Pathways

of Inter‐Basin

Exchange

From the Bellingshausen

Sea to the Amundsen

Sea

M. Mar Flexas

1

, Andrew

F. Thompson

1

, Megan

L. Robertson

1

, Kevin

Speer

2

,

Peter M. F. Sheehan

3

, and Karen

J. Heywood

3

1

California

Institute

of Technology,

Pasadena,

CA, USA,

2

Florida

State University,

Tallahassee,

FL, USA,

3

University

of

East Anglia,

Centre

for Ocean and Atmospheric

Sciences,

School

of Environmental

Sciences,

Norfolk,

UK

Abstract

The West Antarctic

Ice Sheet is experiencing

rapid thinning

of its floating

ice shelves,

largely

attributed

to oceanic

basal melt. Numerical

models

suggest

that the Bellingshausen

Sea has a key role in setting

water properties

in the Amundsen

Sea and further

downstream.

Yet, observations

confirming

these pathways

of

volume

and tracer exchange

between

coast and shelf break and their impact

on inter‐sea

exchange

remain

sparse.

Here we analyze

the circulation

and distribution

of glacial

meltwater

at the boundary

between

the

Bellingshausen

Sea and the Amundsen

Sea using a combination

of glider observations

from January

2020 and

hydrographic

data from instrumented

seals. Meltwater

distributions

over previously

unmapped

western

regions

of the continental

shelf and slope reveal two distinct

meltwater

cores with different

optical

backscatter

properties.

At Belgica

Trough,

a subsurface

meltwater

peak is linked with hydrographic

properties

from

Venable

IceShelf.WestofBelgica

Trough,

thevertical

structure

ofmeltwater

concentration

changes,

withpeak

values occurring

at greater

depths

and denser

isopycnals.

Hydrographic

analysis

suggests

that the western

(deep) meltwater

core is supplied

from the eastern

part of Abbot Ice Shelf, and is exported

to the shelf break via

a previously‐overlooked

bathymetric

trough

(here named

Seal Trough).

Hydrographic

sections

constructed

from seal data reveal that the Antarctic

Coastal

Current

extends

west past Belgica

Trough,

delivering

meltwater

totheAmundsen

Sea.Eachofthesecirculation

elements

hasdistinct

dynamical

implications

fortheevolution

of

ice shelves

and water masses

both locally

and downstream,

in the Amundsen

Sea and beyond.

Plain Language

Summary

Floating

iceshelves

inWestAntarctica

arethinning,

whichislargely

due

to melting

of the ice shelf base by the ocean. Here, measurements

of ocean temperature,

salinity,

and dissolved

oxygen,

collected

by a remotely‐controlled

underwater

vehicle

(a glider),

are used to estimate

the amount

of ice

shelf meltwater

released

in the Bellingshausen

Sea. Distinct

cores of meltwater

can be distinguished

by the

amount

of suspended

material

that is present

in the water, which we attribute

to meltwater

following

different

circulation

pathways

after entering

the ocean. Historical

data from seals equipped

with temperature

and salinity

sensors

provide

additional

information

about how the meltwater

circulates

in the region.

The seal data show the

presence

of a narrow

coastal

current

that bringsmeltwater

from the Bellingshausen

Sea intothe Amundsen

Sea.

The pathways

of meltwater

revealed

in this study suggest

an important

influence

of the Bellingshausen

Sea on

ice shelves

throughout

West Antarctica.

1. Introduction

In past decades,

the Antarctic

Ice Sheet has experienced

rapid thinning

of its floating

ice shelves

and grounding

lineretreat(Konrad

etal.,2018; Pritchard

etal.,2012). Satellite

observations

starting

inthe1990srevealthatboth

thinning

and retreat have recently

accelerated

and are largest

for the West Antarctic

Ice Sheet (WAIS)

(Konrad

et al., 2018; Paolo et al., 2015). Changes

in the thickness

of WAIS ice shelves

are largely

attributed

to oceanic

basal melt (Adusumilli

et al., 2020; Pritchard

et al., 2012) driven

by changes

in the heat transport

of warm

Modified

Circumpolar

Deep Water (MCDW;

Whitworth

et al., 1998) intruding

onto the continental

shelf via

bathymetric

troughs.

Thisheattransport

maydepend

onthevelocity

fieldand/orthethickness

andtemperature

of

the MCDW

layer. This process,

dominant

from the West Antarctic

Peninsula

(WAP)

(Hofmann

& Klinck,

1998;

Klinck

et al., 2004; Moffat

et al., 2009) to the Amundsen

Sea, is triggered

by wind‐driven

cross‐slope

and cross‐

shelf exchange

processes

(Dutrieux

et al., 2014; Jacobs

et al., 2011; Jenkins

et al., 2018; Silvano

et al., 2022;

Thoma

et al., 2008) acting from seasonal

to decadal

time scales (Holland

et al., 2019; Jenkins

et al., 2018; Paolo

et al., 2018; Schodlok

et al., 2012; Wallis

et al., 2023).

RESEARCH

ARTICLE

10.1029/2023JC020080

Key Points:

•

Hydrographic

observations

identify

both shelf‐break

and coastal

meltwater

pathways

from the western

Belling-

shausen

Sea into the Amundsen

Sea

•

Differences

in optical

backscatter

properties

associated

with meltwater

are related

to distinct

coast‐to‐shelf

break pathways

•

The main pathway

to the shelf break is

via Seal Trough,

identified

as the

de

facto

western

boundary

of the

Bellingshausen

Sea

Supporting

Information:

Supporting

Information

may be found in

the online version

of this article.

Correspondence

to:

M. M. Flexas,

marf@caltech.edu

Citation:

Flexas,

M. M., Thompson,

A. F.,

Robertson,

M. L., Speer, K., Sheehan,

P.

M. F., & Heywood,

K. J. (2024).

Pathways

of inter‐basin

exchange

from the

Bellingshausen

Sea to the Amundsen

Sea.

Journal

of Geophysical

Research:

Oceans

,

129

, e2023JC020080.

https://doi.org/10.

1029/2023JC020080

Received

23 JUN 2023

Accepted

9 MAY 2024

Author

Contributions:

Conceptualization:

M. Mar Flexas,

Andrew

F. Thompson,

Kevin Speer, Peter

M. F. Sheehan,

Karen J. Heywood

Formal

analysis:

M. Mar Flexas,

Megan

L. Robertson

Funding

acquisition:

Andrew

F. Thompson,

Kevin Speer, Karen

J. Heywood

Investigation:

M. Mar Flexas,

Andrew

F. Thompson,

Megan

L. Robertson,

Kevin Speer, Peter M. F. Sheehan,

Karen

J. Heywood

Methodology:

M. Mar Flexas,

Andrew

F. Thompson,

Kevin Speer, Peter

M. F. Sheehan,

Karen J. Heywood

Software:

M. Mar Flexas,

Megan

L. Robertson

Validation:

M. Mar Flexas

Visualization:

M. Mar Flexas

Writing

– original

draft:

M. Mar Flexas,

Andrew

F. Thompson,

Kevin Speer

©2024. American

Geophysical

Union.

All

Rights

Reserved.

FLEXAS

ET AL.

1 of 18

More

recently,

studies

have

highlighted

the

importance

of

lateral,

inter‐sea

exchange

on

setting

shelf

water

mass

properties

and

ice

shelf

melt

rates

around

Antarctica.

Recent

modeling

studies

suggest

that

water

masses

from

West

Antarctica

can

reach

all

regions

of

the

Antarctic

continental

shelf

in

only

15

years

(Dawson

et

al.,

2023

;

Nakayama

et

al.,

2020

).

Observational

studies

in

East

Antarctica

show

how

inter‐basin

exchange

preconditions

the

ocean

for

sea

ice

formation,

water

mass

transformation,

and

bottom

water

production

(Silvano

et

al.,

2018

).

In

particular,

glacial

meltwater

reduces

dense

shelf

water

formation

by

suppressing

on‐shelf

convection;

in

the

absence

of

deep

convection,

warm

water

reaches

further

onto

the

continental

shelf,

increasing

basal

melt

(Silvano

et

al.,

2018

).

In

West

Antarctica,

it

has

long

been

known

that

freshwater

transport

from

the

Amundsen

Sea

into

the

Ross

Sea

can

modulate

dense

shelf

water

production

(Jacobs

et

al.,

2022

;

Jacobs

&

Giulivi,

2010

).

Numerical

models

corroborate

the

connection

between

the

Amundsen‐Bellingshausen

seas

and

the

Ross

Sea

(Assmann

&

Timmermann,

2005

;

Beckmann

&

Timmermann,

2001

;

Nakayama

et

al.,

2014

).

Furthermore,

modeling

studies

show

that

glacial

meltwater

can

trigger

a

positive

feedback

mechanism

that

enhances

ice

shelf

melt

through

changes

in

ocean

stratification

at

the

ice

shelf

front

(Flexas

et

al.,

2022

).

Feedbacks

between

the

different

components

of

the

freshwater

balance

and

their

role

in

modifying

the

shelf

overturning

circulation

and

ice

shelf

basal

melt

(e.g.,

Bett

et

al.,

2020

;

Hyogo

et

al.,

2024

;

Jourdain

et

al.,

2017

;

Kimura

et

al.,

2017

;

Moorman

et

al.,

2023

)

are

critical

aspects

that

deserve

further

exploration,

in

particular

in

the

Bellingshausen

Sea.

The

Bellingshausen

Sea

(Figure

1

)

is

a

relatively

unexplored

region

of

West

Antarctica,

especially

when

compared

with

adjacent

regions

like

the

WAP

or

the

Amundsen

Sea.

Recent

observational

efforts

have

high-

lighted

some

distinguishing

characteristics

of

the

Bellingshausen

Sea's

shelf

and

slope

circulation.

The

warmest

waters

of

the

entire

West

Antarctic

shelf

seas

are

found

in

the

Bellingshausen

Sea

(Schmidtko

et

al.,

2014

),

where

MCDW

exhibits

hydrographic

properties

that

are

only

weakly

modified

from

offshore

values.

Warm

intrusions

of

MCDW

enter

the

continental

shelf

at

the

deepest

part

of

the

Belgica

Trough

flowing

toward

the

coast

along

the

eastern

side

of

the

trough

(Schulze

Chretien

et

al.,

2021

).

Water

mass

transformation

peaks

near

the

coast

and

Writing

–

review

&

editing:

M.

Mar

Flexas,

Andrew

F.

Thompson,

Kevin

Speer,

Peter

M.

F.

Sheehan,

Karen

J.

Heywood

Figure

1.

Map

of

the

Bellingshausen

Sea

with

major

bathymetric

features,

floating

ice

shelves,

and

the

2020

TABASCO

glider

expedition

(black

dots

/lines

marking

cross‐slope

sections

86W,

89W,

90W,

92W

and

96W,

and

along

shelf‐break

Sections

I

to

IV).

Seal

dives

from

the

MEOP

data

set

are

shown

as

gray

dots,

with

colored

dives

indicating

those

locations

where

the

maximum

seal

depth

exceeds

the

IBCSO

bathymetry.

Land

is

shown

in

black,

and

the

ice

shelf

edge

derived

from

BedMachine

Antarctica

v2

is

marked

in

violet.

The

480

m,

1,000

m,

2,000

m

and

3,000

m

isobaths

from

the

IBCSO

bathymetry

are

shown

in

gray.

Journal

of

Geophysical

Research:

Oceans

10.1029/2023JC020080

FLEXAS

ET

AL.

2

of

18

21699291, 2024, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2023JC020080 by California Inst of Technology, Wiley Online Library on [14/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

closes

the

shelf

overturning

circulation

as

MCDW

is

glacially‐modified

and

upwells

to

intermediate

(

∼

200

m)

levels

(Ruan

et

al.,

2021

).

Glacially‐modified

MCDW

recirculates

offshore,

toward

the

continental

shelf

along

the

western

side

of

Belgica

Trough

(Sheehan

et

al.,

2023

;

Thompson

et

al.,

2020

).

Similar

circulation

patterns

are

observed

in

Latady

Trough,

with

modified

warm

waters

eventually

flowing

northwards

along

the

western

side

of

the

trough

into

Belgica

Trough

(Schulze

Chretien

et

al.,

2021

).

At

the

shelf

break,

the

conjunction

of

glacially‐modified

MCDW

and

Winter

Water

(WW;

Mosby,

1934

),

the

winter

form

of

Antarctic

Surface

Water

(AASW),

leads

to

the

formation

of

the

Antarctic

Slope

Front

(ASF;

Jacobs,

1991

),

a

key

element

of

the

Antarctic

continental

margins

for

deep

ocean

ventilation

and

Antarctic

Bottom

Water

production.

The

ASF

(and

its

associated

current,

the

Antarctic

Slope

Current;

ASC)

is

a

quasi‐circumpolar

feature

that

originates

in

the

Bellingshausen

Sea

(Thompson

et

al.,

2020

)

and

terminates

in

the

southern

sector

of

the

Scotia

Sea

(Azaneu

et

al.,

2017

;

Heywood

et

al.,

2004

)

where

its

structure

is

largely

determined

by

tides

(Flexas

et

al.,

2015

).

Wind

stress

and

large‐scale

modes

of

climate

variability

play

an

important

role

in

controlling

the

strength

and

variability

of

the

ASC

(Gill,

1973

;

Spence

et

al.,

2014

;

Stewart

&

Thompson,

2012

;

Thompson

et

al.,

2018

).

Strikingly,

the

only

place

around

Antarctica

where

the

ASF

is

not

observed

is

at

the

WAP

(Thompson

et

al.,

2018

).

Close

to

the

coast

flows

the

Antarctic

Coastal

Current

(AACC),

a

boundary

current

that

contributes

to

the

renewal

of

shelf

waters

around

Antarctica

(Heywood

et

al.,

1998

,

2004

;

Jacobs,

1991

).

At

the

WAP,

the

AACC

carries

freshwater

from

the

WAP

(Moffat

et

al.,

2008

)

throughout

the

Bellingshausen

Sea

(Schubert

et

al.,

2021

).

The

AACC

varies

in

response

to

both

buoyancy

(Moffat

et

al.,

2008

)

and

wind

(Holland

et

al.,

2010

;

Sverdrup,

1953

)

forcing.

Numerical

simulations

suggest

that

the

AACC

has

a

key

role

in

setting

the

response

to

climate‐induced

surface

forcing

perturbations

(e.g.,

freshwater

fluxes

from

increased

run‐off)

through

its

role

in

setting

the

stratification

and

rates

of

vertical

heat

transport

in

the

water

column

(Flexas

et

al.,

2022

).

A

stronger

AACC

is

associated

with

enhanced

access

of

MCDW

into

ice

shelf

cavities,

ultimately

enhancing

basal

melt

(Flexas

et

al.,

2022

).

More

generally,

the

AACC

conveys

changes

in

water

properties

occurring

near

the

coast

and

enables

remote

responses

to

localized

perturbations;

these

dynamics

may

occur

broadly

around

Antarctica.

In

this

study

we

combine

hydrographic

data

from

historical

instrumented

seals

from

the

Marine

Mammals

Exploring

the

Oceans

Pole

to

Pole

(MEOP)

data

base

(Roquet

et

al.,

2013

)

with

data

from

an

autonomous

glider

deployed

in

2020

to

map

the

circulation

at

the

boundary

between

the

Bellingshausen

and

Amundsen

seas.

We

analyze

meltwater

fractions

and

backscatter

data

from

the

glider

observations.

We

revisit

the

MEOP

data

base

in

the

Bellingshausen

Sea

to

explore

bathymetry

and

sea

surface

dynamic

height

from

seal

hydrographic

profiles.

We

detect

meltwater

sources

from

different

ice

shelves

and

find

a

previously

overlooked

trough

that

plays

a

key

role

collecting

meltwater

at

the

westernmost

boundary

of

the

Bellingshausen

Sea.

Based

on

these

results,

we

discuss

pathways

of

water

property

exchange

(including

meltwater)

from

the

Bellingshausen

Sea

into

the

Amundsen

Sea.

We

shed

light

on

the

AACC

pathways

from

the

Bellingshausen

Sea

to

the

Amundsen

Sea

and

show

how

coastal

processes

in

the

Bellingshausen

Sea

are

connected

to

the

formation

of

the

ASC/ASF.

We

also

discuss

the

need

to

redefine

the

western

boundary

of

the

Bellingshausen

Sea

west

of

Belgica

Trough.

The

outline

of

the

paper

is

as

follows.

Section

2

details

data

and

methods:

in

Section

2.1

we

describe

the

new

glider

data

set;

in

Section

2.2

we

summarize

the

historical

MEOP

seal

data;

in

Section

2.3

we

present

the

method

used

to

calculate

meltwater

fractions.

Section

3

is

dedicated

to

present

results:

in

Section

3.1

we

present

the

glider

observations

at

the

shelf

break,

separating

the

data

into

cross‐slope

sections

(labeled

by

longitude)

and

along‐

shelf‐break

sections

(labeled

using

roman

numerals)

(Figure

1

);

in

Section

3.2

we

analyze

the

MEOP

database

seal

observations

over

Belgica

Trough

and

Seal

Trough;

in

Section

3.3

we

present

seal

observations

over

Thurston

Plateau.

In

Section

4

we

discuss

the

different

pathways

found

toward

the

shelf

break

and

along

the

coast.

Our

concluding

remarks

appear

in

Section

5

.

2.

Data

and

Methods

2.1.

Glider

Data

As

part

of

the

Transport

of

the

Antarctic

Peninsula

and

Bellingshausen

Sea:

Antarctic

Slope

Current

Origin

(TABASCO)

project,

Seaglider

SG621

was

deployed

in

the

Bellinghsausen

Sea,

in

West

Antarctica,

on

1

February

2020,

and

surveyed

the

western

part

of

the

Bellinghsausen

Sea

until

19

March

2020

(Figure

1

).

The

region,

usually

heavily

covered

in

sea

ice,

had

particularly

low

sea

ice

concentrations

in

early

February

during

this

Journal

of

Geophysical

Research:

Oceans

10.1029/2023JC020080

FLEXAS

ET

AL.

3

of

18

21699291, 2024, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2023JC020080 by California Inst of Technology, Wiley Online Library on [14/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

year.

From

Belgica

Trough

to

Thurston

Plateau,

the

glider

performed

multiple

transects

crossing

the

continental

shelf

and

slope.

The

glider

transects

sampled

progressively

westwards,

following

the

retreat

of

the

sea

ice

edge

(Figure

2

).

This

unique

data

set

reveals

the

interaction

of

shelf

and

slope

waters

at

the

boundary

between

the

Bellingshausen

Sea

and

the

Amundsen

Sea

at

high

horizontal

resolution

(2–4

km

between

profiles).

To

our

knowledge,

this

is

the

first

dedicated

set

of

glider‐based

(or

cruise‐based)

observations

ever

to

be

obtained

west

of

Belgica

Trough

in

the

Bellingshausen

Sea.

Seaglider

SG621

(Ogive

profile;

Kongsberg

Underwater

Technology,

Inc.)

belongs

to

a

family

of

underwater

autonomous

buoyancy‐driven

vehicles

capable

of

profiling

to

a

maximum

depth

of

1,000

m

in

a

sawtooth

(V‐

shape)

pattern

(Eriksen

et

al.,

2001

;

Rudnick,

2016

).

Seaglider

SG621

carried

a

Sea‐Bird

SBE3

temperature

sensor

and

SBE4

conductivity

sensor,

a

pressure

sensor,

an

Aanderaa

4330F

oxygen

optode,

WetLabs

ECOpuck

with

two

wavelengths

of

optical

backscatter

(470

and

700

nm)

and

chlorophyll

fluorescence

sensors.

Following

factory

calibration,

in

situ

temperature,

practical

salinity,

and

dissolved

oxygen

concentrations

are

accurate

to

0.018°C,

0.01,

and

2

mmol

kg

1

,

respectively.

Sensor

precision

is

0.0018°C

and

0.0003

S

m

1

for

temperature

and

conductivity,

respectively,

combining

to

a

salinity

precision

of

0.00121.

Sampling

occurred

approximately

every

5

s,

leading

to

a

0.5

m

vertical

resolution

at

typical

vertical

speeds

of

0.1

m

s

1

.

The

glider

data

set

was

processed

using

the

University

of

East

Anglia

glider

Toolbox

(Queste,

2013

),

which

includes

hydrodynamic

flight

model

corrections

(Frajka‐Williams

et

al.,

2011

)

and

thermal

lag

corrections

(Garau

et

al.,

2011

).

Depth‐averaged

currents

(DACs)

are

calculated

using

the

difference

between

the

expected

and

actual

surface

location

of

the

glider

at

the

end

of

each

profile

and

applying

hydrodynamic

flight

model

corrections.

We

use

potential

temperature,

practical

salinity

and

neutral

density

(γ

n

;

Jackett

and

McDougall

(

1997

))

unless

otherwise

noted.

Velocity

fields

were

constructed

by

calculating

the

geostrophic

shear

from

the

density

field

observed

by

the

glider

and

then

referencing

this

velocity

to

the

DACs.

Hydrographic

sections

were

constructed

by

optimally

interpolating

glider

observations

onto

a

regular

grid

with

horizontal

grid

spacing

of

about

300

m.

The

optimally

interpolated

scheme

uses

three

main

parameters

to

smooth

the

data:

a

given

number

of

grid

points

in

the

horizontal

direction

(

d

),

a

given

distance

in

the

vertical

direction

(

p

),

and

the

relative

error

(

ε

,

for

which

0

<

ε

<

1).

The

parameters

chosen

for

this

study

are

d

=

10,

p

=

15,

and

ε

=

0.2,

which

represent

horizontal

length

scales

of

3

km

and

vertical

length

scales

of

75

m

(given

a

vertical

grid

spacing

of

5

m).

Figure

2.

(a)

Sea

ice

concentration

on

15

February

2020

with

glider

positions

(yellow

dots)

up

to

the

date.

(b)

Same

as

panel

(a),

but

for

17

March

2020.

Land

is

shown

in

black,

and

the

ice

shelf

edge

is

marked

in

violet.

Bathymetry

contours

of

480

m,

1,000

m,

2,000

m

and

3,000

m

isobaths

are

shown

in

white.

For

names

of

major

bathymetric

features

and

floating

ice

shelves,

see

Figure

1

.

Journal

of

Geophysical

Research:

Oceans

10.1029/2023JC020080

FLEXAS

ET

AL.

4

of

18

21699291, 2024, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2023JC020080 by California Inst of Technology, Wiley Online Library on [14/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

2.2.

Seal

Data

Hydrographic

data

complementary

to

the

glider

observations

were

analyzed

from

the

MEOP

data

base

(Roquet

et

al.,

2013

).

We

use

nearly

20,000

seal

profiles

from

the

Bellingshausen

Sea

that

were

originally

analyzed

by

Zhang

et

al.

(

2016

)

and

later

by

Schubert

et

al.

(

2021

).

This

analysis

extends

the

Schubert

et

al.

(

2021

)

analysis

farther

to

the

west,

providing

a

first

look

at

circulation

patterns

over

Thurston

Plateau

and

into

the

Amundsen

Sea.

The

seals

are

equipped

with

Conductivity‐Temperature‐Depth

(CTD)

Oceanography

Satellite

Relay

Data

Log-

gers

with

an

in

situ

accuracy

for

pressure

better

than

5

db

(Boehme

et

al.,

2009

).

The

MEOP

data

set

is

subjected

to

temperature

and

salinity

calibrations,

and

calibrated

data

have

estimated

accuracy

of

better

than

±

0.005°C

for

temperature

and

±

0.02

for

salinity

(Boehme

et

al.,

2009

).

The

data

span

the

years

2007–2010

and

austral

summer

of

2013–2014

(Zhang

et

al.,

2016

).

In

Section

3.2

and

Section

3.3

we

present

the

seal

data

analysis,

including

water

mass

analysis

and

dynamic

height

anomalies,

the

latter

calculated

with

a

400

m

level

of

no

motion,

following

Schubert

et

al.

(

2021

).

The

MEOP

data

base

also

provides

insight

into

the

bathymetry

of

the

western

Bellingshausen

Sea,

as

inferred

from

differences

between

seal

dive

depths

and

the

International

Bathymetric

Chart

of

the

Southern

Ocean

(IBCSO)

(Arndt

et

al.,

2013

).

Because

seal

dives

often

do

not

reach

the

seafloor,

we

focus

on

those

locations

where

the

seal

depth

exceeds

the

IBCSO

bathymetry.

Major

uncertainties

in

water

column

depth

(

>

500

m)

occur

close

to

coast

and

near

ice

shelf

fronts

(Figure

1

).

The

main

reason

for

this

discrepancy

is

the

limited

ship

accessibility

to

coastal

regions,

often

covered

by

sea

ice

even

during

summer

(Padman

et

al.,

2010

),

and

the

associated

lack

of

multibeam

swath

bathymetry

observations.

According

to

the

seal

depth

data,

the

well‐known

troughs

in

the

Bellingshausen

Sea

(Belgica

and

Latady

troughs)

would

be

∼

10–100

m

deeper

than

in

IBCSO.

More

importantly,

differences

between

seal

dive

depths

and

the

IBCSO

bathymetry

point

to

an

overlooked

trough

west

of

Belgica

Trough,

at

∼

90°W

(hereafter

named

Seal

Trough

because

of

the

importance

of

MEOP

data

in

highlighting

this

feature;

Figure

1

),

that

would

provide

a

direct

path

from

the

easternmost

tip

of

Abbot

Ice

Shelf

to

the

shelf

break.

Here,

we

provide

evidence

of

the

role

of

this

trough

in

shaping

the

shelf

circulation

of

the

Bellingshausen

Sea.

Because

the

seal‐based

CTDs

use

ARGOS

(Advanced

Research

and

Global

Observation

Satellite)

telemetry

system,

the

error

in

position

is

±

5

km

(Roquet

et

al.,

2013

),

comparable

to

the

scale

of

the

deformation

radius

over

the

continental

shelf.

However,

the

high

density

of

the

seal

profiles

provides

statistical

confidence

in

the

updates

to

trough

locations

as

well

as

in

the

circulation

derived

from

the

dynamic

height

analysis.

This

circulation

should

be

viewed

as

a

climatology

over

the

period

covered

by

the

observations.

While

the

accuracy

of

the

seal

position

data

is

not

high,

it

is

adequate

to

resolve

a

feature

as

large

as

a

trough

(Seal

Trough

characteristic

width

is

35–70

km),

and

we

have

confidence

in

the

trough

location

because

multiple

seal

profiles

are

located

within

this

trough.

We

acknowledge

that

seal

data

alone

do

not

allow

exact

inference

of

seabed

gradients,

as

seals

do

not

always

dive

to

the

seafloor

(though

typically

they

forage

at

the

sea

bed).

However,

these

data

allow

identification

of

regions

where

IBCSO

is

too

shallow.

The

analysis

presented

here

suggests

that

Seal

Trough

is

deeper

than

shown

by

IBCSO,

and

we

discuss

its

relevance

in

the

shelf

circulation

of

the

Belling-

shausen

Sea.

2.3.

Meltwater

Fractions

Meltwater

fractions

were

calculated

following

the

Optimum

Multiparameter

(OMP)

water

mass

analysis

of

Tomczak

and

Large

(

1989

),

as

described

by

Biddle

et

al.

(

2017

)

for

the

Antarctic

margins.

The

OMP

analysis

optimizes

the

use

of

a

set

of

hydrographic

variables

by

solving

an

over‐determined

linear

set

of

mixing

equations.

The

method

requires

representation

of

water

masses

by

specific

water

types,

finds

the

correct

weights

for

the

hydrographic

variables

(weight

functions

reflect

the

quality

of

each

oceanographic

parameter),

and

solves

the

mixing

equations

by

minimization

of

the

residuals.

We

applied

the

OMP

analysis

to

temperature,

salinity

and

dissolved

oxygen

observations

obtained

from

the

glider.

We

chose

the

water

types

(or

end

members)

of

WW

and

MCDW

using

property‐property

diagrams

and

used

canonical

values

for

glacial

meltwater

(Biddle

et

al.,

2017

;

Schulze

Chretien

et

al.,

2021

;

Sheehan

et

al.,

2023

).

Water

types

and

weight

functions

are

detailed

in

(Text

S1

and

Figure

S1

in

Supporting

Informa-

tion

S1

).

We

explored

the

sensitivity

of

meltwater

to

the

selection

of

WW

and

MCDW

end

members

by

running

the

OMP

analysis

for

the

entire

range

of

MCDW

and

WW

values

found

in

the

glider

observations

and

choosing

the

most

restrictive

solution

(i.e.,

the

one

with

the

lowest

meltwater

content).

The

meltwater

fractions

obtained

Journal

of

Geophysical

Research:

Oceans

10.1029/2023JC020080

FLEXAS

ET

AL.

5

of

18

21699291, 2024, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2023JC020080 by California Inst of Technology, Wiley Online Library on [14/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

with

other

end

members

were

as

high

as

double

the

values

presented

here.

However,

the

distribution

of

meltwater

fractions

was

qualitatively

consistent

throughout

the

different

analyses

(see

Supporting

Information

S1

).

To

trace

the

pathway

of

meltwater

fractions

from

the

ice

shelves

to

the

continental

shelf

break

we

took

advantage

of

instrumented

seal

observations.

Given

that

seal

observations

only

provide

temperature

and

salinity

data,

it

is

important

to

understand

the

role

of

dissolved

oxygen

in

our

estimates

of

meltwater

fractions.

For

this

reason,

we

performed

an

OMP

analysis

only

using

temperature

and

salinity

data

from

the

gliders.

Meltwater

fractions

from

glider

observations

without

dissolved

oxygen

are

similar

in

magnitude

to

estimates

from

glider

observations

using

dissolved

oxygen

(Figure

S2

in

Supporting

Information

S1

).

They

are

also,

importantly,

located

within

similar

density

ranges.

However,

there

are

substantial

differences

in

the

horizontal

locations

of

the

maxima

of

meltwater

fractions.

For

this

reason,

we

limit

our

interpretation

of

meltwater

estimates

from

seals

to

a

qualitative

discussion

of

the

distribution

of

meltwater

properties

on

the

Bellingshausen

Sea

continental

shelf.

3.

Results

3.1.

Shelf

Break

Hydrographic

Properties

The

cross‐slope

glider

sections

span

from

slightly

south

of

the

continental

shelf

break

to

slightly

north

of

the

southern

boundary

(Bdy;

Orsi

et

al.,

1995

)

of

the

Antarctic

Circumpolar

Current

(ACC).

The

Bdy

is

defined

by

the

southernmost

extension

of

Upper

Circumpolar

Deep

Water

(

θ

>

1.5°C,

S

>

34.5)

and

nearly

coincides

with

the

southernmost

eastward

jet

associated

with

the

ACC

(Orsi

et

al.,

1995

).

In

our

data

set,

the

maximum

potential

temperature

of

the

Bdy

is

1.85°C

at

300–400

m

depth;

the

Southern

ACC

front

has

a

temperature

of

2.0°C

at

about

400

m

depth

(Figure

3

).

In

all

the

glider

cross‐slope

sections,

the

Bdy

is

found

over

the

2,000

m

isobath

(Figures

1

and

3a

).

Moving

westward

from

86°W

to

90°W,

the

Bdy

(and

its

associated

eastward

jet;

Figure

3b

)

is

found

progressively

closer

to

the

continental

shelf

as

the

slope

steepens.

The

Bdy

reaches

its

closest

proximity

to

the

shelf

break

at

92°W,

and

it

separates

offshore

again

at

96°W.

The

geostrophic

velocity

field

is

nearly

independent

of

depth

in

the

upper

1,000

m.

South

of

the

Bdy,

the

flow

is

mostly

westward,

with

occasional

flow

reversals

presumably

related

to

eddies

shed

from

the

continental

shelf

to

the

open

ocean.

At

92°W,

the

proximity

of

the

Bdy

to

the

shelf

break

causes

a

reversal

of

the

westward

velocity

field

at

depth

(below

∼

250

m

at

the

shelf

break;

Figure

3b

).

The

meridional

tilting

of

isopycnals

at

the

pycnocline

(27.7

>

γ

n

>

27.9

kg

m

3

)

is

enhanced,

and

isopycnals

intersect

the

seafloor

(Figure

3b

),

showing

the

typical

density

structure

of

the

ASF.

The

Bdy

sets

the

limit

of

the

northward

extension

of

WW.

We

use

the

1.5°C

isotherm

to

estimate

the

thickness

of

the

WW

layer.

The

WW

layer

is

thicker

(150–170

m

thick)

south

of

the

Bdy.

At

92°W

the

Bdy

is

closest

to

the

shelf

break,

and

the

thick

WW

layer

over

the

shelf

provides

the

characteristic

water

mass

configuration

of

the

ASF

in

the

region

(Thompson

et

al.,

2020

).

Meltwater

(MW)

is

found

at

the

main

pycnocline

(Figure

3c

).

Noticeably,

in

all

sections,

the

location

of

the

Bdy

coincides

with

the

northward

extension

of

large,

thick

MW

pools.

Meltwater

fractions

are

largest

(

>

0.5%)

at

90°W

and

92°W.

The

ASF/ASC

system

plays

a

main

role

in

transporting

meltwater

along

the

slope

(Heywood

et

al.,

1998

),

while

filaments

and

eddies

likely

shed

meltwater

toward

the

Bdy

(Sheehan

et

al.,

2023

).

The

glider

sections

taken

along

the

shelf

break

span,

roughly,

10

degrees

of

longitude

(from

86°W

to

96°W).

They

sample

through

four

sills,

one

at

the

western

side

of

each

along‐shelf‐break

section

(Figure

4

).

Sections

I

and

II

are

located

in

Belgica

Trough

(Figure

1

).

Here,

WW

is

relatively

thick

(100

m),

with

its

core

at

∼

100

m.

Near

the

bottom,

the

intrusion

of

warm,

MCDW

is

observed

in

salinity

(

>

34.7)

and

dissolved

oxygen

(

<

140

μmol

kg

m

3

).

At

the

pycnocline

we

observe

two

meltwater

cores

located

close

to

each

sill

(MW

fraction

>

0.5%;

Figure

4d

).

These

two

meltwater

cores

contain

relatively

low

backscatter

(10

3.8

m

1

;

Figure

4e

)

when

compared

with

all

the

other

waters

sampled

along

the

shelf

break

by

the

glider.

Section

III

is

taken

over

a

trough

that

connects

the

eastern

tip

of

Abbot

Ice

Shelf

to

the

continental

shelf

break

(Figure

1

).

Section

IV

is

taken

at

the

shelf

break

off

Thurston

Plateau.

Along

these

two

sections,

between

91.5°W

and

93.5°W,

the

WW

layer

thickens

(150–170

m

thick;

Figure

4a

).

At

the

pycnocline,

we

find

a

large

meltwater

core

(MW

fraction

>

0.5%;

Figure

4d

)

with

high

backscatter

properties

(10

3.7

m

1

;

Figure

4e

),

which

are

larger

than

those

found

in

sections

I

and

II

(Figures

4d

and

4e

).

To

quantify

the

relation

between

meltwater

and

optical

backscatter

(Figures

5a,

5b

,

and

5f

)

we

calculated

the

linear

regression

of

these

two

variables

for

meltwater

values

higher

than

0.3%.

We

found

that

a

linear

regression

explained

only

5%

of

the

variance

in

Section

I,

but

the

relation

Journal

of

Geophysical

Research:

Oceans

10.1029/2023JC020080

FLEXAS

ET

AL.

6

of

18

21699291, 2024, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2023JC020080 by California Inst of Technology, Wiley Online Library on [14/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

Figure

3.

Journal

of

Geophysical

Research:

Oceans

10.1029/2023JC020080

FLEXAS

ET

AL.

7

of

18

21699291, 2024, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2023JC020080 by California Inst of Technology, Wiley Online Library on [14/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

increased

to

30%

in

Section

II,

36%

in

Section

III

and

48%

in

Section

IV.

A

quadratic

fit

or

cubic

fit

led

to

comparable

results.

The

MW

core

in

Sections

III

and

IV

is

also

deeper

by

about

50

m

(Figures

5d–5h

),

lies

at

slightly

greater

density

levels

(about

0.02

kg

m

3

;

Figures

5c–5g

)

and

has

slightly

smaller

dissolved

oxygen

values

(by

about

5

μmol

kg

m

3

;

Figures

5e–5i

),

compared

with

the

MW

core

found

in

Sections

I

and

II.

These

differences

between

MW

cores

were

statistically

tested.

For

instance,

a

two‐sample

t

‐test

gives

a

mean

density

of

27.87

kg

m

3

in

Sections

I

and

II

and

a

mean

density

27.89

kg

m

3

in

Sections

III

and

IV.

The

difference

between

the

two

means

is

statistically

significant

(

p

=

10

26

and

α

=

0.05,

where

p

is

the

probability

of

the

null

hypothesis

at

the

α

*100%

significance

level).

3.2.

Circulation

Pathways

Toward

the

Shelf

Break

Dynamic

height

calculated

from

seal

observations

shows

a

positive

anomaly

(of

0.8–1.2

m

2

s

2

)

along

the

coast

throughout

the

Bellingshausen

Sea

(Figure

6a

).

The

associated

along‐coast

westward

geostrophic

flow,

corre-

sponding

to

the

AACC,

has

a

variable

width

of

50–150

km.

These

values

are

larger

than

observations

of

the

AACC

in

the

WAP,

where

the

AACC

is

observed

as

a

7–20

km

wide

current

(Moffat

et

al.,

2008

).

Such

dif-

ferences

are

likely

due

to

the

nature

of

the

seal

data

(sparse

dives

used

as

a

climatology,

rather

than

a

snapshot

hydrographic

section).

The

AACC

is

correlated

with

fresher

water

near

the

coast

(Figure

6b

)

and

tilting

isopycnals

that

deepen

toward

the

coast

(Figure

6c

).

At

about

90°W,

the

AACC

splits

into

two

branches.

The

main

part

of

the

flow

is

directed

toward

the

shelf

break,

following

the

480‐m

isobath.

A

secondary

branch

continues

west

along

the

coast,

toward

Thurston

Plateau

and

into

the

Amundsen

Sea.

The

dynamic

height

also

indicates

a

cyclonic

recirculation

within

Belgica

Trough,

in

agreement

with

Schulze

Cretien

et

al.

(

2021

)

and

Sheehan

et

al.

(

2023

).

Seal‐based

temperature‐salinity

(T/S)

diagrams,

separated

into

different

regions

of

the

western

Bellingshausen

Sea,

provide

a

complementary

view

of

the

circulation

features

observed

in

dynamic

height.

A

total

of

eight

T/S

clusters

of

seal

profile

data

are

selected

by

location:

three

in

Belgica

Trough,

one

inside

Seal

Trough,

one

in

front

of

Venable

Ice

Shelf,

and

three

off

Abbot

Ice

Shelf

(Figure

7

).

We

color

the

profiles

by

location

(Figure

7d

),

and

manually

group

the

different

profiles

when

they

display

similar

surface‐to‐bottom

temperature

and

salinity

properties

in

temperature‐salinity

space.

This

grouping

by

T/S

properties

captures

both

spatial

differences

and

seasonal

variability

in

the

water

formation

processes.

Water

properties

in

front

of

Venable

Ice

Shelf

(in

yellow)

show

two

different

“families”

with

distinct

T/S

properties

(in

yellow;

Figures

7a

and

7b

).

The

first

family

of

profiles

(Figure

7a

)

contains

WW

with

salinity

around

34.00

and

temperatures

below

1.0°C

and

relatively

eroded

MCDW

properties

(34.65

in

salinity

and

1.0–

1.3°C).

These

profiles

correspond

to

MEOP

data

from

the

months

of

March

and

April,

with

temperature

values

in

AASW

below

0.7°C.

Profiles

east

Abbot

Ice

Shelf

(in

green)

and

in

Seal

Trough

(in

orange)

show

similar

T/S

properties

to

this

family

(Figure

7a

).

We

did

not

separate

the

data

by

years.

The

second

T/S

family

of

profiles

off

Venable

Ice

Shelf

(in

yellow;

Figure

7b

)

contains

WW

with

salinity

of

34.25

and

MCDW

properties

with

salinity

of

34.65

and

temperatures

between

1.0

and

1.2°C.

Profiles

belonging

to

this

T/S

family

correspond

to

MEOP

data

from

the

months

of

December,

January

and

February,

with

temperature

values

in

AASW

between

1.0°C

and

0°C.

They

include

seal

dives

off

Venable

Ice

Shelf

(in

yellow)

and

dives

over

the

western

side

of

Belgica

Trough

(in

red)

(Figure

7b

).

We

note

that

the

temperature

maxima

of

MCDW

in

front

of

Venable

Ice

Shelf

(in

yellow)

is

at

the

seafloor,

while

at

Belgica

Trough

(in

red)

is

at

intermediate

depth

(

∼

400

m).

Last,

we

look

at

profiles

collected

in

different

parts

of

Belgica

Trough

(Figure

7c

).

These

profiles

feature

WW

temperatures

below

1.2°C

(including

near

the

freezing

point)

and

salinity

ranging

between

33.50

and

34.15,

and

Figure

3.

Glider

cross‐slope

sections

along

86°W,

89°W,

90°W,

92°W

and

96°W

showing

(a)

potential

temperature,

θ

,

(b)

geostrophic

velocity

referenced

to

depth‐

average

currents,

ug

,

and

(c)

meltwater

fraction,

MW

(in

%).

In

panel

(a),

the

1.5

and

1.7

isotherms

are

marked

in

white;

the

1.85°C

isotherm

and

the

2.0°C

isotherm

are

marked

in

red.

The

location

of

2,000

m

isobaths

is

marked

with

a

gray

vertical

dashed

line.

We

use

the

1.5°C

isotherm

to

estimate

the

thickness

of

the

WW

layer.

We

use

the

southernmost

location

of

the

1.85°C

isotherm

to

track

the

location

of

the

Bdy,

and

the

southernmost

location

of

the

2.0°C

isotherm

to

track

the

location

of

the

Southern

Antarctic

Circumpolar

Current

(ACC)

front.

For

clarity,

colored

triangles

along

the

top

of

the

panels

mark

the

position

of

the

Bdy

(red),

the

Southern

ACC

front

(black)

and

the

Antarctic

Slope

Front

(blue).

In

panel

(c),

a

black

vertical

dashed

line

indicates

uncertainty

in

offshore

meltwater

estimates

because

of

our

choice

of

WW

and

MCDW

end

members.

In

all

sections,

the

x

‐axis

shows

distance

from

the

glider

dive

closest

to

coast.

For

section

locations,

see

Figure

1

.

Journal

of

Geophysical

Research:

Oceans

10.1029/2023JC020080

FLEXAS

ET

AL.

8

of

18

21699291, 2024, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2023JC020080 by California Inst of Technology, Wiley Online Library on [14/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

Figure

4.

Glider

sections

along

the

shelf

break

(Sections

I

to

IV)

showing

(a)

potential

temperature

(°C),

(b)

salinity,

(c)

dissolved

oxygen,

DO

(

μ

mol

kg

m

3

),

(d)

meltwater

fraction

(%),

and

(e)

optical

backscatter

(m

1

,

in

log

10

).

In

panel

(a),

white

bold

contours

of

1.5°C

and

1.7°C

are

used

to

visualize

the

WW

layer.

In

panel

(b),

red

salinity

contours

of

34.7

are

used

to

identify

MCDW

intrusions

onto

the

continental

shelf.

In

all

panels,

black

contours

of

neutral

density

27.7

>

γ

n

>

27.9

kg

m

3

are

used

to

highlight

the

main

pycnocline.

For

section

locations,

see

Figure

1

.

Journal

of

Geophysical

Research:

Oceans

10.1029/2023JC020080

FLEXAS

ET

AL.

9

of

18

21699291, 2024, 6, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2023JC020080 by California Inst of Technology, Wiley Online Library on [14/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License