except in the deeps. Thermal conductivity likely also varies gradu- ally. We therefore favour an approach similar to that of Martinez & Cochran (1989), who developed a smoothly varying model of sedi- ment conductivity to calculate heat flow, rather than use individual conductivity measurements. Martinez & Cochran (1989) showed one line of interpreted in- dustry seismic reflection data coinciding with one of their heat flow probe transects. Figs 9(a) and (b) show their data and their struc- tural interpretation. We estimated the temperature structure with a simple 1-D calculation assuming steady state and conductivities −1 ◦ −1 of 1.1 and 5.1 W m K for the Plio-Pleistocene sediments and evaporites, respectively, where the former value is an average measurement of Martinez & Cochran (1989) and the latter value is typical of halite (Wheildon et al. 1974). To estimate depths to the 600–800 C isotherms (Fig. 9c; we explain the significance of these isotherms later), we extrapolated temperature gradients using −1 ◦ −1 conductivity below the evaporites of 3.1 W m K , which has been used previously for oceanic lithosphere (Crosby & McKenzie 2009). The above procedure ignores the possibility of fluid circulation within or below the Red Sea sediments (Makris et al. 1991); the available sonar data in these areas do not show pockmarks which are otherwise potential evidence for fluid expulsions observed at one site in the eastern Red Sea (Feldens et al. 2016). Nevertheless, Figure 5. Magnitude-frequency incremental distributions of ISC events the basal evaporite temperatures in Fig. 9(a) are likely to be min- from (a) the Southwest Indian Ridge, (b) the Red Sea 1964–2016 and (c) ima. The depths to isotherms in Fig. 9(c) were calculated assuming the Red Sea 2005 onward. Events are counted every 0.2 magnitude unit. In that heat diffusion at these shallow depths is nearly steady state, (a), grey symbols represent counts of all events within 97 km north–south so geothermal gradients can be extrapolated using known conduc- distance from the ridge axis, while green symbols represent event counts for tivities. Continual uplift of basement with the ongoing extension, the easterly SWIR only where there is no major transform fault (Supporting however, could make geotherms non-steady state and hence tem- Information Fig. S1). Grey dashed line is a least-squares regression through perature gradients through constant conductivity bodies nonlinear. the easterly SWIR data (implying b = 0.89). In (b), blue and red symbols To evaluate this effect, we recorded the depth to the 780 C isotherm represent counts of all events with the ISC catalogue (blue and red dashed lines are least-squares regressions of these data over the magnitude ranges beneath sea level from the numerical rifting model of Buck et al. shown). Dashed grey line is the regression in (a) reproduced for comparison. (1988). Those depths are deeper rather than shallower than the Large red circle is average northern Red Sea frequency over 4 < M ≤ 5.0. In 800 C isotherm depths, so we concluded that rifting is too slow (c), the blue stars represent ISC events for only the restricted 2005–2016 pe- here to distort geotherms greatly away from steady state profiles. riod, whereas solid black and green squares represent frequencies for north and south Red Sea polygons derived from a full catalogue of 2005–2015 events recorded by the Saudi Geological Survey array. Lower blue circle Extents of evaporites is average of ISC events over 2.5 < M ≤ 3.5 and upper blue circle that frequency adjusted for incompleteness of the ISC record. With the average Hydrothermal circulation in oceanic basement typically involves (red circle) from (b), the latter implies b = 1.22 (dashed line). water entering or being expelled where basement outcrops above Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 1514 N.C. Mitchell and I.C.F. Stewart Figure 6. Spatial variations in cumulative seismic moment (Nm). Moment proxies derived from the ISC catalogue for 1964–2016 were summed within 0.1 X0.1 cells. Cross symbols are selected deeps from the compilation of Augustin et al. (2014). Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 Northern Red Sea aseismicity 1515 Figure 7. (a) Full spreading rate (solid circles) and cumulative seismic moment (turquoise-filled circles, from Fig. 2a) variations with distance along the Red Sea rift. Spreading rates are from Chu & Gordon (1998). Green circles show cumulative moment of the rift derived from the full catalogue of the SGS for 2000–2015 (moments X 52/15.5 to account for the shorter recording period; rates roughly coincide with those derived from the ISC catalogue in the northern Red Sea but are smaller in the southern Red Sea because of poor coverage there). Magenta circles show cumulative moment derived by combining data from the ISC catalogue for M > 3.5 and from the SGS 2000–2015 catalogue for M < 3.5 (scaled by 52/15.5 to allow for the shorter recording period). (b) Estimated seismic coupling based on the cumulative moments and the dashed along-rift regression line for spreading rates in (a). Horizontal blue lines are average coupling coefficients over the distance ranges shown. Magenta symbols were derived with the combined SGS and ISC data in (a). Red bar on left is coupling coefficient of the easternmost SWIR for comparison, computed as for the Red Sea. Figure 8. Sediment conductivity (a) and temperature gradient (b) at heat flow measurement sites from the compilation of Hasterok (2013). Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 1516 N.C. Mitchell and I.C.F. Stewart less permeable sediments (Stein & Stein 1994). Given the low per- of Aden spreading centres (Ayele & Hulhanek ´ 1997; Hofstetter meability of halite (Peach 1991), we would expect the structure of &Beyth 2003). Therefore, although these values are uncertain, the evaporites and bedrock in the Red Sea to strongly control circula- frequency distribution for the reported northern Red Sea events sug- tion and any pore fluid overpressure. We have therefore mapped out gests swarm-like behaviour similar to that of other volcanic rifts. the known extents of igneous basement or rock outcrops within the This is perhaps unsurprising given the morphologic, magnetic, heat sea as shown in Fig. 1 from the constraints shown in Supporting flow and rock sampling evidence of likely active volcanism in some Information Fig. S2 and other published data (Guennoc et al. 1988; of the northern Red Sea deeps, such as the Conrad, Shaban and Martinez & Cochran 1988). The axis of the Red Sea has been al- Mabahiss deeps (Cochran et al. 1986; Guennoc et al. 1988;Mar- most continuously surveyed south of 23 N with multibeam sonars tinez & Cochran 1989; Ehrhardt & Hubscher ¨ 2015). (Ligi et al. 2012; Augustin et al. 2014, 2016) so the distribution Figs 3 and 6 provide further details on the spatial distribution of outcrops is well constrained there. The distribution is less well- of events. The smaller magnitudes (M < 3.0) have been better ◦ ◦ known north of 23 N, although published multibeam and seismic recorded beyond 1200 km along-rift distance (north of 23 N). data sets provide some constraints (Cochran 1983, 2005; Pautot In Fig. 6, a band of events runs north from the Vema Deep to 1983; Cochran et al. 1986; Guennoc et al. 1988, 1990; Martinez & Kebrit Deep and Mabahiss Mons, although with subdued moment Cochran 1988; Miller & Barakat 1988;Richter et al. 1991; Rihm release around Kebrit Deep. Immediately north of Mabahiss Mons, et al. 1991; Haase et al. 2000; Ehrhardt et al. 2005; Gordon et al. there is a gap in seismicity of ∼50 km and Fig. 3 suggests that 2010; Ehrhardt & Hubscher ¨ 2015; Mitchell & Augustin 2017). We moment release is subdued for an additional 100 km or so. Beyond have evaluated the incidences of outcrops from those publications, there, another region of greater seismicity occurs around the Shaban as well as from the character of 3.5 kHz records from RV Conrad and Conrad deeps before the Aqaba Transform Fault and other cruise RC2507 archived at Lamont Doherty Earth Observatory, the seismicity associated with the Sinai plate (Fig. 1). distribution of islets (Bonatti et al. 1983;Taviani et al. 1984), and In Fig. 2, the moment release rate of the northern Red Sea is incidences of igneous rocks recovered in dredges in PetDB and from an order of magnitude or more smaller than for either the southern Altherr et al. (1988). Although we cannot rule out the existence of Red Sea or SWIR (comparing with the easterly SWIR similarly some unrecorded outcrops given that this database is not continu- lacking in transform fault activity). The release rate recorded over ous, we would be surprised if the unrecorded outcropping igneous 2000–2015 by the SGS network also shown in Fig. 7(a) (green rocks were greatly larger in extent than that mapped out in Fig. 1. circles) is similar in the northern Red Sea; smaller moments in the southern Red Sea are an artifact of weaker detection by the network in the south, as also suggested by the higher magnitude rollover in their distribution in Fig. 5(c). Some of the difference in SEISMOLOGICAL DATA ASSESSMENT moment release rate between the northern and southern Red Sea Assessment is complicated by the irregular reporting of events. The could be due to the smaller spreading rate of the north, which is spatial distribution of small events in Fig. 3,where M < 3.0 is closer to the pole of opening between Arabia and Nubia (Chu & almost absent for some periods along the whole northern Red Sea, Gordon 1998). However, the seismic coupling coefficients shown suggests that local events have not been reported for approximately with black symbols in Fig. 7(b), which account for the spreading half the post-2005 period. The incremental magnitude-frequency rate differences, are an order of magnitude smaller in the northern graph for the northern Red Sea (Fig. 5b) is irregular also because Red Sea than in the other two areas. The mean coefficients of the of the modest number of events, but is approximately linear for southern and northern Red Sea marked with the blue bars in Fig. M ≥ 2.5 (a least squares regression implies b = 0.76), with a 7(b) are 0.125 and 0.00562, respectively. rollover implying a detection threshold of approximately M = 2.0. Some researchers view the central Red Sea as a different province As the SGS network has recorded small magnitude events over from the northern Red Sea and in a state of transition to full seafloor the past decade, but the global network has done a better job of spreading with stretched continental crust underlying much of it recording larger events over the full period (1964 onward), we com- (Bonatti 1985; Ligi et al. 2011, 2012). We disagree with this division bined estimates of frequencies from the small and large extremes to as some data contradict the widespread presence of continental crust estimate an unbiased value of b. In Fig. 5(c), the lower blue circle in the central Red Sea (Egloff et al. 1991; Izzeldin 1987; Mitchell represents an average of log (N /t) computed over 2.5 ≤ M ≤ 3.5 and Park 2014) and perhaps even in the northern Red Sea (Dyment 10 i for 2005 onward from SGS events obtained from the ISC (where et al. 2013; Tapponnier et al. 2013). Given the small number of N are the interval frequencies). Further adjusting for the likely in- M > 4.0 events, we have combined all data from the northern half completeness of the SGS reported events with a factor of 11/5 (yr of the Red Sea also to ensure that the seismic coupling estimate is −1 yr ), we computed the mean frequency shown with the upper blue the most reliable. However, the average coupling coefficient of the circle in Fig. 5(c). The red circle represents the average log (N /t) northern Red Sea recalculated with only the four 100 km averages 10 i over 4.0 ≤ M ≤ 5.0 for 1964 onward computed from the full ISC in Fig. 7(b) north of 1300 km is 0.0071, which is still only 5.7 per catalogue. Joining the adjusted average frequency for M = 3tothat cent of that in the south. Hence, the precise selection of area for the for M = 4.5 implies a b-value of 1.22. For further verification, a calculation has little effect on our conclusion that the northern Red frequency distribution for the full SGS catalogue (including events Sea has anomalously low coupling. omitted from their ISC submission) also shown in Fig. 5(c) passes Could the apparent northern aseismicity be caused by small through the adjusted frequency and has a similar gradient. events that are poorly reported in the ISC catalogue? As the SGS This b = 1.22 is greater than 1.0 typical of faulting not asso- contributions to the ISC catalogue include more small-magnitude ciated with magmatism (e.g. Tongue et al. 1992) and towards the events in the northern Red Sea than in the southern Red Sea, the larger values of volcanic rifts experiencing swarm-like behaviour, coupling coefficients for the north are less biased by missing small for example, b = 1.7 for the 1978 Krafla eruption (Einarsson & events than the coefficients for the south. Hence, the difference be- Brandsdottir ´ 1980), b = 1.13 for the Ethiopian Rift (Keir et al. tween the regions is a minimum. This argument depends on the 2006)and b = 1.05–1.3 for the Afar, southern Red Sea and Gulf b-value not being greatly different between the two regions, which Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 Northern Red Sea aseismicity 1517 we expect as both are volcanic rifts and where b-values have been (1984). A more recent evaluation with an updated thermal model derived in the south from higher quality data (Ayele & Hulhanek ´ also suggested that 600 C limits earthquake depths in oceanic litho- 1997; Hofstetter & Beyth 2003;Keir et al. 2006)theyhavebeen sphere (McKenzie et al. 2005). comparably high to the b = 1.22 derived here for the north. How- To assess the potential depth extent of seismicity, the evaporites ever, to further assess this, Fig. 7(a) shows (magenta) cumulative are likely too weak to contribute, so depths to 600 and 800 Cbelow moments for the north derived by adding moments for the ISC the evaporites are more effective measures of schizosphere thick- events of M > 3.5 to moments of the SGS events of M < 3.5 (the ness. Estimates of those depths (Fig. 9c) are generally < 20 km and latter scaled by 52/15.5 to allow for the shorter recording period). in places much shallower. We have examined data from two further This allows for small events, which are well recorded to M = 2.0 areas. The easterly ends of the three long lines of heat flow mea- (Fig. 6). From these data, the average coupling coefficient marked surements collected on RV Conrad (Fig. 8b) coincide with seismic by the magenta bar in Fig. 7(b) is 0.007, which is still more than an refraction lines of Rihm et al. (1991). Although the seismic velocity order of magnitude smaller than that in the south. models were constructed from data from only a few ocean bottom seismometers, they were supported by industry seismic reflection data (Richter et al. 1991). Using the average depths of interfaces DISCUSSION—CAUSES OF THE where their models intersect the heat flow measurements and the MODEST SEISMICITY IN THE average of the heat flow data recorded at the easterly ends of the −2 NORTHERN RED SEA three RV Conrad lines in Fig. 8(b) (131 mW m ), the depth to the 800 C isotherm is estimated to be 14 km below the evaporites. This Low rates of seismic moment release at fast-spreading ridges have however is > 50 km distance NE of the rift axis. been suggested to occur because rapid spreading involves a verti- A further group of heat flow measurements was collected far- cally thinner schizosphere (the brittle zone where earthquakes nu- ther south, northeast of Zabargad Island (within the black-outlined cleate) than at slow-spreading ridges (Cowie et al. 1993). Where circle in Fig. 8b). Unfortunately, we have no information on the slower spreading ridges have elevated axes because of the effects of depth of basement or thickness of the evaporites here. If we assume mantle hotspots (Klein & Langmuir 1987), high lithospheric tem- thicknesses of the evaporites of 1–3km and that the Plio-Pleistocene peratures may also explain their low incidences of M > 5.0 events. layer has a typical thickness of 200 m (Mitchell et al. 2017), the However, some other slow- or intermediate-spreading ridges and 800 C isotherm lies 10–13 km beneath the evaporites. rifts also have low M > 5.0 incidences where they are not obviously Given that conductivities of the Plio-Pleistocene sediments are affected by hotspots, but are covered by thick sediments. Parts of the five times smaller than those of the evaporite minerals, their thick- Juan de Fuca Ridge away from Axial Seamount and the seamount ness is important for causing a thermal blanket effect. The thickness chains of the Endeavour Segment provide examples (Nedimovic ´ in the northern half of the Red Sea varies from ∼100 to 400 m, with et al. 2009). On the other hand, the continuation of the Gakkel a typical value of ∼200 m (Mitchell et al. 2017). These values are Ridge beneath sediments of the Russian eastern Arctic shelf is not similar to those in Fig. 9 so we might similarly expect the lithosphere obviously accompanied with diminished seismicity (Nikishin et al. to be hotter by up to ∼100 K more generally. 2017). If the thick sediments of the northern Red Sea are responsi- A 100 K effect of thermal blanketing ignores upper mantle tem- ble for the suppressed seismicity, understanding this area may help peratures. From lava geochemical data, Haase et al. (2000)inferred in understanding the role of sediments more broadly. In the fol- ◦ ◦ a ∼60 K decrease in the source region of axial lavas from 18 N lowing, we explore three mechanisms that may help to explain the to 26 N, opposing the blanketing effect. Their Na values (Na 8.0 apparently modest seismicity of the northern Red Sea. wt per cent corrected for low-temperature fractionation to 8 wt per cent MgO) were interpreted following the methods of Klein and Langmuir (1987). However, that decreasing melting temperature is Lithospheric temperature also expected to be accompanied by a decreasing mean pressure of The length-scale of along-rift segmentation of the gravity field in the melting of a few kb (Klein & Langmuir 1987), so it is unclear if the lithospheric geothermal gradient is changed greatly along the central and northern Red Sea of ∼50–100 km (Cochran & Karner 2007; Mitchell & Park 2014) is not especially different from other Red Sea by this effect. Furthermore, a recent mantle seismic S-wave rift basins or slow-spreading centres, so we do not expect lengths velocity model at 75 and 100 km depth (Chang & Van der Lee 2011; of faults to have limited the earthquake magnitudes. However, high Chang et al. 2011) shows an abrupt northward increase of ∼300 m −1 ◦ −1 lithospheric temperatures potentially limit the depth extent of seis- s beneath the Red Sea at 19 N and a modest ∼100 m s decrease mogenic fault movements, reducing the total fault area available to from ∼20 N toward the northernmost Red Sea. These variations break in an individual rupture. Scholz (2002) summarized knowl- suggest an abrupt decrease in temperature at 19 N and then more edge of temperature effects on the depth extents of seismogenic fault gradual increase northward, which are not obviously reflected in the movements. The transition at depth from unstable sliding to stable seismicity (Figs 1, 2 and 6) if mantle temperature variations were sliding, where plastic mechanisms dominate, depends on strain rate important to brittle lithosphere temperature structure. The ∼100 K as well as temperature. Given that strain rate in deep fault shear thermal blanketing effect is therefore likely more important than zones is poorly known, we follow earlier workers (Solomon et al. mantle temperature variations to lithospheric temperatures. 1988;Cowie et al. 1993) by instead using an apparent isotherm con- Depths to the 800 Cisothermshallowerthan10kminFig. 9(c) trol, based on the known depth extents of earthquakes at mid-ocean suggest that the lithosphere is extremely weak locally at Conrad ridges where temperature structures can be estimated from thermal Deep lying on the right end of that profile. Perhaps such extreme models. weakness and dyke injections (implied by a swarm-like b-value) In their study of mid-ocean ridge earthquakes using body wave- if occurring also at the other deeps in the northern Red Sea have form inversions, Bergman & Solomon (1984) found source depths suppressed tectonic stress more generally, leading to suppressed limited by the 800 C isotherm. In contrast, Cowie et al. (1993)used seismicity. Suppressed tectonic stress would also limit the potential 600 C as the limiting isotherm based on results of Wiens & Stein for bending stresses associated with major fault slip that have been Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 1518 N.C. Mitchell and I.C.F. Stewart Figure 9. Estimates of temperature structure based on data from Martinez & Cochran (1989) for the RV Conrad heat flow transect marked in Fig. 8(b) (distance is from westerly end of that line). (a) Heat flow data (cross symbols) and temperatures at the base of the evaporites assuming conductive heat loss (red circles, see text). (b) Structure interpreted by Martinez & Cochran (1989) from an industry multichannel seismic line almost coincident with the heat flow ◦ ◦ ◦ measurements. (c) Inferred depths to the 600 C and 800 C isotherms assuming conductive heat loss as explained in the text. Dashed line (780 C) is taken from a model of Buck et al. (1988). linked to seismicity in mid-ocean ridge segments with detachment magmatic fluids, reducing effective stress. There have been reports faults (Olive & Escart´ ın 2016). of overpressures experienced during drilling through evaporites and from geophysical data evaluations in the Red Sea and Gulf of Suez (R Swarbrick and MJR Gee, personal communication 2016). Fluid overpressures Evaporites almost continuously cover the basement across the The general correspondence between areas of the Red Sea with more northern Red Sea (Fig. 1). In the central Red Sea, larger areas extensive rock outcrops in Fig. 1 and seismicity leads us to consider of igneous basement are exposed, as illustrated by the multibeam whether fluid overpressures could also be involved in reducing the data in Fig. 10. However, as that figure shows, even in such areas, vertical thickness of the schizosphere. According to Cowie et al. the rift border faults marked in the figure are largely covered by (1993), faults in the shallow crust can be stabilized by low effective evaporite flows, leaving only small outcrops (Mitchell et al. 2010). normal stress and elevated fluid pressures. If the evaporites seal the A similar covering of faults is observed around Atlantis II Deep underlying formations, fluid pressures may increase with progres- (Augustin et al. 2014; Feldens & Mitchell 2015). Further south, sive compaction of underlying pre-evaporite sediments or injected greater lengths of exposed faults are observable in the multibeam Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 Northern Red Sea aseismicity 1519 Figure 10. Shaded relief image of example multibeam bathymetry data from the central Red Sea (artificial illumination from the NE; Mitchell et al. 2010;Ligi et al. 2011; Augustin et al. 2014). Annotation ‘F’ marks where two rift border faults have been covered by evaporite flows (Mitchell et al. 2010). Annotation ‘V’ marks volcanic geomorphology in the floor of the deeps. data (Augustin et al. 2014; Mitchell & Augustin 2017), coinciding of Atlantis II Deep (Feldens & Mitchell 2015). Instead, evaporites with greater seismicity (Fig. 1). appear generally to flow in a viscous-like manner (Mitchell et al. From the change in heat flow away from rock outcrops on the 2010). sedimented Juan de Fuca Ridge, length-scales of cooling by fluid circulation have been estimated to be ∼20 km (Davis et al. 1999). As Serpentinized upper mantle the pressures needed to drive thermal convection are much smaller than those needed to create overpressures significant compared Ravat et al. (2011) estimated the depth of the base of the magnetic with overburden stress, we expect much greater than 20 km for source layer below the mainly Egyptian northern Red Sea from sealing by evaporites to be effective and for overpressures to de- magnetic anomalies. Their depths in places exceed 15 km, and velop. This would seem to rule out this overpressure mechanism for ◦ around 25 N in the western Red Sea exceed 20 km. As this is some aseismic areas, particularly around the deeps. Furthermore, deeper than the Moho here (∼12–15 km below sea level; Gaulier loss of effective stress and hence strength would lead to extensive et al. 1988), Ravat et al. (2011) suggested that they imply that there slumping, which is not generally observed in multibeam and seis- has been widespread serpentinization of the upper mantle. The weak mic data, aside from a > 20-km-scale slump on the eastern side rheology of serpentinite (Escart´ ın et al. 1997a,b; Hirth et al. 1998) Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 1520 N.C. Mitchell and I.C.F. Stewart offers a potential explanation for the suppressed seismicity of the elsewhere within Arabia so the source was more likely landward northern Red Sea, which lies further from the Afar plume than the than seaward. No other historical events were reported around the southern Red Sea and therefore has a colder lithospheric mantle northern Red Sea coasts. Similarly, in his assessment of historical perhaps more prone to hydration (if we ignore the thermal blanket documents of events up to 1900, Ambraseys (2009) found no clear effect for the sake of argument). A colder lithosphere in the north evidence of an earthquake occurring within the northern Red Sea would also be associated with the slow plate spreading, which falls away from the Gulf of Suez and Aqaba fault. More speculatively, −1 below 10 mm yr (Fig. 7a). Interestingly, a year-long experiment Nof & Paldor (1992) considered whether the sea crossing by the Is- with ocean bottom seismometers on amagmatic spreading segments raelites mentioned in Exodus in the Bible was enabled by a tsunami of the similarly ultra-slow spreading SWIR revealed a remarkable but rejected the possibility as less likely than a meteorological cause lack of earthquakes in the upper 15 km of the lithosphere, which and in any case there are many difficulties in interpreting stories of was interpreted as due to serpentinization (Schlindwein & Schmid such antiquity (Segert 1994). Ambraseys (2009) suggested the his- 2016; Schmid & Schlindwein 2016). torical record is effective for events of about M > 6.5, hence the lack However, the varied magnetic character in the northern Red Sea of historical evidence represents around two millenia of no events is at least partly a result of the distribution of volcanic intrusions of M > 6.5 in the northern Red Sea. Although not ruling out smaller and extrusions (Cochran 2005) and laterally extensive shallower events as having occurred, the data mostly point to this area being magnetic sources could also produce the long-wavelength anoma- of low seismic moment release, not a gap with seismic potential. lies interpreted by Ravat et al. (2011) as due to deep sources. Un- fortunately, there are few alternative data that can corroborate the existence of serpentinite. Seismic refraction profile VI of Rihm CONCLUSIONS et al. (1991) overlaps the area of deep magnetic basement of Ravat The seismic catalogue obtained from the ISC was used to charac- −1 et al. (2011). Although 7.5 km s below the Moho in the veloc- terize the seismic coupling of the northern Red Sea, which is an ity model of Rihm et al. (1991) could be consistent with modest order of magnitude smaller than that of the southern Red Sea and serpentinization (Carlson & Miller 1997), the uncertainty of the the Southwest Indian Ridge, a similarly slow-spreading ridge but −1 7.5 km s velocity is unclear. The depths of serpentinization sug- farther from recording stations. Historical documents reviewed by gested by Ravat et al. (2011) overlap with the depths to the 600 others suggest that the lack of large-magnitude seismicity has per- and 800 C isotherms calculated earlier (Fig. 9c), implying that sisted for two millenia in the northern Red Sea. There is therefore serpentinite would not be stable. Furthermore, serpentinization is unlikely to be a gap here with seismic potential. likely to have affected only the upper lithosphere (Schlindwein & Contributions to the ISC catalogue by the Saudi Geological Sur- Schmid 2016; Schmid & Schlindwein 2016), leaving a thickened vey are temporally incomplete but suggest a detection threshold of lower lithosphere capable of generating significant moment release, around M = 2.0. After adjusting for incompleteness, those data as the seismicity data of the serpentinite-dominated SWIR suggest suggest a b-value for the northern Red Sea of 1.22, consistent with (Fig. 2b). Consequently, we reject serpentinization as an explanation swarm-like activity. This is typical of volcanic rifts and is consistent for the northern Red Sea aseismicity. with morphologic data showing volcanic features within the deeps in the north. The low seismic coupling of the northern Red Sea may have more A gap with seismic potential? than one cause. The Plio-Pleistocene and Miocene sediments are es- timated to elevate the lithospheric temperature by ∼100 K because Could the apparently low seismic moment release rate and cou- of a thermal blanketing effect, reducing the schizosphere thickness pling of the northern Red Sea be due to the still limited period thermally. The low permeabilities of the evaporites may have led to of instrument recordings, that is, could the northern Red Sea rep- pore fluid overpressure in the basement underlying them, although resent a major seismic gap that could be filled by a future large basement outcrops (where the evaporite seal is likely breached) magnitude event (McCann et al. 1979)? According to Fig. 7,the occur also in areas of low seismicity. Serpentinized upper mantle X8 seismic moment deficit over the period 1964–2016 is ∼10 may furthermore provide a partial solution, although serpentinite cells or ∼10 N.m. If released in a single event, eq. (3) suggests is stable only below 400 C and a cold lithosphere associated with the earthquake would have M = 6.6. Such an event could involve a serpentinization is most likely to be seismically active, as is the rupture length of ∼10–20 km according to intraplate normal events case on the Southwest Indian Ridge where serpentinites are com- compiled by Scholz et al. (1986). Such a rupture length lies within mon. Such low temperatures are also incompatible with the heat the range of fault block length implied by structures in the free-air flow data. We therefore favour a combination of the temperature gravity field (Mitchell & Park 2014). In reviewing historical evi- and pore pressure explanations for the suppressed seismicity. More dence of major earthquakes in the Middle East, Ambraseys (1970) generally, the strong suppression here suggests that caution should found that active faults typically go through short-lived phases of be exerted when interpreting the tectonics of active rifts where thick activity separated by 75–150 yr of quiescence, so perhaps the post sediments are present. 1964 instrumental record is still inadequate to rule out a gap with seismic potential in the northern Red Sea. However, this calculation likely underestimates the moment ACKNOWLEDGEMENTS deficit of the northern Red Sea if we consider the lack of significant seismicity shown in the earlier instrumental and historical records. Dick Swarbrick and Martin Gee are thanked for informal informa- The reassessment of 1900–1999 earthquakes of Ambraseys (2001) tion on overpressures from industry wells within the northern Red shows none within the northern Red Sea. Historical documents re- Sea. We thank Sang-Mook Lee for advice on conductivities used to viewed by Poirier & Taher (1980) suggest that one earthquake of calculate heat flow data and Sigurjon Jonsson ´ and Jillian Foulger Mercalli intensity VI was felt on the Red Sea coast of Arabia in for discussions on Red Sea seismicity. Giulio di Toro suggested 1068 at Sharm Yanbu, although it was felt with intensities up to IX looking at the Zagros literature. Sergey Sokolov provided Russian Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 Northern Red Sea aseismicity 1521 literature on the eastern Gakkel Ridge. Figures in this article were Carlson, R.L. & Miller, D.J., 1997. A new assessment of the abundance of serpentinite in the oceanic crust, Geophys. Res. Lett., 24, 457–460. produced with the aid of the GMT free software system (Wessel Chang, S.-J. & van der Lee, S., 2011. Mantle plumes and associated flow & Smith 1991). We thank the editor Gabi Laske, Marco Ligi and beneath Arabia and East Africa, Earth planet Sci. Lett., 302, 448–454. an anonymous reviewer for helpful comments that were useful in Chang, S.J., Merino, M., Van der Lee, S., Stein, S. & Stein, C.A., 2011. revising this article. Mantle flow beneath Arabia offset from the opening Red Sea, Geophys. Author contributions: NCM carried out the analysis and wrote Res. Lett., 38. the article. ICFS provided advice on the seismology and regional Chu, D. & Gordon, R.G., 1998. Current plate motions across the Red Sea, seismicity and commented on the article. Geophys. J. Int., 135, 313–328. Cochran, J., Martinez, F., Steckler, M.S. & Hobart, M.A., 1986. Conrad Deep: a northern Red Sea deep. Origin and implication for continental rifting, Earth planet. Sci. Lett., 78, 18–32. REFERENCES Cochran, J.R., 1983. A model for the development of the Red Sea, Am. Al-Almadi, K., Al-Amri, A. & See, L., 2014. A spatial statistical analysis of Assoc. Petrol. Geol. Bull., 67, 41–69. the occurrence of earthquakes along the Red Sea floor spreading: clusters Cochran, J.R., 2005. Northern Red Sea: nucleation of an oceanic spread- of seismicity, Arab. J. Geosci., 7, 2893–2904. ing center within a continental rift, Geochem. Geophys. Geosyst., 6, Al-Amri, A.M.S., 1995. Recent seismic activity in the northern Red Sea, J. doi:03010.01029/02004GC000826. Geodyn., 20, 243–253. Cochran, J.R. & Karner, G.D., 2007. Constraints on the deformation and Allen, M.B., Saville, C., Blanc, E.J.-P., Talebian, M. & Nissen, E., 2013. rupturing of continental lithosphere of the Red Sea: the transition from Orogenic plateau growth: expansion of the Turkish-Iranian Plateau across rifting to drifting, in Imaging, Mapping and Modelling Continental Litho- the Zagros fold-and-thrust belt, Tectonics, 32, 1–20. sphere Extension and Breakup, Spec. Publ. 282, pp. 265–289, eds Karner, Altherr, R., Henjes-Kunst, F., Puchelt, H. & Baumann, A., 1988. Volcanic G.D., Manatschal, G. & Pinheiro, L.M., Geological Society. activity in the Red Sea axial trough - evidence for a large mantle diapir? Cowie, P.A., Scholz, C.H., Edwards, M. & Malinverno, A., 1993. Fault strain Tectonophysics, 150, 121–133. and seismic coupling on mid-ocean ridges, J. geophys. Res., 948(B10), Ambraseys, N.N., 1970. Some characteristic features of the Anatolian Fault 17911–17920. Zone, Tectonophysics, 9, 143–165. Crosby, A.G. & McKenzie, D., 2009. An analysis of young ocean depth, Ambraseys, N.N., 2001. Reassessment of earthquakes, 1900–1999, in the gravity and global residual topography, Geophys. J. Int., 178, 1198–1219. Eastern Mediterranean and the Middle East, Geophys. J. Int., 145, 471– Daggett, P.H., Morgan, P., Boulos, F.K., Hennin, S.F., El-Sherif, A.A., El- Sayod, A.A., Basta, N.Z. & Melek, Y.S., 1986. Seismicity and active Ambraseys, N.N., 2009. Earthquakes in the Mediterranean and Middle East: tectonics of the Egyptian Red Sea margin and the northern Red Sea, A Multidisciplinary Study of Seismicity up to 1900, Cambridge University Tectonophysics, 125, 313–324. Press, 947 pp. Das, R., Wason, H.R. & Sharma, M.L., 2011. Global regression relations ArRajehi, A. et al., 2010. Geodetic constraints on present-day motion of for conversion of surface wave and body wave magnitudes to moment the Arabian Plate: implications for Red Sea and Gulf of Aden rifting, magnitude, Nat. Hazards, 59, 801–810. Tectonics, 29(3). Davis, E.E. et al., 1999. Regional heat flow variations across the sedimented Augustin, N., Devey, C.W., van der Zwan, F.M., Feldens, P., Tominaga, Juan de Fuca Ridge eastern flank: constraints on lithospheric cooling and M., Bantan, R. & Kwasnitschka, T., 2014. The transition from rifting to lateral hydrothermal heat transport, J. geophys. Res., 104, 17675–17688. spreading in the Red Sea, Earth planet. Sci. Lett., 395, 217–230. Dyment, J., Tapponnier, P., Afifi, A.M., Zinger, M.A., Franken, D. & Augustin, N., van der Zwan, F.M., Devey, C.W., Ligi, M., Kwasnitschka, T., Muzaiyen, E., 2013. A new seafloor spreading model of the Red Sea: Feldens, P., Bantan, R. & Basaham, A.S., 2016. Geomorphology of the magnetic anomalies and plate kinematics, American Geophysical Union, central Red Sea Rift: determining spreading processes, Geomorph., 274, Fall Meeting, abstract #T21A-2512. 162–179. Dziewonski, A.M. & Woodhouse, J.H., 1983. An experiment in systematic Ayele, A. & Hulhanek, ´ O., 1997. Spatial and temporal variation of the study of global seismicity - centroid-moment tensor solutions for 201 seismicity in the Horn of Africa from 1960–1993, Geophys. J. Int., 130, moderate and large earthquakes of 1981, J. geophys. Res., 88, 3247–3271. 805–810. Egloff, F. et al., 1991. Contrasting structural styles of the eastern and west- Bakun, W.H., 1984. Seismic moments, local magnitudes, and coda-duration ern margins of the southern Red Sea: the 1988 SONNE experiment, magnitudes for earthquakes of central California, Bull. seism. Soc. Am., 74, 439–458. Tectonophysics, 198, 329–353. Ehrhardt,A.&Hubscher, C., 2015. The northern Red Sea in transition from Becker, J.J. et al., 2009. Global bathymetry and elevation data at 30 arc ¨ rifting to drifting – lessons learned from ocean deeps, in The Red Sea: seconds resolution: SRTM30 PLUS, Mar. Geod., 32, 355–371. The Formation, Morphology, Oceanography and Environment of a Young Bergman, E.A. & Solomon, S.C., 1984. Source mechanisms of earthquakes Ocean Basin, pp. 99–121, eds Rasul, N.M.A. & Stewart, I.C.F., Springer near mid-ocean ridges from body waveform inversion: implications for Earth System Sciences. the early evolution of oceanic lithosphere, J. geophys. Res., 89, 11415– Ehrhardt, A., Hubscher ¨ , C. & Gajewski, D., 2005. Conrad Deep, Northern Red Sea: development of an early stage ocean deep within the axial Bonatti, E., 1985. Punctiform initiation of seafloor spreading in the Red Sea depression, Tectonophysics, 411, 19–40. during transition from a continental to an oceanic rift, Nature, 316, 33–37. Einarsson, P. & Brandsdottir ´ , B., 1980. Seismological evidence for latermal Bonatti, E. et al., 1983. Zabargad (St. John’s) Island: an uplifted fragment magma intrusion during the July 1978 deflation of the Krafla volcano in of sub-Red Sea lithosphere, J. geol. Soc. Lond., 140, 677–690. NE-Iceland, J. Geophys., 47, 160–165. Bormann, P. & Dewey, J.W., 2014. The new IASPEI standards for determin- El-Isa, Z.H. & Al Shanti, A., 1989. Seismicity and tectonics of the Red Sea ing magnitudes from digital data and their relation to classical magnitudes, and western Arabia, Geophys. J. Int., 97, 449–457. in New Manual of Seismological Observatory Practice 2 (NMSOP-2), pp. Endo, E., Zahran, H., Nofal, H. & Hadidy, S., 2007. The Saudi National 1–44, ed Bormann, P., Deutsches GeoForschungsZentrum GFZ. Seismic Network, Seismol. Res. Lett., 78, 439–445. Brune, J.N., 1968. Seismic moment, seismicity, and rate of slip along major Escart´ ın, J., Hirth, G. & Evans, B., 1997a. Effects of serpentinization on the fault zones, J. geophys. Res., 73, 777–784. lithospheric strength and the style of normal faulting at slow-spreading Buck, R., Martinez, F., Steckler, M.S. & Cochran, J.R., 1988. Thermal ridges, Earth planet. Sci. Lett., 151, 181–189. consequences of lithospheric extension: pure and simple, Tectonics, 7, Escart´ ın, J., Hirth, G. & Evans, B., 1997b. Non-dilatant brittle deformation 213–234. of serpentinites: implications for Mohr-Coulomb theory and the strength Cann, J.R. et al., 1997. Corrugated slip surfaces formed at ridge-transform of faults, J. geophys. Res., 102, 2897–2913. intersections on the Mid-Atlantic Ridge, Nature, 385, 329–332. Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 1522 N.C. Mitchell and I.C.F. Stewart Escart´ ın, J., Smith, D.K., Cann, J., Schouten, H., Langmuir, C.H. Klein, E.M. & Langmuir, C.H., 1987. Global correlations of ocean ridge & Escrig, S., 2008. Central role of detachment faults in accre- basalt chemistry with axial depth and crustal thickness, J. geophys. Res., tion of slow-spreading oceanic lithosphere, Nature, 455, 790–794. 92, 8089–8115. http://dx.doi.org/10.1038/nature07333 Ligi, M. et al., 2011. Initial burst of oceanic crust accretion in the Red Sea Fairhead, J.D. & Girdler, R.W., 1970. The seismicity of the Red Sea, Gulf of due to edge-driven mantle convection, Geology, 39, 1019–1022. Aden and Afar triangle, Phil. Trans. Roy. Soc., A267, 49–74. Ligi, M. et al., 2012. Birth of an ocean in the Red Sea: initial pangs, Feldens, P. & Mitchell, N.C., 2015. Salt flows in the central Red Sea, in The Geochem. Geophys. Geosyst., 13, Q08009. . Red Sea: The Formation, Morphology, Oceanography and Environment Lolli, B., Gasperini, P. & Vannucci, G., 2014. Empirical conversion between of a Young Ocean Basin, pp. 205–218, eds Rasul, N.M.A. & Stewart, teleseismic magnitudes (mb and Ms) and moment magnitude (Mw) at I.C.F. Springer Earth System Sciences. the Global, Euro-Mediterranean and Italian scale, Geophys. J. Int., 199, Feldens, P., Schmidt, M., Muck ¨ e, I., Augustin, N., Al-Farawati, R., Orif, M. 805–828. & Faber, E., 2016. Expelled subsalt fluids form a pockmark field in the Mahmoud, S., Reilinger, R., McClusky, S., Vernant, P. & Tealeb, A., 2005. eastern Red Sea, Geo-Mar. Lett., 36(5), 339–352. GPS evidence for northward motion of the Sinai Block: implications for Frohlich, C., 2007. Practical suggestions for assessing rates of seismic- E. Mediterranean tectonics, Earth planet. Sci. Lett., 238, 217–224. moment release, Bull. seism. Soc. Am., 97, 1158–1166. Makris, J., Tsironidis, J. & Richter, H., 1991. Heatflow density distribution Frohlich, C. & Wetzel, L.R., 2007. Comparison of seismic moment release in the Red Sea, Tectonophysics, 198, 383–393. rates along different types of plate boundaries, Geophys. J. Int., 171, Martinez, F. & Cochran, J.R., 1988. Structure and tectonics in the north- 909–920. ern Red Sea: catching a continental margin between rifting and drifting, Gasperini, P., Lolli, B. & Vannucci, G., 2013. Body-wave magnitude m is Tectonophysics, 150, 1–32. a good proxy of moment magnitude M for small earthquakes (M < Martinez, F. & Cochran, J.R., 1989. Geothermal measurements in the north- w b 4.5-5.0), Seismol. Res. Lett., 84, 932–937. ern Red Sea: implications for lithospheric thermal structure and mode of Gaulier, J.M., LePichon, X., Lyberis, N., Avedik, F., Gely, L., Moretti, I., extension during continental rifting, J. geophys. Res., 94, 12239–12265. Deschamps, A. & Hafez, S., 1988. Seismic study of the crustal thickness, McCann, W.R., Nishenko, S.P., Sykes, L.R. & Krause, J., 1979. Seismic gaps northern Red Sea and Gulf of Suez, Tectonophysics, 153, 55–88. and plate tectonics: seismic potential for major boundaries, Pageoph., 117, Gharib, A., Nicioglu, A. & Al-Amri, A., 1997. Seismicity re-evaluation of 1082–1147. the northern Red Sea, Geophys J. Int., 130, 335–348. McClusky, S., Reilinger, R., Mahmoud, S., Ben Sari, D. & Tealeb, A., 2003. Gordon, G., et al., 2010. The hydrocarbon prospectivity of the Egyptian GPS constraints on Africa (Nubia) and Arabia plate motions, Geophys. J. North Red Sea basin, in Petroleum Geology: From Mature Basins to New Int., 155, 126–138. Frontiers - Proceedings of the 7th Petroleum Geology Conference, pp McClusky, S. et al., 2010. Kinematics of the southern Red Sea-Afar Triple 783–789, eds Vining, B.A. & Pickering, S.C., Geol. Soc. Lond. Junction and implications for plate dynamics, Geophys. Res. Lett., 37, Guennoc, P., Pautot, G. & Coutelle, A., 1988. Surficial structures of the L05301. doi:05310.01029/02009GL041127 ◦ ◦ northern Red Sea axial valley from 23 Nto28 N: time and space McKenzie, D., Jackson, J. & Priestley, K., 2005. Thermal structure of oceanic evolution of neo-oceanic structures, Tectonophysics, 153, 1–23. and continental lithosphere, Earth planet. Sci. Lett., 233, 337–349. Guennoc, P., Pautot, G., LeQentrec, M.-F. & Coutelle, A., 1990. Structure of Miller, P.M. & Barakat, H., 1988. Geology of the Safaga Concession, north- an early oceanic rift in the northern Red Sea, Oceanol. Acta, 13, 145–157. ern Red Sea, Egypt, Tectonophysics, 153, 123–136. Gutenberg, B. & Richter, C.F., 1954. Seismicity of the Earth and Associated Mitchell, N.C., 1991. Distributed extension at the Indian Ocean Triple Junc- Phenomena, 2nd edn, Princeton University Press. tion, J. geophys. Res., 96(B5), 8019–8043. Haase, K.M., Muhe, ¨ R. & Stoffers, P., 2000. Magmatism during extension of Mitchell, N.C., 2015. Lineaments in gravity data of the Red Sea, in The Red the lithosphere: geochemical constraints from lavas of the Shaban Deep, Sea: The Formation, Morphology, Oceanography and Environment of a northern Red Sea, Chem. Geol., 166, 225–239. Young Ocean Basin, pp. 123–133, eds Rasul, N.M.A. & Stewart, I.C.F., Hanks, T.C. & Kanamori, H., 1979. A moment magnitude scale, J. geophys. Springer Earth System Sciences. Res., 84, 2348–2350. Mitchell, N.C. & Augustin, N., 2017. Halokinetics and other features of Hasterok, D., 2013. A heat flow based cooling model for tectonic plates, GLORIA long-range sidescan sonar data from the Red Sea, Mar. Pet. Earth planet. Sci. Lett., 361, 34–43. Geol., 88, 724–738. Hirth, G., Escartin, J. & Lin, J., 1998. The rheology of the lower oceanic Mitchell, N.C., Ligi, M., Farrante, V., Bonatti, E. & Rutter, E., 2010. Subma- crust: implications for lithospheric deformation at mid-ocean ridges, in rine salt flows in the central Red Sea, Geol. Soc. Am. Bull., 122, 701–713. Mitchell, N.C., Ligi, M., Feldens, P. & Hubscher, C., 2017. Deformation of a Faulting and Magmatism at Mid-Ocean Ridges, pp. 291–303, eds Buck, ¨ W.R., Delaney, P.T., Karson, J.A. & Lababrielle, Y., American Geophysical young salt giant: regional topography of the Red Sea Miocene evaporites, Union. Basin Res., 29, 352–369. Hofstetter, R. & Beyth, M., 2003. The Afar Depression: interpretation of Mitchell, N.C., Ligi, M. & Rohling, E.J., 2015. Red Sea isolation history the 1960–2000 earthquakes, Geophys. J. Int., 155, 715–732. suggested by Plio-Pleistocene seismic reflection sequences, Earth planet. Huang, P.Y., Solomon, S.C., Bergman, E.A. & Nabelek, J.L., 1986. Focal Sci. Lett., 430, 387–397. depths and mechanisms of Mid-Atlantic Ridge earthquakes from body Mitchell, N.C. & Park, Y., 2014. Nature of crust in the central Red Sea, waveform inversion, J. geophys. Res., 91, 579–598. Tectonophysics, 628, 123–139. Izzeldin, A.Y., 1987. Seismic, gravity and magnetic surveys in the central Nedimovic, ´ M.R., Bohnenstiehl, D.R., Carbotte, S.M., Canales, J.P. & Dziak, part of the Red Sea: their interpretation and implications for the structure R.P., 2009. Faulting and hydration of the Juan de Fuca plate system, Earth and evolution of the Red Sea, Tectonophysics, 143, 269–306. planet. Sci. Lett., 284, 94–102. Jackson, J. & McKenzie, D., 1988. The relationship between plate motions Nikishin, A.M., Petrov, E.I., Malyshev, N.A. & Ershova, V.P., 2017. Eurasia and seismic moment tensors, and the rates of active deformation in the Basin and Gakkel Ridge, Arctic Ocean: crustal asymmetry, ultra-slow Mediterranean and Middle East, Geophys. J., 93, 45–73. spreading and continental rifting revealed by new seismic data, Tectono- Jahani, S., Callot, J.-P., Lamotte, D.Fd., Letouzey, J. & Leturmy, P., 2007. physics, in press. The salt diapirs of the eastern Fars province (Zagros, Iran): a brief outline Nissen, E., Jackson, J., Jahani, S. & Tatar, M., 2014. Zagros “phantom of their past and present, in Thrust Belts and Foreland Basins, pp. 289– earthquakes” reassessed - the interplay of seismicity and deep salt flow 308, eds Lacombe, O., Roure, F., Lave, ´ J. & Verges, ´ J., Frontiers in Earth in the Simply Folded Belt? J. geophys. Res., 119, 3561–3583. Sciences, Springer. Nissen, E., Tatar, M., Jackson, J.A. & Allen, M.B., 2011. New views on Keir, D., Stuart, G.W., Jackson, A. & Ayele, A., 2006. Local earthquake earthquake faulting in the Zagros fold-and-thrust belt of Iran, Geophys. magnitude scale and seismicity rate for the Ethiopian Rift, Bull. seism. J. Int., 186, 928–944. Soc. Am., 96, 2221–2230. Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 Northern Red Sea aseismicity 1523 Nof, D. & Paldor, N., 1992. Are there oceanographic explanations for the Stein, C.A. & Stein, S., 1994. Constraints on hydrothermal heat flux through Israelites’ crossing of the Red Sea, Bull. Am. Meteorol. Soc., 73, 305–314. the oceanic lithosphere from global heat flow, J. geophys. Res., 99, 3081– Ogata, Y. & Katsura, K., 1993. Analysis of temporal and spatial hetero- 3095. geneity of magnitude frequency distribution inferred from earthquake Steinbrugge, K.V., Zacher, E.G., Tocher, D., Whitten, C.A. & Claire, C.N., catalogues, Geophys. J. Int., 113, 727–738. 1960. Creep on the San Andreas fault, Bull. Seism. Soc. Am., 50, 389–415. Olive, J.-A. & Escart´ ın, J., 2016. Dependence of seismic coupling on normal Stewart, I.C.F., 2007. Earthquake risk in western Saudi Arabia and the Red fault style along the Northern Mid-Atlantic Ridge, Geochem. Geophys. Sea from seismic moment, Saudi Geol. Surv., Technical Report SGS-TR- Geosyst., 17, 4128–4152. 2007-4, pp. 41. Pautot, G., 1983. Red Sea deeps – a geomorphological study by Seabeam, Tapponnier, P., Dyment, J., Zinger, M.A., Franken, D., Afifi, A.M., Wylie, Oceanol. Acta, 6, 235–244. A., Ali, H.G. & Hanbal, I., 2013. Revisiting seafloor-spreading in the Peach, C.J., 1991. Influence of deformation on the fluid transport properties Red Sea: basement nature, transforms and ocean-continent boundary, of salt rocks, Geologica Ultraiectina, University of Utrecht, Netherlands, American Geophysical Union, Fall Meeting 2013, San Francisco, CA, Vol. 77, pp. 1–238. 9–13 Dec, abstract #T12B-04. Poirier, J.P. & Taher, M.A., 1980. Historical seismicity in the Near and Taviani, M., Bonatti, E., Colantoni, P. & Rossi, P.L., 1984. Tectonically Middle East, North Africa, and Spain from Arabic documents (VIIth- uplifted crustal blocks in the northern Red Sea: data from the Brothers XVIIth Century), Bull. seism. Soc. Am., 70, 2185–2201. Islets, Mem. Soc. Geol. It., 27, 47–50. Ravat, D., Salem, A., Abdelaziz, A.M.S., Elawadi, E. & Morgan, P., 2011. Tongue, J.A., Maguire, P.K.H. & Burton, P., 1992. Seismicity distribution Probing magnetic bottom and crustal temperature variations along the from temporary earthquake recording networks in Kenya, Tectonophysics, Red Sea margin of Egypt, Tectonophysics, 510, 337–344. 204, 71–79. Reilinger, R., McClusky, S. & ArRajehi, A., 2015. Geodetic constraints on Wiens, D.A. & Stein, S., 1984. Intraplate seismicity and stresses in young the geodynamic evolution of the Red Sea, in The Red Sea: The Formation, oceanic lithosphere, J. geophys. Res., 89, 11442–11464. Morphology, Oceanography and Environment of a Young Ocean Basin, Wesnousky, S.G., Scholz, C.H., Shimazaki, K. & Matsuda, T., 1983. Earth- pp. 135–149, eds Rasul, N.M.A. & Stewart, I.C.F., Springer Earth System quake frequency distribution and the mechanics of faulting, J. geophys. Sciences. Res., 88, 9331–9340. Richter, H., Makris, J. & Rihm, R., 1991. Geophysical observations offshore Wessel, P. & Smith, W.H.F., 1991. Free software helps map and display data, Saudi Arabia: seismic and magnetic observations, Tectonophysics, 198, EOS, Trans. Am. geophys. Un., 72, 441–448. 297–310. Wheildon, J., Evans, T.R. & Girdler, R.W., 1974. Thermal conductivity, Rihm, R., Makris, J. & Moller ¨ , L., 1991. Seismic surveys in the northern density, and sonic velocity measurements of samples of anhydrite and Red Sea: asymmetric crustal structure, Tectonophysics, 198, 279–295. halite from Sites 225 and 227, in Initial Reports of the Deep Sea Drilling Rundquist, D.V. & Sobolev, P.O., 2002. Seismicity of mid-oceanic ridges Project, Vol. 23, pp. 909–911, eds Whitmarsh, R.B., Weser, O.E., Ross, and its geodynamic implications: a review, Earth-Sci. Rev., 58, 143–161. D.A., et al., U.S. Govt. Printing Office. Sauter, D. & Cannat, M., 2010. The ultraslow spreading Southwest Indian Willemann, R.J. & Storchak, D.A., 2001. Data collection at the International Ridge, in Diversity of Hydrothermal Systems on Slow Spreading Ocean Seismological Centre, Seismol. Res. Lett., 72, 440–453. Ridges, Geophysics Monograph Series 188, pp. 153–173 eds Rona, P.A., Zahran, H.M., Sokolov, V., Roobol, M.J., Stewart, I.C.F., El-Hadidy Youssef, Devey, C.W., Dyment, J. & Murton, B.J., American Geophysical Union. S. & El Hadidy, M., 2016. On the development of a seismic source Schettino, A., Macchiavelli, C., Pierantoni, P.P., Zanoni, D. & Rasul, N., zonation model for seismic hazard assessment in western Saudi Arabia, 2016. Recent kinematics of the tectonic plates surrounding the Red Sea J. Seismol., 20, 747–769. and Gulf of Aden, Geophys. J. Int., 207, 457–480. Schlindwein, V. & Schmid, F., 2016. Mid-ocean ridge seismicity re- veals extreme types of ocean lithosphere, Nature, 535, 276–279. http://dx.doi.org/10.1038/nature18277 SUPPORTING INFORMATION Schmid, F. & Schlindwein, V., 2016. Microearthquake activity, lithospheric Supplementary data are available at GJI online. structure, and deformation modes at an amagmatic ultraslow spread- Figure S1: Map of ISC events and Global CMT solutions for the ing Southwest Indian Ridge segment, Geochem. Geophys. Geosyst., 17, Southwest Indian Ridge, for the period 1964–2016. Symbols as Fig. 2905–2921. 1 of the main text, with bathymetry from the GeomapApp database Scholz, C.H., 1994. A reappraisal of large earthquake scaling, Bull. seism. Soc. Am., 84, 215–218. (Ryan et al. 2009). Black polygon outlines seafloor younger than Scholz, C.H., 2002. The Mechanics of Earthquakes and Faulting, Cambridge 15 Ma. Red polygon outlines part of this area that is without major University Press, 471 pp. transform faults. Scholz, C.H., Aviles, C. & Wesnousky, S.G., 1986. Scaling differences Figure S2: Spatial distribution of data used to infer the basement between large interplate and intraplate earthquakes, Bull. seism. Soc. Am., rock exposures in Fig. 1 of the main text. 76, 65–70. Please note: Oxford University Press is not responsible for the con- Segert, S., 1994. Crossing the waters: Moses and Hamilcar, J. Near East. tent or functionality of any supporting materials supplied by the Stud., 53, 195–203. authors. Any queries (other than missing material) should be di- Solomon, S.C., Huang, P.Y. & Meinke, L., 1988. The seismic moment budget rected to the corresponding author for the article. of slowly spreading ridges, Nature, 334, 58–60. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Geophysical Journal International Oxford University Press

The modest seismicity of the northern Red Sea rift

Loading next page...
1
 
/lp/ou_press/the-modest-seismicity-of-the-northern-red-sea-rift-zja6Hjfu4g

References (121)

Publisher
Oxford University Press
Copyright
Copyright © 2022 The Royal Astronomical Society
ISSN
0956-540X
eISSN
1365-246X
DOI
10.1093/gji/ggy176
Publisher site
See Article on Publisher Site

Abstract

Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 Geophys. J. Int. (2018) 214, 1507–1523 doi: 10.1093/gji/ggy176 Advance Access publication 2018 May 05 GJI Marine geosciences and applied geophysics 1 2 Neil C. Mitchell and Ian C. F. Stewart School of Earth and Environmental Sciences, University of Manchester, Williamson Building, Oxford Road, Manchester M13 9PL, United Kingdom. E-mail: neil.mitchell@manchester.ac.uk Stewart Geophysical Consultants Pty. Ltd, Adelaide, South Australia, Australia Accepted 2018 May 4. Received 2018 April 29; in original form 2017 November 29 SUMMARY Inferring tectonic movements from earthquakes (seismotectonics) relies on earthquakes faith- fully recording tectonic motions. In the northern half of the Red Sea, however, events of magnitude 5.0 and above are almost entirely absent from global catalogues, even though GPS −1 and other plate motion data suggest that the basin is actively rifting at ∼10 mm yr . Seismic moments computed here from event magnitudes contributed to the International Seismology Centre (ISC) suggest that the moment release rate is more than an order of magnitude smaller than for the southern Red Sea and for the Southwest Indian Ridge (SWIR), which is spreading at a comparable rate to the central Red Sea and is more remote from recording stations. A smaller moment release rate in the northern Red Sea might be anticipated from its smaller spreading rate, but seismic coupling coefficients, which account for spreading rate variations, are also one order of magnitude smaller than for the other two areas. We explore potential explanations for this apparently reduced seismicity. The northern Red Sea is almost continu- ously covered with thick evaporites and overlying Plio-Pleistocene sediments. These deposits may have reduced the thickness of the seismogenic layer, for example, by elevating lithosphere temperatures by a thermal blanketing effect or by leading to excess pore fluid pressures that reduce effective stress. The presence of subdued seismicity here implies that tectonic move- ments can in places be poorly recorded by earthquake data and requires that alternative data be sought when investigating the active tectonics of sedimented rifts in particular. Key words: Heat flow; Continental neotectonics; statistical seismology; mid-ocean ridge processes; paleoseismology; seismicity and tectonics. northern Red Sea lithosphere is more cold and rigid than it is in the INTRODUCTION southern Red Sea closer to the Afar mantle plume and that rigidity Earthquakes provide valuable information on the distribution, mag- has prevented the northern Red Sea proceeding to a sea floor spread- nitude and orientation of strain in the lithosphere (Scholz 2002), but ing stage. However, as we show below, the apparently suppressed the tectonic information is rendered incomplete by varied efficiency seismicity could instead be interpreted as indicating a locally thin of seismic release (seismic coupling) and by the possibility of some brittle layer and hence a weak lithosphere. Studies of this region’s faults slipping stably without generating major earthquakes (e.g. seismicity may therefore contribute to a more general geodynamic Steinbrugge et al. 1960). Characterizing the reduced seismicity in understanding of how continental rifting proceeds by identifying regions such as the northern Red Sea and investigating possible causes of lithospheric weakening. causes may ultimately provide general insight into how regions of The low seismicity in the Zagros Mountains of Iran illustrates suppressed seismicity arise and help us to consider how or whether why the Red Sea may be more generally important. Jackson & tectonic information can be reliably extracted from seismological McKenzie (1988) estimated that less than 15 per cent of plate- data in such regions. Knowledge of the conditions in such areas may tectonic shortening is accommodated co-seismically in an area also help to assess whether regions presently lacking seismicity are where the Hormuz evaporites reached 1–4 km in original depo- persistently aseismic or whether they are merely gaps with seismic sitional thickness (Jahani et al. 2007). Although the seismic strain potential and are hence more hazardous. estimate has since been revised to 30 per cent, centroid depths de- The Red Sea is considered to be one of the best examples of a rived from body wave inversion in the Southern Fold Belt occur rifted continental shield actively proceeding to the sea floor spread- within the lower part of the strata overlying the salt and not in ing stage of ocean basin formation (Buck et al. 1988). Relevant the basement underlying the sedimentary sequence (Nissen et al. to that issue, Cochran & Karner (2007) have suggested that the 2011; Allen et al. 2013; Nissen et al. 2014). Because the tectonic The Author(s) 2018. Published by Oxford University Press on behalf of The Royal Astronomical Society. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 1508 N.C. Mitchell and I.C.F. Stewart plates are converging under the Zagros, the low basement seismic- (2016) study, instead following the Solomon et al. (1988) method ity could potentially be caused by folding of the basement rather for reasons given below, so our values are not directly comparable than faulting (Nissen et al. 2011). In contrast, as the northern Red with those of the former study. Nevertheless, the results are useful Sea is extending, basement folding cannot explain the suppressed for comparing rates between the areas we have characterized. The seismicity there, so it forces us to seek mechanisms that lead to coupling is found to be at least one order of magnitude smaller in the widespread aseismic slip on faults. northern Red Sea than it is in the southern Red Sea and Southwest While small magnitude events have been recognized to occur in Indian Ridge, and coincides with a decrease in plate spreading rate −1 the northern Red Sea (Daggett et al. 1986;Gharib et al. 1997), to below 10 mm yr . This difference contrasts with global trends, as a few authors have remarked previously on the low frequency of seismic coupling typically increases toward slow spreading rates at magnitude M > 4.0 earthquakes compared with other active rifts divergent plate boundaries (Rundquist & Sobolev 2002; Frohlich & (Fairhead & Girdler 1970; El-Isa & Al Shanti 1989;Al-Amri 1995; Wetzel 2007), as would be expected from axial lithosphere increas- Al-Almadi et al. 2014;Zahran et al. 2016). In one earlier study, this ingly thickened by conductive cooling. This extremely low coupling was attributed to a poor distribution of seismic stations in the region in the northern Red Sea casts doubt on the ability of earthquakes to (Fairhead & Girdler 1970). record tectonic motions in sedimented rifts unless corroborated by Al-Amri (1995) speculated that the apparently limited seismic- independent data, such as from space geodesy. ity could be due to deformation occurring outside the Red Sea rift. However, although some outcrops have been interpreted as suggest- ing intraplate deformation within Arabia (e.g. Schettino et al. 2016), DATA SETS AND METHODS there is little seismicity outside the rift (Fig. 1). GPS measurements have been made around the Red Sea since 1992 (Mahmoud et al. Seismic data 2005; McClusky et al. 2003, 2010; ArRajehl et al. 2010; Reilinger et al. 2015). Reilinger et al. (2015) described the most recent eval- Event data uation of continuous and survey mode measurements carried out at the sites located in Fig. 1. Their site velocities within Arabia show Earthquakes of all magnitudes reported for the Red Sea region oc- no deformation of the plate that is resolvable within their 1σ uncer- curring from 1964 January to 2016 July were obtained from the ISC −1 tainties of 0.4–1.0 mm yr (which are small compared with the Red event catalogue including events not reviewed by the ISC (On-line −1 Sea spreading velocities of 10–16 mm yr , Chu and Gordon 1998). Bulletin, http://www.isc.ac.uk, Internat. Seismol. Cent., Thatcham, There have been fewer measurements made on the western side of United Kingdom, 2016; Willemann & Storchak 2001). The spatial the Red Sea, though measurements on the western side of the Gulf distribution of events of magnitude (typically m ) 4.0 and above is of Suez in Mahmoud et al. (2005) show no significant movement shown in Fig. 1, along with events for which seismic moments were of those sites relative to the rest of Nubia. Reilinger et al. (2015) available from the Global Centroid-Moment-Tensor database (CMT, found that GPS-measured rates of movement across the northern http://www.globalcmt.org/; Dziewonski & Woodhouse 1983). Fo- Red Sea were within 10 per cent of geologic rates, whereas their cal mechanisms of the CMT database (not shown) suggest normal measurements in the southern Red Sea were within 6 per cent of faulting events within the Red Sea, except around and south of 17 geologic rates. The lack of seismicity within the northern Red Sea N, where strike-slip events are associated with the boundaries of the is therefore not caused by the deformation occurring significantly Danakil tectonic plate (Chu & Gordon 1998) [a strike-slip mech- ◦ ◦ outside the sea. anism reported for a 1967 event at 19.79 N, 38.82 E(Fairhead This study has been made possible by the submission of lo- &Girdler 1970) has not since been repeated at that latitude]. For cal magnitude data from the Saudi Geological Survey (SGS) to comparison, we have obtained similar event and CMT data for the the International Seismology Centre (ISC). From around 2004, the Southwest Indian Ridge (Supporting Information Fig. S1). mainly university-run stations were supplemented with 13 broad- One event (1967 March 28) had an extreme m = 6.7. Although band stations (Endo et al. 2007) and the Saudi Arabian network has such a magnitude is physically possible [the seismic moment cal- culated using the procedure below of 4.9 × 10 Nm could occur since been extended to more than 200 stations distributed across the kingdom. The utility of the data to studies of the Red Sea on a fault ∼100 km long (Scholz 1994), which is within the spatial rift is somewhat compromised, as the sea lies outside the network, scale of segmentation of the Red Sea free air gravity field (Mitchell making epicentre locations uncertain by amounts that vary with lo- &Park 2014; Mitchell 2015)], it appears anomalous compared with cation and magnitude (Stewart 2007). The velocity model used for the other seismicity data. Information on this event originally sub- locations has varied over time and is more appropriate for shield mitted to the ISC by the US Geological Survey (USGS) suggests lithosphere. Attenuation-distance relationships were also developed that only one seismometer recording was used in estimating this for the Arabian shield so magnitudes may have a bias that is difficult magnitude and the USGS no longer lists the event on their web- to quantify, though potentially of up to a few tenths of a magnitude site (www.usgs.gov). The seismic coupling calculations described unit. Nevertheless, more than 1000 events have been contributed by below were therefore carried out without this event. the SGS to the ISC catalogue and represent a significant resource, According to Bergman & Solomon (1984), deformation about which we examine here. mid-ocean ridges generating observable seismicity is concentrated In the following, we describe our calculations of seismic moments in oceanic lithosphere up to 15 m.y. old. The red and blue polygons from magnitudes in the ISC catalogue. We then estimate rates of in Fig. 1 outline crust that would be 15 m.y. or younger in the moment release. This allows us to estimate a minimal seismic cou- Red Sea if it were all oceanic and formed with the spreading rate pling (the proportion of plate tectonic movement occurring during and direction of the central Red Sea according to Chu & Gordon earthquakes) for the first time for the Red Sea, following similar (1998). Although the nature of the crust in the Red Sea has been calculations in earlier studies of mid-ocean ridges (Solomon et al. controversial and not necessarily all oceanic (Bonatti 1985; Cochran 1988;Cowie et al. 1993; Olive & Escart´ ın 2016). Our calculations 2005; Cochran & Karner 2007; Ligi et al. 2011; Ligi et al. 2012; are somewhat different from those used in the Olive & Escart´ ın Augustin et al. 2014; Mitchell & Park 2014), we find these polygons Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 Northern Red Sea aseismicity 1509 Figure 1. Seismicity of the Red Sea region. Red-filled circles scaled by magnitude (key) represent event locations obtained from the International Seismological Centre for the period 1964–2016 (only M > 4.0 shown). Black-filled squares locate seismic recording stations according to the ISC. Black-filled diamonds with white outlines are GPS stations of Reilinger et al. (2015). Location of the Danakil triple junction (TJ) is based on data in Chu & Gordon (1998). Light green fill in areas seaward of coastlines represent various indications of exposed basement rocks from sonar data (see main text) and igneous and metamorphic rocks reported in PetDB (www.geomapapp.org) and by Altherr et al. (1988). Background hill-shaded topography in this and Figs 6 and 8 is from the Becker et al. (2009) compilation. Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 1510 N.C. Mitchell and I.C.F. Stewart usefully encompass most seismicity within the rift. The polygon thresholds of data sets from the different regions and reporting or- edges also run close to the GPS stations shown to be moving with ganizations, and it allows a simple adjustment for incompleteness Arabia and Nubia (Fig. 1; Reilinger et al. 2015). explained below. The along-rift variations in magnitudes and moments are shown The grey dashed line in Fig. 5(a) is a least squares regression over in Fig. 2(a), in which ISC events and CMTs within the polygons the data range shown and is reproduced in Fig. 5(b) for comparison. ◦ ◦ ◦ ◦ were projected onto a line from 14 N, 42.5 Eto28 N, 34 E(zero The b-value implied by this line is 0.89. Similar plots were made distance corresponds to a point southeast of the Red Sea). Latitudes for the Red Sea over the full period and a restricted period of 2005– given along the top were computed for points along the Red Sea 2016 as shown. Dashed regression lines were similarly computed axis and the vertical grey bar locates the Danakil Triple Junction. A over the ranges of the data shown. Other features of Fig. 5 are similar graph was generated for the Southwest Indian Ridge (Fig. explained below. 2b) by projecting events within the 15 Ma polygon for the ridge ◦ ◦ ◦ ◦ from 25 S, 70 Eto41 S, 42 E [zero distance corresponds with the Rodrigues or Indian Ocean Triple Junction (Mitchell 1991)]. Seismic moments ISC event occurrences within the Red Sea polygons are shown by Converting magnitudes to seismic moments is complicated by the distance and date in Fig. 3, colour-coded by magnitude to highlight diversity of magnitude types. In their study of seismic coupling on varied reporting of the small magnitude events to the ISC. Fig. 4 the Mid-Atlantic Ridge, Olive & Escart´ ın (2016) used a relation similarly reveals the varied reporting of events for the northern Red of Das et al. (2011) to estimate moment magnitudes (M ) from Sea polygon by magnitude range over time. The increase with time w in the 2466 reported 1.0 < M < 3.0 events is mostly due to the body-wave magnitudes (m ), attempting to correct the tendency for improvements in the Saudi Arabian networks, with responsibility magnitudes to be underestimated from body waves at large mag- passing from the Saudi National Seismic Network (764 events) to nitude. However, under-reporting of small M in the ISC database the SGS (1497 events) in 2007 (Endo et al. 2007). Other authors likely biases such relations (Gasperini et al. 2013). We chose to use included the King Saud University, National Research Institute of the following relations to estimate M , which are from Lolli et al. Astronomy and Geophysics (NRIAG), Helwan, Cairo, Egypt, the (2014) and account for European Mediterranean moment-tensor Geophysical Institute of Israel and Institute for Petroleum Research data extending to smaller magnitude than the global data sets. Only and Geophysics, Israel. The majority of magnitudes are listed as M = 3.5 and above were adjusted. Other magnitude types were local (M ), although some other types were also reported in this considered sufficient to represent M for the purpose of our study L w range (52 M ,62 M ,2 M ). Of the 56 events of M ≥ 4.0, the and were left unchanged. b d m majority (42) were M , with lesser M and M The author of those b L d. M = exp (2.133 + 0.063M ) − 6.205, (2a) w s events was mainly the ISC (38 events). Events of 3.0 < M < 4.0 have varied types and originating institutions. M = exp (0.719 + 0.212m ) − 0.737. (2b) w b M was then converted to seismic moment (M in Nm) using the w 0 Magnitude-frequency distributions Hanks & Kanamori (1979) relation: Magnitude-frequency incremental distributions of the seismicity in log M = 1.5 (M + 6.033) . (3) 10 o w each polygon are shown in Fig. 5. Event frequencies (every 0.2 magnitude interval) were computed for all magnitude types and A similar equation to eq. (3) was found by Bakun (1984)us- normalized by measurement period and polygon area as shown. For ing a wide range of local magnitudes from southern California the Southwest Indian Ridge, a distribution was computed for data (3 < M < 6). The equation is also similar to the recommendations covering only the eastern end of the ridge, where major transform of Bormann and Dewey (2014), differing only by the constant 6.033, faults are absent (Sauter & Cannat 2010) because this may be the which their equation suggests would be 6.0667 in their scheme. A best analogue for the central and southern Red Sea where multibeam 6.0667 constant would change the derived moments by a uniform and gravity data suggest a lack of large transform faults (Augustin factor of 1.12. Hence, the seismic coupling values derived below et al. 2014; Mitchell & Park 2014; Augustin et al. 2016). The north may be underestimated by 12 per cent, although this bias has no and central Red Sea also lack strike-slip focal mechanisms in the effect on our comparisons of moment release between areas. CMT catalogue. ◦ ◦ The estimated seismic moments were summed within 0.1 X 0.1 Many earthquake cumulative distributions obey a law of the form geographical cells to produce the map in Fig. 6. Note that the greater (Gutenberg & Richter 1954): reporting of small events by the SGS in the northern Red Sea gives the impression of a broad zone of low activity (blue-green colours Log (N /t ) = a − bM, (1) in the figure) that is absent in the southern Red Sea, but its absence where N/t is the cumulative frequency of events of magnitude M in the latter is more a result of underrecording of small events (the (frequency of events of that magnitude and larger). Equation (eq. 1) scale in Fig. 6 is logarithmic—notice that the higher cumulative represents an exponential distribution, so (N/t) can be represented by moments in red and orange form a comparably narrow band around a cumulative probability density function P(m > M) that is propor- the spreading axis in the north as in the south). − β M tional to e ,where β= 2.3b (e.g. Ogata and Katsura 1993). Con- With event locations within the two polygons projected along- sequently, the (non-cumulative) probability density function p(M), rift as described earlier, the seismic moments of those events were which is obtained by differentiating P(m > M), is also proportional summed within 100 km along-axis increments to produce the graphs − β M to e . Therefore, if event data follow equation (eq. 1), their graph in Fig. 2(a) (connected turquoise circles). That is, the summed mo- of the logarithm of incremental frequencies against magnitude will ments in Fig. 2(a) represent total 1964–2016 moment release every also have a straight line with gradient −b. We prefer to use an incre- 100 km along the rift and within the polygons (Fig. 1). A similar mental distribution here as it more clearly highlights the detection calculation was carried out for the SWIR (Fig. 2b). Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 Northern Red Sea aseismicity 1511 Figure 2. ISC earthquake magnitudes (red symbols) are plotted versus along-rift distance for the Red Sea and Southwest Indian Ridge. Purple-filled stars with blue outlines represent seismic moments derived from the Global Centroid-Moment-Tensor database. Only data within the rift polygons marked in Figs 1 and Supporting Information Fig. S1 are shown. Distances of lines of latitudes and position of Zabargad Fracture Zone (ZFZ) in (a) were computed along the rift axis. Grey line in (a) is the approximate limit of the Danakil tectonic plate and triple junction according to Chu & Gordon (1998). Grey bands in (b) locate major fracture zones. Light blue-filled circles represent cumulative moment every 100-km along-rift derived from the magnitude data (see text for details). Seismic coupling Our approach has instead been to compute average seismic mo- ment release rates for the Red Sea simply from the ISC magnitude- The seismic coupling coefficient is defined as the ratio: derived moments and describe the coupling coefficient derived from them as ‘apparent’ to highlight that some unrecorded small- χ = , (4) ˙ magnitude seismicity is not accounted for. Nevertheless, the con- tribution of small magnitude events from the northern local seis- ˙ ˙ where M is the seismic moment release rate and M is the mo- 0 0 mometer arrays, but not in the south, suggests that the southern ment rate calculated from plate motions (Scholz 2002). The in- Red Sea and SWIR moment release rates are more greatly under- completeness of the seismic record causes problems in applying estimated (i.e. there is a greater downward bias from unreported eq. (4). Cowie et al. (1993) used the observation that total seismic small events in the south). This implies that the northern Red Sea moment release rate is typically about twice the moment released deficit of moment release compared with the south suggested by the by the largest earthquake in a region (Wesnousky et al. 1983)to averages in Fig. 2(a) is a useful minimum. We return to this issue adjust their estimate derived from the CMT catalogue. Olive & Es- again later. cart´ ın (2016) used a method that relies on the power-law slope and We have returned to the method used by Solomon et al. (1988)to range of the Guttenburg–Richter magnitude-frequency relationship estimate χ for slow-spreading ridges by assuming extension occurs to correct for incompleteness (Frohlich 2007). Given the difficul- ◦ on inward-dipping normal faults of average ∼45 dip (Huang et al. ties in characterizing the slope of the relationship here (Fig. 5)and 1986). Although low-angle faults have been found at the seabed only weak knowledge of maximum magnitude, we considered this over slow-spreading ridges (Cann et al. 1997), none has so far been unreliable in the Red Sea. Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 1512 N.C. Mitchell and I.C.F. Stewart Figure 3. Distribution of ISC events with along-rift distance by year colour-coded by magnitude for (a) northern and (b) southern Red Sea polygons (Fig. 1). Figure 4. Number of events reported for the northern Red Sea polygon by year and magnitude range, revealing increasing reporting of events below magnitude 4.0 with time but likely increasingly irregular reporting to the ISC. found in multibeam data from the Red Sea (Mitchell et al. 2010; was derived for situations in which all plate spreading is ac- Augustin et al. 2014), and in practice such faults are expected to commodated by slip on normal faults, ignoring aseismic slip evolve by slipping at steeper angles within the crust (Escart´ ın et al. and dyke intrusion (which both contribute to seismic coupling 2008). Using the relation between total moment of a shear zone deficits). and total mean slip on faults (Brune 1968) and rift fault geometry, To estimate v as a function of along-rift distance, we regressed the Solomon et al. (1988) showed that spreading full rates from Chu & Gordon (1998,fig. 9)asshown by the dashed line in Fig. 7(a). Combined with μ = 30 GPa, L = 10 m M = 4μLvh, (5) and h = 3000 m [from the half depth of mid-ocean ridge seismicity where μ is the shear modulus (usually assumed to be 30 GPa), as explained by Solomon et al. (1988)], we estimated M and thus L is length of deforming zone being characterized, v is seafloor the coupling coefficients χ shown with black symbols in Fig. 7(b). spreading half-rate and h is the seismogenic layer thickness. Eq. (5) Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 Northern Red Sea aseismicity 1513 A χ -value computed similarly for the easterly SWIR without major transform faults (red bar on left in Fig. 7b) is similar to that of the southern Red Sea. Heat flow data The conductivity and temperature gradient data in Fig. 8 [from the Hasterock (2013) compilation] were mainly derived from heat flow probe measurements, supplemented with a few borehole (mainly Deep Sea Drilling Project) data. Makris et al. (1991) noted the great variability between heat flow values, which is comparable with that around mid-ocean ridges and interpreted by them as caused by ad- vection of fluids within the sediments or crustal rocks. However, heat flow data are derived from measurements of sediment temper- ature gradient and conductivity, so heat flow variability could also arise from variability of the conductivity measurements. Outside the deeps, sediment profiler data typically reveal a simple draping sequence of reflections (Mitchell et al. 2015; Mitchell et al. 2017), suggesting that the sediment has a spatially uniform or gradually varying composition, rather than one with many abrupt changes except in the deeps. Thermal conductivity likely also varies gradu- ally. We therefore favour an approach similar to that of Martinez & Cochran (1989), who developed a smoothly varying model of sedi- ment conductivity to calculate heat flow, rather than use individual conductivity measurements. Martinez & Cochran (1989) showed one line of interpreted in- dustry seismic reflection data coinciding with one of their heat flow probe transects. Figs 9(a) and (b) show their data and their struc- tural interpretation. We estimated the temperature structure with a simple 1-D calculation assuming steady state and conductivities −1 ◦ −1 of 1.1 and 5.1 W m K for the Plio-Pleistocene sediments and evaporites, respectively, where the former value is an average measurement of Martinez & Cochran (1989) and the latter value is typical of halite (Wheildon et al. 1974). To estimate depths to the 600–800 C isotherms (Fig. 9c; we explain the significance of these isotherms later), we extrapolated temperature gradients using −1 ◦ −1 conductivity below the evaporites of 3.1 W m K , which has been used previously for oceanic lithosphere (Crosby & McKenzie 2009). The above procedure ignores the possibility of fluid circulation within or below the Red Sea sediments (Makris et al. 1991); the available sonar data in these areas do not show pockmarks which are otherwise potential evidence for fluid expulsions observed at one site in the eastern Red Sea (Feldens et al. 2016). Nevertheless, Figure 5. Magnitude-frequency incremental distributions of ISC events the basal evaporite temperatures in Fig. 9(a) are likely to be min- from (a) the Southwest Indian Ridge, (b) the Red Sea 1964–2016 and (c) ima. The depths to isotherms in Fig. 9(c) were calculated assuming the Red Sea 2005 onward. Events are counted every 0.2 magnitude unit. In that heat diffusion at these shallow depths is nearly steady state, (a), grey symbols represent counts of all events within 97 km north–south so geothermal gradients can be extrapolated using known conduc- distance from the ridge axis, while green symbols represent event counts for tivities. Continual uplift of basement with the ongoing extension, the easterly SWIR only where there is no major transform fault (Supporting however, could make geotherms non-steady state and hence tem- Information Fig. S1). Grey dashed line is a least-squares regression through perature gradients through constant conductivity bodies nonlinear. the easterly SWIR data (implying b = 0.89). In (b), blue and red symbols To evaluate this effect, we recorded the depth to the 780 C isotherm represent counts of all events with the ISC catalogue (blue and red dashed lines are least-squares regressions of these data over the magnitude ranges beneath sea level from the numerical rifting model of Buck et al. shown). Dashed grey line is the regression in (a) reproduced for comparison. (1988). Those depths are deeper rather than shallower than the Large red circle is average northern Red Sea frequency over 4 < M ≤ 5.0. In 800 C isotherm depths, so we concluded that rifting is too slow (c), the blue stars represent ISC events for only the restricted 2005–2016 pe- here to distort geotherms greatly away from steady state profiles. riod, whereas solid black and green squares represent frequencies for north and south Red Sea polygons derived from a full catalogue of 2005–2015 events recorded by the Saudi Geological Survey array. Lower blue circle Extents of evaporites is average of ISC events over 2.5 < M ≤ 3.5 and upper blue circle that frequency adjusted for incompleteness of the ISC record. With the average Hydrothermal circulation in oceanic basement typically involves (red circle) from (b), the latter implies b = 1.22 (dashed line). water entering or being expelled where basement outcrops above Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 1514 N.C. Mitchell and I.C.F. Stewart Figure 6. Spatial variations in cumulative seismic moment (Nm). Moment proxies derived from the ISC catalogue for 1964–2016 were summed within 0.1 X0.1 cells. Cross symbols are selected deeps from the compilation of Augustin et al. (2014). Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 Northern Red Sea aseismicity 1515 Figure 7. (a) Full spreading rate (solid circles) and cumulative seismic moment (turquoise-filled circles, from Fig. 2a) variations with distance along the Red Sea rift. Spreading rates are from Chu & Gordon (1998). Green circles show cumulative moment of the rift derived from the full catalogue of the SGS for 2000–2015 (moments X 52/15.5 to account for the shorter recording period; rates roughly coincide with those derived from the ISC catalogue in the northern Red Sea but are smaller in the southern Red Sea because of poor coverage there). Magenta circles show cumulative moment derived by combining data from the ISC catalogue for M > 3.5 and from the SGS 2000–2015 catalogue for M < 3.5 (scaled by 52/15.5 to allow for the shorter recording period). (b) Estimated seismic coupling based on the cumulative moments and the dashed along-rift regression line for spreading rates in (a). Horizontal blue lines are average coupling coefficients over the distance ranges shown. Magenta symbols were derived with the combined SGS and ISC data in (a). Red bar on left is coupling coefficient of the easternmost SWIR for comparison, computed as for the Red Sea. Figure 8. Sediment conductivity (a) and temperature gradient (b) at heat flow measurement sites from the compilation of Hasterok (2013). Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 1516 N.C. Mitchell and I.C.F. Stewart less permeable sediments (Stein & Stein 1994). Given the low per- of Aden spreading centres (Ayele & Hulhanek ´ 1997; Hofstetter meability of halite (Peach 1991), we would expect the structure of &Beyth 2003). Therefore, although these values are uncertain, the evaporites and bedrock in the Red Sea to strongly control circula- frequency distribution for the reported northern Red Sea events sug- tion and any pore fluid overpressure. We have therefore mapped out gests swarm-like behaviour similar to that of other volcanic rifts. the known extents of igneous basement or rock outcrops within the This is perhaps unsurprising given the morphologic, magnetic, heat sea as shown in Fig. 1 from the constraints shown in Supporting flow and rock sampling evidence of likely active volcanism in some Information Fig. S2 and other published data (Guennoc et al. 1988; of the northern Red Sea deeps, such as the Conrad, Shaban and Martinez & Cochran 1988). The axis of the Red Sea has been al- Mabahiss deeps (Cochran et al. 1986; Guennoc et al. 1988;Mar- most continuously surveyed south of 23 N with multibeam sonars tinez & Cochran 1989; Ehrhardt & Hubscher ¨ 2015). (Ligi et al. 2012; Augustin et al. 2014, 2016) so the distribution Figs 3 and 6 provide further details on the spatial distribution of outcrops is well constrained there. The distribution is less well- of events. The smaller magnitudes (M < 3.0) have been better ◦ ◦ known north of 23 N, although published multibeam and seismic recorded beyond 1200 km along-rift distance (north of 23 N). data sets provide some constraints (Cochran 1983, 2005; Pautot In Fig. 6, a band of events runs north from the Vema Deep to 1983; Cochran et al. 1986; Guennoc et al. 1988, 1990; Martinez & Kebrit Deep and Mabahiss Mons, although with subdued moment Cochran 1988; Miller & Barakat 1988;Richter et al. 1991; Rihm release around Kebrit Deep. Immediately north of Mabahiss Mons, et al. 1991; Haase et al. 2000; Ehrhardt et al. 2005; Gordon et al. there is a gap in seismicity of ∼50 km and Fig. 3 suggests that 2010; Ehrhardt & Hubscher ¨ 2015; Mitchell & Augustin 2017). We moment release is subdued for an additional 100 km or so. Beyond have evaluated the incidences of outcrops from those publications, there, another region of greater seismicity occurs around the Shaban as well as from the character of 3.5 kHz records from RV Conrad and Conrad deeps before the Aqaba Transform Fault and other cruise RC2507 archived at Lamont Doherty Earth Observatory, the seismicity associated with the Sinai plate (Fig. 1). distribution of islets (Bonatti et al. 1983;Taviani et al. 1984), and In Fig. 2, the moment release rate of the northern Red Sea is incidences of igneous rocks recovered in dredges in PetDB and from an order of magnitude or more smaller than for either the southern Altherr et al. (1988). Although we cannot rule out the existence of Red Sea or SWIR (comparing with the easterly SWIR similarly some unrecorded outcrops given that this database is not continu- lacking in transform fault activity). The release rate recorded over ous, we would be surprised if the unrecorded outcropping igneous 2000–2015 by the SGS network also shown in Fig. 7(a) (green rocks were greatly larger in extent than that mapped out in Fig. 1. circles) is similar in the northern Red Sea; smaller moments in the southern Red Sea are an artifact of weaker detection by the network in the south, as also suggested by the higher magnitude rollover in their distribution in Fig. 5(c). Some of the difference in SEISMOLOGICAL DATA ASSESSMENT moment release rate between the northern and southern Red Sea Assessment is complicated by the irregular reporting of events. The could be due to the smaller spreading rate of the north, which is spatial distribution of small events in Fig. 3,where M < 3.0 is closer to the pole of opening between Arabia and Nubia (Chu & almost absent for some periods along the whole northern Red Sea, Gordon 1998). However, the seismic coupling coefficients shown suggests that local events have not been reported for approximately with black symbols in Fig. 7(b), which account for the spreading half the post-2005 period. The incremental magnitude-frequency rate differences, are an order of magnitude smaller in the northern graph for the northern Red Sea (Fig. 5b) is irregular also because Red Sea than in the other two areas. The mean coefficients of the of the modest number of events, but is approximately linear for southern and northern Red Sea marked with the blue bars in Fig. M ≥ 2.5 (a least squares regression implies b = 0.76), with a 7(b) are 0.125 and 0.00562, respectively. rollover implying a detection threshold of approximately M = 2.0. Some researchers view the central Red Sea as a different province As the SGS network has recorded small magnitude events over from the northern Red Sea and in a state of transition to full seafloor the past decade, but the global network has done a better job of spreading with stretched continental crust underlying much of it recording larger events over the full period (1964 onward), we com- (Bonatti 1985; Ligi et al. 2011, 2012). We disagree with this division bined estimates of frequencies from the small and large extremes to as some data contradict the widespread presence of continental crust estimate an unbiased value of b. In Fig. 5(c), the lower blue circle in the central Red Sea (Egloff et al. 1991; Izzeldin 1987; Mitchell represents an average of log (N /t) computed over 2.5 ≤ M ≤ 3.5 and Park 2014) and perhaps even in the northern Red Sea (Dyment 10 i for 2005 onward from SGS events obtained from the ISC (where et al. 2013; Tapponnier et al. 2013). Given the small number of N are the interval frequencies). Further adjusting for the likely in- M > 4.0 events, we have combined all data from the northern half completeness of the SGS reported events with a factor of 11/5 (yr of the Red Sea also to ensure that the seismic coupling estimate is −1 yr ), we computed the mean frequency shown with the upper blue the most reliable. However, the average coupling coefficient of the circle in Fig. 5(c). The red circle represents the average log (N /t) northern Red Sea recalculated with only the four 100 km averages 10 i over 4.0 ≤ M ≤ 5.0 for 1964 onward computed from the full ISC in Fig. 7(b) north of 1300 km is 0.0071, which is still only 5.7 per catalogue. Joining the adjusted average frequency for M = 3tothat cent of that in the south. Hence, the precise selection of area for the for M = 4.5 implies a b-value of 1.22. For further verification, a calculation has little effect on our conclusion that the northern Red frequency distribution for the full SGS catalogue (including events Sea has anomalously low coupling. omitted from their ISC submission) also shown in Fig. 5(c) passes Could the apparent northern aseismicity be caused by small through the adjusted frequency and has a similar gradient. events that are poorly reported in the ISC catalogue? As the SGS This b = 1.22 is greater than 1.0 typical of faulting not asso- contributions to the ISC catalogue include more small-magnitude ciated with magmatism (e.g. Tongue et al. 1992) and towards the events in the northern Red Sea than in the southern Red Sea, the larger values of volcanic rifts experiencing swarm-like behaviour, coupling coefficients for the north are less biased by missing small for example, b = 1.7 for the 1978 Krafla eruption (Einarsson & events than the coefficients for the south. Hence, the difference be- Brandsdottir ´ 1980), b = 1.13 for the Ethiopian Rift (Keir et al. tween the regions is a minimum. This argument depends on the 2006)and b = 1.05–1.3 for the Afar, southern Red Sea and Gulf b-value not being greatly different between the two regions, which Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 Northern Red Sea aseismicity 1517 we expect as both are volcanic rifts and where b-values have been (1984). A more recent evaluation with an updated thermal model derived in the south from higher quality data (Ayele & Hulhanek ´ also suggested that 600 C limits earthquake depths in oceanic litho- 1997; Hofstetter & Beyth 2003;Keir et al. 2006)theyhavebeen sphere (McKenzie et al. 2005). comparably high to the b = 1.22 derived here for the north. How- To assess the potential depth extent of seismicity, the evaporites ever, to further assess this, Fig. 7(a) shows (magenta) cumulative are likely too weak to contribute, so depths to 600 and 800 Cbelow moments for the north derived by adding moments for the ISC the evaporites are more effective measures of schizosphere thick- events of M > 3.5 to moments of the SGS events of M < 3.5 (the ness. Estimates of those depths (Fig. 9c) are generally < 20 km and latter scaled by 52/15.5 to allow for the shorter recording period). in places much shallower. We have examined data from two further This allows for small events, which are well recorded to M = 2.0 areas. The easterly ends of the three long lines of heat flow mea- (Fig. 6). From these data, the average coupling coefficient marked surements collected on RV Conrad (Fig. 8b) coincide with seismic by the magenta bar in Fig. 7(b) is 0.007, which is still more than an refraction lines of Rihm et al. (1991). Although the seismic velocity order of magnitude smaller than that in the south. models were constructed from data from only a few ocean bottom seismometers, they were supported by industry seismic reflection data (Richter et al. 1991). Using the average depths of interfaces DISCUSSION—CAUSES OF THE where their models intersect the heat flow measurements and the MODEST SEISMICITY IN THE average of the heat flow data recorded at the easterly ends of the −2 NORTHERN RED SEA three RV Conrad lines in Fig. 8(b) (131 mW m ), the depth to the 800 C isotherm is estimated to be 14 km below the evaporites. This Low rates of seismic moment release at fast-spreading ridges have however is > 50 km distance NE of the rift axis. been suggested to occur because rapid spreading involves a verti- A further group of heat flow measurements was collected far- cally thinner schizosphere (the brittle zone where earthquakes nu- ther south, northeast of Zabargad Island (within the black-outlined cleate) than at slow-spreading ridges (Cowie et al. 1993). Where circle in Fig. 8b). Unfortunately, we have no information on the slower spreading ridges have elevated axes because of the effects of depth of basement or thickness of the evaporites here. If we assume mantle hotspots (Klein & Langmuir 1987), high lithospheric tem- thicknesses of the evaporites of 1–3km and that the Plio-Pleistocene peratures may also explain their low incidences of M > 5.0 events. layer has a typical thickness of 200 m (Mitchell et al. 2017), the However, some other slow- or intermediate-spreading ridges and 800 C isotherm lies 10–13 km beneath the evaporites. rifts also have low M > 5.0 incidences where they are not obviously Given that conductivities of the Plio-Pleistocene sediments are affected by hotspots, but are covered by thick sediments. Parts of the five times smaller than those of the evaporite minerals, their thick- Juan de Fuca Ridge away from Axial Seamount and the seamount ness is important for causing a thermal blanket effect. The thickness chains of the Endeavour Segment provide examples (Nedimovic ´ in the northern half of the Red Sea varies from ∼100 to 400 m, with et al. 2009). On the other hand, the continuation of the Gakkel a typical value of ∼200 m (Mitchell et al. 2017). These values are Ridge beneath sediments of the Russian eastern Arctic shelf is not similar to those in Fig. 9 so we might similarly expect the lithosphere obviously accompanied with diminished seismicity (Nikishin et al. to be hotter by up to ∼100 K more generally. 2017). If the thick sediments of the northern Red Sea are responsi- A 100 K effect of thermal blanketing ignores upper mantle tem- ble for the suppressed seismicity, understanding this area may help peratures. From lava geochemical data, Haase et al. (2000)inferred in understanding the role of sediments more broadly. In the fol- ◦ ◦ a ∼60 K decrease in the source region of axial lavas from 18 N lowing, we explore three mechanisms that may help to explain the to 26 N, opposing the blanketing effect. Their Na values (Na 8.0 apparently modest seismicity of the northern Red Sea. wt per cent corrected for low-temperature fractionation to 8 wt per cent MgO) were interpreted following the methods of Klein and Langmuir (1987). However, that decreasing melting temperature is Lithospheric temperature also expected to be accompanied by a decreasing mean pressure of The length-scale of along-rift segmentation of the gravity field in the melting of a few kb (Klein & Langmuir 1987), so it is unclear if the lithospheric geothermal gradient is changed greatly along the central and northern Red Sea of ∼50–100 km (Cochran & Karner 2007; Mitchell & Park 2014) is not especially different from other Red Sea by this effect. Furthermore, a recent mantle seismic S-wave rift basins or slow-spreading centres, so we do not expect lengths velocity model at 75 and 100 km depth (Chang & Van der Lee 2011; of faults to have limited the earthquake magnitudes. However, high Chang et al. 2011) shows an abrupt northward increase of ∼300 m −1 ◦ −1 lithospheric temperatures potentially limit the depth extent of seis- s beneath the Red Sea at 19 N and a modest ∼100 m s decrease mogenic fault movements, reducing the total fault area available to from ∼20 N toward the northernmost Red Sea. These variations break in an individual rupture. Scholz (2002) summarized knowl- suggest an abrupt decrease in temperature at 19 N and then more edge of temperature effects on the depth extents of seismogenic fault gradual increase northward, which are not obviously reflected in the movements. The transition at depth from unstable sliding to stable seismicity (Figs 1, 2 and 6) if mantle temperature variations were sliding, where plastic mechanisms dominate, depends on strain rate important to brittle lithosphere temperature structure. The ∼100 K as well as temperature. Given that strain rate in deep fault shear thermal blanketing effect is therefore likely more important than zones is poorly known, we follow earlier workers (Solomon et al. mantle temperature variations to lithospheric temperatures. 1988;Cowie et al. 1993) by instead using an apparent isotherm con- Depths to the 800 Cisothermshallowerthan10kminFig. 9(c) trol, based on the known depth extents of earthquakes at mid-ocean suggest that the lithosphere is extremely weak locally at Conrad ridges where temperature structures can be estimated from thermal Deep lying on the right end of that profile. Perhaps such extreme models. weakness and dyke injections (implied by a swarm-like b-value) In their study of mid-ocean ridge earthquakes using body wave- if occurring also at the other deeps in the northern Red Sea have form inversions, Bergman & Solomon (1984) found source depths suppressed tectonic stress more generally, leading to suppressed limited by the 800 C isotherm. In contrast, Cowie et al. (1993)used seismicity. Suppressed tectonic stress would also limit the potential 600 C as the limiting isotherm based on results of Wiens & Stein for bending stresses associated with major fault slip that have been Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 1518 N.C. Mitchell and I.C.F. Stewart Figure 9. Estimates of temperature structure based on data from Martinez & Cochran (1989) for the RV Conrad heat flow transect marked in Fig. 8(b) (distance is from westerly end of that line). (a) Heat flow data (cross symbols) and temperatures at the base of the evaporites assuming conductive heat loss (red circles, see text). (b) Structure interpreted by Martinez & Cochran (1989) from an industry multichannel seismic line almost coincident with the heat flow ◦ ◦ ◦ measurements. (c) Inferred depths to the 600 C and 800 C isotherms assuming conductive heat loss as explained in the text. Dashed line (780 C) is taken from a model of Buck et al. (1988). linked to seismicity in mid-ocean ridge segments with detachment magmatic fluids, reducing effective stress. There have been reports faults (Olive & Escart´ ın 2016). of overpressures experienced during drilling through evaporites and from geophysical data evaluations in the Red Sea and Gulf of Suez (R Swarbrick and MJR Gee, personal communication 2016). Fluid overpressures Evaporites almost continuously cover the basement across the The general correspondence between areas of the Red Sea with more northern Red Sea (Fig. 1). In the central Red Sea, larger areas extensive rock outcrops in Fig. 1 and seismicity leads us to consider of igneous basement are exposed, as illustrated by the multibeam whether fluid overpressures could also be involved in reducing the data in Fig. 10. However, as that figure shows, even in such areas, vertical thickness of the schizosphere. According to Cowie et al. the rift border faults marked in the figure are largely covered by (1993), faults in the shallow crust can be stabilized by low effective evaporite flows, leaving only small outcrops (Mitchell et al. 2010). normal stress and elevated fluid pressures. If the evaporites seal the A similar covering of faults is observed around Atlantis II Deep underlying formations, fluid pressures may increase with progres- (Augustin et al. 2014; Feldens & Mitchell 2015). Further south, sive compaction of underlying pre-evaporite sediments or injected greater lengths of exposed faults are observable in the multibeam Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 Northern Red Sea aseismicity 1519 Figure 10. Shaded relief image of example multibeam bathymetry data from the central Red Sea (artificial illumination from the NE; Mitchell et al. 2010;Ligi et al. 2011; Augustin et al. 2014). Annotation ‘F’ marks where two rift border faults have been covered by evaporite flows (Mitchell et al. 2010). Annotation ‘V’ marks volcanic geomorphology in the floor of the deeps. data (Augustin et al. 2014; Mitchell & Augustin 2017), coinciding of Atlantis II Deep (Feldens & Mitchell 2015). Instead, evaporites with greater seismicity (Fig. 1). appear generally to flow in a viscous-like manner (Mitchell et al. From the change in heat flow away from rock outcrops on the 2010). sedimented Juan de Fuca Ridge, length-scales of cooling by fluid circulation have been estimated to be ∼20 km (Davis et al. 1999). As Serpentinized upper mantle the pressures needed to drive thermal convection are much smaller than those needed to create overpressures significant compared Ravat et al. (2011) estimated the depth of the base of the magnetic with overburden stress, we expect much greater than 20 km for source layer below the mainly Egyptian northern Red Sea from sealing by evaporites to be effective and for overpressures to de- magnetic anomalies. Their depths in places exceed 15 km, and velop. This would seem to rule out this overpressure mechanism for ◦ around 25 N in the western Red Sea exceed 20 km. As this is some aseismic areas, particularly around the deeps. Furthermore, deeper than the Moho here (∼12–15 km below sea level; Gaulier loss of effective stress and hence strength would lead to extensive et al. 1988), Ravat et al. (2011) suggested that they imply that there slumping, which is not generally observed in multibeam and seis- has been widespread serpentinization of the upper mantle. The weak mic data, aside from a > 20-km-scale slump on the eastern side rheology of serpentinite (Escart´ ın et al. 1997a,b; Hirth et al. 1998) Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 1520 N.C. Mitchell and I.C.F. Stewart offers a potential explanation for the suppressed seismicity of the elsewhere within Arabia so the source was more likely landward northern Red Sea, which lies further from the Afar plume than the than seaward. No other historical events were reported around the southern Red Sea and therefore has a colder lithospheric mantle northern Red Sea coasts. Similarly, in his assessment of historical perhaps more prone to hydration (if we ignore the thermal blanket documents of events up to 1900, Ambraseys (2009) found no clear effect for the sake of argument). A colder lithosphere in the north evidence of an earthquake occurring within the northern Red Sea would also be associated with the slow plate spreading, which falls away from the Gulf of Suez and Aqaba fault. More speculatively, −1 below 10 mm yr (Fig. 7a). Interestingly, a year-long experiment Nof & Paldor (1992) considered whether the sea crossing by the Is- with ocean bottom seismometers on amagmatic spreading segments raelites mentioned in Exodus in the Bible was enabled by a tsunami of the similarly ultra-slow spreading SWIR revealed a remarkable but rejected the possibility as less likely than a meteorological cause lack of earthquakes in the upper 15 km of the lithosphere, which and in any case there are many difficulties in interpreting stories of was interpreted as due to serpentinization (Schlindwein & Schmid such antiquity (Segert 1994). Ambraseys (2009) suggested the his- 2016; Schmid & Schlindwein 2016). torical record is effective for events of about M > 6.5, hence the lack However, the varied magnetic character in the northern Red Sea of historical evidence represents around two millenia of no events is at least partly a result of the distribution of volcanic intrusions of M > 6.5 in the northern Red Sea. Although not ruling out smaller and extrusions (Cochran 2005) and laterally extensive shallower events as having occurred, the data mostly point to this area being magnetic sources could also produce the long-wavelength anoma- of low seismic moment release, not a gap with seismic potential. lies interpreted by Ravat et al. (2011) as due to deep sources. Un- fortunately, there are few alternative data that can corroborate the existence of serpentinite. Seismic refraction profile VI of Rihm CONCLUSIONS et al. (1991) overlaps the area of deep magnetic basement of Ravat The seismic catalogue obtained from the ISC was used to charac- −1 et al. (2011). Although 7.5 km s below the Moho in the veloc- terize the seismic coupling of the northern Red Sea, which is an ity model of Rihm et al. (1991) could be consistent with modest order of magnitude smaller than that of the southern Red Sea and serpentinization (Carlson & Miller 1997), the uncertainty of the the Southwest Indian Ridge, a similarly slow-spreading ridge but −1 7.5 km s velocity is unclear. The depths of serpentinization sug- farther from recording stations. Historical documents reviewed by gested by Ravat et al. (2011) overlap with the depths to the 600 others suggest that the lack of large-magnitude seismicity has per- and 800 C isotherms calculated earlier (Fig. 9c), implying that sisted for two millenia in the northern Red Sea. There is therefore serpentinite would not be stable. Furthermore, serpentinization is unlikely to be a gap here with seismic potential. likely to have affected only the upper lithosphere (Schlindwein & Contributions to the ISC catalogue by the Saudi Geological Sur- Schmid 2016; Schmid & Schlindwein 2016), leaving a thickened vey are temporally incomplete but suggest a detection threshold of lower lithosphere capable of generating significant moment release, around M = 2.0. After adjusting for incompleteness, those data as the seismicity data of the serpentinite-dominated SWIR suggest suggest a b-value for the northern Red Sea of 1.22, consistent with (Fig. 2b). Consequently, we reject serpentinization as an explanation swarm-like activity. This is typical of volcanic rifts and is consistent for the northern Red Sea aseismicity. with morphologic data showing volcanic features within the deeps in the north. The low seismic coupling of the northern Red Sea may have more A gap with seismic potential? than one cause. The Plio-Pleistocene and Miocene sediments are es- timated to elevate the lithospheric temperature by ∼100 K because Could the apparently low seismic moment release rate and cou- of a thermal blanketing effect, reducing the schizosphere thickness pling of the northern Red Sea be due to the still limited period thermally. The low permeabilities of the evaporites may have led to of instrument recordings, that is, could the northern Red Sea rep- pore fluid overpressure in the basement underlying them, although resent a major seismic gap that could be filled by a future large basement outcrops (where the evaporite seal is likely breached) magnitude event (McCann et al. 1979)? According to Fig. 7,the occur also in areas of low seismicity. Serpentinized upper mantle X8 seismic moment deficit over the period 1964–2016 is ∼10 may furthermore provide a partial solution, although serpentinite cells or ∼10 N.m. If released in a single event, eq. (3) suggests is stable only below 400 C and a cold lithosphere associated with the earthquake would have M = 6.6. Such an event could involve a serpentinization is most likely to be seismically active, as is the rupture length of ∼10–20 km according to intraplate normal events case on the Southwest Indian Ridge where serpentinites are com- compiled by Scholz et al. (1986). Such a rupture length lies within mon. Such low temperatures are also incompatible with the heat the range of fault block length implied by structures in the free-air flow data. We therefore favour a combination of the temperature gravity field (Mitchell & Park 2014). In reviewing historical evi- and pore pressure explanations for the suppressed seismicity. More dence of major earthquakes in the Middle East, Ambraseys (1970) generally, the strong suppression here suggests that caution should found that active faults typically go through short-lived phases of be exerted when interpreting the tectonics of active rifts where thick activity separated by 75–150 yr of quiescence, so perhaps the post sediments are present. 1964 instrumental record is still inadequate to rule out a gap with seismic potential in the northern Red Sea. However, this calculation likely underestimates the moment ACKNOWLEDGEMENTS deficit of the northern Red Sea if we consider the lack of significant seismicity shown in the earlier instrumental and historical records. Dick Swarbrick and Martin Gee are thanked for informal informa- The reassessment of 1900–1999 earthquakes of Ambraseys (2001) tion on overpressures from industry wells within the northern Red shows none within the northern Red Sea. Historical documents re- Sea. We thank Sang-Mook Lee for advice on conductivities used to viewed by Poirier & Taher (1980) suggest that one earthquake of calculate heat flow data and Sigurjon Jonsson ´ and Jillian Foulger Mercalli intensity VI was felt on the Red Sea coast of Arabia in for discussions on Red Sea seismicity. Giulio di Toro suggested 1068 at Sharm Yanbu, although it was felt with intensities up to IX looking at the Zagros literature. Sergey Sokolov provided Russian Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 Northern Red Sea aseismicity 1521 literature on the eastern Gakkel Ridge. Figures in this article were Carlson, R.L. & Miller, D.J., 1997. A new assessment of the abundance of serpentinite in the oceanic crust, Geophys. Res. Lett., 24, 457–460. produced with the aid of the GMT free software system (Wessel Chang, S.-J. & van der Lee, S., 2011. Mantle plumes and associated flow & Smith 1991). We thank the editor Gabi Laske, Marco Ligi and beneath Arabia and East Africa, Earth planet Sci. Lett., 302, 448–454. an anonymous reviewer for helpful comments that were useful in Chang, S.J., Merino, M., Van der Lee, S., Stein, S. & Stein, C.A., 2011. revising this article. Mantle flow beneath Arabia offset from the opening Red Sea, Geophys. Author contributions: NCM carried out the analysis and wrote Res. Lett., 38. the article. ICFS provided advice on the seismology and regional Chu, D. & Gordon, R.G., 1998. Current plate motions across the Red Sea, seismicity and commented on the article. Geophys. J. Int., 135, 313–328. Cochran, J., Martinez, F., Steckler, M.S. & Hobart, M.A., 1986. Conrad Deep: a northern Red Sea deep. Origin and implication for continental rifting, Earth planet. Sci. Lett., 78, 18–32. REFERENCES Cochran, J.R., 1983. A model for the development of the Red Sea, Am. Al-Almadi, K., Al-Amri, A. & See, L., 2014. A spatial statistical analysis of Assoc. Petrol. Geol. Bull., 67, 41–69. the occurrence of earthquakes along the Red Sea floor spreading: clusters Cochran, J.R., 2005. Northern Red Sea: nucleation of an oceanic spread- of seismicity, Arab. J. Geosci., 7, 2893–2904. ing center within a continental rift, Geochem. Geophys. Geosyst., 6, Al-Amri, A.M.S., 1995. Recent seismic activity in the northern Red Sea, J. doi:03010.01029/02004GC000826. Geodyn., 20, 243–253. Cochran, J.R. & Karner, G.D., 2007. Constraints on the deformation and Allen, M.B., Saville, C., Blanc, E.J.-P., Talebian, M. & Nissen, E., 2013. rupturing of continental lithosphere of the Red Sea: the transition from Orogenic plateau growth: expansion of the Turkish-Iranian Plateau across rifting to drifting, in Imaging, Mapping and Modelling Continental Litho- the Zagros fold-and-thrust belt, Tectonics, 32, 1–20. sphere Extension and Breakup, Spec. Publ. 282, pp. 265–289, eds Karner, Altherr, R., Henjes-Kunst, F., Puchelt, H. & Baumann, A., 1988. Volcanic G.D., Manatschal, G. & Pinheiro, L.M., Geological Society. activity in the Red Sea axial trough - evidence for a large mantle diapir? Cowie, P.A., Scholz, C.H., Edwards, M. & Malinverno, A., 1993. Fault strain Tectonophysics, 150, 121–133. and seismic coupling on mid-ocean ridges, J. geophys. Res., 948(B10), Ambraseys, N.N., 1970. Some characteristic features of the Anatolian Fault 17911–17920. Zone, Tectonophysics, 9, 143–165. Crosby, A.G. & McKenzie, D., 2009. An analysis of young ocean depth, Ambraseys, N.N., 2001. Reassessment of earthquakes, 1900–1999, in the gravity and global residual topography, Geophys. J. Int., 178, 1198–1219. Eastern Mediterranean and the Middle East, Geophys. J. Int., 145, 471– Daggett, P.H., Morgan, P., Boulos, F.K., Hennin, S.F., El-Sherif, A.A., El- Sayod, A.A., Basta, N.Z. & Melek, Y.S., 1986. Seismicity and active Ambraseys, N.N., 2009. Earthquakes in the Mediterranean and Middle East: tectonics of the Egyptian Red Sea margin and the northern Red Sea, A Multidisciplinary Study of Seismicity up to 1900, Cambridge University Tectonophysics, 125, 313–324. Press, 947 pp. Das, R., Wason, H.R. & Sharma, M.L., 2011. Global regression relations ArRajehi, A. et al., 2010. Geodetic constraints on present-day motion of for conversion of surface wave and body wave magnitudes to moment the Arabian Plate: implications for Red Sea and Gulf of Aden rifting, magnitude, Nat. Hazards, 59, 801–810. Tectonics, 29(3). Davis, E.E. et al., 1999. Regional heat flow variations across the sedimented Augustin, N., Devey, C.W., van der Zwan, F.M., Feldens, P., Tominaga, Juan de Fuca Ridge eastern flank: constraints on lithospheric cooling and M., Bantan, R. & Kwasnitschka, T., 2014. The transition from rifting to lateral hydrothermal heat transport, J. geophys. Res., 104, 17675–17688. spreading in the Red Sea, Earth planet. Sci. Lett., 395, 217–230. Dyment, J., Tapponnier, P., Afifi, A.M., Zinger, M.A., Franken, D. & Augustin, N., van der Zwan, F.M., Devey, C.W., Ligi, M., Kwasnitschka, T., Muzaiyen, E., 2013. A new seafloor spreading model of the Red Sea: Feldens, P., Bantan, R. & Basaham, A.S., 2016. Geomorphology of the magnetic anomalies and plate kinematics, American Geophysical Union, central Red Sea Rift: determining spreading processes, Geomorph., 274, Fall Meeting, abstract #T21A-2512. 162–179. Dziewonski, A.M. & Woodhouse, J.H., 1983. An experiment in systematic Ayele, A. & Hulhanek, ´ O., 1997. Spatial and temporal variation of the study of global seismicity - centroid-moment tensor solutions for 201 seismicity in the Horn of Africa from 1960–1993, Geophys. J. Int., 130, moderate and large earthquakes of 1981, J. geophys. Res., 88, 3247–3271. 805–810. Egloff, F. et al., 1991. Contrasting structural styles of the eastern and west- Bakun, W.H., 1984. Seismic moments, local magnitudes, and coda-duration ern margins of the southern Red Sea: the 1988 SONNE experiment, magnitudes for earthquakes of central California, Bull. seism. Soc. Am., 74, 439–458. Tectonophysics, 198, 329–353. Ehrhardt,A.&Hubscher, C., 2015. The northern Red Sea in transition from Becker, J.J. et al., 2009. Global bathymetry and elevation data at 30 arc ¨ rifting to drifting – lessons learned from ocean deeps, in The Red Sea: seconds resolution: SRTM30 PLUS, Mar. Geod., 32, 355–371. The Formation, Morphology, Oceanography and Environment of a Young Bergman, E.A. & Solomon, S.C., 1984. Source mechanisms of earthquakes Ocean Basin, pp. 99–121, eds Rasul, N.M.A. & Stewart, I.C.F., Springer near mid-ocean ridges from body waveform inversion: implications for Earth System Sciences. the early evolution of oceanic lithosphere, J. geophys. Res., 89, 11415– Ehrhardt, A., Hubscher ¨ , C. & Gajewski, D., 2005. Conrad Deep, Northern Red Sea: development of an early stage ocean deep within the axial Bonatti, E., 1985. Punctiform initiation of seafloor spreading in the Red Sea depression, Tectonophysics, 411, 19–40. during transition from a continental to an oceanic rift, Nature, 316, 33–37. Einarsson, P. & Brandsdottir ´ , B., 1980. Seismological evidence for latermal Bonatti, E. et al., 1983. Zabargad (St. John’s) Island: an uplifted fragment magma intrusion during the July 1978 deflation of the Krafla volcano in of sub-Red Sea lithosphere, J. geol. Soc. Lond., 140, 677–690. NE-Iceland, J. Geophys., 47, 160–165. Bormann, P. & Dewey, J.W., 2014. The new IASPEI standards for determin- El-Isa, Z.H. & Al Shanti, A., 1989. Seismicity and tectonics of the Red Sea ing magnitudes from digital data and their relation to classical magnitudes, and western Arabia, Geophys. J. Int., 97, 449–457. in New Manual of Seismological Observatory Practice 2 (NMSOP-2), pp. Endo, E., Zahran, H., Nofal, H. & Hadidy, S., 2007. The Saudi National 1–44, ed Bormann, P., Deutsches GeoForschungsZentrum GFZ. Seismic Network, Seismol. Res. Lett., 78, 439–445. Brune, J.N., 1968. Seismic moment, seismicity, and rate of slip along major Escart´ ın, J., Hirth, G. & Evans, B., 1997a. Effects of serpentinization on the fault zones, J. geophys. Res., 73, 777–784. lithospheric strength and the style of normal faulting at slow-spreading Buck, R., Martinez, F., Steckler, M.S. & Cochran, J.R., 1988. Thermal ridges, Earth planet. Sci. Lett., 151, 181–189. consequences of lithospheric extension: pure and simple, Tectonics, 7, Escart´ ın, J., Hirth, G. & Evans, B., 1997b. Non-dilatant brittle deformation 213–234. of serpentinites: implications for Mohr-Coulomb theory and the strength Cann, J.R. et al., 1997. Corrugated slip surfaces formed at ridge-transform of faults, J. geophys. Res., 102, 2897–2913. intersections on the Mid-Atlantic Ridge, Nature, 385, 329–332. Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 1522 N.C. Mitchell and I.C.F. Stewart Escart´ ın, J., Smith, D.K., Cann, J., Schouten, H., Langmuir, C.H. Klein, E.M. & Langmuir, C.H., 1987. Global correlations of ocean ridge & Escrig, S., 2008. Central role of detachment faults in accre- basalt chemistry with axial depth and crustal thickness, J. geophys. Res., tion of slow-spreading oceanic lithosphere, Nature, 455, 790–794. 92, 8089–8115. http://dx.doi.org/10.1038/nature07333 Ligi, M. et al., 2011. Initial burst of oceanic crust accretion in the Red Sea Fairhead, J.D. & Girdler, R.W., 1970. The seismicity of the Red Sea, Gulf of due to edge-driven mantle convection, Geology, 39, 1019–1022. Aden and Afar triangle, Phil. Trans. Roy. Soc., A267, 49–74. Ligi, M. et al., 2012. Birth of an ocean in the Red Sea: initial pangs, Feldens, P. & Mitchell, N.C., 2015. Salt flows in the central Red Sea, in The Geochem. Geophys. Geosyst., 13, Q08009. . Red Sea: The Formation, Morphology, Oceanography and Environment Lolli, B., Gasperini, P. & Vannucci, G., 2014. Empirical conversion between of a Young Ocean Basin, pp. 205–218, eds Rasul, N.M.A. & Stewart, teleseismic magnitudes (mb and Ms) and moment magnitude (Mw) at I.C.F. Springer Earth System Sciences. the Global, Euro-Mediterranean and Italian scale, Geophys. J. Int., 199, Feldens, P., Schmidt, M., Muck ¨ e, I., Augustin, N., Al-Farawati, R., Orif, M. 805–828. & Faber, E., 2016. Expelled subsalt fluids form a pockmark field in the Mahmoud, S., Reilinger, R., McClusky, S., Vernant, P. & Tealeb, A., 2005. eastern Red Sea, Geo-Mar. Lett., 36(5), 339–352. GPS evidence for northward motion of the Sinai Block: implications for Frohlich, C., 2007. Practical suggestions for assessing rates of seismic- E. Mediterranean tectonics, Earth planet. Sci. Lett., 238, 217–224. moment release, Bull. seism. Soc. Am., 97, 1158–1166. Makris, J., Tsironidis, J. & Richter, H., 1991. Heatflow density distribution Frohlich, C. & Wetzel, L.R., 2007. Comparison of seismic moment release in the Red Sea, Tectonophysics, 198, 383–393. rates along different types of plate boundaries, Geophys. J. Int., 171, Martinez, F. & Cochran, J.R., 1988. Structure and tectonics in the north- 909–920. ern Red Sea: catching a continental margin between rifting and drifting, Gasperini, P., Lolli, B. & Vannucci, G., 2013. Body-wave magnitude m is Tectonophysics, 150, 1–32. a good proxy of moment magnitude M for small earthquakes (M < Martinez, F. & Cochran, J.R., 1989. Geothermal measurements in the north- w b 4.5-5.0), Seismol. Res. Lett., 84, 932–937. ern Red Sea: implications for lithospheric thermal structure and mode of Gaulier, J.M., LePichon, X., Lyberis, N., Avedik, F., Gely, L., Moretti, I., extension during continental rifting, J. geophys. Res., 94, 12239–12265. Deschamps, A. & Hafez, S., 1988. Seismic study of the crustal thickness, McCann, W.R., Nishenko, S.P., Sykes, L.R. & Krause, J., 1979. Seismic gaps northern Red Sea and Gulf of Suez, Tectonophysics, 153, 55–88. and plate tectonics: seismic potential for major boundaries, Pageoph., 117, Gharib, A., Nicioglu, A. & Al-Amri, A., 1997. Seismicity re-evaluation of 1082–1147. the northern Red Sea, Geophys J. Int., 130, 335–348. McClusky, S., Reilinger, R., Mahmoud, S., Ben Sari, D. & Tealeb, A., 2003. Gordon, G., et al., 2010. The hydrocarbon prospectivity of the Egyptian GPS constraints on Africa (Nubia) and Arabia plate motions, Geophys. J. North Red Sea basin, in Petroleum Geology: From Mature Basins to New Int., 155, 126–138. Frontiers - Proceedings of the 7th Petroleum Geology Conference, pp McClusky, S. et al., 2010. Kinematics of the southern Red Sea-Afar Triple 783–789, eds Vining, B.A. & Pickering, S.C., Geol. Soc. Lond. Junction and implications for plate dynamics, Geophys. Res. Lett., 37, Guennoc, P., Pautot, G. & Coutelle, A., 1988. Surficial structures of the L05301. doi:05310.01029/02009GL041127 ◦ ◦ northern Red Sea axial valley from 23 Nto28 N: time and space McKenzie, D., Jackson, J. & Priestley, K., 2005. Thermal structure of oceanic evolution of neo-oceanic structures, Tectonophysics, 153, 1–23. and continental lithosphere, Earth planet. Sci. Lett., 233, 337–349. Guennoc, P., Pautot, G., LeQentrec, M.-F. & Coutelle, A., 1990. Structure of Miller, P.M. & Barakat, H., 1988. Geology of the Safaga Concession, north- an early oceanic rift in the northern Red Sea, Oceanol. Acta, 13, 145–157. ern Red Sea, Egypt, Tectonophysics, 153, 123–136. Gutenberg, B. & Richter, C.F., 1954. Seismicity of the Earth and Associated Mitchell, N.C., 1991. Distributed extension at the Indian Ocean Triple Junc- Phenomena, 2nd edn, Princeton University Press. tion, J. geophys. Res., 96(B5), 8019–8043. Haase, K.M., Muhe, ¨ R. & Stoffers, P., 2000. Magmatism during extension of Mitchell, N.C., 2015. Lineaments in gravity data of the Red Sea, in The Red the lithosphere: geochemical constraints from lavas of the Shaban Deep, Sea: The Formation, Morphology, Oceanography and Environment of a northern Red Sea, Chem. Geol., 166, 225–239. Young Ocean Basin, pp. 123–133, eds Rasul, N.M.A. & Stewart, I.C.F., Hanks, T.C. & Kanamori, H., 1979. A moment magnitude scale, J. geophys. Springer Earth System Sciences. Res., 84, 2348–2350. Mitchell, N.C. & Augustin, N., 2017. Halokinetics and other features of Hasterok, D., 2013. A heat flow based cooling model for tectonic plates, GLORIA long-range sidescan sonar data from the Red Sea, Mar. Pet. Earth planet. Sci. Lett., 361, 34–43. Geol., 88, 724–738. Hirth, G., Escartin, J. & Lin, J., 1998. The rheology of the lower oceanic Mitchell, N.C., Ligi, M., Farrante, V., Bonatti, E. & Rutter, E., 2010. Subma- crust: implications for lithospheric deformation at mid-ocean ridges, in rine salt flows in the central Red Sea, Geol. Soc. Am. Bull., 122, 701–713. Mitchell, N.C., Ligi, M., Feldens, P. & Hubscher, C., 2017. Deformation of a Faulting and Magmatism at Mid-Ocean Ridges, pp. 291–303, eds Buck, ¨ W.R., Delaney, P.T., Karson, J.A. & Lababrielle, Y., American Geophysical young salt giant: regional topography of the Red Sea Miocene evaporites, Union. Basin Res., 29, 352–369. Hofstetter, R. & Beyth, M., 2003. The Afar Depression: interpretation of Mitchell, N.C., Ligi, M. & Rohling, E.J., 2015. Red Sea isolation history the 1960–2000 earthquakes, Geophys. J. Int., 155, 715–732. suggested by Plio-Pleistocene seismic reflection sequences, Earth planet. Huang, P.Y., Solomon, S.C., Bergman, E.A. & Nabelek, J.L., 1986. Focal Sci. Lett., 430, 387–397. depths and mechanisms of Mid-Atlantic Ridge earthquakes from body Mitchell, N.C. & Park, Y., 2014. Nature of crust in the central Red Sea, waveform inversion, J. geophys. Res., 91, 579–598. Tectonophysics, 628, 123–139. Izzeldin, A.Y., 1987. Seismic, gravity and magnetic surveys in the central Nedimovic, ´ M.R., Bohnenstiehl, D.R., Carbotte, S.M., Canales, J.P. & Dziak, part of the Red Sea: their interpretation and implications for the structure R.P., 2009. Faulting and hydration of the Juan de Fuca plate system, Earth and evolution of the Red Sea, Tectonophysics, 143, 269–306. planet. Sci. Lett., 284, 94–102. Jackson, J. & McKenzie, D., 1988. The relationship between plate motions Nikishin, A.M., Petrov, E.I., Malyshev, N.A. & Ershova, V.P., 2017. Eurasia and seismic moment tensors, and the rates of active deformation in the Basin and Gakkel Ridge, Arctic Ocean: crustal asymmetry, ultra-slow Mediterranean and Middle East, Geophys. J., 93, 45–73. spreading and continental rifting revealed by new seismic data, Tectono- Jahani, S., Callot, J.-P., Lamotte, D.Fd., Letouzey, J. & Leturmy, P., 2007. physics, in press. The salt diapirs of the eastern Fars province (Zagros, Iran): a brief outline Nissen, E., Jackson, J., Jahani, S. & Tatar, M., 2014. Zagros “phantom of their past and present, in Thrust Belts and Foreland Basins, pp. 289– earthquakes” reassessed - the interplay of seismicity and deep salt flow 308, eds Lacombe, O., Roure, F., Lave, ´ J. & Verges, ´ J., Frontiers in Earth in the Simply Folded Belt? J. geophys. Res., 119, 3561–3583. Sciences, Springer. Nissen, E., Tatar, M., Jackson, J.A. & Allen, M.B., 2011. New views on Keir, D., Stuart, G.W., Jackson, A. & Ayele, A., 2006. Local earthquake earthquake faulting in the Zagros fold-and-thrust belt of Iran, Geophys. magnitude scale and seismicity rate for the Ethiopian Rift, Bull. seism. J. Int., 186, 928–944. Soc. Am., 96, 2221–2230. Downloaded from https://academic.oup.com/gji/article/214/3/1507/4993249 by DeepDyve user on 15 July 2022 Northern Red Sea aseismicity 1523 Nof, D. & Paldor, N., 1992. Are there oceanographic explanations for the Stein, C.A. & Stein, S., 1994. Constraints on hydrothermal heat flux through Israelites’ crossing of the Red Sea, Bull. Am. Meteorol. Soc., 73, 305–314. the oceanic lithosphere from global heat flow, J. geophys. Res., 99, 3081– Ogata, Y. & Katsura, K., 1993. Analysis of temporal and spatial hetero- 3095. geneity of magnitude frequency distribution inferred from earthquake Steinbrugge, K.V., Zacher, E.G., Tocher, D., Whitten, C.A. & Claire, C.N., catalogues, Geophys. J. Int., 113, 727–738. 1960. Creep on the San Andreas fault, Bull. Seism. Soc. Am., 50, 389–415. Olive, J.-A. & Escart´ ın, J., 2016. Dependence of seismic coupling on normal Stewart, I.C.F., 2007. Earthquake risk in western Saudi Arabia and the Red fault style along the Northern Mid-Atlantic Ridge, Geochem. Geophys. Sea from seismic moment, Saudi Geol. Surv., Technical Report SGS-TR- Geosyst., 17, 4128–4152. 2007-4, pp. 41. Pautot, G., 1983. Red Sea deeps – a geomorphological study by Seabeam, Tapponnier, P., Dyment, J., Zinger, M.A., Franken, D., Afifi, A.M., Wylie, Oceanol. Acta, 6, 235–244. A., Ali, H.G. & Hanbal, I., 2013. Revisiting seafloor-spreading in the Peach, C.J., 1991. Influence of deformation on the fluid transport properties Red Sea: basement nature, transforms and ocean-continent boundary, of salt rocks, Geologica Ultraiectina, University of Utrecht, Netherlands, American Geophysical Union, Fall Meeting 2013, San Francisco, CA, Vol. 77, pp. 1–238. 9–13 Dec, abstract #T12B-04. Poirier, J.P. & Taher, M.A., 1980. Historical seismicity in the Near and Taviani, M., Bonatti, E., Colantoni, P. & Rossi, P.L., 1984. Tectonically Middle East, North Africa, and Spain from Arabic documents (VIIth- uplifted crustal blocks in the northern Red Sea: data from the Brothers XVIIth Century), Bull. seism. Soc. Am., 70, 2185–2201. Islets, Mem. Soc. Geol. It., 27, 47–50. Ravat, D., Salem, A., Abdelaziz, A.M.S., Elawadi, E. & Morgan, P., 2011. Tongue, J.A., Maguire, P.K.H. & Burton, P., 1992. Seismicity distribution Probing magnetic bottom and crustal temperature variations along the from temporary earthquake recording networks in Kenya, Tectonophysics, Red Sea margin of Egypt, Tectonophysics, 510, 337–344. 204, 71–79. Reilinger, R., McClusky, S. & ArRajehi, A., 2015. Geodetic constraints on Wiens, D.A. & Stein, S., 1984. Intraplate seismicity and stresses in young the geodynamic evolution of the Red Sea, in The Red Sea: The Formation, oceanic lithosphere, J. geophys. Res., 89, 11442–11464. Morphology, Oceanography and Environment of a Young Ocean Basin, Wesnousky, S.G., Scholz, C.H., Shimazaki, K. & Matsuda, T., 1983. Earth- pp. 135–149, eds Rasul, N.M.A. & Stewart, I.C.F., Springer Earth System quake frequency distribution and the mechanics of faulting, J. geophys. Sciences. Res., 88, 9331–9340. Richter, H., Makris, J. & Rihm, R., 1991. Geophysical observations offshore Wessel, P. & Smith, W.H.F., 1991. Free software helps map and display data, Saudi Arabia: seismic and magnetic observations, Tectonophysics, 198, EOS, Trans. Am. geophys. Un., 72, 441–448. 297–310. Wheildon, J., Evans, T.R. & Girdler, R.W., 1974. Thermal conductivity, Rihm, R., Makris, J. & Moller ¨ , L., 1991. Seismic surveys in the northern density, and sonic velocity measurements of samples of anhydrite and Red Sea: asymmetric crustal structure, Tectonophysics, 198, 279–295. halite from Sites 225 and 227, in Initial Reports of the Deep Sea Drilling Rundquist, D.V. & Sobolev, P.O., 2002. Seismicity of mid-oceanic ridges Project, Vol. 23, pp. 909–911, eds Whitmarsh, R.B., Weser, O.E., Ross, and its geodynamic implications: a review, Earth-Sci. Rev., 58, 143–161. D.A., et al., U.S. Govt. Printing Office. Sauter, D. & Cannat, M., 2010. The ultraslow spreading Southwest Indian Willemann, R.J. & Storchak, D.A., 2001. Data collection at the International Ridge, in Diversity of Hydrothermal Systems on Slow Spreading Ocean Seismological Centre, Seismol. Res. Lett., 72, 440–453. Ridges, Geophysics Monograph Series 188, pp. 153–173 eds Rona, P.A., Zahran, H.M., Sokolov, V., Roobol, M.J., Stewart, I.C.F., El-Hadidy Youssef, Devey, C.W., Dyment, J. & Murton, B.J., American Geophysical Union. S. & El Hadidy, M., 2016. On the development of a seismic source Schettino, A., Macchiavelli, C., Pierantoni, P.P., Zanoni, D. & Rasul, N., zonation model for seismic hazard assessment in western Saudi Arabia, 2016. Recent kinematics of the tectonic plates surrounding the Red Sea J. Seismol., 20, 747–769. and Gulf of Aden, Geophys. J. Int., 207, 457–480. Schlindwein, V. & Schmid, F., 2016. Mid-ocean ridge seismicity re- veals extreme types of ocean lithosphere, Nature, 535, 276–279. http://dx.doi.org/10.1038/nature18277 SUPPORTING INFORMATION Schmid, F. & Schlindwein, V., 2016. Microearthquake activity, lithospheric Supplementary data are available at GJI online. structure, and deformation modes at an amagmatic ultraslow spread- Figure S1: Map of ISC events and Global CMT solutions for the ing Southwest Indian Ridge segment, Geochem. Geophys. Geosyst., 17, Southwest Indian Ridge, for the period 1964–2016. Symbols as Fig. 2905–2921. 1 of the main text, with bathymetry from the GeomapApp database Scholz, C.H., 1994. A reappraisal of large earthquake scaling, Bull. seism. Soc. Am., 84, 215–218. (Ryan et al. 2009). Black polygon outlines seafloor younger than Scholz, C.H., 2002. The Mechanics of Earthquakes and Faulting, Cambridge 15 Ma. Red polygon outlines part of this area that is without major University Press, 471 pp. transform faults. Scholz, C.H., Aviles, C. & Wesnousky, S.G., 1986. Scaling differences Figure S2: Spatial distribution of data used to infer the basement between large interplate and intraplate earthquakes, Bull. seism. Soc. Am., rock exposures in Fig. 1 of the main text. 76, 65–70. Please note: Oxford University Press is not responsible for the con- Segert, S., 1994. Crossing the waters: Moses and Hamilcar, J. Near East. tent or functionality of any supporting materials supplied by the Stud., 53, 195–203. authors. Any queries (other than missing material) should be di- Solomon, S.C., Huang, P.Y. & Meinke, L., 1988. The seismic moment budget rected to the corresponding author for the article. of slowly spreading ridges, Nature, 334, 58–60.

Journal

Geophysical Journal InternationalOxford University Press

Published: Sep 1, 2018

There are no references for this article.