Elsevier

Earth and Planetary Science Letters

Volume 430, 15 November 2015, Pages 387-397
Earth and Planetary Science Letters

Red Sea isolation history suggested by Plio-Pleistocene seismic reflection sequences

https://doi.org/10.1016/j.epsl.2015.08.037Get rights and content

Highlights

  • Red Sea Pleistocene sediment density correlated with sea level.

  • Chirp reflections correlated with rapid changes in sea level.

  • Sea level effect observed at least to marine isotope stage 16.

  • Seismic data also suggest history of Red Sea connection to Indian Ocean.

Abstract

High evaporation rates in the desert climate of the Red Sea ensure that, during glacial sea level lowstands when water exchange with the Indian Ocean was more restricted, water salinity and δ18O became unusually extreme. Modeling of the effect on Red Sea sedimentary δ18O has been used previously to reconstruct relative sea level to 500 ka and now poses the question of whether that sea-level model could be extended if continuous core material of older sediment became available. We attempt to address this question here by examining seismic reflection data. The upper Pleistocene hemipelagic sediments in the Red Sea contain intervals of inorganic aragonite precipitated during supersaturated conditions of sea-level lowstands. Seismic impedance changes associated with boundaries to those aragonite-rich layers appear to explain seismic reflection sequences. A segment of Chirp sediment profiler data from the central Red Sea reveals prominent reflections at ∼1, ∼5, ∼23, ∼26 and ∼36 ms two-way travel time (TWT) from the seabed. Based on depths to the glacial marine isotope stages (MIS) in cores, we relate the upper three reflections to the tops of aragonite-rich layers and hence the sea level rises immediately following MIS 2, 6 and 12. The reflection at 26 ms is related to an unusually rapid fall into MIS 12 predicted by one sea level reconstruction, which may have created an abrupt lower boundary to the MIS 12 aragonite-rich layer. With the aid of seismogram modeling, we tentatively associate the ∼36 ms reflection with the top of an aragonite-rich layer formed during MIS 16. Furthermore, some segments of lower frequency (airgun and sparker) seismic data from the central and southern Red Sea show a lower (earlier) Plio-Pleistocene (PP) interval that is less reflective than the upper (late) PP interval. This implies less variability in sediment impedance and that extreme variability in water salinity did not develop; water exchange with the Indian Ocean likely continued throughout this interval. These results suggest that viable relative sea level reconstructions may be recovered from Red Sea sediment δ18O data to at least MIS 16 and perhaps even as far back as the early Pliocene.

Introduction

The Red Sea is an enclosed water body with a narrow and shallow connection to the Indian Ocean at the Bab el Mandab Strait (Fig. 1), with water over the shallowest point at Hanish Sill only 137 m deep at present (Siddall et al., 2003, Siddall et al., 2004, Lambeck et al., 2011), limiting exchange of water with the Indian Ocean. Along with extreme net evaporation (2 m yr−1), the limited replenishment with global ocean water has caused the sea to become more saline (present-day salinity, S>40; Sofianos and Johns, 2007) than the global ocean. Hypersalinity is prevented by loss of saline Red Sea water and gain of less-saline Indian Ocean water through the strait, the exchange occurring in two layers of differing density, with the Indian Ocean water at the surface and denser Red Sea water beneath it (Maillard and Soliman, 1986, Siddall et al., 2004). Recent assessments of global sea level lowering and regional isostatic responses suggest that water depth over Hanish Sill fell to only 25–33 m during the Last Glacial Maximum (LGM, ∼20 ka) (Biton et al., 2008, Lambeck et al., 2011). The consequential further reduced water exchange with the Indian Ocean caused the Red Sea to become more saline and supersaturated for aragonite, leading to widespread inorganic aragonite precipitation (Almogi-Labin et al., 1986, Fenton et al., 2000, Milliman et al., 1969, Rohling et al., 1998; and references therein). As the effects of seawater evaporation were less compensated by Indian Ocean inflow, this also caused foraminiferal δ18O to increase beyond 2‰, leading to net glacial–interglacial variations 2–3 times those of the global oceans (Hemleben et al., 1996). Estimates from δ18O measurements suggest that 50<S<57 during the LGM (Biton et al., 2008, Hemleben et al., 1996), or even S>60 (Siddall et al., 2003), in agreement with salinities explaining the disappearance of planktonic foraminifera (S>49) but continuation of euryhaline pteropods and benthic foraminifera (S<70; Fenton et al., 2000, Fernandes et al., 2006, Rohling et al., 1998, hence water exchange did not stop entirely during the LGM). Late Pleistocene sea-level lowstands in the Red Sea are thus characterized by ‘aplanktonic zones’, which lack planktonic foraminifera, contain only euryhaline pteropods and benthic foraminifera, and where inorganic aragonite cement is commonplace (Almogi-Labin et al., 1986, Fenton et al., 2000, Milliman et al., 1969, Rohling et al., 1998). In the Red Sea, bottom-water movements are modest (Sofianos and Johns, 2007), aragonite dissolution occurs in interglacials but there is no widespread calcite dissolution (Almogi-Labin et al., 1986, Almogi-Labin et al., 1998), and productivity (and its variability) is low (Trommer et al., 2011).

The impact on sedimentary δ18O has been exploited to reconstruct relative sea level (RSL) at Hanish Sill, by coupling hydraulic models for exchange fluxes with the Indian Ocean with known effects of evaporation on isotopic ratios in the sea (Biton et al., 2008, Rohling et al., 2009, Siddall et al., 2003, Siddall et al., 2004). As the sensitivity of δ18O to sea level is greatest when the sea is more isolated, the results are particularly useful for reconstructing sea level around lowstands, which are not so well documented from margin sequences (e.g., Miller et al., 2005), while they are less accurate for highstands. Hanish Sill lies adjacent to the volcanically active Hanish Island and potentially active faults, but a simple adjustment of ∼0.02 m kyr−1 (representing steady sill uplift; Rohling et al., 1998, Rohling et al., 2009, Siddall et al., 2003) and a further adjustment for hydro-isostasy align the Red Sea (Hanish Sill) RSL with global sea level estimates from (i) an independent approach of similar philosophy that uses Mediterranean δ18O records (Rohling et al., 2014), (ii) correcting deep-sea benthic foraminiferal δ18O for bottom-temperature changes (Elderfield et al., 2012) and (iii) model-based assessment of global deep-sea benthic δ18O (Bintanja and van de Wal, 2008), as well as with other sea level estimates such as from some coral elevations and marginal unconformities (see Rohling et al., 2014 and Fig. 2). While there are short-lived (∼10 ky) deviations between the Red Sea and the other sea level reconstructions, there are no permanent shifts (Fig. 2), so any geological change in elevation of Hanish Sill, if any occurred, would need to have been short-lived and quickly reversed. However, most geological processes are unlikely to reverse within a few ky; we know of no examples of normal fault movements reversing over such timescales in extensional environments such as that considered here. Furthermore, currents measured over Hanish Sill are ∼1 m s−1 (Pratt et al., 1999), which is too small to cause erosion of bedrock (Mitchell et al., 2012), so any emplaced volcanic materials are unlikely to been fully removed by currents. We therefore prefer to consider the short-period deviations from the other sealevel models in Fig. 2 as potentially genuine global variations until proven otherwise. This is important for our interpretation of MIS 12, where there is one such divergence.

Since no continuous Red Sea sediment cores reach beyond 500 ka, it is unknown whether new deep coring could extend the sea-level method to earlier times. The Red Sea method would be attractive because the sea level estimates can be achieved with smaller analytical uncertainty margins than other methods around lowstands (Rohling et al., 2014). Even though systematic uncertainties in the Red Sea results may be enlarged prior to 500 ka (e.g., due to sill elevation uncertainty), a longer Red Sea record would still be useful because the uncertainties are independent of those of the other methods. An approach to narrowing those uncertainties could involve identifying durations in the Red Sea reconstruction where sea level variations are similar to those based on coral, marginal unconformities and deep-sea δ18O. If those sea level segments are offset but otherwise similar in character, corrections for biases due to geological processes at Hanish Sill can be worked out and applied to the reconstructed RSL, thus yielding intervals older than 500 ky with potential millennial-scale resolution.

In the present article, we describe an assessment of various forms of seismic data collected in the Red Sea, where rapid changes in seawater chemistry associated with sea level variation (Biton et al., 2008, Rohling et al., 2009, Siddall et al., 2003, Siddall et al., 2004) are suggested here to have led to sharp changes in sediment physical properties, producing prominent seismic reflections. In particular, reports of the lowstand aragonite-rich layers describe them as “hard layers”, where interconnected fibrous aragonite crystals have left the bulk sediment indurated (Gevirtz and Friedman, 1966, Milliman et al., 1969, Stoffers and Ross, 1974). In our experience, the LGM layer in particular has the consistency of brittle concrete. Such materials can be expected to be more rigid and thus to have higher seismic velocity and impedance than surrounding unconsolidated sediment.

The principal data used are from a Chirp sediment profiler. Such systems create records of sub-metre resolution by transmitting long pulses of swept frequency and correlating the recorded echoes with a replica of the transmitted pulse (Dal Forno and Gasperini, 2008, Schock et al., 1989; also see the electronic supplement to this article). We show that many of the variations in reflectivity with depth in the Chirp data are as expected from the estimated history of late Pleistocene global and regional sea level change (Bintanja and van de Wal, 2008, Elderfield et al., 2012, Rohling et al., 2014) and with their anticipated effects on sediment density and velocity.

Section snippets

Variations in sediment physical properties

Central Red Sea sediments are predominantly hemipelagic carbonate oozes, with varied aeolian dust contributions (Stoffers and Ross, 1974). In order to interpret reflection sequences in the Chirp data, we need indications of how bulk density and P-wave velocity (Vp) vary with depth and age in these sediments. Sediment bulk density measurements for core GeoTü-KL11 (hereafter KL11; Fig. 1) shown in Fig. 3 reveal high densities around times of the global sea level lowstands MIS 2, 6, 8 and 10

Sediment profiler datasets

Sediment profiler data were collected on RV Urania in 2005 with a 16-transducer Benthos CAP-6600 CHIRP-II system. A Chirp system operates by transmitting a long-duration pulse (here 20 ms) with swept frequency (here 2 to 7 kHz) and processing the reflected signals by correlating them with a replica of the transmitted pulse. The result is a record in which the pulse is effectively “compressed” so that the data have very high theoretical resolution (here ∼0.1 ms), with higher energy than would be

Observations

The Chirp data in Fig. 6 all display reflections with a draping morphology typical of pelagic sediments (Mitchell, 1995); sub-bottom reflections follow the seabed reflection at nearly uniform sub-bottom two-way time in each panel. We have deliberately chosen segments of data with this characteristic, but relatively few (roughly <20%) of the data outside the Thetis deep show evidence for slumping or other forms of mass-movement. The data do not typically show disrupted stratigraphy

Interpretation

The regional deposition pattern of the hemipelagic sediments is now well characterized by many decades of studies with cores, so we use those results to guide our interpretation. The almost constant sedimentation rate with time at KL11 (Fig. 3) may be unusual as some variability can be expected given the large dust component of the sediment, which is unlikely to be constant everywhere in the Red Sea (e.g., Fig. 4). Fig. 10 shows the sediment depths to various glacial isotope stages in the

Seismogram characterization and modeling

Seismogram modeling provides a demanding test of the interpretations because gradients in impedance generate reflections, so small errors in data used to generate an impedance series produce exaggerated effects in the models. We consider only relative amplitude changes with TWT. We first examine the reflection amplitudes as they help to inform the seismogram modeling.

Utility of the characteristic Chirp reflection sequences

The presence of common reflection sequences in the Chirp data that apparently correspond with the sea level reconstructions suggests that these data could be useful to guide future long-coring campaigns in the Red Sea. For example, if earlier (MIS16-age) samples are sought, these are more likely to be recoverable where reflection R5 is closer to the bed and within coring depth.

The chronostratigraphic nature of the reflections will also allow spatial variations in deposition rate across the area

Conclusions

“Hard layers” that contain inorganic aragonite precipitated during sealevel lowstands are suggested to explain sequences of Chirp reflections in the central Red Sea. We have developed an interpretation based on depths to such layers recorded in sediment core data, with apparent correspondence between the reflection sequence and times of rapid changes in sea level around lowstands, which likely marked the transitions to/from aragonite precipitation. A seismogram model developed to reproduce the

Acknowledgments

Permissions of the governments of Egypt, Sudan and Saudi Arabia to carry out the survey on RV Urania are gratefully acknowledged. The Urania cruise was funded by the Consiglio Nazionale delle Ricerche under project LEC-EMA21F of the European Science Foundation programme EUROMARGINS (contract ERAS-CT-2003-980409 of the European Commission, DG Research FP6). Figures in this article were created with the “GMT” software system. The density and carbonate data for KL11 (Fig. 3) were kindly provided

References (49)

  • M. Siddall et al.

    Understanding the Red Sea response to sea level

    Earth Planet. Sci. Lett.

    (2004)
  • W.H. Berger et al.

    Distribution of carbonate in surface sediments of the Pacific Ocean

    J. Geophys. Res.

    (1976)
  • R. Bintanja et al.

    North American ice-sheet dynamics and the onset of 100,000-year glacial cycles

    Nature

    (2008)
  • E. Biton et al.

    Red Sea during the last glacial maximum: implications for sea level reconstruction

    Paleoceanography

    (2008)
  • J.M. Bull et al.

    Magmatic and tectonic history of Iceland's western rift zone at Lake Thingvallavatn

    Geol. Soc. Am. Bull.

    (2005)
  • W.G. Deuser et al.

    O18/O16 and C13/C12 ratios of fossils from the hot-brine deep area of the central Red Sea

  • A. Ehrhardt et al.

    The northern Red Sea in transition from rifting to drifting – lessons learned from ocean Deeps

  • H. Elderfield et al.

    Evolution of ocean temperature and ice volume through the Mid-Pleistocene Climate Transition

    Science

    (2012)
  • C.A. Fernandes et al.

    Absence of post-Miocene Red Sea land bridges: biogeographic implications

    J. Biogeogr.

    (2006)
  • I.G. Gass et al.

    Volcanic islands of the Red Sea

    J. Geol. Soc.

    (1973)
  • J.L. Gevirtz et al.

    Deep-sea carbonate sediments of the Red Sea and their implications on marine lithification

    J. Sediment. Petrol.

    (1966)
  • K.M. Grant et al.

    Sea-level variability over five glacial cycles

    Nat. Commun.

    (2014)
  • E.L. Hamilton

    Sound velocity-density relations in sea-floor sediments and rocks

    J. Acoust. Soc. Am.

    (1978)
  • C. Hemleben

    Three hundred eighty thousand year long stable isotope and faunal records from the Red Sea: influence of global sea level change on hydrography

    Paleoceanography

    (1996)
  • Cited by (18)

    • The evolution of seafloor environmental conditions in the southern Red Sea continental shelf during the last 30 ka

      2022, Marine Micropaleontology
      Citation Excerpt :

      Enhanced acoustic reflection marks the glacial (high DBD) section, while the (low DBD) post-glacial sediments are acoustically transparent (Figs. 1c, 2). A similar contrast has been observed in seismic profiles from the central Red Sea (Mitchell et al., 2015). This indicates consistent sediment density changes throughout (at least) the southern and central Red Sea during the Quaternary.

    • Sea-level history and tectonic uplift during the last-interglacial period (LIG): Inferred from the Bab al-Mandab coral reef terraces, southern Red Sea

      2018, Journal of African Earth Sciences
      Citation Excerpt :

      The Red Sea is an enclosed water body extending some 2000 km SE-NW, between latitudes 12° N and 28° N, from the Gulf of Aden to the Gulf of Aqaba. The Red Sea connection with Indian Ocean lies at Hanish Sill in the Bab al-Mandab at depth of 137 m compare to the depth of a full glacial lowstand of 120 m (Siddall et al., 2003, 2004; Lambeck et al., 2011; Mitchell et al., 2015). Many researchers focus on Hanish sill as a dynamic coastal landscape and a potential migration corridor involving a crossing of the Bab al-Mandab Straits (Bailey et al., 2007; Rosenberg et al., 2011).

    • Halokinetics and other features of GLORIA long-range sidescan sonar data from the Red Sea

      2017, Marine and Petroleum Geology
      Citation Excerpt :

      Whereas mid-grey tone in the imagery likely indicates an unconsolidated sedimentary seabed, black areas occurring behind bright areas can be caused by acoustic shadows cast by high-standing terrain (Mitchell, 1991), helping the interpreter identify areas of relief. Dredging of seabed areas in the Red Sea that have extremely low backscattering in multibeam acoustic data by one of us (NA) has recovered porous carbonate encrusted sediment, so such material (likely deposited during the Last Glacial Maximum (Almogi-Labin et al., 1998; Milliman et al., 1969)) probably occurs in many of the areas of low GLORIA backscattering, overlain by a ∼1 m drape of Holocene sediment (Mitchell et al., 2015). We interpreted the data in the central and southern Red Sea first with the aid of the new multibeam sonar data.

    View all citing articles on Scopus
    View full text