Content uploaded by Ulysse Lebrec
Author content
All content in this area was uploaded by Ulysse Lebrec on Dec 13, 2021
Content may be subject to copyright.
Earth-Science Reviews 224 (2022) 103864
Available online 9 November 2021
0012-8252/© 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Morphology and distribution of submerged palaeoshorelines: Insights from
the North West Shelf of Australia
Ulysse Lebrec
a
,
b
,
*
, Rosine Riera
b
, Victorien Paumard
a
, Michael J. O’Leary
a
, Simon C. Lang
a
a
Centre for Energy Geoscience, School of Earth Sciences, The University of Western Australia, 35 Stirling Highway, Perth, WA 6009, Australia
b
Norwegian Geotechnical Institute, 40 St Georges Terrace, Perth, WA 6000, Australia
ARTICLE INFO
Keywords:
Palaeoshoreline
Seabed
Continental Shelf
Reef
Carbonate
Geomorphology
Terrace
Beach ridges
ABSTRACT
Palaeoshorelines and associated palaeo-coastal features are studied to reconstruct past sea level, climate, and
depositional environments. Their identication typically depends on direct eld observations and is therefore
challenging in the marine environment, where the interpretation mostly relies on sparse geophysical data. This
review presents, based on 118 published case studies, a summary of morphological evidence that can be used to
identify submerged relict coastal features worldwide, using only geophysical data. Four coastal feature categories
that can be used as palaeoshoreline indicators were identied: (1) beach ridges of wind and wave origin; (2)
shoreface strata; (3) marine terraces; and (4) coral-reef terraces.
In light of this proposed classication, an area of ~200,000 km
2
was investigated along the Rowley Shelf
(North West Shelf, Australia), a carbonate-dominated platform, based on the integration of high-resolution ba-
thymetry (i.e., seismic-derived bathymetry, satellite-derived bathymetry, multibeam echosounder bathymetry,
spot depth soundings) and 2D reection seismic lines. Relict features were discriminated from modern bedforms
using ve criteria: (1) stratigraphic position; (2) emersion features; (3) similarity with modern and published
analogues; (4) integration of modern ocean conditions; and (5) evidence of early cementation. In total, over 500
submerged relict coastal features were identied, making this review the most comprehensive catalogue pub-
lished to date.
Relict features are concentrated over specic depths, referred to as modal sea-level depths (MSLDs), which
correspond to depths where the relative sea level remained stable over long periods of time. Nine MSLDs are
observed at 20, 35, 50, 60, 70, 80, 90, 105 and 125 m below sea level. Each MSLD is the result of the accu-
mulation of coastal features through multiple glacial/ interglacial cycles. Most of the features may nevertheless
be related to the last glacial sea-level fall and were likely formed between Marine Isotopes Stages (MIS) 5 and 2.
The analysis of the submerged coastal features indicates that the overall shelf morphology is controlled by the
distribution of these features, and that, while in a carbonate province, their formation is related to wind, tide,
uvial and wave processes. The higher concentration of relict uvial-inuenced features at shallower depths and
of relict tide-inuenced features at greater depths suggest that uvial runoffs were limited during glacial periods.
This, in turn, supports the hypothesis of a prevalent dry climate during glacial periods and in contrast, of a humid
climate during interglacial periods. Finally, the study reveals that most modern coral reefs of the Rowley Shelf
are growing on top of relict coastal features and that seabed ridges previously interpreted as drowned coral reefs
are, in fact, likely to be relict coastal features formed through clastic processes.
Results from this study will support the identication of submerged palaeoshorelines on continental shelves
around the globe and highlight the inuence of associated relict coastal features on shelf morphologies. Addi-
tionally, this study provides new insights on processes shaping carbonate provinces.
* Corresponding author at: Centre for Energy Geoscience, School of Earth Sciences, The University of Western Australia, 35 Stirling Highway, Perth, WA 6009,
Australia.
E-mail address: ulysse.lebrec@ngi.no (U. Lebrec).
Contents lists available at ScienceDirect
Earth-Science Reviews
journal homepage: www.elsevier.com/locate/earscirev
https://doi.org/10.1016/j.earscirev.2021.103864
Received 23 May 2021; Received in revised form 14 August 2021; Accepted 5 November 2021
Earth-Science Reviews 224 (2022) 103864
2
1. Introduction
Coastal areas are dynamic environments shaped through marine,
terrestrial and atmospheric processes (Davidson-Arnott, 2010; Wood-
roffe, 2002). Resulting coastal features have distinctive morphologies
that form in response to the dominant meteorological and oceano-
graphic (metocean) processes, extreme events, and longer-term climate.
Shoreline position is xed at the interface between the land and water
domains (Bird, 2008; Oertel, 2005) and changes with relative sea level
and local sediment supply.
The identication, characterisation and age dating of relict coastal
features, and of associated palaeoshorelines, provide substantial infor-
mation to reconstruct the evolution of past relative sea level, which in
turn can be used to highlight past eustasy, tectonic displacements and
subsidence (Laws et al., 2019; Taylor, 2019; Williams et al., 2018).
Additionally, past coastal environments and climate played a central
role in the development of early human communities through the
exploitation of coastal resources (Erlandson and Fitzpatrick, 2006; Rick
and Fitzpatrick, 2012). In that regard, mapping palaeoshorelines could
help in the identication of indigenous underwater cultural heritage
sites and shed light on past human migration routes (Benjamin et al.,
2020; Braje et al., 2019; O’Leary et al., 2020; Wiseman et al., 2021).
Finally, coastal features act as depocenters for sediments eroded from
adjacent hinterlands and may therefore locally exhibit economically
viable concentration of heavy minerals (Roy, 1999) or gemstones
(Phillips et al., 2018; Runds et al., 2019; Spaggiari et al., 2006). Pres-
ently, the sea level is higher than during most of the Quaternary (Grant
et al., 2014; Miller et al., 2020; Miller et al., 2005) suggesting that, in
tectonically stable areas, the majority of relict coastal features are now
submerged.
Palaeoshoreline features can be identied from morphological, bio-
logical, sedimentological, stratigraphical or archaeological evidence
(Evelpidou and Pirazzoli, 2015). However, most of these indicators
require either direct observation of coastal features or access to in-situ
measurements (e.g., sediment or rock samples), which make their
detection and interpretation challenging in the marine environment.
The identication of submerged coastal features often relies on ba-
thymetry, shallow seismic surveys and identication of similar
morphological, stratigraphical, and sedimentological patterns on mod-
ern analogues (e.g., modern coastlines). The interpretation of these
patterns and the nomenclature used to describe them varies signicantly
depending on the approach chosen by the investigator. For example,
some investigators consider, based on eld observations, that beach
ridges are restricted to the backshore and foreshore (i.e., supra- to
intertidal, Otvos, 2000), whereas others, following the inclusion of
geophysical data, also include shoreface deposits (i.e., subtidal) as part
of the beach ridges (Ainsworth et al., 2019; Ainsworth et al., 2011;
Tamura, 2012; Vakarelov and Ainsworth, 2013). Similarly, the term reef
is used either to describe a topographic high lying beneath the surface of
a natural body of water (e.g., nautical reef, rocky reef), a marine topo-
graphic high colonised by corals, or a topographic high that is built
exclusively by corals and algae (Fagerstrom, 1987; Finkl, 2013; James
and Macintyre, 1985; Montaggioni, 2001; Roberts et al., 2006). As a
result, coastal terms like beach ridges, coastal dunes, barriers, foredunes
and reefs are used interchangeably, or the same term is used to describe
different features (Mauz et al., 2013; Otvos, 2000; Tamura, 2012), hence
demonstrating the need for a standardised approach that would support
the identication and characterisation of morphological markers asso-
ciated with submerged palaeoshorelines.
Australia’s North West Shelf (NWS) is located in the far eld (i.e., far
from the former ice margins, sensu Lambeck et al. (2014)), making it the
perfect laboratory to study past sea level and the associated sedimentary
features (Brooke et al., 2017; Haworth et al., 2002). Based on scattered
high-resolution surveys and regional low-resolution bathymetry, several
possible palaeoshorelines were previously identied along the shelf
(Hengesh et al., 2011; James et al., 2004; Jones, 1973). However,
mainly due to the limited extent and/or resolution of the geophysical
data available to the authors, the extent and the nature of these features
remained unknown. Recently, Lebrec et al. (2021), complemented the
existing data collection with newly processed seismic-derived and
satellite-derived bathymetry improving by up to 800 times the resolu-
tion of the regional bathymetry. This new dataset now allows for a more
detailed investigation of submerged palaeoshorelines on the NWS and
offers the opportunity to establish the processes that led to their for-
mation and preservation.
We rst present, using published sources, a global synthesis of sub-
merged coastal features and associated morphological indicators of past
sea level, with a particular focus on late Pleistocene and Holocene fea-
tures where sufcient data coverage and resolution is available. A more
detailed investigation of submerged relict coastal features covering the
NWS of Australia is then presented, where the newly processed
geophysical surveys help to: (1) fully characterise submerged coastal
features at the scale of a continental shelf; (2) describe processes leading
to their preservation post inundation; and (3) highlight the role of relict
coastal features on modern-days environments. This paper also presents
morphological criteria to help submerged palaeoshorelines and associ-
ated relict coastal features around the globe and discusses the processes
driving tropical carbonate shelf morphologies.
2. Palaeoshoreline indicators described worldwide
Climate, sediment source and supply, main coastal and metocean
processes, and uctuations of relative sea level all inuence the nature
of coastal features, their position on the continental shelf, and their
potential for preservation, resulting in a large array of submerged
palaeoshoreline features deposited at various depths on the world’s
continental shelves. The present review integrates the descriptions from
118 studies whose authors have unambiguously identied late Quater-
nary submerged shoreline(s) from geophysical surveys (Fig. 1).
Coastal and marine features, which are not identiable from
geophysical surveys alone (e.g., tidal notches, peat layers), or whose
relationship to the sea level cannot be conrmed solely based on their
morphologies (e.g., aeolian parabolic and transverse dunes), were not
included in the review. In this scope, coral reefs, which can thrive in a
large range of water depths and typically require species determination
to reconstruct the relative sea level (Hibbert et al., 2016), were not
considered as a morphological relative sea-level marker. A notable
exception are coral reef structures forming distinctive terraces or plat-
forms, suggesting that a reef is vertically constrained by sea level, that
have been used as relative sea-level morphologic indicators by several
authors (references thereafter).
In an attempt to propose a unied and standard nomenclature, the
original descriptions from the authors were converted into categories
reecting the nature and the genesis of the coastal features. While this
approach may lead to an over-simplication of the descriptions and
interpretations performed by the authors, it allows the identication of
key characteristics and discriminating morphologies. The review led to
the identication of four categories of direct morphological indicators of
palaeoshorelines: (1) beach ridges; (2) shoreface strata; (3) marine ter-
races; and (4) coral-reef terraces. The list of all studies integrated in this
review is available in supplementary material.
2.1. Beach ridges
Beach ridges are elongated sub-parallel coastal features formed
through wind (i.e., backshore foredunes) or wave (i.e., foreshore to
backshore berm ridges) processes (Hesp et al., 2005; Otvos, 2000;
Tamura, 2012). Individual beach ridges are often developed in pro-
gradational sequences resulting in the formation of regressive strand-
plains. Beach ridges are a robust indicator of local palaeo sea level
because their morphology is directly constrained by the maximum
height of constructive waves. In order to use beach ridges as a relative
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
3
Fig. 1. Distribution and nature of submerged relict coastal features reported in the review. Numbering and associated references are listed in supplemen-
tary material.
Fig. 2. Relict coastal features and associated palaeoshoreline indicators identiable from geophysical surveys. A) Wave-built beach ridges, modied from Tamura
(2012); B) Wind-built beach ridges, modied from Tamura (2012); C) Barrier complex formed from wave-built beach ridges; D) Barrier complex formed from wave-
built beach ridges, with an aeolian cap; E) Barrier complex formed from a drowned relict wind-built beach ridge; F) Shoreface strata, modied from Patruno and
Helland-Hansen (2018); G) Marine terrace, modied from Blanchon and Jones (1995); H) Coral-reef terrace, modied from Blanchon and Jones (1995).
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
4
sea-level indicator it is however necessary to understand the genesis of
the ridges as their position, with respect to the sea level, varies
depending on the coastal processes involved in their formations
(Tamura, 2012).
2.1.1. Wave-built ridges
Wave-built ridges can be formed through the accumulation of gravel
during storm events or sand in fair-weather conditions (Tamura, 2012).
Occasionally, wave-built beach ridges can be formed through longshore
bar welding. In any case, the top of the ridge represents the maximum
height of constructive waves (Fig. 2A, Tamura, 2012). Submerged,
relict, wave-built beach ridges have been identied: (1) in the Medi-
terranean Sea (De Falco et al., 2015; De Santis et al., 2020; Maselli et al.,
2011; Storms et al., 2008; Ulzega et al., 1986; Zecchin et al., 2015); (2)
in the Marmara Sea (Ergin et al., 1997); (3) in the Baltic Sea (Bennike
et al., 2000; Hansson et al., 2018; Jensen and Stecher, 1992); (4) along
the Northern Atlantic Ocean (Belknap and Kraft, 1981; Fern´
andez Salas
et al., 2016; Forbes et al., 1991; Gayes and Bokuniewicz, 1991;
McMaster and Garrison, 1967; Mellett et al., 2012; Oldale, 1985; Oldale
et al., 1993; Shaw et al., 2009; Wellner et al., 1993; Westley et al., 2011);
(5) along the western Canadian Shelf (Barrie and Bornhold, 1989; Barrie
et al., 1991); and (6) offshore Namibia (Runds et al., 2019). The large
majority of wave-built beach ridges have a relief of less than 5 m. They
are generally preserved through drowning, also referred to as over-
stepping (see supplementary material), when sediments are coarse
enough to resist wave erosion. In such instance the sea level rises quickly
enough, probably due to meltwater pluses, so that the beach ridges end
up below the limit of wave energy, hence reducing their erosion and
increasing their potential for preservation. In some cases, the formation
of beachrock through vadose cementation can facilitate preservation
(Alc´
antara-Carri´
o et al., 2013; Liquete et al., 2007; Pretorius et al.,
2017). The age of submerged wave-built ridges has been linked through
radiocarbon dating (or inferred) with the period of post glacial sea-level
rise (see supplementary material), possibly because the morphology of
older features is not fully preserved or has been overprinted with
younger features. The only notable exception is from Wellner et al.
(1993) who reported dates of 55 ka BP along the US Atlantic coast.
However, these dates were inferred based on seismic stratigraphy, and
the ridge was almost fully abraded, so the interpretation mostly relies on
vibrocore samples description instead of seismic morphologies.
2.1.2. Wind-built ridges
Wind-built ridges are formed behind the beach whenever winds are
of sufcient velocity to transport sediments from the beach to its lee
(Short, 2020). They can form as either transgressive or regressive coastal
deposits (Abegg et al., 2001). The base of the ridge typically represents
the maximum height of constructive waves (Fig. 2B, Tamura, 2012).
Extensive submerged wind-built ridges were reported in: (1) the Medi-
terranean Sea (Goff et al., 2018; Gzam et al., 2016; Mart and Belknap,
1991; Micallef et al., 2013; Sade et al., 2006); (2) eastern South Africa
(Cawthra et al., 2013; Green et al., 2012; Green et al., 2013; Green et al.,
2017; Martin and Flemming, 1986; Pretorius et al., 2016; Pretorius
et al., 2019; Ramsay, 1994; Salzmann et al., 2013); (3) Mozambique
(Wenau et al., 2020); (4) Australia (Beaman et al., 2005; Brooke et al.,
2010; Brooke et al., 2014; Nicholas et al., 2014; O’Leary et al., 2020;
Passos et al., 2019; Sprigg, 1979); (5) India (Rao et al., 2001); (6) Florida
(Finkl and Andrews, 2008; Gardner et al., 2007; Jarrett et al., 2005;
Locker et al., 1996); (7) Bermuda (Stanley and Swift, 1967); and (8)
Turkey (Ocako˘
glu et al., 2018). Submerged wind-built beach ridges
exhibit a large variety of sizes, with heights ranging from a few to tens of
meters, spanning over areas of hundreds to thousands of meters. Such
features were exclusively preserved through early cementation as aeo-
lianites, suggesting that un- or poorly consolidated ridges are eroded
during submersion. A possible explanation is that wind-built ridges are
typically composed of ner grains than their wave counterparts, given
that smaller grains can be transported through aeolian processes and are
therefore more susceptible to remobilisation and erosion (Arens et al.,
2002). Interestingly, the distribution of submerged wind-build ridges
around the globe matches closely the distribution of emerged carbonate
aeolianites reported by Brooke (2001) but differs from the distribution
of modern coastal wind-built dune presented by Martínez and Psuty
(2004). Such discrepancy between the modern coastal dunes and their
relict counterparts would tend to conrm that, while coastal dunes are
present globally, they can only be preserved locally where early
cementation is possible. Wind-built ridges are reported through the rst
ve marine isotope stages during both relative sea-level rise and fall,
hence covering larger time periods than wave-built ridges (see supple-
mentary material). The preservation of relict wind-built beach ridges
across glacial-interglacial cycles highlights the important role that
cementation plays in preserving these seabed features.
2.1.3. Discriminating wind-built and wave-built ridges
The discrimination between wave-built and wind-built ridges can be
problematic without cores, as both types of ridges can have similar
beddings (Tamura, 2012). Aeolian ridges are expected to exhibit steeper
bedding than wave-built ridges. However, exceptions to this rule are
common, especially along reective or coarse grained beaches (Otvos,
2000) and potentially in the case of low-angle established foredunes (e.
g., Hesp, 1988). Several authors reported the presence of an erosional
boundary at the base of the upper shoreface associated with the
migration of longshore trough (Fraser et al., 2004; Tamura, 2012;
Tamura et al., 2007). Such surfaces were not reported in submerged
cases, however, their identication would allow to conrm the wave
origin of a feature. Alternatively, the identication of blow outs or
washover deposits would suggest that the ridge was above sea level
during its formation, and therefore is likely of aeolian origin (Gardner
et al., 2007). Finally, reported submerged wind-built ridges seem to be
signicantly taller than their wave counterpart (less than 5 m vs up to
tens of meters). While this can be used to categorise the largest dunes as
having an aeolian origin, there is no clear boundary for
intermediate-size objects.
A more systematic approach to discriminate between submerged
wind-built and wave-built ridges could be achieved through the iden-
tication of the processes involved in the preservation of submerged
ridges. Indeed, all reported submerged wind-built ridges are preserved
through early cementation, whereas most of the reported wave-built
ridges have been argued to be preserved through drowning (see sup-
plementary material). The identication of diagenetic features from the
seismic would therefore suggest an aeolian origin. In any case, it may not
always be possible to infer the origin of beach ridges without samples.
2.1.4. Barrier complexes
In several instances, beach ridges are detached from the mainland to
form a barrier enclosing lagoons, bays or estuaries. Depending on
whether a barrier is attached by one, two or none of its ends to the
mainland it will be referred to as a barrier spit, barrier beach or barrier
island respectively (Cope, 2004). Barriers are generally associated with
wave-built beach ridges (Fig. 2C, Alc´
antara-Carri´
o et al., 2013; Barrie
and Bornhold, 1989; Belknap and Kraft, 1981; Bennike et al., 2000; De
Falco et al., 2015; De Santis et al., 2020; Forbes et al., 1991; Hansson
et al., 2018; Jensen and Stecher, 1992; McMaster and Garrison, 1967;
Mellett et al., 2012; Oldale, 1985; Oldale et al., 1993; Runds et al., 2019;
Storms et al., 2008; Wellner et al., 1993), but can also occur in associ-
ation with wind-built ridges. For the latest, two congurations are
possible: (1) wind-built ridges are developed at the same time as the
barrier (Fig. 2D, Gardner et al., 2007; Gardner et al., 2005; Green et al.,
2012; Green et al., 2013; Jarrett et al., 2005; Pretorius et al., 2016); or
(2) a relict cemented wind-built beach ridge becomes a barrier following
a relative sea-level rise (Fig. 2E, O’Leary et al., 2020; Passos et al., 2019).
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
5
2.2. Shoreface strata
Shoreline clinoforms as dened by Patruno and Helland-Hansen
(2018) and Pellegrini et al. (2020) are ubiquitous inclined strata, 5 to
40 m high, extending over hundreds to thousands of meters seaward.
The formation of shoreline clinoforms can be subject to a variable degree
of uvial, wave or tidal inuence, resulting in several types of clino-
thems such as deltaic mouth bars, wave-built beach ridges, and barriers
(Patruno and Helland-Hansen, 2018; Vakarelov and Ainsworth, 2013).
The position of the shoreline is typically dened by the topset (proximal
low angle section) to foreset (intermediate, steep section) rollover point,
which corresponds to the point of maximum curvature (Fig. 2F, Pelle-
grini et al., 2020). The term ‘shoreface strata’ is hereafter used to refer to
shoreface inclined strata of unspecied origin typically corresponding to
the foreset of shoreline clinoforms. While terms such as shoreline cli-
noform and foreset encompass coastal features formed through various
processes, they are mostly used in the literature to describe coastal
features associated with delta fronts. Delta fronts, as palaeo relative sea-
level indicators, were described in: (1) the Mediterranean Sea (Díaz
et al., 1990; Rabineau et al., 2006); (2) the Gulf of Mexico (Gardner
et al., 2007; Gardner et al., 2005; Rogers et al., 2009); (3) the Marmara
Sea (G¨
okas¸an et al., 2005); (4) offshore Brazil (Rangel et al., 2020); and
(5) South Africa (Engelbrecht et al., 2020). Through all reported cases,
shoreface strata is preserved via drowning of the deposits during a rapid
relative sea-level rise.
2.3. Marine terraces
Marine terraces are erosive features formed through the repeated
mechanical action of the waves along the coast (Anderson et al., 1999;
Bradley, 1958; Kennedy, 2015) and dominantly developed during
relative sea-level stillstands (Chaytor et al., 2008). Such terraces being
purely erosional, they can form under any climatic conditions and are
observed at all latitudes (e.g., Barboza and Tomazelli, 2003; Barrett,
1962; Barrie et al., 1991; Bezore et al., 2016; Chang Hwan et al., 2013;
Galparsoro et al., 2010; Gomes et al., 2020; Lebesbye and Vorren, 1996;
Moawad, 2013; Ricchi et al., 2018; Savini et al., 2012; Siddiquie, 1975).
As a result, marine terraces were extensively described in the literature
and represent nearly half of the identied submerged shorelines (Fig. 1;
see supplementary material for details). Marine terraces are referred to
interchangeably as terraces (Arai et al., 2016; Fairbridge and Stewart,
1960), marine terraces (Bilbao-Lasa et al., 2020; Ricchi et al., 2018),
abrasion platforms (Laws et al., 2019; Pepe et al., 2014), wave-cut
platforms (Barboza and Tomazelli, 2003; Carter and Johnson, 1986),
shore platform (Kennedy, 2015) or wave ravinement surfaces (Zecchin
et al., 2011). These features are typically composed of a low angle
seaward dipping platform, corresponding to the swash or surf zones,
backed by a sub-vertical cliff, or riser (Fig. 2G, Braje et al., 2019). The
intersection between the platform and the cliff, often referred to as the
beach angle, represents the palaeoshoreline (Jara-Mu˜
noz et al., 2016;
Lajoie, 1986). Direct observation of the beach angle often reveals the
presence of tidal notches (Blanchon and Jones, 1995). Such features are
however not observable on typical geophysical data without the use of
dedicated equipment, including measurements from dive computers
(Collina-Girard, 2002). Sea stacks, isolated remnant of eroded cliff, can
sometimes be observed on the terrace (Bezore et al., 2016). Special care
should be taken when interpreting palaeo sea level from marine terraces
using bathymetry surveys alone, as a terrace can be overlain by modern
sediments resulting in an underestimation of the palaeoshoreline depth
(Lajoie, 1986; Rovere et al., 2018).
2.4. Coral-reef terraces
Coral-reef terraces are bioconstructed features formed through ag-
gradational or progradational reef accretion, whose vertical position is
constrained by the position of the sea level (Blanchon and Jones, 1995).
Given that coral reefs can thrive in a large range of water depth (Hibbert
et al., 2016), coral reefs that do not exhibit a clear terrace morphology
and therefore whose morphology cannot be precisely linked with the sea
level using geophysical data alone are not considered as palaeoshoreline
morphological indicators and are excluded from this review. The
morphology of such bioconstructed terraces have similarities with
erosive terraces, but coral-reef terraces typically exhibit a raised crest, a
generally horizontal reef at and a central or leeward depression rep-
resenting a palaeolagoon. In contrast erosive terraces are gently dipping
seaward (Fig. 2G–H, Blanchon and Jones, 1995; Wagle et al., 1994).
In practice, it is difcult to differentiate erosive terraces from coral-
reef terraces based on morphology alone, and it is not excluded that
geological features interpreted as coral-reef terraces may actually be
erosive terraces (Beaman et al., 2008; Coulbourn et al., 1974; Fürstenau
et al., 2009; Harris and Davies, 1989; Kennedy et al., 2012; Lewis, 1968)
or that some level of erosion was involved in their formation (Flemming,
1986; Wagle et al., 1994). Additionally, in some instances, corals may
colonise a pre-existing terrace (Blanchon and Jones, 1995; Webster
et al., 2006) and therefore the resulting feature does not qualify as
proper coral-reef terrace. Palaeo sea level were reconstructed from the
morphology of relict coral-reef terraces (sensu stricto) in intertropical
zones including: (1) the Great Barrier Reef (Abbey et al., 2011); (2)
French Polynesia (Montaggioni et al., 2019); (3) the Gulf of Mexico
(Khanna et al., 2017); (4) the Mozambique channel (Jorry et al., 2016);
and (5) the Maldives (Rovere et al., 2018).
Blanchon and Jones (1995) suggested that coral-reef terraces are
likely to be preserved during relative sea-level fall as they would be
isolated from further coral growth or marine erosion. This hypothesis
tends to be conrmed as most true coral-reef terraces are reported as
having formed during highstand intervals (Montaggioni et al., 2019;
Rovere et al., 2018). This theory could explain the limited number of
reported submerged coral-reef terraces compared to marine terraces.
Indeed, during subsequent relative sea-level rise, the terrace is exposed
once again to coastal erosion and becomes morphologically similar to
purely erosive terraces. Alternatively, Khanna et al. (2017) reported
transgressive coral-reef terraces preserved through successive over-
stepping events related to meltwater pulses. In such case, the sea-level
rise is rapid enough to limit the effect of coastal erosion on the terrace
morphology. Interestingly, emerged coral-reef terraces, which have
been extensively described from Barbados and Huon (Chappel, 1974;
Chappell and Polach, 1991; Mesolella et al., 1970; Schellmann and
Radtke, 2004; Yokoyama et al., 2001b), are often cited as analogues for
submerged coral-reef terraces (Beaman et al., 2008; Blanchon and Jones,
1995; Fürstenau et al., 2009; Montaggioni et al., 2019). However, these
sites are tectonically uplifted and have therefore not been exposed to
multiple transgression/ regression cycles, which could explain their
level of preservation.
Corals forming coral-reef terraces can grow few meters below the sea
surface and are, therefore, not direct morphological indicators of past
relative sea level (Rovere et al., 2016). To work around that, authors
have used modern analogues to estimate the water depth at which a
terrace could have developed (Camoin et al., 2004; Rovere et al., 2018).
As such palaeo relative sea level derived from coral-reef terraces
morphology should be used with caution in the absence of samples,
especially when attempting to reconstruct regional sea-level variations.
3. The Rowley Shelf
3.1. Geographic extent
The NWS of Australia is a geographic province located on the tropical
north-western portion of the Australian continental margin that extends
over ~2400 km between the North West Cape and the Melville Islands,
from ~22◦S to ~11◦S (Fig. 3, James et al., 2004; Purcell and Purcell,
1988). While in its original sense the term NWS designates the north-
western Australian oil and gas province, which includes the
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
6
continental shelf together with its marginal platforms and plateaux out
to the ~2000 m isobath (Purcell and Purcell, 1988), the term is often
used to designate the continental shelf alone, particularly in studies
investigating modern oceanography and Quaternary geology (Baker
et al., 2008; James et al., 2004). In that context, the NWS can be divided
in two parts: (1) the Rowley Shelf stretching from Exmouth to the Cape
Leveque, north of Broome; and (2) the Sahul Shelf extending in the north
east toward Melville Islands (Fig. 3, Carrigy and Fairbridge, 1954;
Wilson, 2013). The present study focuses on the Rowley Shelf, which
covers an area ~1000 km long and ~200 km wide. The coast bordering
the Rowley Shelf is divided in two sub-regions located on either side of
the De Grey Delta (Figs. 3, 4), namely: (1) the Canning coast, spanning
east of the delta and developed along the Phanerozoic Canning Basin
and, westward, (2) the Pilbara coast, a wave-dominated coastal plain fed
by seasonal rivers draining high-reliefs of the Pilbara region (Brocx and
Semeniuk, 2011; Semeniuk, 1992).
3.2. Geological setting
The Rowley Shelf, and more largely the NWS, is formed over Pre-
cambrian massifs (Exon and Colwell, 1994) that were rifted and
dismantled during the breakup of Gondwanaland and then Pangea
(Keep et al., 2007; Purcell and Purcell, 1988). During the Mesozoic,
several rifting and aborted rifting events formed margin-parallel sedi-
mentary basins (I’Anson et al., 2019; Longley et al., 2002). Rifting ended
during the Lower Cretaceous, following the nal break-up between
Fig. 3. Area of interest. The North West Shelf of Australia is composed of the
Rowley and Sahul shelves. The present study focuses on the Rowley Shelf.
Fig. 4. Geophysical datasets integrated in the present study (modied after Lebrec et al., 2021).
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
7
Australia and Greater India (Gibbons et al., 2012; Keep et al., 2007;
Longley et al., 2002; Marshall and Lang, 2013; Müller et al., 1998;
Paumard et al., 2018; Yeates et al., 1987). Since then, the NWS became a
passive margin with its accommodation dominantly controlled by
thermal subsidence and local tectonic reactivation, in particular during
the Late Cretaceous breakup between Australia and Antarctica (Cathro
and Karner, 2006; Direen et al., 2007; Driscoll and Karner, 1998; Keep
et al., 2007; Romine et al., 1997; Tindale et al., 1998). During the
Cenozoic, rift-induced basins were progressively lled and capped by
prograding wedges of shales, marls and shelfal carbonates, locally
interbedded with sandstones, resulting in a homogenisation of the NWS
area (Apthorpe, 1988; Longley et al., 2002; Romine et al., 1997).
Nowadays, the shelf is gently steeping from the shore to the shelf
edge at the ~200 m isobath, and has been described as a bathymetrical
gentle ramp (James et al., 2004). It can be divided in three depth zones:
(1) the inner-ramp between 0 and 50 m; (2) the mid-ramp between 50
and 120 m; and (3) the outer-ramp between 120 m and the shelf break
(Dix et al., 2005; James et al., 2004; Wilson, 2013).
3.3. Present-day climate and water circulation
The continental climate of the coast bordering the Rowley Shelf is
dry through most of the year under the inuence of the southeast trade
winds and of the Subtropical Ridge. During summer months, from
December to March, a monsoonal-type climate, with rains carried by
cyclones, thunderstorms and tropical depressions, can prevail (Suppiah,
1992). While monsoon rains can occur in the vicinity of the North West
Cape, at a latitude of ~22◦S (Hesse et al., 2004), most of the summer
rainfall occurs along the Sahul Shelf in the northern Kimberley region
(McRobie et al., 2015) and the annual amount of rainfall along the
Rowley Shelf, south of 17◦S, is highly variable (Sharmila and Hendon,
2020). Overall, the climate is dryer in the south and becomes gradually
wetter toward the North (Wilson, 2013).
Water circulation along the Rowley Shelf is inuenced by shallow
currents that originate in the West Pacic Warm Pool, including the
shallow (i.e., <300 m deep) and warm, low salinity Leeuwin Current,
which carries warm waters southward along the coast of Western
Australia and suppresses coastal upwellings (Gallagher et al., 2018;
Haller et al., 2018; Hatcher et al., 1991; Holloway, 1995; James et al.,
2004; Stephen et al., 2009). Water circulation is also inuenced by tide,
with the tidal range reaching up to 6 m along the coast and generating
strong tide-induced currents in the cross-shelf direction, which in turn
generate internal tides (Condie and Andrewartha, 2008; Holloway,
2001; Katsumata, 2006). There are also long period waves and swells
during winter months (James et al., 2004) and internal waves during
summer months (Baines, 1981). The combination of these water uxes
impacts the morphology of the sedimentary features present along the
Rowley Shelf, with for example the development of sand waves (Belde
et al., 2017; James et al., 2004; Jones et al., 2009).
3.4. Sedimentology and Quaternary climate of the Rowley Shelf
Along the Pilbara coast, there are intermittent siliciclastic inputs
from rivers that locally form coastal deltaic complexes (Semeniuk,
1996). Most of the Rowley Shelf is, however, dominated by carbonate
sedimentation (James et al., 2004; Short, 2020; Wilson, 1975). Overall,
the variability and distribution of seabed features and of the associated
supercial sediments is linked to the Quaternary evolution of deposi-
tional environments, tied to uctuating climate, oceanography and in
particular relative sea-level changes (Baker et al., 2008).
The climate along the shelf evolved in conjunction with glacial cycles
through the Quaternary. During glacial periods, the monsoon is limited
and the climate dryer, whereas during interglacial intervals, the
monsoon is strengthened, leading to a wetter climate (Hesse et al.,
2004). Those alternations of dry and wet climates may have had a direct
impact on the sedimentation of the shelf, with increased siliciclastic
inux during wet interglacial intervals, and, in contrast, an increased
production of aragonite and ooids during dry glacial periods (Gallagher
et al., 2018; Gallagher et al., 2014). In this context, large accumulations
of ooids were described on the seabed in water depth ranging from 50 to
150 m below sea level (m bsl; James et al., 2004). They are interpreted
as being the result of important episodes of ooid-peloid and aragonitic
mud production following the Last Glacial Maximum (LGM). The pro-
duction of the ooids ended at around 12 ka BP (Hallenberger et al., 2019;
James et al., 2004). At about the same time, the Australian Summer
Monsoon re-established. However, the exact timing of the transition is
still debated, and it is unclear if the establishment of humid conditions
was abrupt (i.e., transition to wetter conditions at 14 ka BP; Hesse et al.,
2004; Van Der Kaars and De Deckker, 2002; Wyrwoll and Miller, 2001)
or progressive (i.e., shift to a humid climate at 11.6 ka BP in the north-
eastern part of the NWS, and at 10.1 ka BP in the south-western part;
Hallenberger et al., 2019). More arid conditions re-established after ~6
ka BP, leading to the present day relatively dry climate with variable
amount of monsoonal rains, that contrasts with previous humid in-
terglacials (Hesse et al., 2004). In any case, when precipitation events
occur, it is usually the result of a cyclonic event resulting in intense
precipitation within the coastal catchment area which appears sufcient
enough for the rivers to supply sediments to the coast (Semeniuk, 1992).
In response to late Pleistocene glacial-interglacial climate cycles, the
relative sea level around Australia oscillated between up to +9 m (MIS 5,
O’Leary et al., 2013) and −125 m (MIS2, Yokoyama et al., 2001a), when
compared with the present sea level, resulting in multiple exposure or
submergence events affecting much of the current shelf. Several po-
tential submerged palaeoshorelines, stemmed from such events, are
identied in the form of submerged terraces and ridges along the shelf
from low-resolution bathymetry datasets (Dix et al., 2005; Gallagher
et al., 2014; Hengesh et al., 2011; James et al., 2004; Jones, 1973).
However, mainly due to the limited coverage and/or resolution of the
geophysical data available to the authors, the nature of the features
forming those submerged topographic reliefs, as well as their total
extent, remain unknown.
Finally, the Rowley Shelf is part of the Western Australian coral-reef
province, where coral-reef development has been observed at latitude as
south as 29◦S (Collins, 2002; Collins et al., 2004; Collins et al., 1993;
Hatcher et al., 1991; McCulloch and Mortimer, 2008). Along the Rowley
Shelf, a multitude of shelfal islands are fringed by coral reefs such as
Barrow Island, Montebello Islands, the 64 reef-fringed Pilbara islands
and the Dampier Archipelago (Bonesso et al., 2020; Collins, 2002;
Grifth, 2004; Moustaka et al., 2019). Similarly, the North West Cape is
fringed by the 200 km-long Ningaloo Reef (Collins et al., 2003). Addi-
tionally, submerged topographic highs such as Glomar Shoal (~40–60 m
bsl) and Rankin Bank (~30–70 m bsl) are colonised by modern meso-
photic coral ecosystems (Abdul Wahab et al., 2018). Signicant reef
development likely occurred during the Pleistocene, with seabed topo-
graphic highs interpreted from geophysical data as drowned coral reefs
in water depth of ~60 m around Barrow Island (Gallagher et al., 2017c;
Gallagher et al., 2014).
4. Datasets and methods
An extensive analysis of the seabed features was conducted over the
Rowley Shelf between Exmouth (Fig. 3) and Broome. The area of interest
stretches along the coast over more than 1000 km and encompasses an
area of nearly 200,000 km
2
. Data interpretation is based on multiple
geophysical datasets including multi-sourced bathymetry and high-
resolution 2D seismic surveys (Table 1).
Bathymetry datasets are mainly sourced from the bathymetry
compilation produced by Lebrec et al. (2021), which includes: (1)
45,000 km
2
of satellite-derived bathymetry processed from over a
thousand Sentinel-2 satellite images acquired over three years in water
depth of less than 30 m; (2) 100,000 km
2
of seismic-derived bathymetry
produced through the integration of 3D seismic datasets with navigation
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
8
recordings and hydrographic depth soundings; and (3) 100,000 km
2
of
Multibeam Echosounder (MBES) bathymetry (Fig. 4).
The bathymetry compilation was further complemented with
seismic-derived bathymetry produced using the workow from Lebrec
et al. (2021) and proprietary seismic surveys provided by TGS and PGS
(seismic acquisition and processing companies). In coastal areas with
water depth of less than 30 m, nearly 5000 km
2
of Light Detection and
Ranging (LiDAR) surveys were sourced from the Western Australian
Department of Transport (WA DOT) in the area surrounding the
Dampier Archipelago and Barrow Island. In addition, data acquired by
the Australian Institute of Marine Sciences (AIMS) and Tenix LADS,
respectively, were accessed as PDF maps in the vicinity of: (1) Glomar
Shoal and the Rankin Plateau (Abdul Wahab et al., 2018); and (2) Port
Hedland (SKM, 2011). Voids between surveys were subsequently lled
using the 2009 Australian Bathymetry and Topography Grid (Whiteway,
2009). All surveys depths are (as reported) reduced to the Mean Sea
Level (MSL) and corrected for tide, roll, pitch and heave effects. How-
ever, it is common to observe vertical offsets of several meters between
overlapping surveys, or even between acquisition lines within the same
survey. Such discrepancies were corrected, using depth soundings from
the Australian Hydrographic Ofce (AHO) as a reference datum, and by
shifting vertically datasets accordingly. While this approach reduced
vertical uncertainties, it was not possible to obtain a perfect match as the
vertical offsets between overlapping surveys are generally not consistent
across survey areas.
2D high-resolution surveys were sourced as SGY les from Geo-
science Australia and include 18,000 km of sub-bottom proler lines
acquired through multiple cruises between 2005 and 2017. The data is
of very variable quality and typically only penetrates through surcial
unconsolidated sediments. In several cases, only the seabed reection is
visible on the data. 3D seismic surveys publicly available in the area
were also reviewed but their vertical sampling rate (usually 4 ms) and
resolution were not considered sufcient to interpret late Pleistocene to
recent strata.
All datasets were integrated within a Geographic Information System
(GIS) database and IHS Kingdom software project and were systemati-
cally reviewed to identify and map submerged palaeoshoreline features
and more generally submerged coastal landforms. The identication and
characterisation of seabed features, and more specically of relict fea-
tures, was performed iteratively through all datasets. Identied relict
features were then classied depending on the key processes (i.e., wind,
tide, uvial, wave) involved in their formation. In most cases, features
were associated with a given process not solely based on their
morphology, but also considering surrounding features.
5. Discriminating modern and relict bedforms along the Rowley
Shelf
A key aspect of the seabed interpretation is the discrimination be-
tween modern marine and relict coastal features. While the identica-
tion of submerged terraces can appear straightforward (i.e., sharp sub-
horizontal surface associated with a vertical step), dunes and clino-
forms can develop in a large range of depositional environments
(Patruno and Helland-Hansen, 2018; Rubin, 2012) and their general
morphology alone is not always sufcient to classify them as relict
coastal features. This can lead, especially in areas where data coverage is
sparse, to the misidentication of modern features as relict coastal de-
posits (Martin and Flemming, 1986). The discrimination of modern and
relict bedforms is performed through ve criteria.
(1) Stratigraphic order. Seabed bedforms visible on the bathymetry
are often partly overlain by younger sediments, suggesting that
they are inactive and can therefore be categorised as relict
(Fig. 5B). The identication from seismic data of a surcial
sediment veneer can therefore be used to discriminate relict
features.
(2) Presence of sub-aerial exposure features. Emersion features,
including dissolution features (e.g., karsts) and sub-aerial
erosion, indicate that the underlying feature is partly cemented
and was formed prior to the last sea-level rise (Jacques, 1996; Kan
et al., 2015; Micallef et al., 2013; Suri´
c, 2002; Taviani et al.,
2012) and should therefore be considered as relict. Dissolution
features are characterised by circular depressions, generally a few
meters deep, that can reach up to hundreds of meters in diameter
(Fig. 6). Due to their size, dissolution features can typically only
be identied from high-resolution surveys. Such depressions are
generally located on top of cemented ridges and are therefore
interpreted as karsts. Additionally, relict features are often
partially eroded and can have an irregular surface, and, in ba-
thymetry, they can appear as segmented objects with a rough
texture, which contrasts with modern seabed features that are
non-segmented and have a smooth texture (Fig. 5D).
(3) Similarity with analogues and associated discrimination
criteria. Seabed features were systematically compared with
modern analogues observed through aerial photographs and case
studies reported in the literature. For example, the morphology of
the seabed ridges presented in Fig. 5B is nearly identical to the
morphology of relict barrier spits described by Jarrett et al.
(2005). Additionally, these ridges are partly covered by younger
Table 1
Geophysical data available within the area of interest. Data extent presented in Fig. 4.
Data type Dataset Data provider Coverage Resolution Vertical
accuracy
Spatial
accuracy
Satellite-derived
bathymetry
NWS SDB Lebrec et al. (2021) 45,000 km
2
10 ×10 m 5% 16 m
Seismic-derived
bathymetry
NWS SDB Lebrec et al. (2021) 100,000
km
2
30 ×30 m 1% 200 m
Seismic-derived
bathymetry
Proprietary seismic surveys. Includes
Capreolus, Polly and Zeester.
TGS 33,954 km
2
12.5 ×25 m 1% Na
Seismic-derived
bathymetry
Proprietary seismic surveys. Includes
Carnavon MegaSurvey
PGS 49,717 km
2
25 ×25 m 1% Na
MBES Surveys GA0296, GA2390, GA2408,
GA2437, GA2476 and GA4811
Geoscience Australia (data request Sept. 2019) 100,000
km
2
30 ×30 m <1% Na
LiDAR Barrow Island Western Australia Department of Transport
(WA DOT, personal communication)
4200 km
2
10 ×10 m <1% Na
Dampier Archipelago 250 km
2
1 ×1 m <1% Na
LiDAR (PDF charts) Glomar Shoal & Rankin Bank Abdul Wahab et al. (2018) 800 km
2
Na Na Na
Port Hedland SKM (2011) 3300 km
2
Na Na Na
DEM compilation Australian Topography and
Bathymetry Grid
Geoscience Australia (data portal, Whiteway
(2009))
1000,000
km
2
0.25 ×0.25
arc sec
Variable 250+m
2DHR Survey GA096, GA2390, GA2408,
GA2437 and GA2476
Geoscience Australia (data request Sept. 2019) 18,000 km Variable Variable Variable
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
9
(caption on next page)
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
10
sediments. They are therefore considered as submerged palaeo-
barrier spits.
(4) Bedform orientation. Modern tidal current direction and wave
climate data from the CSIRO (Herzfeld et al., 2020) are integrated
with the geophysical datasets. If the orientation of a bedform
cannot be explained by modern ocean conditions, this bedform is
considered as relict. East of the Dampier Archipelago, for
example, it is possible to identify two sets of bedforms from the
satellite-derived bathymetry that exhibit similar height, width,
and sinuosity (Fig. 5C). Yet, one subset is perpendicular to the
main modern tidal constituent direction (constituent M2) and
appears to overlap the second one, which is oblique to the tidal
current direction and parallel to the shore (Fig. 5C). Based on this
approach, the rst set is considered as modern while the second
one is interpreted as relict.
(5) Seismic attenuation. Relict ridges identied through the above
methods are all exhibiting seismic attenuation (e.g., blanking).
Such behaviour can result from the absorption and scattering of
seismic pulses due to a change in rock properties (Lamprecht and
Plekker, 1991; Penrose et al., 2006; Stevenson et al., 2002). In
this case, the seismic attenuation is interpreted as resulting from
the cementation of the bedforms. Such cementation could be
either of marine or meteoritic origin (James and Jones, 2015;
Scholle and Ulmer-Scholle, 2005). However, given that modern
marine early cementation has not been reported along the NWS,
with the exception of coastal beachrock (James et al., 2004;
Semeniuk, 1992; Semeniuk, 1996; Semeniuk, 2008), it is likely
that submerged cemented bedforms are relict. Amplitude atten-
uation was subsequently used, in combination with the other
criteria, to identify relict features (Fig. 5D).
Ideally, both high-resolution bathymetry and 2D seismic lines should
be used in seabed characterisation, effectively allowing the visualization
of seabed features in three dimensions, as well as the identication of
multiple criteria. However, in most cases, only bathymetry surveys of
variable resolution or scattered seismic lines were available. To work
around data limitations, the investigation focused on areas where
multiple high-resolution datasets overlap to identify discriminating
criteria, before being extended to areas with lower data coverage/
quality.
6. Catalogue of submerged coastal features on the Rowley Shelf
The analysis of the Rowley Shelf datasets led to the identication of
over ve hundred potential submerged relict coastal features that are
either direct or indirect morphological indicators of palaeoshoreline
positions, and which are related to three of the four palaeoshoreline
categories identied in the literature (i.e., beach ridges, shoreface strata
and marine terraces). Features associated with each category exhibit a
range of morphological characteristics depending on whether they were
formed through aeolian, wave, tidal or uvial coastal processes. The
relation between submerged coastal features, coastal processes and the
associated shoreline category is presented in Table 2. Relict coral-reef
terraces that could be used as a morphologic palaeoshoreline indicator
were not formally identied within this area of interest. This however
does not exclude the potential presence of other types of relict coral
reefs.
6.1. Beach ridges
6.1.1. Isolated foredunes and associated aeolian features
Along the Rowley Shelf, submerged isolated foredunes (i.e., wind-
built beach ridges) are characterised by linear shore-parallel relict
ridges, typically 150 to 500 m wide and 5 to 15 m high that can be
tracked over tens of kilometres on the bathymetry (Fig. 7A-B). The term
isolated is here used to discriminate such foredunes from wind-built
ridges that are part of prograding strandplains. They are particularly
well developed in 16 to 20 m of water depth between the De Grey Delta
and Broome, and between Muiron Islands (Exmouth Gulf) and Barrow
Island. In some instances, circular or semi-circular U-shaped de-
pressions, concave landward, up to 500 m wide can be observed on the
seaward ank of the ridges (Fig. 7C–D). Isolated foredunes of similar
width, height and morphology, and associated depressions, are
described along the coast (Semeniuk, 1992; Semeniuk, 1996; Semeniuk,
Fig. 5. Illustration of the workow to discriminate modern and relict bedforms. A) Location map; 6A and 6B indicate the location of Fig. 6 illustrations. B) Seabed
ridges morphology observed from bathymetry are similar to relict barrier spits described in Jarrett et al. (2005) and the seismic section (B′B′′) indicates that ridges are
overlain by younger sediments conrming their relict nature. C) Two sets of seabed ridges can be identied from the bathymetry. The ocean current direction
explains the morphology and distribution of the blue set of ridges but not of the red set of ridges suggesting that the latter could be composed of relict features. Where
the resolution of the bathymetry is not t for purpose, the interpretation relies on the identication of key seismic facies. The seismic section (D′D′′ ) displays two
types of ridges: one is associated with seismic blanking while the other is not. Blanking is interpreted as resulting from early cementation and its identication can
therefore be used to discriminate modern and relict features. Geophysical data courtesy of: (B) WA DOT; (C) Lebrec et al. (2021); and (B’-B’’,D,D-D’’) Geoscience
Australia. Regional bathymetry (A) from Whiteway (2009). (For interpretation of the references to colour in this gure legend, the reader is referred to the web
version of this article.)
Fig. 6. Illustration of the workow to discriminate modern and relict bedforms. The identication of sub-aerial exposure indicators such as dissolution features
suggests that the underlying feature is relict. Dissolution features were identied in the area of interest in sizes ranging from few meters (A) to hundreds of meters (B).
Inset locations are displayed in Fig. 5A. Bathymetry courtesy of WA DOT (A) and Geoscience Australia (B).
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
11
2008; Short, 2020), with the depressions representing blow outs formed
through the continuous action of the sea breeze (Eliot et al., 2013; Payne
et al., 2004). The presence of blow outs was therefore regarded as a
distinctive morphology to identify relict foredunes. However, in most
cases, the resolution of the bathymetry is not sufcient to observe them.
In which case the general morphology of the ridge (i.e., contour parallel
relict ridge continuous over tens of kilometres, 5 to 15 m high) was used
as a discriminating factor and it can not be excluded that some of the
smaller features may correspond to beachrock (i.e., cemented wave-
built ridge).
Parabolic dunes, which can evolve from blow outs (Hesp, 2011), are
identied east of the De Grey Delta and off Eighty Miles Beach
(Fig. 7E–F). These features are not considered as a morphological evi-
dence of palaeo sea level because they can migrate landward over
several kilometres (Psuty, 2008). However, their presence indicates that
a coastal area is prone to the formation of aeolian features like
foredunes. Parabolic dunes, and more generally sand dunes, can be
formed both in marine and aeolian settings (Andreotti and Claudin,
2013; Cardinale et al., 2014; Rubin, 2012). Therefore, the parabolic
morphology alone is not sufcient to classify a parabolic feature as an
aeolian dune. Submerged relict aeolian parabolic dunes are discrimi-
nated from their marine counterpart using the ocean current data from
the CSIRO: if the orientation of a parabolic feature cannot be explained
by the present marine conditions, it is considered as a relict aeolian
feature. Submerged relict parabolic dunes of the Rowley Shelf exhibit
slightly divergent hemicyclic lobes suggesting, based on the classica-
tion from Pye and Tsoar (2009), that they were formed under seasonal
wind directions.
6.1.2. Strandplains
Strandplains, which correspond to sets of closely spaced sub-parallel
beach ridges (Otvos, 2000), are the most common relict features on the
Table 2
Submerged palaeoshoreline indicators of the Rowley Shelf.
Indicator
category
Coastal features Morphology/ Metrics Associated features Dominating
processes
Figure
Beach ridges Isolated foredunes Ridges. 5–15 m high, 150–500 m wide, shore parallel and
continuous over 10s to 100 s of km
Blow outs, parabolic dunes, beach shoreface Aeolian 7
Strandplains Multiple prograding parallel ridges (often 10+). Shore
parallel, 2–5 m high, 150–500 m wide
Fluvial channels and associated mouth bars,
beach shoreface
Wave, aeolian 8
Tombolos Isthmus between mainland and an island Strandplains, beach shoreface Wave 8
Barrier complexes Ridges. 2–15 m high, 150–500 m wide, continuous over
10s of km, close an embayment
Tidal inlets, lagoon, ebb-tidal deltas,
washover deposits, recurved spits
Wave, aeolian 9
Shoreface
strata
Beach shoreface
deposits
Inclined strata. 1–4 m high, few hundred meters long Beach ridges Wave 10
Ebb-tidal deltas Hemi circular lobes. 5–10 km in diameter, 5–10 m thick,
intersected in the middle by channel(s)
Tidal inlets, lagoons, barrier complexes Tidal 11
Mouth bars Lobes. <10 km long. 5–10 m thick Fluvial channels, strandplains Fluvial 11
Marine terraces Seaward dipping terraces backed by a cliff-like feature Cliffs (terrace riser) Wave 12
Fig. 7. Illustration of aeolian relict features identied within the area of interest using satellite-derived bathymetry from Lebrec et al. (2021). Feature locations are
presented in Fig. 8 except for inset F, located outside of the area of interest, at the northern end of the Dampier Peninsula. A–B) Foredunes; C–D) Blow outs; E–F)
Parabolic dunes. Imagery B, D and F courtesy of Google Earth, based on data from SIO, NOAA, U.S. Navi, NGA, GEBCO © Maxar Technologies.
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
12
shelf. Individual ridges are several kilometres long, 2 to 5 m high and 50
to 150 m wide. Each set is sub-parallel to the bathymetry contours and
may include dozens of ridges for a total width of up to 5 km (Fig. 8B).
Three main congurations are identied: (Type 1) all ridges are
parallel to the bathymetry contours and have a low sinuosity (Fig. 8I–J);
(Type 2) all ridges are slightly oblique to the coast and are presenting
sinusoidal shapes (Fig. 8G–H); and (Type 3) ridges are prograding
seaward from an estuarine or uvial channel (Fig. 8E–F). These con-
gurations are interpreted as developed under slightly different coastal
conditions. The rst conguration may have been formed in the absence
of longshore current, enabling the accumulation of shore parallel ridges,
whereas the second case may highlight the presence of a current that
would have draped sediments along the coast. Finally, the third case
highlights the presence of a uvial inuence leading to the coast pro-
gradation. The local presence of blow outs suggests that ridges forming
the strandplain morphology are likely of aeolian origin. However, if the
sea level is stable, the progradation of the coast which enables the
development of the strandplains may be related to wave processes
(Otvos, 2000). Hence, strandplains are considered built from mixed
wind/ wave processes. Overall, it was not possible to fully discriminate
wave-built and wind-built components of strandplains based on the
bathymetry alone.
6.1.3. Tombolo
Tombolos typically form in the lee of a nearshore island with wave
refraction causing sediments to be deposited from two opposite di-
rections and eventually to form an isthmus bridging the gap between the
mainland and an island (de Mahiques, 2016; Flinn, 1997). They are rare
along the Rowley Shelf and are only identied 5 km west of Rosemary
Island, within the Dampier Archipelago (Fig. 8C) and along the Rankin
Bank. The interpretation is based on the identication of a distinctive
cuspate morphology associated with a topographic high. This
morphology is also observed along the modern coast near Port Smith
(Fig. 8D).
6.1.4. Barrier complexes
Barrier complexes along the Rowley Shelf are identied through the
presence of: (1) recurved spits; (2) washover deposits; and (3) multi-
kilometric lagoons and tidal inlets.
Recurved spits, which are mainly formed under the inuence of
longshore currents (Evans, 1942), are associated within the area of in-
terest with multi-kilometric ridges that are 150 to 500 m wide and 2 to 5
m high (Fig. 9B). These ridges, which are often grouped together to form
horizontal stacks of ridges 1500 to 2500 m wide, are generally con-
nected by one end to another palaeoshoreline feature, while the other
end is advancing seaward (Fig. 9B–C). Additionally, these ridges are
either straight or recurved landward, and often appear to be enclosing
an embayment hence forming a barrier spit. This interpretation is sup-
ported by the presence of similar barrier spits along the modern coast of
the NWS (Fig. 9C). The resolution of the bathymetry is often not suf-
cient to discriminate the aeolian cap from the wave-built ridge, where
seismic data is not available. In such instance, barrier spits are only
regarded as an indirect palaeo relative sea-level marker used to conrm
sea-level measurements made from nearby relict features.
Along the coast of Eighty Miles Beach, in 18 m of water depth,
overwash features can be identied across relict barriers (Fig. 9D). Such
features, which are formed when the ow of water and sediments ex-
ceeds the barrier height (Donnelly et al., 2006), are marked by barrier
breaches with sediment bodies dragged landward (Fig. 9E). The nature
of these features is inferred based on their resemblance with storm-
induced washover deposits (Chantal et al., 2006; Hudock et al., 2014;
Schwartz, 1982). Similar features have also been extensively described
in the Exmouth Gulf (May et al., 2017). While washover deposits are not
relative sea-level morphological indicators, they are used to classify the
underlying beach ridges as barriers.
Extensive linear contour-parallel ridges can be observed offshore
from Barrow Island and off Port Hedland, and are backed on their
landward side by a multi-kilometric depression (Fig. 9G). Ridges, typi-
cally 5 to 15 m high and 500 m wide, are often discontinuous with
channels connecting their landward and seaward sides. O’Leary et al.
(2020) described the features identied offshore Barrow Island as bar-
rier complexes composed of coastal dunes and lagoons, using the inlets
associated with Collins Pool, along the south west coast of Western
Australia, as the best analogue for such features (Fig. 9G). The same
interpretation is used for similar submerged features identied else-
where along the shelf.
With the exception of recurved spits, which exhibit clear morpho-
logical patterns characteristic of wave processes, there are generally not
enough morphological evidences to determine if the Rowley Shelf bar-
riers are dominantly wave or wind built. The formation of the barriers
could be: (1) contemporaneous to the formation of the lagoon
(Fig. 2C–D); or (2) a pre-existing ridge, such as a foredune, which is
inundated during a relative seal-level rise and subsequently becomes a
barrier (Fig. 2E). The latter can be observed along the modern coast and
was extensively described by Semeniuk (1992, 1996) and is therefore,
when recurved spits are not observed, the preferred interpretation.
However, this interpretation remains speculative. In this context, we
consider that these features have potentially either a wave or aeolian
origin.
6.2. Shoreface strata
6.2.1. Beach shoreface
Beach shoreface strata are identied through 2D seismic lines as
seaward dipping clinothems and were observed offshore Dampier and
Broome in water depth of 30 m and 125 m, respectively. Individual
clinothems are 1 to 5 m thick and extends over 600 m (Fig. 10). Such
morphologies correspond to the description of “shoreline clinoforms”
(Patruno and Helland-Hansen, 2018), which encompasses delta front
and beach shoreface deposits. Given that no major uvial systems are
identied in their direct vicinity, we interpret these features as beach
shoreface strata accumulated by wave processes. This interpretation is
supported by the clinoforms angles, <0.1 degree, which are below the
expected threshold of delta front mouth bars (Cosgrove et al., 2018;
Paumard et al., 2019; Paumard et al., 2020) and by the presence of relict
beach ridges landward of the clinothems.
6.2.2. Mouth bar
Multiple uvial channels can be observed along the shelf in associ-
ation with Type 3 strandplains, as dened in Fig. 8, E–F. A channel is
characterised as uvial dominated, as opposed to tidal dominated, based
on its landward extent, its relationship with modern uvial system, and
the characteristic progradation of the associated shoreline (Fig. 11B–E).
Offshore, between Dampier and Barrow Island, in 60 m of water depth, a
sediment lobe appears to be related to a meandering channel. The exact
morphology of the feature is uncertain due to the lack of high-resolution
bathymetry and 2D seismic lines. Based on the comparison of the lobe
morphology with the digital elevation model from the De Grey Delta and
the Ashburton Delta (Fig. 11B–E), the sediment lobe is tentatively
interpreted as a delta front mouth bar. While uvial channels are not
regarded in this study as a morphological marker of past relative sea
level, they can be used to infer the position of relict mouth bars
(Fig. 11B–D). Mouth bars, on the other hand, are a subtype of shoreline
clinoforms, and their seaward rolling point can be used as a morpho-
logical marker of past relative sea level (Patruno and Helland-Hansen,
2018).
6.2.3. Ebb-tidal deltas
A dozen ebb-tidal deltas, sand shoals located seaward of a tidal inlet
(Hayes and FitzGerald, 2013), can be observed in water depths ranging
from 80 to 110 m belowsea level offshore port Hedland (Fig. 11F) and
east of the Rankin Plateau (Fig. 11G). Ebb-tidal deltas are typically 5 to
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
13
Fig. 8. Illustration of strandplain morphologies within the area of interest using bathymetry (A,B) and 2D seismic (B’B’’) from Geoscience Australia; seismic-derived
bathymetry (E) based on TGS proprietary datasets; and satellite-derived bathymetry (C, G, I) from Lebrec et al. (2021). A) Location map; 7A-E indicate the location of
Fig. 7 illustrations. Strandplains are identied as a succession of closely spaced linear ridges, which typically exhibit seismic blanking interpreted as resulting from
early cementation (B, B
′B′′). In contrast, unconsolidated sediments are reworked to form modern bedforms (I). Three main strandplain congurations are identied:
Type 1: all ridges are contour parallel (I, J); Type 2: ridges are slightly oblique to the contours and have a sinusoidal shape, interpreted as resulting from the action of
longshore current (G, H); Type 3: ridges are prograding seaward due to localised sediment input from riverine systems (E, F). In some instances, sediments can create
an isthmus between the mainland and an island to form a tombolo (C, D). Imagery (D) source: Esri, Maxar, GeoEye, Earthstar Geographics, CNES/ Airbus DS, USDA,
USGS, AeroGRID, IGN, and the GIS User Community.
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
14
Fig. 9. Illustration of barrier complex morphologies along the Rowley Shelf using satellite-derived (B,D) and seismic-derived (G) bathymetry from Lebrec et al.
(2021). Regional bathymetry (A) from Whiteway (2009). Two types of barriers were identied: (1) barrier spits, which are attached to the coast by one end and
exhibit recurved spit on the other (B, C); and (2) generic relict barriers, which are identied by the presence of a back-barrier lagoon (F, G, modied from O’Leary
et al. (2020)); F is outside the area of interest and corresponds to Collins Pool, along the southern coast of Western Australia. Where high-resolution bathymetry is
available, barriers can be identied through the presence of overwash deposits and breaches (D, E). Imagery (E, C, F) source: Esri, Maxar, GeoEye, Earthstar
Geographics, CNES/ Airbus DS, USDA, USGS, AeroGRID, IGN, and the GIS User Community.
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
15
10 km in diameter and are incised in the middle by a tidal channel. Ebb-
tidal deltas are backed, on their at landward side, by a series of bar-
riers, located on either side of a tidal inlet. Two distinct ridge mor-
phologies can be observed: (1) barrier spits asymmetric on either side of
the main inlet (Fig. 11G); and (2) perfectly straight barriers that can be
tracked from one side of the inlet to the other (Fig. 11F). In both cases,
the inlet is connected to a lagoon located further landward. Rarely (i.e.,
one occurrence observed), tidal bars are present within the inlet. The
interpretation of these features is based on the similarity of their mor-
phologies with tidal inlets, tidal bars and ebb-tidal deltas observed along
the modern coast of the NWS (Fig. 11H–J). The most prominent modern
feature is located off Port Smith, where an ebb-tidal delta is connected to
a lagoon through a tidal inlet (Fig. 11H). There is no evidence of a uvial
feeder system, hence conrming that the formation of such sedimentary
body can be related to tidal processes alone.
Additionally, the presence of barrier spits on either side of a tidal
inlet, similar to those presented in Fig. 11G, can be observed along the
modern coast and appear to be associated with an active barrier
(Fig. 11I), suggesting that the formation of the barrier and of the tidal
systems are contemporary. On the other hand, we consider that ridges
that do not exhibit recurved spits (Fig. 11F) are probably antecedent
features that were subsequently breached during a transgressive event.
This interpretation is supported by the presence of erosive features along
the straight barriers that are observed along the modern coast (Fig. 11J).
Ebb-tidal deltas correspond to a subtype of shoreface strata and their
shallowest point can be regarded as a morphological marker of past
relative sea level.
6.3. Marine terraces
Sub-horizontal terraces are observed from both the bathymetry and
the 2D seismic data at depths of 17, 35 and 90 m. They are characterised
by gentle seaward dipping surfaces and are backed by a terrace riser
(Fig. 12B–E). Both criteria are distinctive morphologies of marine ter-
races formed through wave erosion (Anderson et al., 1999; Blanchon
and Jones, 1995; Bradley, 1958). Similar features can be observed along
the modern coast (Fig. 12E), conrming the nature of the terraces.
Within the Dampier Archipelago, a terrace identied at 17 m of water
depth from LiDAR bathymetry is rimmed by a <1 m ridge over a section
of 100 m, which represents less than 5% of the entire terrace (Fig. 12D).
In isolation, and following the denition from Blanchon and Jones
(1995), such morphology could tend to indicate the presence of a coral-
reef terrace. However, given the limited spatial extend of the ridge, it is
assumed that the ridge was developed after the formation of the terrace
and therefore that the terrace morphology is related to erosive processes
(i.e., marine terrace).
The terrace riser is often not very well expressed, and the presence of
a terrace is subsequently interpreted from a change in the shelf slope
(Fig. 12B). In such instance, the interpretation of the geophysical data is
tentative only and the interpreted terrace is only used to conrm the
presence of other coastal features.
7. Distribution of the shorelines
Submerged palaeoshoreline features have been identied along the
Rowley Shelf in water depths ranging from 5 m to 145 m below sea level
(Fig. 13) and provide a nearly continuous record of past relative sea
level. While coastal features were virtually identied at any depth, as sea
level and therefore a shoreline regressed and transgressed across the
entire shelf, they appear to be concentrated along specic depth values
that can be tracked across the Rowley Shelf (Figs. 13 and 14). Such
behaviour is interpreted as the result of modal sea levels, when a given
elevation is occupied by shorelines over longer periods of time than
others (Brooke et al., 2017; Ribo et al., 2020), hence providing time for
coastal sedimentary deposits or erosional features to develop. Each
modal sea level may comprise multiple coastal features observed at
slightly different depths, but part of the same trend. For example, a
bathymetry step along which palaeoshorelines features are cascading
from a depth to a depth +x meters will be considered as representing
one modal sea level. Typically, a shoreline morphology representing
modal sea level will be visible from regional datasets, whereas indi-
vidual shoreline features can only be observed on high-resolution sur-
veys. Along the Rowley Shelf, 9 modal sea levels, hereafter referred to as
modal sea-level depth X or ‘MSLD X’, with X representing the average
water depth of a given trend, were observed at 20 m, 35 m, 50 m, 60 m,
70 m, 80 m, 90 m, 105 m and 125 m below present sea level (Figs. 13, 14
and 15). It should be noted that variable vertical offsets of several meters
were locally observed between surveys sharing the same datum along
the Rowley Shelf (Lebrec et al., 2021) and therefore some level of un-
certainty, which can not be efciently quantied, exists when
Fig. 10. Illustration of beach shoreface deposits. Location of the seismic line displayed in Fig. 11A. Along the Rowley Shelf, beach shoreface deposits are charac-
terised by low angle 1–4 m high clinoforms located seaward of relict beach ridges. Deposits typically extend over 150 to 800 m. Seismic line courtesy of Geo-
science Australia.
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
16
Fig. 11. Illustration of relict tidal and uvial features along the Rowley Shelf. Fluvial features are identied through the presence of prograding sedimentary bodies,
associated with a channel, comparable to what is observed along the modern coast (B, E). In most cases, beach ridges (strandplains) can be identied (D, E). Relict
mouth bars were not rmly identied. Relict tidal features are characterised by the presence of ebb-tidal deltas (F, G, H) associated with barriers, tidal inlets and
lagoons (F, G, H, I). In rare cases, tidal bars can be observed (F, J). Seismic data used to produce the bathymetry displayed in D and F courtesy of TGS. Bathymetry (A,
C, G) from Lebrec et al. (2021). Imagery (B, E, H, I, J) source: Esri, Maxar, GeoEye, Earthstar Geographics, CNES/ Airbus DS, USDA, USGS, AeroGRID, IGN, and the
GIS User Community.
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
17
attempting regional correlation between submerged coastal features.
7.1. Modal sea-level depth 20
The MSLD 20 is characterised by three sub-parallel groups of beach
ridges and barriers of which the base was identied at 22, 18 and 16 m
below sea level. Overall, all three groups of beach ridges form a seaward
dipping plateau. The transition with deeper MSLDs is marked by a step
of 2–3 m on the outer edge of the plateau (Fig. 15C).
Between the North West Cape and the Dampier Archipelago, the
MSLD is highlighted by one main isolated foredune, at 16 to 20 m of
water depth (Fig. 7A), that is often part of strandplains and is locally
associated with a tombolo (Fig. 8C). Within the Dampier Archipelago,
palaeoshorelines are marked by highly karstied barrier spits and by
shoreface strata (Fig. 10). East of the Dampier Archipelago, submerged
coastal features are well preserved and all three subgroups can be fol-
lowed almost continuously toward Broome for nearly 800 km (Fig. 16).
Over this segment, three sub-areas can be dened. First, west of the De
Grey Delta, the relict features consist principally of Types 1 and 2
strandplains, which highlight the presence of numerous channels. In
nearshore areas, individual isolated foredunes can be identied but are
generally not well expressed. Second, over the De Grey Delta area, most
of the seabed is covered by modern bedforms. Strandplains and barriers
spits can nevertheless be identied on the outer border of the plateau,
associated in some cases with parabolic dunes. Third, east of De Grey
Delta, the seabed is characterised by barriers and Types 2 and 3
strandplains. Within this area, bedforms are regularly modied by blow
outs, and more rarely affected by overwash features (Figs. 7C and 9D).
Across MSLD 20, it was not always possible to observe clear wind/ wave
discriminating criteria and therefore some of the smallest ridges that
were classied as wind-built for consistency (e.g., Fig. 16, r2) may in fact
be wave-built.
Individual isolated foredunes and barriers are also identied locally
in water depth as shallow as 3 m (Figs. 14 and 16) and are referred to as
MSLD 7. These features, which are only visible from high-resolution
surveys, are often very localised and can not be tracked continuously
across the area of interest. In that context they were not discriminated
from MSLD 20.
7.2. Modal sea-level depth 35
The MSLD 35 is identied through a series of marine terraces that are
well expressed on 2D HR seismic lines in the vicinity of Broome, Port
Hedland (Fig. 12B) and Dampier. A few ridges interpreted as isolated
foredunes and strandplains were observed between Barrow Island and
the Dampier Archipelago.
7.3. Modal sea-level depth 50
The MSLD 50 is generally not very well expressed. It can be identied
through scattered relict isolated foredunes and marine terraces. The
most prominent bathymetry change is associated with the De Grey
Delta, where the 50 m contour line advances signicantly seaward
(Fig. 13). North west of the Dampier Archipelago, a potential mouth bar
was identied in association with a uvial channel.
7.4. Modal sea-level depth 60
The MSLD 60 is characterised by relict isolated foredunes and marine
terraces. Most of the features appear to be developed on top of ridges
Fig. 12. Illustration of marine terrace (MT) morphologies along the Rowley Shelf. A) Location map. Marine terraces are characterised by a seaward dipping terrace,
backed by a topographic high. Such objects can be identied at small scale (D, WA DoT LiDAR bathymetry) from high-resolution surveys but are mostly observed at
larger scale and are often associated with a change in the trend of the regional slope (B, C, seismic data courtesy of Geoscience Australia). Regional bathymetry (A)
from Lebrec et al. (2021). Imagery (E) source: Esri, Maxar, GeoEye, Earthstar Geographics, CNES/ Airbus DS, USDA, USGS, AeroGRID, IGN, and the GIS
User Community.
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
18
associated with the MSLD 70 features (see Section 7.5.) which would
have then acted as barriers (Fig. 15C). Such features are observed west of
Broome, north west of Port Hedland and north west of Barrow Island.
Lastly a large V shaped embayment, tied to a channel was observed
between the Dampier Archipelago and the Rankin Bank (Fig. 13).
Fig. 13. Distribution and nature of submerged coastal features identied along the Rowley Shelf, and of the associated modal sea-level depth shorelines which
correspond to areas exhibiting higher concentration of relict features. D: Direct morphological sea-level indicator. I: Indirect morphological sea-level indicator.
Regional bathymetry from Whiteway (2009)
Fig. 14. Depth distribution of direct morphological sea-level indicator along the Rowley Shelf. Submerged features appear to be concentrated over specic depths,
referred to as modal sea-level depths. The number of features identied is dependent on the data coverage.
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
19
Fig. 15. Morphology of the Rowley Shelf. A) Location map. The shelf morphology is marked by several steps (C-C
′, D-D′) which can be followed from west to east.
These steps result from the stacking of several relict coastal features such as beach ridges on top of each other (D, Inset E) and in turn correspond to MSLDs. In some
cases, and especially in more than 100 m of water depth, relict coastal features are overlain by younger sediments (B-B′, D-D′). Water depths (m) are derived using a
bathymetry prole along the seismic lines. Geophysical datasets courtesy of Geoscience Australia and Whiteway (2009).
Fig. 16. Illustration of MSLD 20 features using satellite-derived bathymetry and regional DEM from Lebrec et al. (2021). At least three groups of sub-parallel relict
features can be identied at depth of 16 m (r2), 18 m (r3) and 24 m (r4). Locally in nearshore area ridges can be observed in 3 m of water depth (r1). Unconsolidated
relict sediments are often reworked to form modern bedforms attached to the relict ridges (rw). Relict ridges are continuous between Dampier and Broome except
near uvial systems such as the De Grey Delta where they are overlaid by modern sediments (m).
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
20
7.5. Modal sea-level depth 70
The MSLD 70 is characterised by large embayments spread across the
shelf that are locally backed in the vicinity of Barrow Island, north east
of Port Hedland and west of Broome by extensive relict isolated fore-
dunes, forming relief up to 15 m high (Figs. 13 and 15C–D). This
interpretation is consistent with the results from O’Leary et al. (2020)
who interpreted ridges located west of Barrow Island as relict coastal
dunes. These features would have acted as barriers during subsequent
sea level rise. North of Port Hedland, a linear erosive surface, which
underlies both the MSLD 80 and MSLD 90 features, appears to extend
from the MSLD 70 beach ridges. Finally, north of Exmouth Gulf, the
MSLD 70 corresponds to the base of an extensive plateau located in 70 to
50 m of water depth. Seabed features, similar to beach ridges, can be
observed on top of the plateau. However, it is not possible to be
conclusive on their origin due to the presence of extensive dissolution
features (Fig. 6B).
7.6. Modal sea-level depth 90 & 80
The MSLD 90 is marked by a steep rise from 90 to 95 m of water
depth to about 75 m of water depth that can be tracked over most of the
area of interest (Fig. 15C). Between the Rankin Plateau and Glomar
Shoal, the rise is segmented by isolated barrier complexes associated
with barrier spits, ebb-tidal deltas and tidal inlets (Fig. 11G). Locally, it
is possible to identify an intermediate step, referred to as the MSLD 80.
The section between Glomar Shoal and the latitude of De Grey Delta is
characterised by extensive strandplains developed at the back of the rise
(Fig. 17). In some rare cases, the 80 m modal shoreline is visible along
the rise as an isolated ridge. Eastward, the MSLD 90 features are mainly
inferred due to the lack of data, with the exception of the area directly
west of Broome, where scattered MBES and 2D seismic lines reveal the
presence of submerged strandplains.
7.7. Modal sea-level depth 105
The MSLD 105 is one of the most prominent features of the shelf and
is often marked by a regional increase of the slope (Figs. 15C and 17). On
the eastern part of the shelf, between the North West Cape and the
Rankin Plateau, relict features are buried under modern sediments and
are only observed on one seismic line as a ridge (Fig. 15B). Similarly,
only one ridge is visible on the bathymetry between Rankin plateau and
Glomar Shoal. Eastward, the MSLD 105 becomes one of the most
prominent features of the shelf and its outer boundary marks a regional
slope break on the bathymetry. The sector between Glomar Shoal and
the latitude of the De Grey Delta is marked by an extensive coastal plain,
which extends up to 25 km inland, that includes on its western part an
interplay of beach ridges and uvial channels (Fig. 17), and on its
eastern part, a barrier complex characterised by barriers, up to 15 m
high, lagoons, tidal inlets, tidal bars and ebb-tidal deltas (Fig. 11F). East
of this point, the modal shoreline is less visible on the seabed, mainly
due to the lack of high-resolution surveys. Scattered MBES lines never-
theless reveal large embayments backed by beach ridges.
7.8. Modal sea-level depth 125
The MSLD 125 is mainly characterised by sets of beach ridges and
truncating surfaces that are systematically overlapped by 5 to 15 m of
younger sediments and are therefore only observable from seismic data
(Fig. 15B). East of Barrow Island, the modal depth is characterised by
sets of beach ridges and potential tidal inlets. However, due to the sparse
coverage of high-resolution seismic, the relation between the ridges and
the channels is unknown. In the central part of the area of interest,
offshore Dampier, the MSLD seems to be marked by several erosive
surfaces. Near Broome, the seismic data display a set of ridges that
transition into clinoforms, which are interpreted as a strandplain and
shoreface strata, respectively.
8. Discussion
8.1. Age and preservation of MSLD shorelines
In the absence of geological samples, the majority of published
studies have inferred the age of submerged palaeoshorelines and asso-
ciated coastal features based on their morphologies, stratigraphic order
and observed water depths, subsequently calibrated using relative sea-
level curves derived from oxygen isotopes (e.g., Gardner et al., 2005;
Goff et al., 2018; Liquete et al., 2007; Micallef et al., 2013; O’Leary et al.,
2020; Rangel et al., 2020; Wellner et al., 1993). In tectonically stable far
eld domains the observed relative sea level corresponds to the com-
bination of the eustatic sea level with glacial-hydro-isostatic variations
(Lambeck, 2014). Along the north western Australian region, hydro-
Fig. 17. Illustration of MSLD 90 and 105. Seismic data used to produce the displayed bathymetry (B) courtesy of TGS. Regional bathymetry (A) from Whiteway
(2009) and Lebrec et al. (2021). Both MSLDs are characterised by extensive strandplains. The transition between MSLD 90 and 105 is marked by a step of 10–15 m.
MSLD 105 is composed of at least 4 intermediate palaeoshorelines (1) to (4) which are intersected by channels (c) and illustrate the progradation of the coast. On the
outer edge of the MSLD 105, beach ridges appear to be truncated, indicating later reworking.
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
21
isostatic effects were estimated to amount for 10% of the relative sea
level variations (Lambeck and Nakada, 1990; Yokoyama et al., 2001a).
Similar values were reported in the Red Sea (Grant et al., 2014), where a
relative sea-level curve was constructed from U/Th-dated speleothem.
Thus, the relative sea-level curve from the Red Sea (Fig. 18, Grant et al.,
2012; Grant et al., 2014) could be theoretically used to build an age
model to associate each of the Rowley Shelf MSLD features to specic
marine isotope stages, sensu Lisiecki and Raymo (2005).
However, there are inherent difculties in constraining the precise
position of palaeo sea level using oxygen isotope records due to the
confounding effects of salinity, temperature, carbonate ion effects, and
carbonate diagenesis on very small isotopic and trace element signals
(Adkins et al., 2002; Eldereld et al., 2006; Lambeck et al., 2014;
Lisiecki and Raymo, 2009; Skinner and Shackleton, 2005; Spratt and
Lisiecki, 2016). In that regard, Siddall et al. (2003) reported, based on
the work from Adkins et al. (2002), that a misestimation of the water
temperature by ±1 degrees can result in inferred sea-level uncertainties
of ±30 m. More recently, Spratt and Lisiecki (2016) reported that most
sea-level curves have uncertainties in the order of 10 to 20 m. In addi-
tion, considerations need to be made regarding post depositional verti-
cal displacement of palaeoshorelines as a result of glacial-hydro-isostatic
effects (Lambeck and Nakada, 1990; Yokoyama et al., 2001a), dynamic
topography (Mitrovica et al., 2020), subsidence (Montaggioni et al.,
2019), and/or neotectonics (Lambeck et al., 2011; Sandstrom et al.,
2020). Recent work has demonstrated that even along the North West
Shelf, considered as a passive margin, vertical displacements of up to 8
m have been locally observed since the last interglacial (Whitney et al.,
2016). In that context a feature identied at -70 m could well have been
formed during MIS 3, 4, 5 or during older glacial/ interglacial cycle
(Fig. 18).
Lastly, MSLD shorelines along the Rowley Shelf exhibit complex
architectures that are not fully resolved by the existing dataset and
prevent doing such age assumptions. For example, north of Port Hed-
land, beach ridges associated with the MSLD 90 appear to be strati-
graphically younger than beach ridges from the MSLD 80 and, in turn,
stratigraphically older than beach ridges of the MSLD 105 (Figs. 15C and
17). Such pattern tends to indicate that all three MSLDs are part of the
same relative sea-level fall trend. However, locally, both MSLD 90 and
105 include ebb-tidal deltas which are locally overprinting the MSLD
105 beach ridges (Fig. 11F). This indicates that MSLD 90 contains fea-
tures that are stratigraphically younger and older (i.e., stratigraphically
above and below) MSLD 105 features. Such observation suggests that
MSLDs may contain diachronous features, potentially formed through
multiple eustatic cycles. While in this case it is possible to identify two
separate events that resulted in the formation of beach ridges and
overlapping tidal complexes, the coverage and the resolution of the
geophysical data is often not sufcient to discriminate individual fea-
tures along a given MSLD (Fig. 15E) or to fully resolve the stratigraphic
relationship between adjacent MSLD shorelines (Fig. 15B and D).
Moreover, most of the beach ridges identied through MSLD 20 to MSLD
125 exhibit some level of cementation, highlighted by seismic blanking,
which likely increased their resistance to coastal erosion and hence
increased the possibility that they could be preserved through multiple
glacial cycles.
The above observations emphasize that extreme care should be used
when attempting to age date submerged coastal features using relative
sea-level curves as a reference. Additionally, given that MSLD may
contain diachronous features, this study also highlights the importance
of having a detailed stratigraphic model, even when suitable age-dating
material is available.
This being said, a few comments can be made on the age and pres-
ervation of the Rowley Shelf MSLD shorelines. First, it should be noted
that the LGM was identied at depths comprised between 121 and 125
m below sea level in the Bonaparte Basin (Sahul Shelf), further north
along the NWS (Yokoyama et al., 2001a), which suggests that MSLD 125
could correspond to the LGM. This interpretation is further supported by
age dating performed offshore Barrow Island, at IODP site U1461, that
identied the LGM 14 m below the seabed, in water depth of 127 m
(Ishiwa et al., 2019b), in line with the observation of MSLD 125 offshore
Broome (Fig. 15D and B). Recent work on the Great Barrier Reef found
that for most of the LGM the sea level was lower by about 100 m relative
to the modern level and only plunged by an additional 20 m for a brief
period of time between 21,900 and 20,500 years BP (Yokoyama et al.,
2018). Similar results were also obtained along the Sahul Shelf, in the
Bonaparte Basin (Ishiwa et al., 2019a). The fact that MSLD 105 is the
most prominent feature of the Rowley Shelf and that MSLD 125 is often
poorly expressed or even not visible (e.g., Fig. 17) suggests that MSDL
105 could correspond to the LGM sea-level plateau described by the
above authors. Such interpretation would tend to support their results.
Beach ridges associated with MSLD 20 and MSLD 70 exhibit disso-
lution features suggesting that they were exposed to subaerial dissolu-
tion after their deposition and cementation. Subaerial exposure could
only have happened if beach ridges were deposited and cemented prior
to the inundation post LGM. However, there is not enough information
to estimate whether this happened during the last glacial cycle (MIS 5 to
MIS 3) or previous ones. Finally, ebb-tidal deltas identied within MSLD
Fig. 18. Relative sea-level curve and the associated uncertainties from Grant et al. (2012) and Grant et al. (2014). MIS boundaries from Lisiecki and Raymo (2005).
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
22
90 and 105 appear to be formed on top of the MSLD 105 and 125,
suggesting that they were formed after the LGM. This interpretation is in
line with radiocarbon dating performed on ooids, sampled from the
lagoon associated with the MSLD 105 ebb delta, that yielded an age
between 12.7 ka BP and 15.4 ka BP (James et al., 2004). Given that these
features do not appear to be cemented, their preservation could be
related with the meltwater pulse 1a at circa 14.5 ka BP, which corre-
sponds to a rapid rise of the sea level of approximately 20 m from −100
to −80 m bsl (Deschamps et al., 2012; Lambeck et al., 2014). Overall,
most of the submerged coastal features observed on the seabed do not
appear to be overprinted by younger features and were therefore likely
formed during the last glacial cycle. The presence of diachronous fea-
tures within MSLDs may illustrate either interstadial uctuations of the
sea level or the succession of older glacial/ interglacial cycles. Addi-
tional age dating should be conducted in order to develop a detailed and
reliable age model for the MSLD shorelines of the Rowley Shelf.
8.2. Inuence of palaeoshorelines on shelf morphology
The Rowley Shelf is extensively covered by submerged relict coastal
features (Fig. 13). Areas with a lesser density of reported submerged
coastal features correspond to areas with data gaps, typically where only
regional low-resolution bathymetry is available. There is little doubt
that additional surveys will reveal additional features and allow a more
precise reconstruction of past relative sea level and climate. Neverthe-
less, it is already possible to note that a large portion of major topo-
graphic changes along the shelf are related to relict coastal features. In
that regard, bathymetry steps highlighted in Fig. 15 and by previous
publications, such as 30 m cliffs (James et al., 2004), steps (Jones, 1973)
and terraces (Hengesh et al., 2011), all represent relict coastal features.
One notable exception is the presence of antecedent topographic highs
such as Barrow Island, remnant of Miocene anticlines (McNamara and
Kendrick, 1994) or the Burrup Peninsula (Dampier), made of Archean
rocks (Hickman and Strong, 2003; Kojan, 1994).
In that context, one could therefore question whether relict coastal
features are shaping the shelf or if they simply stabilise along pre-
existing topography. While the latter likely plays a part, given that
antecedent topographies can inuence the rate at which a shoreline can
migrate laterally, seismic sections reveal that, often, antecedent topog-
raphies are themselves composed of older stacked coastal features
(Fig. 15E), hence conrming their preponderant role on the shelf
morphology.
The combination of relative sea-level changes, and associated coastal
deposits, with the antecedent topography over multiple glacial/ inter-
glacial cycles, would therefore lead to the accumulation of multiple
diachronous coastal features over limited areas, resulting in the devel-
opment of MSLD shorelines that represent the main morphological
features of the shelf. Such behaviour is strengthened by the shelf sedi-
mentation pattern: the shelf is mainly an area of sediment transport
rather than deposition and sediments which are not transported along
the coast to form coastal features, primarily bypass the inner and middle
shelf to accumulate on the outer shelf, leaving relict features outcrop-
ping on most of the seabed (James et al., 2004; Jones, 1971; Jones,
1973). Interestingly, no kilometre-scale relict bioconstructed reefs,
including coral-reef terraces, were observed along the Rowley Shelf
during this study. This is in stark contrasts to the eastward facing Great
Barrier Reef, along which relict linear reefs several km-long were
identied at water depths of 80 m, 90 m, 100 m and 110 m (Webster
et al., 2018). Along the Rowley Shelf, kilometre-scale linear features
seem to be systematically formed by relict coastal features, as no mor-
phologies characteristic of true coral reefs (e.g., atoll morphologies,
spurs and grooves, reef pass; see Khanna et al., 2017) were observed,
while morphologies characteristic of coastal features are abundant. It is
not excluded that relict submerged coral reefs may be locally present
along the Rowley Shelf either in between or on top of palaeoshorelines.
In any case, such features would have only a limited impact on the
regional shelf morphology.
8.3. Shaping processes along carbonate ramps
The Rowley Shelf and more largely the NWS is a carbonate-
dominated platform and seabed sediments typically exhibit a carbon-
ate content in excess of 90% (Dix, 1989; Dix et al., 2005; James et al.,
2004). Fluvial inputs are limited both in time and space (Carrigy and
Fairbridge, 1954; James et al., 2004), yet almost all of the submerged
coastal features identied within the area (i.e., strandplains, barrier
complexes and shoreface strata) are features typically associated with
siliciclastic coasts. Coastal classications based on wave, uvial and tide
processes have been developed for siliciclastic coasts and were not
intended to be used on coasts dominated by carbonate sediments (e.g.,
Ainsworth et al., 2011; Nyberg and Howell, 2016). However, it is
possible to apply those classications to the different submerged
shorelines of the shelf, which would then be categorised as wave
dominated (i.e., primary process) and uvial-inuenced or tidal-
inuenced (i.e., secondary process). The main difference with purely
siliciclastic coasts is that the sediment budget does not solely rely on
uvial inputs but also benets from the marine carbonate production.
In warm-water environments with a dominant carbonate sedimen-
tation, ridges observed from geophysical surveys tend to be interpreted
as submerged/ drowned bioconstructed reefs (Hinestrosa et al., 2016;
McCaffrey et al., 2020; Nichol and Brooke, 2011; Rosleff-Soerensen
et al., 2012; Vora et al., 1996; Wu et al., 2009). In contrast, along the
modern Rowley Shelf, all major morphological changes and associated
ridges appear to be created by submerged coastal features or controlled
by structural geology (e.g., anticlines). Indeed, along the shelf, sub-
merged relict coastal features that are described here, can form ridges up
to 15 m high as the result of wave and aeolian processes. As such, sub-
merged ridges in the vicinity of Barrow Island and the Rankin Plateau
(Fig. 11G), which were previously interpreted as coral reefs (Gallagher
et al., 2017a; Gallagher et al., 2017b; Gallagher et al., 2017c; Gallagher
et al., 2014) are here interpreted as cemented coastal features including
barrier spits, tidal inlets and ebb-tidal deltas. Similarly, submerged
topographic highs west of Barrow Island (Fig. 9G) – also interpreted as
bioconstructed reefs by Gallagher et al. (2014), are here interpreted as
submerged barrier complexes, in line with O’Leary et al. (2020). These
interpretations are consistent with observations made along the modern
coast (Semeniuk, 1992, 1996) and shows that topographic highs formed
in warm-water, carbonate-dominated environments are not always
bioconstructed. Submerged relict strandplains and barriers were also
described along carbonate shelves in the US, Israel or South Africa (Finkl
and Andrews, 2008; Jarrett et al., 2005), suggesting that the features of
the Rowley Shelf are not an exception.
Modern fringing and mesophotic coral reefs are nevertheless present
along the Rowley Shelf where they locally impact the seabed
morphology (Abdul Wahab et al., 2018; Bonesso et al., 2020). In that
regard, modern fringing coral reefs are reported along the coast, most
notably in the Dampier Archipelago and Muiron Islands, but also near
smaller islands such as Thevenard, Bedout, and Montebello islands
(Bonesso et al., 2020; Kobryn et al., 2013; Moustaka et al., 2019). Sys-
tematically, these coral reefs appear to be aligned with underlying relict
beach ridges identied from the bathymetry (Fig. 19). Onshore obser-
vations from the Dampier Archipelago revealed that Legendre and
Kendrew Islands, that are also aligned with modern reefs, are relict
dunes (Kojan, 1994; Wilson and Marsh, 1974), which tends to conrm
the causal link between the presence of submerged palaeoshorelines and
bioconstructed reef development. Additionally, modern mesophotic
corals were described over the Rankin Plateau and Glomar Shoal
(Fig. 20, Abdul Wahab et al., 2018). Interestingly, the bathymetry ac-
quired by AIMS reveals that the corals are developed on top of linear
ridges that could be interpreted as submerged beach ridges. The above
suggests that shelfal coral reefs from the Rowley Shelf would have
preferentially developed along the pre-existing topographic highs
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
23
formed by relict coastal features. Similar settings were reported in
Florida, South Africa, Gulf of Papua and India, where fringing corals are
developed on top of submerged beach ridges (Jarrett et al., 2005; Mal-
larino et al., 2021.; Ramsay, 1994; Rao et al., 2001).
Several studies point out that antecedent topography has a critical
role in the development of coral reefs (Cabioch et al., 1995; Chevalier,
1973; Coudray, 1976; Taisne, 1965; Veron, 1995). In that context, most
modern shelfal reefs provinces, such as the Australian Great Barrier
Reef, Belize Barrier Reef, Florida Keys Barrier Reef, and the New Cale-
donian Barrier Reef are developed on coastal terrigenous deposits
(Cabioch et al., 1995; Cunningham, 1998; Davies et al., 1989; Ferro
et al., 1999), including submerged beach ridges and delta deposits
accumulated during lowstands (Droxler and Jorry, 2013; Franchi et al.,
2018; Hartman et al., 2015) hence supporting our interpretations.
Recently, it has been demonstrated that at least two Miocene palaeo
deltas in the South China Sea provided favourable substratum of
elevated sand bars for the emergence of transgressive carbonate build
ups (Mathew et al., 2020) further highlighting the importance of better
understanding the link between palaeoshorelines and coral develop-
ment. Our results also show that bioconstructed reefs do not always
Fig. 19. Inuence of relict coastal features on modern environments along the Rowley Shelf, examples using satellite-derived bathymetry (B-E) from Lebrec et al.
(2021) and LiDAR bathymetry provided by WA DOT (B). Regional bathymetry (A) from Whiteway (2009). Most of the islands of the Rowley Shelf (e.g., Muiron,
Rosily West, Legendre, Bedout) appear to be aligned with relict coastal features. This is especially visible in the Dampier Archipelago (B), where outer islands and
associated fringing corals are directly on top of a relict coastal feature, while inner islands are composed of Archean rocks (Kojan, 1994). Similarly, islands offshore
Onslow all appear to be somewhat related to the submerged relict coastal features (C, E). The same pattern is observed offshore of the De Grey Delta with Bedout
Island (D).
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
24
affect the morphology of tropical carbonate provinces at a regional
scale, even when they are present. The dominant forcing factors to
tropical carbonate shelf morphology can therefore be, as for siliciclastic
coasts, wind, wave, tide and uvial processes.
Finally, these observations encourage caution when interpreting
drowned topographic highs as coral reefs, because signicant carbonate
accumulation can be formed by clastic (sensu composed of reworked
sediments) coastal features. In particular, it has long been known that
aeolian deposits can rival coral reefs in term of potential for carbonate
accumulation (e.g., Darwin, 1842), and reach signicant heights, with
for examples coastal barriers up to 200 m high observed in South Africa
(Bateman et al., 2011) and coastal aeolianite deposits up to 60 m high
reported in South Australia (James and Bone, 2015). For comparison, it
is estimated that the total thickness of the bioconstructed coral reef from
Ningaloo Reef (i.e., including both Pleistocene and Holocene reefs), off
Western Australia coast, is less than 40 m (Collins et al., 2003). In the
absence of data with a resolution good enough to identify morphologies
characteristic of non-reefal coastal features (e.g., beach ridges) or coral
reefs (e.g., discrete patch reefs, passes), it may not be possible to
differentiate drowned bioconstructed reefs from submerged relict
coastal features. Such considerations are in line with previous work from
Moustaka et al. (2019) and Schlager (2005).
8.4. Implication for palaeoenvironment reconstruction
Submerged coastal features observed along the Rowley Shelf have
recorded the dominant processes involved in the formation of the MSLD
shorelines and in turn, provide insights on prevalent palaeoenviron-
ments. While all observed features accumulated along wave-dominated
coasts, some are tide-inuenced (e.g., barriers, ebb-tidal deltas and
associated features; Figs. 9 and 11) and others are uvial-inuenced (e.
g., strandplains type 3, mouth bar; Figs. 8 and 11). In both cases, uvial-
inuenced and tide-inuenced features appear to be restricted to spe-
cic MSLD shorelines.
In that regard, the largest relict ebb-tidal deltas, characteristic of
tidal environments, belong to MSLD 105 and MSLD 80 shorelines
(Fig. 11F–G). They are associated with lagoons, lled by large accu-
mulations of ooids and peloids, dated 15.7 to 12.4 ka BP (James et al.,
2004), and extensive barriers, over 15 m high. Such accumulation of
ooids and peloids were previously used to infer that the shelf was once
bordered by a tide-inuenced ‘Bahama-like’ platform (Dix et al., 2005;
Gallagher et al., 2018), indicating a dry, sediment starved environment.
Tide-inuenced features are observed in shallower water depths, like in
the vicinity of Port Smith lagoon, where ooids were described in a
modern tidal inlet (Hearty et al., 2006) in association with an ebb-tidal
delta (Fig. 11H), but their extent appears limited compared to the sub-
merged tidal complexes observed at depths 80 to 105 m bsl.
In contrast, the palaeo-delta of the De Grey River appears more
developed in water depths shallower than 50 m than in deeper waters.
Similarly, a large uvial channel and associated mouth bar is identied
north west of Dampier in 50 to 60 m of water depth. Unfortunately, due
to the scarcity of high-resolution surveys between 30 and 70 m bsl, this
interpretation is based on low-resolution regional bathymetry and re-
mains uncertain. If conrmed, this would support the view that uvial
runoff, and hence the Australian Summer Monsoon, were reduced dur-
ing glacial periods along the Rowley Shelf, and would re-establish dur-
ing the glacial to interglacial transition (Denniston et al., 2013; Hesse
et al., 2004). The presence of isolated uvial channels and of uvial-
inuenced strandplains along the MSLD 105 shoreline suggests that
uvial runoff was nevertheless not completely stopped during such
periods.
Overall, tidal features are more abundant along deeper MSLD
shorelines (>70 m), accumulated during glacial periods, while uvial
features appear more abundant along shallower MSLD shorelines (<60
m), accumulated during warmer periods. This pattern is interpreted as
representing climatic variations: humid climates favour the develop-
ment of uvial-inuenced features through an increase of uvial run-
offs, whereas arid climates enable the onset of tide-inuenced features.
Fig. 20. Inuence of relict coastal features on modern environments along the Rowley Shelf. Figure modied from Abdul Wahab et al. (2018). A study from Abdul
Wahab et al. (2018) describes the presence of mesophotic corals over Rankin Bank (A) and Glomar Shoal (B). The bathymetry presented in their study reveals, under
the coral mounds, the presence multiple beach ridges.
U. Lebrec et al.
Earth-Science Reviews 224 (2022) 103864
25
As such, this trend supports the well-established view that the Pleisto-
cene continental climate of north-west Australia was arid during glacial,
lowstand periods, and humid during interglacial, highstand periods
(Gallagher et al., 2018; Gallagher et al., 2014; Hallenberger et al., 2019;
Hesse et al., 2004), in relation to the development of the monsoon
(Hallenberger et al., 2019; Hesse et al., 2004; Van Der Kaars and De
Deckker, 2002; Wyrwoll and Miller, 2001). The above observations also
suggest that, along the Rowley Shelf, carbonate production dominates
dry lowstand sedimentation while siliciclastic mostly accumulates dur-
ing highstands, in line with previous work (Dix et al., 2005).
9. Conclusions
Submerged coastal features are key indicators of past relative sea
level and environments. The information recorded by each relict feature
varies depending on the coastal processes involved in their genesis and it
is often challenging to interpret them using geophysical data only. Based
on the review of 118 published case studies around the world, four main
categories of relict coastal features, which can be characterised as
palaeoshorelines indicators, were dened: (1) beach ridges of either
wind or wave origin; (2) shoreface strata; (3) marine terraces; and (4)
coral reef terraces. Each category can be identied from geophysical
surveys alone and exhibits clear morphological indicators allowing
reconstruction of past relative sea level and depositional environments.
Such features are typically preserved through drowning during a rela-
tive sea-level rise or via early cementation, the latter having a prepon-
derant role in sub-tropical to tropical latitudes.
In order to trial the review, geophysical surveys, including newly
processed high-resolution bathymetry and 2D seismic data, were inter-
preted over an area of 200,000 km
2
along the Rowley Shelf, the southern
half of the North West Shelf of Australia, a tropical carbonate platform.
The discrimination of modern and relict features was conducted using
ve main criteria, namely the: (1) stratigraphic position of the feature;
(2) presence of emersion indicators; (3) similarity with modern and
published analogues; (4) comparison of the feature orientation with the
modern ocean currents; and (5) identication of seismic attenuation,
highlighting cemented intervals. While relict features can be identied
from bathymetry alone, high-resolution seismic should be used to pre-
cisely identify palaeo sea level as relict features are often overlain by
modern sediments, hence altering the measurements. Additionally,
vertical offsets of several meters were observed between high-resolution
surveys indicating that extreme care should be used when comparing
palaeo sea-level values across surveys and regions.
The interpretation of the Rowley Shelf data led to the identication
of over 500 submerged relict coastal features spread evenly across the
shelf from 5 to 145 m of water depth. Relict features of the Rowley Shelf
belong to each of the categories identied from the review, apart from
relict coral-reef terraces (i.e., coral reefs with a clear terrace morphology
allowing location of palaeo sea level using geophysical data), which
were not identied within the area of interest. Such a conclusion does
not exclude the local presence of other types of drowned coral reefs that
are out of the scope of this study. Relict features appear to be concen-
trated at specic water depths, which represent modal sea-level depths,
at which the shoreline remained stable over longer periods of time. Over
the Rowley Shelf, nine modal sea-levels were identied at depths of 20,
35, 50, 60, 70, 80, 90, 105 and 125 m. Each modal sea-level depth
shoreline is composed of stacked features most likely formed through
numerous glacial/ interglacial cycles. On the seabed, most outcropping
relict features are related to the sea-level fall of the last glacial cycle and
more rarely to the post LGM sea-level rise. The level of preservation of
the relict features is attributed to early and syn-deposition cementation.
The majority of morphological changes identied along the shelf
corresponds to modal sea-level depths suggesting that the shelf was
shaped by relict coastal features accumulated through successive
glacial/ interglacial cycles and associated uvial, tide, and wave coastal
processes. The