Sediment dynamics and geomorphology of a submarine carbonate platform canyon system situated in an arid climate setting

Changes in neritic carbonate production and sediment transport off platforms are related to climate variations, sea‐level fluctuations and tectonic processes. Canyon systems marking the platform slopes represent critical source‐to‐sink pathways transporting shallow‐water sediments basinward. However, these export systems and related processes are primarily studied on platform slopes in humid to tropical climate settings. A newly discovered canyon system on the leeward margin of the Al Wajh platform (north‐east Red Sea) represents the ideal laboratory to investigate source‐to‐sink pathway dynamics in an arid climate that prevailed since the Late Pleistocene. A high‐resolution bathymetry map was established to characterize the slope morphology. The system displays a U‐shaped, 10 km long main channel dominantly sourced by the north‐west/south‐east running outer channel and two smaller 2 to 3 km long canyons. The latter are positioned perpendicular to the main canyon. A 4 km wide head scarp at the reef edge and dozens of amphitheatre‐shaped scarps along the mid to lower slope suggest significant slope failures over time. The analysis of four sediment cores collected on a profile down the canyon revealed sedimentation rates of 26 cm/ka at the mid‐slope to 9.4 cm/ka in the basin. Three main sediment‐export processes were identified: (i) sandy and neritic component‐poor turbidites; (ii) winnowing of strontium‐rich carbonate fines through surface currents; and (iii) remobilized carbonate fines on the upper slope. As of the Last Glacial, turbidites are predominantly deposited during times of significant sea‐level instability, both rises and falls, whereas their flat‐topped‐tropical counterparts show a higher turbidite frequency during highstands. Strontium‐rich carbonate fines are exported similarly through time in both climate settings. Overall, sediment export is controlled by: (i) the platform morphology (flat‐topped versus rimmed lagoon); (ii) variations in sediment production; (iii) sea‐level variations (exposure or flooding of sediment production areas); and (iv) the interaction between the sedimentary system and atmospheric changes (sediment production and delivery).


INTRODUCTION
Sediment production and export of carbonate platforms in response to sea-level fluctuations, tectonic processes and climate variations have been discussed in many studies.The link between shallow-water sediment production, sediment export and sedimentation processes on the adjacent slopes and in the surrounding sedimentary basins has been evaluated extensively (e.g.Droxler et al., 1983;Schlager et al., 1994;Andresen et al., 2003;Paul et al., 2012;Webster et al., 2012;Counts et al., 2018).
Variations in sediments characterizing carbonateproducing sedimentary systems depend on the type of carbonate factory.In the present-day ocean they comprise: (i) the tropical (T-factory); (ii) cool-water (C-factory); (iii) microbial (M-factory); (iv) cold-water coral (CWC-factory); and (v) pelagic (P-factory) systems (Schlager, 2000(Schlager, , 2003(Schlager, , 2005;;Reijmer, 2021).The skeletal and nonskeletal components of the individual carbonate factories reflect the response of the sedimentary system to environmental variations, but also depend on the influence of fish and echinoderms eroding the skeletal elements and (micro-)algae and other biota inhabiting the sedimentary system.Environmental factors that are important for the overall sediment production include: (i) light; (ii) temperature; (iii) nutrients; (iv) salinity; (v) substrate; and (vi) carbonate saturation (Schlager, 2000(Schlager, , 2003;;Pomar, 2001).Regional sitespecific factors that may affect the carbonate factory population are the: (i) ocean currents; (ii) presence of upwelling systems; (iii) oceanatmospheric system; (iv) overall atmospheric regime; (v) shallow-water dynamics; and (vi) terrestrial sediment and water input (Schlager, 2005;Reijmer, 2021).Hence, carbonate sedimentary settings mirror site-specific environmental controls that govern the type of carbonate factory and associated style and type of sediment production and export.
The sediment export out of shallow-water environments differs depending on the type of carbonate producers present, the overall setting with leeward and windward variations, varying responses to changes in local wind and current systems, and other factors (Betzler et al., 1995;Andresen et al., 2003;Rendle-B€ uhring & Reijmer, 2005;Reijmer & Andresen, 2007;Mulder et al., 2017;Wunsch et al., 2017;Jorry et al., 2020).However, in present-day carbonate systems, sediment export also relies on the type of carbonate factory, with a carbonate rim and a protected lagoon characterizing the T-factory, a Cfactory dominated by current and wave action, or a deep-water CWC-factory that depends upon the supply of nutrients.Sea-level fluctuations play a decisive role in T-factory depositional systems, as small-scale variations (between 20 m and 50 m) already expose the main sediment production realm because this system is linked to the depth of the photic zone (Schlager, 2005).Hence, flooding and exposure cycles of the shallow-water sedimentary environment associated with eustatic sea-level fluctuations are reflected in sediment production and export over time (Schlager et al., 1994).These fluctuations are documented on the slopes and basins surrounding the shallow-water sediment production environment.The contrast for T-factories between full sediment production during sea-level highstands and sharply reduced production during sea-level lowstands is enhanced when the T-factory is positioned on steep-sided horst blocks lacking inhabitable slopes, as known from the Red Sea coastal areas.Emmermann et al. (1999) and Emmermann (2000) studied T-factory systems positioned on horst blocks, the Sanganeb and Abington Reef (western Red Sea, Sudan), and assessed this scenario in detail.The present study aligns with latter studies but details variation in sediment export.In addition, a detailed evaluation of the facies distribution of the shallow-water realm is available (Petrovic et al., 2022) that allows for an in-depth assessment of the sediment-export variations and along-slope sediment sorting processes.
The study area of the land-attached Al Wajh carbonate platform is situated at 26°N, at the northern boundary at which T-factory carbonate systems usually occur (Schlanger, 1981;Grigg, 1982).Al Wajh is positioned near the eastern coast of the north-east Red Sea, bordered by the Saudi Arabian Peninsula (Fig. 1A).The Red Sea is a young, narrow, land-locked marine rift basin with active axial zone seafloor spreading (Purser & Bosence, 1998;Augustin et al., 2021).The occurrence, morphology and hydrography of the present-day reefs are directed by extensional tectonic processes and salt diapir formation (Montaggioni et al., 1986;Dullo & Montaggioni, 1998;Bosence, 2005) and salt withdrawal (Petrovic et al., 2023).The landattached reefs are also influenced by the input of clastics delivered by flash floods through wadis entering the marine realm (Piller & Pervesler, 1989;Piller & Mansour, 1994;Petrovic et al., 2022).
Canyon and gully systems at carbonate platform slopes represent critical source-to-sink pathways.(Puga-Bernab eu et al., 2011), display another slope morphology with steep exponential slopes and large shelf-incised canyons that reach up to the cemented barrier reef at the shelf edge.The differences in slope morphology are attributed to the overall shape of the continental slope and the presence (Ribbon Reef) or absence (Noggin Passage region) of shelf-edge barrier reefs.For the Ribbon Reef, the barrier reefs form physical barriers, severely limiting the amount of sediment export, resulting in the development of steep exponential slopes.Slope systems at the Great Bahama Bank (GBB; Principaud et al., 2015Principaud et al., , 2018;;Wunsch et al., 2017) and Cay Sal Bank to the west (Wunsch et al., 2017) exhibit the interaction between slope adjustments during sea-level changes (GBB) and tectonic processes (Cay Sal Bank) and turbidity and contour currents, producing furrows and slope-parallel ridges.The series of canyons at Little Bahama Bank (Tournadour et al., 2017) results from an initial toe-of-slope failure and successive retrogressive headward erosion.Etienne et al. (2021) extensively discussed the steep slopes around the Lansdowne Bank (south-west Pacific Ocean), demonstrating a large variety of canyons, channels and gullies incising the slope.The Lansdowne Bank is also marked by a scalloped margin resulting from bank-margin collapse processes and slope failure.In analogue to the studies by Puga-Bernab eu et al. (2011) and Tournadour et al. (2017), Etienne et al. (2021) proposed that the upslope propagation of sediment bypass morphologies due to retrogressive headward erosion may have caused the bank-margin collapse.They suggested that slope adjustment processes linked to sea-level changes might have played a role.
The literature has extensively discussed that the slope stability of carbonate platforms is strongly linked to the composition of the slope deposits with mud-rich sediment exhibiting concave-upward slopes with angles of repose of below 4°and grain-rich deposits reaching values up to 35° (Kenter, 1990) along straight slope profiles.For carbonate factories, the slope morphology profiles and sediment composition display a strong link to Gaussian profiles for C-factories, exponential slopes for T-factories and linear profiles for M-factories (Adams & Kenter, 2014).In summary, variations in slope morphology along carbonate platforms rely on the: (i) variability in grain sizes of the sediments exported from the platform and deposited on the slope; (ii) effects of contour currents redistributing exported sediments; and (iii) sediment cementation processes stabilizing the platform margin (Reolid et al., 2014(Reolid et al., , 2017) ) and slope deposits (Grammer et al., 1993).
This study presents the analysis of a recently discovered canyon system positioned at the leeward margin of the land-attached Al Wajh platform (Petrovic et al., 2022;Fig. 1A and B).The study aims to: (i) analyse the canyon system morphology; (ii) document sediment export processes and variations over time; and (iii) assess the overall source-to-sink pathway of this tropical, arid climate carbonate sedimentary system (T-factory, sensu Schlager, 2005).

Geological context
The Red Sea is a young and narrow ocean basin with slow spreading rates ranging from 10 mm/ year in the north to around 15.5 mm/year in the south (DeMets et al., 2010).The initial basin opening was initiated during the Early Oligocene, whereas a second rifting phase occurred in the Late Miocene (Rasul et al., 2015).After flooding with seawater during the Early Miocene, the subsequent restricted conditions led to the deposition of a 2 to 4 km thick evaporite sequence until the Late Mioceneone of the largest evaporite basin deposits worldwide (Evans, 2006).The evaporite succession is unconformable and overlain by continental siliciclastics several kilometres thick and shallow to deep marine carbonate deposits (Hughes & Johnson, 2005;Afifi et al., 2014).
The modern morphology of the Red Sea margins is primarily influenced by fault blocks, salt tectonics and siliciclastic input derived from onshore drainage systems (Bosence, 2005;Smith & Santamarina, 2022;Petrovic et al., 2023).Narrow and gently inclined shelves in the northern Red Sea connect to steep-sloped shallow-water carbonate platforms and fringing reefs.Their formation was previously interpreted to be exclusively related to rift tectonics, salt diapirs and fan deltas (Dullo & Montaggioni, 1998;Bosence, 2005;Rowlands et al., 2014;Rowlands & Purkis, 2015).However, a recent study highlighted the additional influence of salt withdrawal on carbonate platform evolution (Petrovic et al., 2023).The land-attached Al Wajh carbonate platform is in the north-eastern part of the NNE to SSWoriented Zabargad fracture zone, which is interpreted to be an extensive transform fault system (Crane & Bonatti, 1987;Ligi et al., 2018).The platform has a nearly rectangular shape, with a length of around 55 km at the ocean-facing side and an average width of 27 km.It hosts a lagoon with a maximum depth of 43 m (Petrovic et al., 2022).The Al Wajh platform is almost entirely rimmed by a coral reef belt, including carbonate islands, while a 400 m wide main inlet channel separates the northern and southern parts (Fig. 1A and B).The northern reef belt has a maximum width of 4 km and is characterized by two additional tens of metres wide inlets (Fig. 1), small sand shoals and tidal channels tens of metres wide dissecting the reef belt.The southern rim is wider with a maximum width of 6 to 8 km and is characterized by an inner and an outer reef belt separated by an up to 10 m deep internal lagoon and several larger islands (Fig. 1; Petrovic et al., 2022).Larger tidal channels are only present towards the inner reef belt (Fig. 1).Based on its location and sediment composition, especially the abundance of coral reefs, the Al Wajh carbonate platform represents a T-factory carbonate system (sensu Schlager, 2005;Reijmer, 2021).

Climate and oceanography
Due to its location in the African-Arabian desert belt, the northern Red Sea is subjected to arid conditions and the southern Red Sea is characterized by arid-humid conditions related to the influence of the Asian Monsoon system.Northwesterly winds dominate the northern Red Sea with average wind speeds of 5 m/s (Dasari et al., 2018;Patlakas et al., 2019).During the winter, intensive south-west and eastern winds intermittently occur (up to 15 m/s; Raitsos et al., 2013), whereas local west-jet streams that are highly variable in velocity, appear in the summer months (Langodan et al., 2017;Davis et al., 2019).East winds and west-jet streams are considered to deliver iron-rich aeolian sediment to the Red Sea from Sudan and Egypt and the Arabian Peninsula, respectively (Jiang et al., 2009;Prakash et al., 2015).
The narrow (150 km wide) and shallow (À137 m deep) Strait of Bab-El-Mandeb connects the Red Sea with the Indian Ocean (Murray & Johns, 1997;Sofianos & Johns, 2002).Due to the shallow-water depths, the Red Sea is isolated from the large current systems in the Indian Ocean, leading to poor water exchange and a very slow surface-water renewal (every six years; Maillard & Soliman, 1986).During the summer months, the average surface-water temperature is 29°C AE 1°C and during the winters the values are at 23°C AE 1°C.The sea-surface salinities range from 36.5& in the south to 41& in the northern Red Sea (Sofianos & Johns, 2002).The Red Sea surfacewater circulation is controlled by winds producing eddies and sub-gyres (Eshel & Naik, 1997;Sofianos & Johns, 2007;Shaltout, 2019), as the overall Red Sea water body is characterized by an anti-estuarine-like circulation (Eshel & Naik, 1997;Peters et al., 2002;Sofianos & Johns, 2015).The prominent surface Eastern Boundary Current is flowing northward along the Arabian Shelf in the northern Red Sea (Fig. 1).The water body can be subdivided into Red Sea surface water (0 to 150 m) and Red Sea deep water (>150 m), which are both host to a large internal thermohaline cell (Eshel & Naik, 1997;Sofianos & Johns, 2015).During the winter, the Red Sea intermediate water appears (50 to 200 m; Peters et al., 2002;Yao et al., 2014;Zhai et al., 2015) and separates the Red Sea surface water from Red Sea deep water.The Red Sea intermediate water, also called Red Sea outflow water, is one of the most saline water masses in modern oceans.It is formed by mixed-layer deepening in wintertime caused by sea-surface buoyancy loss in the northern Red Sea (Eshel & Naik, 1997;Zhai et al., 2015) and flows towards the south through the Strait of Bab-El-Mandeb into the Gulf of Aden.Traces of the Red Sea intermediate water were recently detected around the Saya de Malha plateau in the south-west Indian Ocean (Betzler et al., 2021).

Palaeoclimate and palaeoceanography
The largely enclosed basin of the Red Sea makes it particularly sensitive to glacio-eustatic sea-level changes (Rohling et al., 2008(Rohling et al., , 2009) ) and orbitalcontrolled climate shifts, which have repeatedly influenced the basin during the Pleistocene glacials and interglacials (Siddall et al., 2004;Grant et al., 2014;Nicholson et al., 2020).During the Last Glacial Maximum (LGM; 22 to 19 ka), the sealevel drop of À120 m below the modern sea level restricted the water-mass exchange with the Gulf of Aden (Indian Ocean) via the Strait of Bab-El-Mandeb (Siddall et al., 2002).This limited exchange resulted in salinities of around 50 psu, exceeding the tolerance level of many planktonic organisms (Winter et al., 1983;Almogi-Labin et al., 1998).These variations also caused the formation of so-called aplanktonic zones (Fenton et al., 2000;Arz et al., 2003a;Legge et al., 2008) that lack a one-to-one relationship with sea-level fluctuations but associate with a set of processes linked to the influence of the north-east monsoon circulation within the Red Sea (Locke & Thunell, 1988;Almogi-Labin et al., 1991, 1998;Hemleben et al., 1996;Geiselhardt, 1998).These processes include the: (i) restructuring of the hydrography (stratification of the water column); (ii) productivity variations that depend on food availability; and (iii) variations in the subsurface oxygenation of the reproduction habitats.
Two prominent sapropels were formed in the subsequent transgression (at 14.5 to 13 ka and 11 to 10.5 ka, respectively; Arz et al., 2003a).Previous studies relate their formation to the combined warming and freshening of the surface water due to the post-LGM sea-level rise that stopped deep-water formation and, thus, obstructed the thermohaline circulation (Arz et al., 2003b).However, whether the sapropel formation is linked to global sea-level pulses or increased local freshwater influxes is still debated.
The so-called pluvial periods (or humid phases) represent time intervals of higher rainfall and increased river discharge in the African-Arabian desert belt during the Pleistocene (Nicholson et al., 2020).These periods are the consequence of orbital-controlled shifts of the Intertropical Convergence Zone (ITCZ) towards the north and their effect on the African and Indian Ocean monsoons.These events are welldocumented from archaeological sites, palaeolakes, speleothem records and marine cores (Grant et al., 2017;Nicholson et al., 2020).The latest documented pluvial period occurred during the Holocene (9.25 to 7.25 ka; Arz et al., 2003b).In the northward shift of the Intertropical Convergence Zone, the results of previous studies indicate a second moisture source originating from the Mediterranean area (Arz et al., 2003b;Cheddadi et al., 2021).

Satellite-based wind data
The Cross-Calibrated Multi-Platform (CCMP V2.0) was used to; produce satellite-based surface vector wind maps with a 0.25°spatial, a six-hourly (L3.0) and monthly (L3.5) temporal resolution (Atlas et al., 1996(Atlas et al., , 2011)).The zonal and meridional components, including the wind speed variable of the monthly L3.5 CCMP V2.0 data (Wentzel et al., 2015), were used to calculate seasonal mean wind directions and wind speeds for December 2017 to November 2018 (December 2017 to February 2018winter; March to May 2018spring; June to August 2018summer; and September to November 2018autumn).The resulting data were employed to produce seasonal mean wind pattern maps for the Al Wajh offshore area (Fig. 2A to D).In addition, the daily L3.0 CCMP V2.0 data with a six-hour resolution were used to analyse the prevailing wind direction and wind speed frequencies from the grid cells (minimum/maximum latitudes: 25.375/25.875;minimum/maximum longitudes: 36.375/37.125).The wind speed and direction were calculated for each grid cell and time based on the zonal and meridional components provided by the data set.Wind directions and speeds were used to create wind rose plots for all times of the year (Fig. 2E).

Multibeam data
Multibeam data were collected with the hullmounted system EM 710-MK2 (Kongsberg Gruppen, Kongsberg, Norway) during two research cruises in 2019 with the King Abdullah University of Science and Technology (KAUST) vessel R/V THUWAL.the data were processed using Qimera 2.2.5 software (QPS, Zeist, the Netherlands) to calculate a bathymetric grid with a resolution of 30 m.In addition, the Red Sea development company provided a processed LIDAR-multibeam data set covering the Al Wajh lagoon platform (À100 to 3 m).Both data sets were merged, and a dyschromatopsia-friendly colour palette was applied for map visualization (Crameri et al., 2020).The data analysis (Figs 3 to 6) was performed using Fledermaus pro 8.2 software (QPS).

Sediment core analysis
Four sediment cores were collected with a 3 m long gravity coring system (KC Denmark; recovery rate 85 to 285 cm; Table 1) and cut after recovery into 1 m sections.Sediment core PERC 001008-1 (577 m water depth) was collected from the canyon head and core PERC 001009-1 (674 m water depth) was taken from the central canyon.Sediment core PERC 0010010-1 (820 m Platform canyon system in an arid climate 2247 water depth) was positioned at the canyon mouth and sediment core PERC 0010011-1 (843 m water depth) was located in the basin in front of the channel.All cores were split into a working and an archive half.The cores were described regarding their boundaries, colour, bioturbation, sediment sorting and other aspects, and the sediments were classified according to Dunham (1962).Samples were taken every 5 cm and divided into three equal portions: one portion was archived, and the other two portions were used for X-ray powder diffraction (XRD) and stable isotope analysis.

X-ray fluorescence
Archive halves were scanned using an X-ray fluorescence (XRF) Core Scanner III (Avaatech Serial No. 12; Dodewaard, The Netherlands) with a 1 cm resolution at MARUM (Center for Marine Environmental Science at the University of Bremen, Germany).The split core surface was covered with a 4 lm thin SPEXCerti Prep Ultralene 1 foil (SPEXCerti Prep, Metuchen, NJ, USA) to avoid contamination of the XRF measurement unit and desiccation of the sediment during scanning and storage.The reported data were acquired using a silicon drift detector (Model SiriusSD â D65133Be-INF with 133 eV X-ray resolution; SGX Sensortech, Corcelles-Cormondr eche, Switzerland), a Topaz-X highresolution digital multichannel analyser, and an 100 W Neptune x-ray tube with rhodium (Rh) target material (Oxford Instruments, Abingdon, UK).The downcore and cross-core slit sizes were 10 mm and 12 mm, respectively.Settings for the first run were 30 kV and 0.5 mA with an 8 s sampling time and a Pd-thick filter.The second run was done without a filter at 10 kV, 0.08 mA and 8 s.The average dead times were 24% and 38% for the 30 kV and 10 kV runs, respectively.Raw data spectra were processed by analysing the X-ray spectra using the iterative least square software (WIN AXIL) package from Canberra Eurisys.In addition, the data were corrected for dead time.

X-ray powder diffraction
Bulk samples (n = 161) were wet-sieved to separate the fine fraction (63 lm) from the coarser grains.Subsequently, the fine fraction was dried at 45°C and powdered with a mortar to homogenize the sample.The mineralogy was measured using a D8 Quest (Bruker, Billerica, MA, USA) and the powder was scanned using a 2Θ angle between 20°and 80°with a speed of 0.040°/s.The results were analysed using Profex 4.0.2software (https://www.profex-xrd.org/).The alignment of the instrument was checked using Al 2 O 3 as a standard.
The stable isotope samples were wet-sieved using a 63 lm mesh sieve.The fraction <63 lm was collected in small bowls and left to settle.After settling, water was decanted, and the samples were dried at 45°C.The coarse fraction was also dried, and five to seven specimens of the planktonic Globigerinoides sacculifer was selected and stored in vials.In addition, two to five specimens of the epipelagic pteropod species Creseis acicula was selected in parts of core PERC 001011-1 in which planktonic foraminifera were absent; the pteropods were also stored in vials.A total of 121 planktonic foraminifera and 50 pteropod samples were analysed for carbon and oxygen isotope variations at the Earth Sciences Stable Isotope Laboratory of the Vrije Universiteit Amsterdam, The Netherlands.Values are reported relative to VPDB.Powdered samples were reacted on a Thermo Finnigan GasBench II (Thermo Fisher Scientific, Waltham, MA, USA) and analysed using the continuous flow mode on a Thermo Finnigan DeltaPlus isotope ratio mass spectrometer (Thermo Fisher Scientific).The sample material, three to four planktonic foraminifera or one to two pteropods, was inserted in 10 ml Exetainerâ vials and sealed and flushed with helium (He).The samples were heated to a temperature of 45°C, followed by injection of anhydrous phosphoric acid.After a 24 h reaction time, the CO 2 -He gas mixture from the headspace of the sample vial was introduced into a GasBench II (Thermo Fisher Scientific) in a He stream.Water was separated using a Nafion trap and CO 2 was purified using a PoraPLOT Q capillary column (Agilent Technologies, Santa Clara, CA, USA) before introduction into the DeltaPlus spectrometer.
A CO 2 reference gas with a known isotopic ratio bracketed each analysis during every measurement run.This reference gas is used to determine the d 13 C (VPDB) and d 18 O (VPDB) values of the sample.A sample size correction was performed for every run (10 runs, each consisting of 60 measurements) using an in-house carbonate standard (VICS).Accuracy was monitored using International Atomic Energy Agency reference material (IAEA-603; with official values of +2.46& for d 13 C and À2.37& for d 18 O), which is measured as a control standard (60 measurements for each of the 10 runs).The standard deviation of the measurements for d 13 C and

Radiocarbon dating
Three accelerator mass spectrometer radiocarbon (AMS-14 C) dates were used to generate an age framework.Samples were selected to date three organic-rich layers (Mollenhauer et al., 2021).The AMS-14 C dating of sediment total organic matter (Table 2) was performed at the Alfred Wegener Institute for Polar and Marine Research (Bremerhaven, Germany).

Al Wajh Outer Channel and south-west slope morphology
The north-west/south-east running Outer Channel in front of the Al Wajh carbonate platform has a length of 27 km and is bounded to the south-west by the outer reef complex (Fig. 3A).
Confined by the reefs, channel walls can reach a maximum angle of 50°.On its north-western end, the Outer Channel is connected to the open ocean via two smaller inlets (400 m and 650 m wide) bounding small, elongated patch reefs, which are part of the northern reef complex (Fig. 3A).The maximum water depths in the north-western part vary from 20 to 25 m.Towards the south-east, the seafloor morphology is relatively smooth, as the water depth increases to an average depth of 33 m.The Outer Channel width increases to 2.7 to 3.9 km.The main inlet channel of up to 400 m wide of the Al Wajh platform connects the lagoon with the Outer Channel (Fig. 3A).The inlet channel exit is characterized by a sharp, tongueshaped seafloor morphology followed by a scour of 1.4 to 0.9 km in size and 4.5 m deep.Between the southern end of the scour and the outer reef complex, an array of submarine sediment waves are present over a length of 1.5 km with individual heights of up to 0.3 m, and wavelengths varying between 30 m and 35 m (Fig. 3B).Due to the S-reef complex, the Outer Channel narrows to an average width of 650 m (minimum width: 500 m).It widens again on its south-eastern end (1.7 to 2.2 km).The narrowest part of the Outer Channel hosts a few elongated patch reefs at water depths of around 20 m (confirmed by scuba diving; see also Chalastani et al., 2020).A smaller reef complex on the south-east Outer Channel exit leads to the shallowing of the channel, ranging from 13 to 18 m in water depth (Fig. 3A).The south-east exit connects the shallow-water Outer Channel with the platform slope and an associated canyon head.
At the Outer Channel exit on the platform top (Fig. 4A to C), the upper slope (31 to 260 m water depth) has a maximum angle of 67°(P1, Fig. 4D).Continuing towards the south-east, along the modern reef crest (Fig. 4B), the upper slope steepens to 72°to 74°(P2 and P3, Fig. 4D).Towards the mid-slope (260 to 700 m water depth), the morphology along the entire southern platform margin flattens to an angle between 3°a nd 1°, and the slope widens towards the southeast (Figs 4C and 5).The slope forms a crescent, pool-like shape that is up to 2 km wide, 3.5 km long, 26 to 29 m deep, and bounded towards the south-west by a ridge (Fig. 5A to C).The reef crest above has a crescent-shaped and 4 km wide head scarp (Fig. 5).Moving downward, the mid-slope morphology displays steep terraces with angles of up to 26°(Fig.5).
The mid-slope has an exponential shape with a decline in slope angle from 67°to 4°(Fig.4D) and a deep-incised canyon system disconnected from the bank edge.A 10 km long main canyon running north-west to south-east connects to 100 m wide gullies ranging in length from 1 to 3 km (Fig. 4A).All gullies start at water depths ranging from 330 to 350 m.Three shorter canyons, 2 to 3 km in length, are connected to short (hundreds of metres) and narrow (30 m) U-shaped gullies.These gullies run perpendicular (north-east to south-west) to the main canyon along with disconnected and U-shaped gullies.The gullies are several hundreds of metres long and are set at a water depth ranging from 400 to 490 m.The morphology of the main canyon varies from a kilometres-wide canyon head with gently incised U-shaped gullies to an up to 100 m deep and hundreds of metres wide U-shaped channel with steep slopes on both sides (24°to 28°; Fig. 6).
Compared to the inclination of 7°to 2°in the thalweg of the main canyon (Fig. 4), both smaller canyons have a steeper inclination of 3°to 11°.One of the canyons displays a prominent morphological feature with a seafloor inclination of 13°(P3, Fig. 4D).Towards the canyon mouth, the morphology changes from symmetrical to asymmetrical, with a gently inclined canyon wall in the west (8°) and a steeper wall in the east (28°; Fig. 6).At the lower slope (>750 m water depth), the morphology flattens until it reaches the basin (Fig. 4).The mouth of the main canyon is marked by three depressions, 5 to 15 m deep (D1 to D3, Fig. 4B).A slightly elevated sediment lobe (2 to 3 m high) connects the distal site to the canyon mouth (Fig. 4C).Collapse features such as reef blocks are absent on the slope and in the basin.However, the entire mid to lower slope is marked by dozens of prominent amphitheatre-shaped scarps facing the main canyon and the basin in the south (Fig. 4C).These scarps occur in all water depths on the mid-slope and vary in size.
On the eastern side of the main canyon, the scarp widths range from 200 to 850 m with a height of up to 65 m.At the western canyon side and on the basin-facing side, the scarp widths range from 670 to 1800 m, with heights varying between 50 m and 120 m.

Sediment core analysis
Based on the known ages of the Red Sea sapropels (RS1a: 14.5 to 13 ka and RS1b: 11 to 10.5 ka; Arz et al., 2003a), the new radiocarbon dates (Table 2) and d 18 O isotope data, a robust stratigraphic framework has been established that was used to distinguish Late Pleistocene and Holocene sediments (Figs 7 and 8).

Late Pleistocene succession
Late Pleistocene sediments were recovered in core PERC 001011-1, with a total thickness of 190 cm (285 to 95 cm; Fig. 7 and Table 2).Generally, the major interval (285 to 135 cm) comprises a beige to light-olive-coloured bioturbated and moderately sorted wackestone succession with common echinoderm spines, benthic foraminifera and pteropod shells, as planktonic foraminifera only occur in the lower part (285 to 250 m).
The interval displays relatively constant strontium (Sr)/calcium (Ca), potassium (K)/(Ca/Ca) and Ca/titanium (Ti) ratios (Fig. 8).At 255 cm core depth (Fig. 7), the aragonite content decreases by 22% and the calcite content increases by 15%, whereas the quartz content remains constant (19.2% average).The major interval also contains a series of beige to greyishcoloured, moderately to poorly sorted wackestone to packstone intervals and two floatstone layers (Figs 8, 9A and 9B) with an erosive base.These bioclastic sand layers often display grading and comprise dominantly pteropod shells, common echinoderm fragments, undefined bioclasts, benthic and planktonic foraminifera, minor calcareous algae, bryozoan fragments and quartz grains (Fig. 9C and D).These layers display positive Sr/ Ca peaks, negative K/(Ca/Ca) peaks, and positive Ca/Ti peaks (Fig. 8), which often agree with an increase in the aragonite content of around 20%.
In addition, a beige-coloured, 9 cm thick layer appears in the centre of the section (220 to 211 cm) with pteropod shells embedded in several 2 cm thick aragonitic crusts surrounded by fine-grained carbonate mud.A layer with similar characteristics, measuring 25 cm in thickness, completes the top of the 285 to 135 cm interval.
An array of four dark-beige to greyish and moderately to poorly sorted floatstone layers overlie the cemented layer (at the 135 to 95 cm interval) and alternate with beige-coloured, bioturbated and moderately sorted wackestones.The wackestone layers range in thickness from 1 to 2 cm.They contain pteropod shells, benthic foraminifera, echinoderm fragments, undefined shell fragments, planktonic foraminifera, minor calcareous algae and bryozoan fragments, as the upper two layers are slightly bioturbated (Fig. 9B).The wackestones have element ratio characteristics comparable to the coarse-grained layers in the lower interval (Fig. 8).In this array, the quartz content decreases towards the top to a maximum of >1%, as the aragonite content increases by 20%.The Late Pleistocene succession is completed (120 to 103 cm) by an olive to blackish-coloured, organic-rich layer (RS1a, 17 cm thick, Fig. 9E), displaying a positive Sr/ Ca peak and a negative K/(Ca/Ca) peak (Fig. 8).The organic-rich layer is succeeded by 6 cm thick, olive-grey-coloured, slightly bioturbated wackestone (Figs 7 and 8).Planktonic foraminifera occur more frequently in the upper part of the core (110 to 0 cm).Among the prominent element ratio signals of the coarse layers in the Pleistocene succession, similar prominent element peaks (including high Sr/Ca ratio) indicates several event layers.The element ratio black) and mineralogy data of the investigated cores.Stratigraphic correlation (dashed lines) based on isotope curve and identified sapropels RS1a (10.5 to 11 ka) and RS1b (13 to 14.5 ka; Arz et al., 2003a,b).Turbidites (yellow bars), ghost layers (grey bars) and lithified layers (cyan bars) are defined via the core description and X-ray fluorescence core data (also see Fig. 8).3).In contrast to the coarse layers characterized by positive Ca/Ti peaks, these 'ghost' layers have negative Ca/Ti peaks.
The pteropod-based d 18 O (VPDB) isotope curve (blue in Fig. 7) displays a gently ascending trend from 3.8& at the base to a maximum of 7.6& in the upper section just above the second cemented layer (150 to 135 cm in the core).Towards the top, the values decline and drop to 4& at the Pleistocene-Holocene boundary.Two major negative peaks (1.1& at 175 cm and 1.9& at 125 cm) agree with two graded layers, as the prominent negative peak (blue, d 18 O 0.9&) and positive peak (black, d 18 O 0.7&) appear at the top of the organic-rich layer (105 cm).

Holocene succession
The Holocene succession is recovered in almost all cores, except for PERC 001010-1, where the base is missing (Fig. 7; Table 2).From the shallowest core, the total sediment thickness decreases from 263 cm (PERC 001008-1, 577 m WD) in the canyon head in an SSE direction to 95 cm in the basin (PERC 001011-1, 843 m WD; Fig. 7).The calculated sedimentation rates (Table 4) range from a maximum of 26.0 cm/ka (PERC 001008-1) in the canyon head to a minimum of 9.4 cm/ka (PERC 0010011-1) in the basin (Table 2).
The Holocene section generally comprises a beige-coloured, bioturbated (Fig. 9F) and moderately sorted wackestone fabric succession with common planktonic and benthic foraminifera, echinoderm fragments and pteropod shells.The sediments exhibit relatively constant Sr/Ca ratios, as K/(Ca/Ca) values decrease and Ca/Ti values increase (Fig. 8).At the base, the succession starts with an olive to blackish coloured, organic-rich, wackestone to packstone layer (RS1b, 5 cm thick, Fig. 9E), displaying a positive Sr/Ca peak and negative K/(Ca/Ca) peak (Fig. 8).In the two shallowest cores (263 to 220 cm for PERC 001008-1 and 177 to 130 cm for PERC 001009-1), an increase in aragonite (from 30 to 48%) occurs in the basal part accompanied by a decrease in quartz (from 8 to <1%) and clay minerals (from 50 to 8%) towards the top.In the deepest core (PERC 001011-1), the aragonite content slightly increases (from 35 to 40%) and quartz slightly decreases (from 2 to <1%) in the same section.The entire upper part of all cores only presents a minor variation in mineral content (Fig. 7).
Individual bioclastic sand layers are only present in the shallowest (41 to 47 cm for PERC 001008-1) and the deepest (81 to 71 cm and 51 to 59 cm for PERC 001011-1) cores.The layers are beige to greyish-coloured, moderately to poorly sorted and have a wackestone to packstone texture.Similar to the layers described for the Pleistocene, the layers agree with positive Sr/Ca peaks, negative K/(Ca/Ca) peaks and positive Ca/Ti peaks (Fig. 8).In addition, layers with a negative Ca/Ti peak (Table 3) are present in all four cores.

Seasonal wind data
During an annual cycle, the prevailing wind direction is NW-NNW in the Al Wajh area (December 2017 to November 2018; Fig. 2).During the winter Platform canyon system in an arid climate 2257 period (Fig. 2A), the wind blows dominantly from the north, with average wind speeds ranging from 5 to 6 m/s.In spring, the wind changes towards the north-west, as the average wind speeds decrease to 4 to 5.5 m/s (Fig. 2B).The average wind speeds remain low during the summer, varying between 3.5 m/s and 4.5 m/s (Fig. 2C).During autumn, the wind direction changes towards the NNW, as the average wind speeds increase to 4 to 5.5 m/s.The average daily wind data also reveal maximum wind speeds of up to 12 m/s in relation to the north-west (NNW and WNW) winds, while ENE winds have minor wind speeds (Fig. 2E).Winds from the south and east seem to play a minor role during the annual cycle.However, from the ENE, the winds can reach maximum speeds of up to 12 m/s.

Morphological indications for sediment transport in the Al Wajh Outer Channel
The various elongated reefs and sand waves encountered in the Al Wajh outer channel (Fig. 3), in combination with the prevailing north-west wind direction (Fig. 2), point to a dominant southeast-directed current flow pattern and associated sediment transport.The carbonate sediments originating within the outer reef complex and the Al Wajh reef belt are transported directly to the canyon head situated at the southern Al Wajh slope.Steinhauff et al. (2021) referred to the southern Al Wajh slope as a windward margin, although they state a dominant north-west wind direction.This interpretation requires a careful re-evaluation, because the annual wind data (Fig. 2) unequivocally illustrate that the southern margin of the Al Wajh platform represents the leeward margin.
The scour in front of the reef belt inlet channel suggests a strong interaction between tidal and/ or backflow-initiated outflow water from the lagoon and the south-east flowing surface current in the main channel (Fig. 10).This interpretation is supported by the presence of sand-sized carbonate bioclasts and the absence of carbonate fines in the scour itself (Petrovic et al., 2022).In addition, sediments are largely absent within the inlet channel.Instead, sand-sized carbonate bioclasts are present at the channel exit towards the lagoon (Petrovic et al., 2022).The water outflow of the inlet varies seasonally.
During winter, intensive north-west winds increase the evaporation rate in the Red Sea (up to 2 m/year), cooling the Red Sea surface waters in the northern Red Sea (Maillard & Soliman, 1986;Sofianos et al., 2002;Sofianos & Johns, 2002;Abdulaziz, 2012), a process especially active in restricted areas, such as the Al Wajh lagoon (20 to 23°C; Zhan et al., 2021;Petrovic et al., 2022).Due to the increase in density of the lagoon water resulting from the evaporation process, the lagoon outflow water (backflow and tidal currents) sinks after leaving the main inlet channel.The dense water is also deflected towards the south by the south-east flowing currents.This water transport process leads to erosion at the base of the outlet, resulting in the formation of the scour (Fig. 10).Mulder et al. (2012a,b) and Betzler et al. (2014) described a similar mechanism of downward-cascading flows leading to erosion for the leeward slope of the Great Bahama Bank.Denser inner platform water initiates a hyperpycnal off-bank flow (Wilson & Roberts, 1992;Hickey et al., 2000).In comparison, the salinity of the water in the Al Wajh lagoon and the open ocean is comparable during the spring to autumn.However, the outflow currents, especially during spring tides and special oceanographic configurations (for example, flood tide plus westerlies), can still be strong (Zhan et al., 2021).

Downslope pathways and indications for slope failure
The concave (exponential) shape of the south Al Wajh slope, with a significant decrease in slope angle at the mid-slope, agrees with the morphology typical of a leeward carbonate platform margin (Grammer & Ginsburg, 1992) of a T-factory carbonate system (Rendle & Reijmer, 2002;Adams & Kenter, 2014).This slope shape relates to the sediment-export pattern of a T-factory in which sediment export links to sediment production in the shallow-water realm.The T-factory is typified on the leeward margins by an increased export of fines during sea-level highstands and a relatively coarse-grained export system during lowstands when the main sediment production areas, reefs and lagoon are exposed (Rendle & Reijmer, 2002;Rendle-B€ uhring & Reijmer, 2005).In combination with the mud-dominated slope sediments and lack of sediment structures linked to slope parallel currents in the cores, the concave shape suggests that contour currents have a minor influence on the sediment distribution patterns (Adams & Schlager, 2000).
Three gullies are directly connected to the main canyon head.They start at a water depth of 330 to 350 m (Fig. 4) and are disconnected from the platform edge, which might indicate that they were formed by the erosional activity of dense plunging currents (see Betzler et al., 2014;Etienne et al., 2021).The shorter south-west/northeast striking canyons (P2 and P3, Fig. 4A and D) are steeper and are only partially fed by gullies.The gullies on the south-west margin start at water depths ranging from 400 to 480 m, which is deeper than the gullies on the neighbouring slope in the west, suggesting a different origin.The gullies occur in combination with a head scarp along the reef crest, a pool-like feature and steep terraces on the mid-slope (Fig. 5).All of these features suggest that a large slump caused a partial cover and cut-back of the shorter canyons.After the slope failure, the slump moved downward, which led to the break-up of the slump mass into two pieces (Fig. 11).The remaining gap set the foundation for the formation of the Platform canyon system in an arid climate 2259 short canyons and possibly altered the thalweg of the north-west/south-east running main canyon (Fig. 11).This interpretation agrees with the asymmetrical shape of the main canyon in the lower part (Fig. 6), with steeper walls on the slope-facing side.Slope failure that resulted in the formation of downslope pathways was also described from the Lansdowne Bank (south-west Pacific Ocean; Etienne et al., 2021) and through retrogressive erosion along the Great Barrier Reef (north-east Australia; Puga-Bernab eu et al., 2011) and Little Bahama Bank (west Atlantic; Tournadour et al., 2017).
Plunge pools are common features within carbonate canyon systems (Principaud et al., 2015;Mulder et al., 2018) but are not exclusive to them (Lamb et al., 2007;Bourget et al., 2011).They are the product of cascading density flows (Hickey et al., 2000) moving down the canyon or slope eroding into bedrock (Schnyder et al., 2018;Mulder et al., 2022).The three depressions identified in front of the main canyon mouth (Fig. 4B to D) have no erosion features associated with a plunge pool, such as knickpoints, as a product of enhanced scouring (Whitaker, 1974;Bourget et al., 2011).Therefore, they are interpreted instead to result from the dynamics of mass movement forming depressions between the slump and debris-flow deposits.However, subbottom profiler or seismic data are needed to clarify their ultimate origin.
The occurrence of amphitheatre-shaped scarps on the mid to lower slope is a common feature on carbonate platform margins associated with slope failure (Mullins & Hine, 1989;Grammer & Ginsburg, 1992;Tournadour et al., 2017;Etienne et al., 2021;Recouvreur et al., 2021).Combined with the sediment lobe in the basin (Fig. 4C), their frequent appearance along the south Al Wajh margin points to a process that regularly occurs along the mid to lower slopes.The scarps are distributed above one another over the full water-depth range of the mid-slope at a water depth between 260 m and 700 m.This distribution pattern contrasts with scarps reported from other carbonate platform margins, where they often occur at defined water-depth intervals (for example, the Bahamian archipelago; Jo et al., 2015;Principaud et al., 2015;Tournadour et al., 2017).This wide water-depth distribution strongly suggests that the entire slope is at a critical angle of repose (6°to 15°).The ocean current exposure, which also can contribute to slope failure and the subsequent formation of channels (Betzler et al., 2014;Principaud et al., 2017;Wunsch et al., 2017), may play a minor role in the study area.This interpretation is supported by the absence of morphological (moats and drifts) and sedimentological evidence for slope-parallel currents.
In addition, the amphitheatre-shaped scarps on the south Al Wajh slope did not evolve into gullies or smaller canyons as a result of retrogressive erosion driven by shallow-water sediment export from the platform top, as observed for the slope at the Great Barrier Reef (Webster et al., 2012;Puga-Bernab eu et al., 2013) and Little Bahama Bank (Tournadour et al., 2017).Typically, sea-level fluctuations, sediment influx from shallow water (Wunsch et al., 2017) and a steepening stratigraphic trend (Busson et al., 2021) are mechanisms controlling slope failures.However, the non-retrogressive trend combined with the other observations suggests tectonic processes triggering slope failures on the southern Al Wajh slope.This finding matches the recently documented tectonic activities in the Zabargad fracture zone offshore Saudi Arabia over the last 4000 years (Babiker et al., 2015;Mitchell & Stewart, 2018).In addition, the succession in sediment core PERC 001011-1 (basin) displays a series of thin-layered turbidites.It is unlikely that these thin turbidites are linked to massive slope failures, which indicates that the amphitheatre-shaped scarps are probably a relic of tectonic events or larger slope adjustment processes in the past.

Sediment transport processes
The core analysis reveals two distinct kinds of event deposits in the canyon system: (i) bioclastic sand layers; and (ii) 'ghost' layers (Table 3).Due to their characteristics (erosive base and grading), the bioclastic sand layers are interpreted as gravity flows (i.e.calciturbidites; Fig. 9).The dominance of planktonic organisms and the minor presence of neritic bioclastic grains (for example, calcareous algae) points to limited off-platform transport and strongly suggests the upper and mid-slope as the main sediment source area.The continuous modern reef crest along the Al Wajh margin (Fig. 4C) most likely acts as a natural barrier to the wider reef belt with its internal lagoon.Significant tidal channels are not present that could serve as sediment-export pathways.This finding to some extent contrasts with flat-topped carbonate platforms, such as the Great and Little Bahama Bank, where shallow-water bioclasts, and/or  (Haak & Schlager, 1989;Reijmer et al., 1992Reijmer et al., , 2012)), because the platform margin displays numerous inlets.In addition, the calciturbidites are accompanied by negative Ca/Sr peaks and increased aragonite content, pointing to a shallow-water source (Great Bahama Bank; Le Goff et al., 2021).
The aforementioned differences in sediment export between the Al Wajh and the Bahamian sedimentary systems most likely depend on: (i) the present-day morphology of the sedimentary system in which the following parameters play a decisive role: (a) the antecedent topography (e.g.Gischler & Hudson, 2004;Droxler & Jorry, 2021); which also determines (b) the occurrence of shelf-edge barrier reefs; and (c) the overall distribution of intra-reef passages (Puga-Bernab eu et al., 2011, 2013;Webster et al., 2012); as well as (d) the depth of the lagoon (Schlager, 1981;Schlager & Purkis, 2013;Rankey, 2021).Together with (ii) variations in sediment production at the platform slope, margin and the lagoon (Schlager, 2000(Schlager, , 2003;;Pomar, 2001) and accompanying differences in the grain-size spectrum that can be produced and subsequently transported (Chazottes et al., 2008;Perry et al., 2011Perry et al., , 2015)).The final parameter is (iii) the interaction between the sedimentary system and seasonal to millennial changes in atmospheric (monsoon, trade winds and hurricane activity) and oceanographic conditions (shallow and deep-water ocean currents, tides).The overall climate setting which is dominantly arid for the Al Wajh carbonate platform and humid for the Bahamas, most likely plays a minor role in sediment export patterns.However, year-round variations in wind directions and associated ocean-current circulation patterns influence the sediment production, transport and export patterns leading to variations in the infill of the lagoon (Schlager & Purkis, 2013;Rankey, 2021) and differences in sediment export to the slopes and basins; compare for example the leeward and windward settings for Great Bahama Bank (Rendle-B€ uhring & Reijmer, 2005).At the same time wind directions and ocean-current circulation patterns govern the overall reef distribution (e.g.Verstappen, 1968).Hence, the balance between sediment production, the sediment Platform canyon system in an arid climate 2261 spectrum produced, and the main sediment transport direction are important as these processes ultimately, together with the sea-level variations and tectonic processes, will determine the loci of sediment deposition and thus the infill of accommodation in the lagoon, on the slope and in the basin.
The Sr-rich, aragonitic carbonate fines from the shallow water of the T-factory (Milliman, 1974;Gischler et al., 2013) are continuously exported offshore through waves and current-induced winnowing (Fig. 12; Milliman et al., 1993;Roth & Reijmer, 2004, 2005;Purkis et al., 2017).Therefore, the outer Al Wajh channel acts as the main feeder channel for the canyon head, which is supported by the presence of gullies situated directly below the channel exit (Fig. 4C).The exported Sr-rich carbonate fines settle and mix with the pelagic background sediments on the upper and mid-slopes, forming a typical periplatform ooze, leading to the negative Ca/Sr peak in the calciturbidites.Turpin et al. (2011) described a similar periplatform export of Sr-rich fines for the Miocene of the Bahamas and Jorry et al. (2020) described this for the present-day Glorieuses Archipelago (south-west Indian Ocean).
Ghost layers are characterized by a negative Ca/ Sr peak and an increased aragonite content, suggesting Sr-rich carbonate fines from the shallow-water factory as a source.In combination with the absence of any lithological evidence in the sediment cores (potentially due to bioturbation), these ghost layers are interpreted to result from winnowing of Sr-rich carbonate fines (Rankey, 2004;Jorry et al., 2020) during events, such as intensive storms (Fig. 12).However, export of carbonate fines can be just as important during fairweather periods (Dierssen et al., 2009;Lopez-Garamundi et al., 2022).The off-platform transport may result in density cascading, as observed for the Great Bahama Bank (Wilson & Roberts, 1992, 1995;Hickey et al., 2000;Roth & Reijmer, 2004, 2005;Dierssen et al., 2009).The asynchronous deposition of these layers in all four cores suggests localized deposition instead of a large-scale event influencing the entire canyon.

Sediment dynamics and control mechanisms through time
A correlation with a calibrated d 18 O (VPDB) isotope curve (Red Sea; Grant et al., 2012) indicates a maximum age of around 38 to 40 ka at the base of core PERC 001011-1 (Fig. 13).Hence, the sediments at the base of the core were deposited during Marine Isotope Stage (MIS) 3, a period marked by rapid climate changes (Siddall et al., 2008).During MIS 3 and MIS 2 (glacial period; Late Pleistocene), the climate in the Red  et al., 2009).Black bar: Aplanktic zone period (Fenton et al., 2000;Arz et al., 2003a;Legge et al., 2008).
Sea area was arid to hyper-arid with increased dust input from the Sahara Desert and the Arabian Peninsula (Hartmann et al., 2020).This climate is reflected in the core by varying quartz contents from 10 to 21% (Fig. 7).The mid to late MIS 3 is characterized by a slow but continuous sea-level fall from 60 to 80 m in the Red Sea (Rohling et al., 2009).During this period, the Al Wajh carbonate platform was fully exposed.However, the time-equivalent core interval has four turbidites (Figs 8 and 13).The dominance of planktonic organisms and minor neritic grains in the sandy layer is interpreted to result from the reworking of upper slope deposits in response to a falling sea level and a downward shift of water masses.During the succeeding MIS 2, sea level dropped further until it reached the lowest level at the LGM (À120 m; Rohling et al., 2009).The associated core section is dominated by background sedimentation with abundant ghost layers (Table 4) and well-defined turbidites in the centre of the section.As the Al Wajh platform was fully exposed during the LGM, aragonite-rich ghost layers are most likely sourced by carbonate fines remobilized from the upper slope during storm events.The basal part of this interval and the top are marked by lithified crusts (Fig. 13), indicating low sedimentation rates (Emmermann, 2000).These crusts support the substitution of organic carbonate production by inorganic precipitation of aragonite and Mg-calcite during a phase of stagnant water circulation in the Red Sea (Brachert, 1999;Emmermann, 2000).After the LGM, sea level rises rapidly, corresponding to a sequence of consecutive turbidites and increased aragonite content of background sediments in the core (Fig. 7).The turbidites occur in response to the destabilization of the slope sediments through upward shifting water mass boundaries and an increasing water load (Spence & Tucker, 1997;Reijmer et al., 2015).During the late MIS 2, the rise in sea level slows.In this period, the sapropels RS1a and RS1b formed (Arz et al., 2003a), and ghost layers were deposited.The sea level rises further in the subsequent MIS 1 (Holocene).Turbidites appear only in the first half of the Holocene.In the second-most recent period, only ghost layers are present (Fig. 13).The occurrence of the turbidites aligns with the return to normal planktonic production after restricted conditions in the Red Sea during the glacial period (Fenton et al., 2000).Slope readjustment processes of the mud-dominated slope Platform canyon system in an arid climate 2263 was re-started similar to the northern Little Bahamas Bank slope (Harwood & Towers, 1988;Kenter, 1990).
Sediment  , 2014).In the aforementioned studies, turbidites predominantly comprise neritic components, which agrees with the sediment composition of the turbidites dominated by planktonic biota from the south-west Al Wajh canyon system.In the Al Wajh canyon system, the turbidite frequency (in total 11 turbidites) is the highest during glacial periods, particularly during sea-level fall (n = 4, MIS 3) and rise (n = 4, MIS 2).In turn, the sediments are generally coarser during glacial periods and finer during interglacial periods (Fig. 13).This pattern aligns with earlier studies of tropical leeward carbonate slopes (e.g.Rendle-B€ uhring & Reijmer, 2005) in which the sediment composition is controlled by the platform-top-derived carbonate fines (Milliman et al., 1993;Reijmer et al., 2012).The increased input of fines also agrees with the frequent occurrence of amphitheatre-shaped scarps on the slope as finegrained muddy sediments can maintain low slope angles only (Kenter, 1990).
The overall aragonite content of the Holocene succession in the studied cores (Fig. 7; 40 to 42%) is lower when compared to cores obtained from the leeward slope of Great Bahama Bank (57 to 93%; Rendle-B€ uhring & Reijmer, 2005).In addition, in the Bahamas, the sedimentation rates are much higher on the upper (98 cm/ka) and mid-slope (93 cm/ka) compared to the south-west Al Wajh slope (26 cm/ka; Table 4).This difference presumably aligns with the continuous rimmed platform margin and the relatively disconnected canyon system at Al Wajh, restricting the off-platform export of fines.In contrast, the Great Bahama Bank carbonate mud displays a fairly continuous sediment export over time (Wilber et al., 1990;Milliman et al., 1993;Rendle-B€ uhring & Reijmer, 2005); this also holds for the Glorieuses Archipelago (Counts et al., 2019(Counts et al., , 2021)).The sediment-export pattern and associated slope morphology with an exponential slope agree with the characteristics discussed by Puga-Bernab eu et al. (2011) for the Ribbon Reef (Great Barrier Reef, north-east Australia), displaying a continuous shelf-edge barrier reef and associated steep exponential slope.
By analysing the differences in sedimentation patterns on the leeward and windward side of a carbonate platform it becomes clear that aforementioned differences probably relate to the interaction of: (i) water-energy (currents, long and short waves, and tides); (ii) atmospheric changes (monsoon, trade winds, upwelling); and (ii) the overall platform morphology (presentday, inherited from the last glacial).Waterenergy may vary throughout seasons, that is, monsoon variations with changing current directions and intensity (Maldives, Al Wajh), or may have a preferential orientation, that is, trade winds and associated currents (Bahamas).Sealevel fluctuations exposing and flooding carbonate platforms combined with karst processes shape the overall topography, see discussion by Schlager (1981) and recent discussion by Droxler & Jorry (2021), and will result in differences in production rates for various platform realms (lagoon, platform rim and slope), but also the grain-size spectrum (sand, silt and mud) and overall sediment budget (sand/silt/mud ratio).Combined with the inherited morphology of the platform (atoll, empty bucket, flat-top), sediment variations (grain-size spectrum and overall budget) may result in a differentiation in the infill of accommodation with either the infill of the lagoon or the preferential sedimentation on the slope-to-basin trajectory or a combination of both sedimentation patterns.These variations in sedimentation arrangements ultimately may lead to the complete infill of the shallow-water accommodation (flat-top platform; Bahamas, Great Barrier Reef) or the reduced infill (e.g.Rankey, 2021;Petrovic et al., 2022), the empty bucket scenario of Schlager (1981), as represented by atolls (Maldives, Indian and Pacific oceans; Droxler & Jorry, 2021).Hence, the carbonate platform sedimentation in arid or humid-tropical climate zones may differ, but the way in which the climate variations are translated in varying sedimentation patterns may diverge for each carbonate platform setting.

CONCLUSIONS
This paper presents the first description of a canyon system in the Red Sea.The studied system is situated on the leeward southern margin of the land-attached Al Wajh carbonate platform and is connected to a channel running parallel to the platform margin.The south-west Al Wajh margin canyons occur in an almost purely carbonate system (siliciclastic only from dust input) disconnected from the shallow-water platform sediment production sites.Bathymetry data and four gravity cores were analysed to characterize the downslope transport and sediment dynamics on the south-west Al Wajh margin as of the Late Pleistocene.The sedimentary system displayed following characteristics: 1 The north-west/south-east running outer Al Wajh channel has a length of 27 km and ends above the head of the studied canyon.An inlet connects the deep Al Wajh lagoon with the Al Wajh main channel.Dense tidal and winddriven backflow currents cascade down to form a scour at the mouth of the inlet channel.The south-west-directed sediment transport (winnowing) is controlled by surface currents generated by the prevailing north-west winds. 2 The main canyon system starts at the mouth of the Al Wajh channel, consists of a 10 km long main canyon up to 400 m wide and is connected to short gullies and two shorter canyons (2 to 3 km in length), which run perpendicular to the main canyon.All three canyons are Ushaped, with the main canyon shape changing from symmetrical in the upper section to asymmetrical in its lower part.3 Major slope failure was identified on the north-west/south-east running slope characterized by a half-round and 4 km wide head scarp.The associated massive slump altered the lower thalweg of the main canyon and ensured the formation of the two shorter canyons.4 The entire mid to lower slope morphology displays amphitheatre-shaped scarps, hundreds of metres to kilometres in width, indicating significant slope failures in the past and continuous slope adjustment.
5 Two main sediment-export processes from the platform top are identified: (i) sandy and neritic component-poor turbidites; and (ii) the winnowing of carbonate fines (ghost layers).6 In the Last Glacial Period (Late Pleistocene), turbidites are predominantly deposited in the basin during significant sea-level falls and rises.Carbonate crusts suggest ocean circulation stagnancy.The Sr-rich sediment layers (ghost layers) occur frequently and are interpreted to result from reworking and remobilizing patches of carbonate fines on the upper slope.7 The early Holocene sedimentary record is marked by the occurrence of two thick, sandy turbidites in the basin succession consisting of pelagic components.These deposits correlate with the return to normal pelagic sediment production and the ongoing sea-level rise after the Last Glacial Period.The Holocene succession is dominated by Sr-rich sediment layers (ghost layers).Sedimentation rates decrease from 26 cm/ ka at the mid-slope to 9.4 cm/ka in the basin.8 Carbonate sediment export in platform canyon systems situated in arid climate settings is overall similar to their humid-tropical counterparts.However, instead of the climate setting: (i) platform morphology; (ii) variations in sediment production; (iii) sea-level changes; and (iv) interaction between the sedimentary system and atmospheric changes (monsoon, trade winds, upwelling and hurricane activity) on seasonalscale to millennium-scale are the main controlling mechanisms.

Fig. 1 .
Fig. 1. (A) Satellite image of the Al Wajh carbonate platform and surrounding area (Google Earth, 2022), including prevailing wind (blue arrows) directions and flow direction of the Eastern Boundary Current.(B) Overview bathymetry (res.30 9 30 m).Dotted-line boxes mark the study area and cover of Figs 3 and 4, while the position of the four sediment cores 8-1 (PERC 001008-1) to 11-1 (PERC 001011-1) are marked with a red dots.Inset box: overview map of the Red Sea region, including surrounding states.

Fig. 2 .
Fig. 2. Satellite-based wind data from December 2017 to November 2018 presented as average wind speeds and directions within the seasons: (A) winter -December to February; (B) spring -March to May; (C) summer -June to August; and (D) autumn -September to November.(E) Average daily wind direction and wind speed data from December 2017 to November 2018.Black dotted box in (A) marks the study area.

Fig. 3 .
Fig. 3. (A) Merged bathymetry-lidar data set (res. 30 9 30 m) illustrating the north-west/south-east oriented channel between the Al Wajh reef belt and the outer reef complex.The channel connects the Al Wajh lagoon via the reef belt main inlet channel with the open ocean.A search in the Saudi Arabian marine archive and British Navy archives by Prof. Andrew Lambert (Laughton Professor of Naval History) suggests no anthropogenic origin of the channel.White arrows indicate the sediment transport direction towards the south-west.(B) Close-up displays submarine sand waves (res.5 9 5 m), including the north-west/south-east profile.

Fig. 4 .
Fig. 4. Slope and canyon morphology including: (A) north-west/south-east (P1-P3, below) and south-west/northeast (P1a-P1e, Fig. 5) profiles of the south-west Al Wajh canyon system.(B) The slope is characterized by dozens of amphitheatre-shaped scarps (AS) ranging from hundreds of metres to kilometres in width and three depressions (D1-D3) in front of the canyon mouth.A black arrow marks the sediment export from the shallow-water channel.The white arrows indicate the sediment influx from the shallow-water carbonate system.The larger white arrows indicate the main sediment import.(C) The seafloor inclination map highlights the gullies along the upper slope feeding the canyon system.(D) Profiles of the south-west Al Wajh canyon system (P1-P3), including gravity core locations.Large-scale mass wasting deposits such as reef blocks are absent on the slope and in the basin.

Fig. 5 .
Fig. 5. (A) Seafloor inclination map of the wide mid-slope in the south-east.Green dashed lines mark the collapsed reef crest; green arrows indicate the movement direction of the large-scale slump.(B) and (C) The pool-like shape is 26 to 29 m deep and bounded towards the south-east by a ridge feature.The subsequent slope towards the basin shows steep terraces.

Fig. 8 .
Fig.8.Stratigraphic frame including element ratio measurements of the investigated cores.Stratigraphic correlation (dashed lines) based on the isotope curve and identified sapropels RS1a (10.5 to 11 ka) and RS1b (13 to 14.5 ka;Arz et al., 2003a,b).Turbidites are marked by yellow bars.Ghost layers are marked in grey and lithified layers are highlighted in cyan.

Fig. 10 .
Fig. 10.Schematic illustration of the sediment dynamics in the outer Al Wajh channel.Denser lagoon outflow water (cyan arrows) cascades downward after leaving the outer rim inlet channel and deflects towards the south by the lighter south-east flowing current (blue arrows).The cascading of the outflow water leads to erosion and forms a scour in front of the channel mouth, whereas reworked sediments are winnowed towards the side (thin orange arrows).

Fig. 11 .
Fig. 11.Schematic illustration of a slope failure.(A) Pristine platform margin.(B) Collapsed margin: slumped slope pieces moved downward.Remaining gaps between the fragmented slope pieces set the foundation for the formation of short canyons.Blue arrows mark movement direction.

Ó
2023 The Authors.Sedimentology published by John Wiley & Sons Ltd on behalf of International Association of Sedimentologists., Sedimentology, 70, 2241-2271 fines, are frequently exported during sea-level highstands, as shown for Tongue of the Ocean and Exuma Sound

Fig. 12 .
Fig. 12. Schematic depositional model for sediment processes on a leeward platform margin.Background sedimentation is characterized by pelagic sediments and winnowed aragonite from the outer Al Wajh channel.Ghost turbidites are formed during increased aragonite export from the platform, such as during storm events and cascading density currents.Turbidites mainly consist of pelagic sediments, whereas neritic sediments are absent.RSDW, Red Sea deep water; RSSW, Red Sea surface water.

Table 1 .
(Rohling et al., 2009)ed sediment cores.Platform canyon system in an arid climate 2251 001011.The shape of the longest isotope curve related to core PERC 001011 fits well to published curves of the Red Sea(Rohling et al., 2009)and peaks can be correlated.The higher values below RS1a are characteristic for the last glacial period as the lower isotope values above RS1b agree with those marking the Holocene time interval in the Red Sea Ó 2023 The Authors.Sedimentology published by John Wiley & Sons Ltd on behalf of International Association of Sedimentologists., Sedimentology, 70, 2241-2271

Table 3 .
Turbidite and ghost layer identification.

Table 4 .
Sedimentation rates during the Holocene.

Table 5 .
Generic comparison of downslope carbonate sediment export of flat-topped and barrier platforms (Tfactory) situated in arid and humid-tropical climate settings.