Environ. Res. Lett. 11 (2016) 094013 doi:10.1088/1748-9326/11/9/094013

LETTER

Projected changes to South Atlantic boundary currents andconfluence region in the CMIP5models: the role of wind and deepocean changes

GMPontes1, A SenGupta2 andASTaschetto2

1 School of Oceanography, StateUniversity of Rio de Janeiro (UERJ), Rio de Janeiro, Brazil2 Climate Change ResearchCentre andAustralian ResearchCouncil (ARC)Centre of Excellence for Climate System Science, The

University ofNew SouthWales, Australia

E-mail: gabriel.pontes@uerj.br

Keywords: South Atlantic ocean, climate change, Climate Model Intercomparisson Project-CMIP, Sverdrup’s dynamics, Brazil current,Brazil–Malvinas confluence, watermasses

Supplementarymaterial for this article is available online

AbstractThe SouthAtlantic (SA) circulation plays an important role in the oceanic teleconnections from theIndian, Pacific and Southern oceans to theNorthAtlantic, with inter-hemispheric exchanges of heatand salt. Here, we show that the large-scale features of the SA circulation are projected to changesignificantly under ‘business as usual’ greenhouse gas increases. Based on 19models from theCoupledModel Intercomparison Project phase 5 there is a projectedweakening in the upper ocean interiortransport (<1000m) between 15° and∼32°S, largely related to aweakening of thewind stress curlover this region. The reduction in ocean interior circulation is largely compensated by a decrease inthe net deep southward ocean transport (>1000m), mainly related to a decrease in theNorthAtlanticdeepwater transport. Between 30° and 40°S, there is a consistent projected intensification in the Brazilcurrent strength of about 40% (30%–58% interquartile range) primarily compensated by anintensification of the upper interior circulation across the Indo-Atlantic basin. The Brazil–Malvinasconfluence is projected to shift southwards, driven by aweakening of theMalvinas current. Such achange could have important implications for the distribution ofmarine species in the southwesternSA in the future.

1. Introduction

The southwestern South Atlantic (SA) is one of theworld’s most energetic oceanic regions. This is a resultof the meeting of two intense opposing flows: thepoleward Brazil current (BC) and the equatorwardMalvinas current (MC), to form the most importantoceanographic front in the SA: the Brazil–Malvinasconfluence (BMC; figure S1).

The BMC strongly influences the local climate andbiological environment. The large sea surface temper-ature (SST) gradient affects local surface winds andheat advection to adjacent coastal areas (De Camargoet al 2013). Frontal zones are characterized by high pri-mary production that enhances the local food chain(Alemany et al 2014).

This study will examine projected changes anddriving factors for the SA circulation. The large-scaleocean circulation can, to some extent, be broken downinto a surface wind-driven circulation and a deep den-sity-driven overturning circulation. The surface circu-lation can reach 1000–2000 m depending on theregion. Of particular importance are the westernboundary currents (WBCs) that transport largeamounts of salt, heat, and other tracers poleward. Thebasin interior surface flow usually opposes the WBC,partially compensating its poleward mass flux. Theseflows are also affected by air-sea fluxes, inter-basintransport and vertical transport from below associatedwith themeridional overturning circulation.

In the SA, the BC extends between about 20 and40°S (figure S1) with considerable transport variation

OPEN ACCESS

RECEIVED

27April 2016

REVISED

10August 2016

ACCEPTED FOR PUBLICATION

19August 2016

PUBLISHED

14 September 2016

Original content from thisworkmay be used underthe terms of the CreativeCommonsAttribution 3.0licence.

Any further distribution ofthis workmustmaintainattribution to theauthor(s) and the title ofthework, journal citationandDOI.

© 2016 IOPPublishing Ltd

along its length. Generally, its transport north of 25°Sis less than 10 Sv (Peterson and Stramma 1991)increasing substantially south of 30°S. Depth inte-grated transport to 1400–1500 m around 38°S mightreach 19 Sv based on geostrophic estimates (Gordonand Greengrove 1986). Table S1 summarizes previoustransport estimates.

The BMC is situated at around 38°S and showsseasonal variability with meridional shifts of about 5°(Wainer et al 2000, Goni et al 2011). A number ofmechanisms have been proposed to explain the seaso-nal BMC movements, including variations in BCtransport (Olson et al 1988), MC transport(Matano 1993), Antarctic Circumpolar current (ACC)transport (Gan et al 1998), and seasonal wind stresschanges (Fetter and Matano 2008). There is also evi-dence of longer-term change in the confluence loca-tionwith both observational data (Goni et al 2011) andnumerical simulations (Combes and Matano 2014)indicating a southward shift at a rate of between 0.39°and 0.81° per decade over the last two decades. Theprimary drivers of this change remain a topic ofresearch.

Many studies have used coupled atmosphere-ocean general circulation models (AOGCM) to inves-tigate changes in oceanic and atmospheric circulationin Global Warming scenarios. One robust change evi-dent in both observations (Hill et al 2008, Wuet al 2012) and climate simulations (Wainer et al 2004,Sen Gupta et al 2009, 2016) is an intensification andpoleward shift of portions of the subtropical gyresassociated with an intensification of theWBCs or theirextensions. This has been related to long-term changesin the wind stress curl (WSC), associated with changesin ozone and CO2 forcing (e.g. Saenko et al 2005, Pol-vani et al 2011). Robust changes in the large-scaleatmosphere circulation are also projected by climatemodels including a widening and slowing down of theHadley cell and poleward shift of the Southern Hemi-spherewesterlies (Lu et al 2007,Harvey et al 2014).

The aim of this study is to examine the projectedchanges to SA ocean circulation by analyzing outputfrom the Coupled Model Intercomparison Projectphase 5 (CMIP5) models (Taylor et al 2012). In part-icular, we examine changes in the SA basin circulationin relation to projected changes in surface winds andthe large-scale overturning circulation and their con-sequences for theWBCs andBMC.

2. Climatemodels andmethods

CMIP5 models include state-of-the-art AOGCM andEarth System Models that were used to inform theIntergovernmental Panel on Climate Change FifthAssessment Report. Typically, ocean horizontal reso-lution is approximately 1.0°×1.0° although there isconsiderable inter-model differences (table S2). As aresult, thesemodels do not explicitly resolvemesoscale

processes, although the effects of these processes areincluded through parameterisation. The vertical oceanresolution varies between 31 and 70 layers.

We used output from 19CMIP5models (table S2),which have the available variables and scenarios for thecurrent study. To examine projected changes we havemade comparisons between the historical simulationsthat covers 1850–2005 and a high emissions scenario(RCP8.5) that extends the historical simulation to2100. Historical simulations are forced by observa-tionally derived data of the major climate forcings(greenhouse gases, anthropogenic and volcanic aero-sols, ozone and solar irradiance). In the RCP8.5 sce-nario greenhouse gas forcing increases throughout the21st century reaching an additional radiative forcingof about 8.5Wm−2, equivalent to∼1370CO2-equiva-lent at the end of the century (Taylor et al 2012). A sin-gle ensemblemember is used from eachmodel (i.e. thefirst member with all available variables; usuallyr1i1p1). Variables from both ocean and atmosphericcomponents were used: mass transport (umo andvmo) and wind stress (tauu and tauv). When masstransport variables were not available, we calculatedmass transport from velocity fields (uo and vo).Despite long spin-up integrations some model driftmay remain in the subsurface fields, however this islikely small compared to forced trends, particularlywhen considering ensemble averages (Sen Guptaet al 2016) and so has been neglected.

To estimate the state of the ocean during the 20thcentury the data was averaged over the period1900–2000 from the historical simulations. In order toexamine the projected changes, the averaged2050–2100 period from the RCP8.5 simulations wascomparedwith the 20th century state.

The calculation of the WBC transports is proble-matic as there are differences in their representationacross models due to different bathymetries, coast-lines, and resolutions. As a result we have manuallyselected the eastern boundary of the WBC for eachmodel as the longitude where meridional polewardflow and any associated weak offshore counter flowvanishes in the long termmean velocities (see examplein figure S2; following Sen Gupta et al 2016). Thecounter current is included as we are interested inidentifying the western limit of the interior flow (thatextends across the remainder of the basin) that can berelated to Sverdrup transport. In most of the modelsthe BC does not extend deeper than 1000 m, as a resultwe use this depth to define upper ocean transports.This is consistent with observations reported by Gor-don and Greengrove (1986) who showed that most ofthe transport occurs in the upper 800 m.

The statistical significance of projected changeshas been determined by applying the Robust RankTest (Fligner and Policello 1981) to examine if two dis-tributions have equal medians and are indicated intable S5. Unless otherwise stated, all values presented

2

Environ. Res. Lett. 11 (2016) 094013

are multi-model medians (MMM) and associatedinterquartile range (IQR).

3. Results

3.1. Basin interior andWSCThe upper-ocean interior circulation in the SA isestimated as the meridional volume transport shal-lower than 1000 m integrated from the eastern edge ofWBC to the African coast (or east to the next landboundary south of Africa). The zonally integrated 20thcentury MMM equatorward interior transportdecreases almost linearly towards the north from 27.5Sv (23.7 to 29.1 Sv) at 30°S to zero at about 16°S(figure 1(a); individual model tranports, inter-quartileranges and MMM are collated in table S5 for allresults). North of this, the interior flow is weaklysouthwards.

Between 17 and∼30°S, all but onemodel project adecrease in the basin interior transport strength. TheMMM decrease is statistically significant at the 95%level between 13°S and 32°S (figure 1(b)). The changein the interior transport is largest at 23°S (21°–25°S).In this region, there is a mean projected weakening intransport of ∼3 Sv (2.4–3.4 Sv) corresponding to a22% decrease (16%–24%). In this same region(between 21° and 25°S), there is no significant changein the BC transport (figures 1(b) and 2(a); seesection 3.3). Thus, the MMM weakening in the oceaninterior transport is not compensated by an associatedweakening of theWBC.

To help understand the cause of the circulationchange we examine the zonally averaged WSC inte-grated over the SA basin (or east to the next landboundary south of Africa; figures 1(c) and (d)). TheWSC can be related via Sverdrup theory to the basininterior transport (positive/negative WSC implies anet northward/southwards interior transport). As theCMIP5 models have coarse resolution, we have line-arly interpolated the zonally averaged WSC to 0.1°resolution and estimated the latitude of the maximumWSC by calculating the WSC weighted mean latitudebased on latitudes whereWSCwas greater than 50%ofits maximum. The WSC maximum is located around38.4°S in themodels (36.4°–39.4°S; figure 1(c)). Com-parison with observations (ERA-Interim reanalysisfrom 1979 to 2013) shows that the latitude of theMMM WSC maximum is biased polewards by ∼2.9°(figure 1(c)). This is despite the fact that further souththere is a distinct equatorwards bias in the maximumzonal winds, consistent with previous studies (i.e.Barnes and Polvani 2013, Bracegirdle et al 2013).

All of the models project a southward shift in bothmaximum WSC and WSC zero line. The polewardshift in the position of the maximum WSC is 1.4°(1.2°–2.4°; figure 1(c)). The MMM WSC zero line islocated at about 51.2°S (50.2°–52.2°S) and is projectedto move southward by about 1.6° (1.1°–2.1°;

consistent with Cai et al 2005, Saenko et al 2005, Wuet al 2012). The latter identified a 2° southward shift inthe circumpolarly averaged WSC, associated with adoubling of CO2 in a single climatemodel.

Figure 1(b) reveals a strong correspondancebetween interior and Sverdrup transport changes(north of the southern tip of Africa), associated with aWSC weakening over this region (figure 1(d)). Astrong inter-model correlation (r =0.8, p<0.05)between Sverdrup transport at 25°S and the associatedmodel interior transport changes indicates that inter-model differences in the projected changes are wellexplained by differences in the projected surface windchanges (table 1).

3.2.Deep circulationThe 20th century deep meridional transport (inte-grated net transport below 1000 m) is almost constantalong the SA (figure 1(a)) with a mean transport of−15.9 Sv (−18.2 to−12.9 Sv). Given that there is a netprojected change in the upper ocean circulation (i.e.the BC and interior changes do not compensate eachother north of the southern tip of Africa) and theAtlantic Basin is closed to the north, there must be acompensating change in the deep ocean circulation.Indeed, all models project a weakening of the deeptransport in the 21st century. The net MMM SA deepcirculation decreases significantly by approximately3.4 Sv (∼2–5 Sv; figure 1(b)) corresponding to a 19%weakening (14%–28%).

We examine whether this weakening can be rela-ted to changes in the southward flowing North Atlan-tic deep water (NADW) and northward flowingAntarctic bottom water (AABW) entering the Atlanticbasin at 33°S. To quantify this we have definedNADWas the southward transport below 1000 m between theSouth American coast and 40°W (figure 2(b)). TheAABW transport is defined as any northward flux onthe western side of the basin (west of 20°W) and below3000 m (figure 2(b)). Based on these definitions, theNADW transport is −20.8 Sv (−22.8 to −19.3 Sv;figure S3) for the historical period. This value is similarto the estimated by Garzoli et al (2015) of −19.5 Sv(5.4) based on observations at 35°S. All but one model(NorESM1-M) suggest that NADW transport willdecrease by 2.4 Sv (1.7–3.5 Sv; table S5) correspondingto a 13% reduction (8%–14%).

The AABW transport for the 20th century is 5.5 Sv(3.8–8.5 Sv; figure S3). While there is a MMM pro-jected decrease of 0.6 Sv (−1.4 to 0.1 Sv), the change isnot statistically significant with 13 out of 19 modelsprojecting aweakening.

As such, theNADWplays the dominant role in thenet deep southward transport decrease and in balan-cing most of the mismatch between surface interiorand BC transport changes. It is interesting that mostmodels also show a net northward projected change atabout 10°Wand 25°W (figure 2(c); albeit considerably

3

Environ. Res. Lett. 11 (2016) 094013

weaker than the WBC change). This may relate to aweakening in parts of the NADW that have separatedaway the South American coastline (figure 2(b)) as aresult of interaction with Vitoria-Trindade Ridge at20°S (Garzoli et al 2015).

3.3.Western boundary currentsAll models simulate the BC between approximately20° and 40°S. The MMMmaximum transport (based

on each model’s maximum transport which occurs atdifferent latitudes) is –20.1 Sv (−23.0 to –13.2 Sv).Here, we have applied the same interpolation as forWSC (see section 3.1) for the integrated transportalong the western boundary. The maximum transportlatitude is 33.3°S (31.8–34.9°S, figure S4). Observa-tional estimates around this region liewithin themodelrange (19.2 Sv, Stramma 1989; 16 Sv, Lentini et al 2006,23 Sv,Garzoli 1993, seefigure S4 and table S1).

Figure 1. (a)Multi-modelmedian transport of BC, basin interior (below the southern tip of Africa it is integrated to the next landmass—Australia, NewZealand or SouthAmerica), wind stress-derived Sverdrup transport, and deep circulation (>1000 m) and the sumof BC, interior and deep transports. Averaged from1900 to 2000 (solid lines) and from 2050 to 2100 (dashed lines). (b)Multi-modelmedian transport change (RCP8.5minus historical), solid line indicates where the change is significant at the 95% level. (c)Multi-modelmedian zonally averagedwind stress curl, averaged from 1900 to 2000 (20th), and from2050 to 2100 (21st), and ERA-Interimaveraged from1979 to 2013.Horizontal lines indicate the associatedmaximumwind stress curl position calculated from theweightedaveragedmethod. (d)Zonally averaged (in between landmasses when south of the tip of Africa)wind stress curl change, solid lineindicates where the change is significant at the 95% level. Banding indicates interquartile range based on 19models (table S2).

4

Environ. Res. Lett. 11 (2016) 094013

There is a projected increase in the strength of theBC along its length (figures 1(a) and (b)). Between 20°and 30°S, the BC intensification is weak and not statis-tically significant, evenwith 14 of 19models projectingan increased transport. South of 30°S all models agree

on the intensification although the magnitude variesconsiderably. Averaged between 30° and 40°S theincrease ranges between 4.8 and 8.1 Sv (IQR) with aMMM increase of ∼6 Sv. This corresponds to a sub-stantial 40% (30%–58%) increase in mean transport.

Figure 2. (a)Multimodelmedian transport and projected changes at selected latitudes. Black arrows represent upper ocean(<=1000 m) transport, green arrows represent deep ocean (>1000 m) transport. Black numbers indicate 20th centurymediantransports (Sv). Interior transports are calculated across all basins for 45°S and 50°S, across Indo-Atlantic for 38°S and across theAtlantic for<=33°S. The length of the arrows ismerely schematic. (b)Multi-modelmeanmeridional velocity at 33°S (m s−1) and (c)associated projected change. Stippling on panel (c) indicates regions where there is 70% agreement on sign of change.

Table 1.Correlation coefficients betweenWestern boundary current (NBC, BCorMC) transport and compen-sating transports (columns 3–6) and between upper ocean interior transport and associated Sverdrup transport(column7). Bold coefficientsmeans a significant correlation at the 95% level (p<0.05). H: historical simula-tion.Δ: RCP8.5 projected change.

WBC correlation

Interior NADW >1000 m ITF Interior-Sverdrup correlation

10°SNBC H 0.7 0.86 0 — 0.7

Δ 0.1 0.95 0.5 — 0.74

25°S BC H 0.57 0.65 0.05 — 0.82

Δ 0.61 0.57 0.4 — 0.81

33°S BC H 0.78 0.19 0.58 — 0.52

Δ 0.81 0.34 0.1 — 0.66

38°S BC H 0.96 −0.69 0.81 0.25 0.34

Δ 0.95 0.12 0 −0.3 0.18

45°SMC H 0.95 — 0.44 — 0.15

Δ 0.95 — 0.1 — 0.1

50°SMC H 0.96 — 0.49 — 0.08

Δ 0.67 — 0.1 — 0.05

5

Environ. Res. Lett. 11 (2016) 094013

The projected change is greatest at around 39°S, pole-ward of the maximum mean state BC transport, withan intensification of 8.8 Sv (∼5–11 Sv; figure 1(b)). Inaddition, 18 out of 19 models show a southward shiftin the latitude of the maximum BC transport of ∼1.2°(0.7°–1.5°; figure S4).

At 33°S, where the historical BC transport is stron-gest, the MMM projected intensification in the BC islargely compensated by a decrease in the NADWtransport (figures 2(a) and (c)). However, even thoughthe upper interior MMM change is not significant,inter model differences in BC transport are stronglyrelated to differences in interior transport (r=0.81,p<0.05; table 1). At 38°S, close to where the BCintensification is strongest, the projected MMM BCintensification is associated with a complex set ofchanges to NADW, interior, other deep ocean trans-ports and Indonesian throughflow (ITF) (figure 2(a)).However, the inter-model differences in the BCchange is most strongly associated with changes in theupper interior transport across the Indo-Atlanticbasin (r=0.95, p<0.05, table 1).

The North Brazil current (NBC) is the northwardflowing branch of the South Equatorial current. In thehistorical simulations it shows an approximately lin-ear increase from ∼20°S to ∼11°S reaching up to 20.5Sv towards the north (17.5–26.7 Sv; figure 1(a)). Thisrange lies within observational estimates at 11°S of23–26 Sv (Hummels et al 2015; table S1). In the regionnorth of 15°S the 21st century simulation shows a sig-nificant (>95%) decrease on NBC strength(figure 1(b)). Themean transport for 15°S–20°S is∼19Sv (16–24.3 Sv; table S5). All but two models project adecrease in transport with aMMMof∼2.5 Sv (−4.7 to−1.4 Sv; table S5). At 10°S, the projected NBC

decrease is largely compensated by changes in theNADW (∼3 Sv; figure 2(a)) with a strong inter-modelcorrelation (r=0.95, p<0.05; table 1).

At the southern extent of the Atlantic basin thepowerful Malvinas current flows equatorwards fed bypart of the ACC thatmigrates northwards as a result oftopographic steering after transiting the Drake Passage(DP; figure S1). In the historical simulations the MChas a transport of 37.4 Sv (30–50.4 Sv) averagedbetween 50°S and 42°S. All but two models project adecrease in transport with a reduction of 3.7 Sv (6.7–1Sv). This corresponds to a weakening of 8% (−12% to−3%). At 45°S the MC is projected to decrease by ∼5Sv, largely associated with a weakening of the upperocean interior transport integrated across all basins(r=0.95, p<0.05;figure 2(a); table 1).

3.4. Brazil–Malvinas confluence (BMC)Next we quantify the projected changes to the BMCand examine drivers thatmay give rise to changes in itsposition. We define the BMC position as the latitudewhere the meridional transport vanishes along thewestern boundary of the Atlantic basin. In order tobetter determine this latitude, the integrated westernboundary meridional transport data was linearlyinterpolated to 0.1°. The simulated BMC is located at39.6°S (38.8°–41.6°S; figure 3(a); equivalent to thelatitude of zero transport in figure 1(a)—red solidline). By comparison the observed location is 38°S(from 1993 to 2008—Goni et al 2011, from 1992 to2007—Lumpkin and Garzoli 2011). As such the BMCis systematically simulated too far to the south. In thewarming scenario, allmodels project a southward shiftin the mean BMC position of 1.2° (0.7°–1.6°;figure 3(a)).

Figure 3. (a)Upper panel: confluence position for 20th century (black) and 21st century (red). Lower panel: confluence positionchange (RCP8.5minus historical). Horizontal solid lines indicatemulti-modelmedians and dashed lines indicate interquartile range.(b) Scatter plot betweenMC transport change (Sv) and confluence position change (°).

6

Environ. Res. Lett. 11 (2016) 094013

To help understand the cause of the BMC latitu-dinal change we examine the various drivers that havebeen proposed in the literature: basin wide WSC(Lumpkin and Garzoli 2011), BC transport (Olsonet al 1988), MC transport (Matano 1993, Combes andMatano 2014), and DP transport (Gan et al 1998). Toexamine the influence of theWSC, we look at the pro-jected change in the latitude of the WSC zero line.Sverdrup theory suggests an interior divergence andan associated zonal flow fed from the WBCs at thislatitude. To analyze the influence due to BC and MCtransport, the meridional transport at latitudes 5°north and south, respectively, of each model’s 20thcentury BMC position was related to the BMC posi-tion. For latitudes closer than this there may be alias-ing of the two current as they can both exist at the samelatitude (see schematic figure S1). The DP transport iscalculated at 68°W.

Changes in theWSC zero line (which shifts south-wards in all models), BC transport (which increases inall models) and MC transport (which decreases in 16out of 19 models) are all consistent with the robustsouthward shift in the confluence position. However,based on a multiple linear regression (MLR) analysisusing these potential drivers as predictors of meanBMC latitude (table S3) we find that the MC transportplays a primary role in explaining intermodel differ-ences in themean BMCposition (r=0.79, p<0.05),followed by BC transport (r=0.5, p<0.05). This isin agreement with Matano (1993) who suggested thatthe BMC position is mainly related to the MCstrength.

MLR is also used to examine inter-model differ-ences in projected BMC latitude changes (table S4).Only the MC transport change has a significant rela-tionship with the BMC position change (r=0.49,p<0.05; figure 3(b)). As such, intermodel differencesin the projected shift in BMC can be partly associatedwith differences in projected MC transport changes.Adding the other variables as predictors of BMCchange provides little additional benefit for predictinginter-model differences in the confluence positionchange (table S4). We would note that south of theBMC at 45°S there is no relationship between MCtransport and the strength of the DP transport(although there is a weak relationship further southr=0.44, p=0.05 at 50°S). As noted, there is a strongrelationship between projected MC change and thechange in interior transport across all basins(r=0.95; table 1; figure 2(a)), however the interiortransport changes are unrelated to WSC changes viaSverdrup dynamics (r<0.1).

4.Discussion and conclusions

Sverdrup theory relates WSC changes to changes inthe basin interior meridional transport, in the absenceof deep density driven circulation changes. Here we

find that a projected decrease in theWSC between 15°S and 30°S indeed explains the ∼20% reduction innorthward upper ocean interior transport. However,there is no significant reduction in theWBC transportat this region, and the reduced interior transport islargely compensated by a reduction in the deep(>1000 m) southward transport. This is consistentwith a weakening of NADW transport resulting fromhigh latitude North Atlantic warming and freshening(Dong et al 2014), and with the projected northwardanomalies in the abyssal circulation (figure 2(c)).

The southward shift in the WSC zero line and theweakening in the upper-ocean basin interior transportare likely to have important consequences to theworld’s ocean dynamics. The change in the WSC pat-tern drives a polewards shift of the frontal zonebetween the subtropical gyre circulation and the ACC,that leads an increased Agulhas leakage (Biastochet al 2009). However, this change in the Agulhas leak-age may not reflect an increased inter-hemisphericheat transport, as there is a projected decrease in bothbasin interior and North Brazil current volume trans-ports: the primary pathways of this heat exchange.

There is a consistent and substantial intensifica-tion of the BC of almost 40% between 30° and 40°S.North of the southern tip of Africa the relatively weakBC increase is largely compensated by the NADWreduction with inter-model differences mainly asso-ciated with large differences in upper ocean interiortransport related to changes in WSC. South of Africa,despite a decrease in the water entering the combinedIndo-Atlantic region via ITF (figure 2(a); Sen Guptaet al 2016) there is a very large increase in the BC. Thisincreased BC transport is primarily compensated byincreases in the upper ocean interior Indo-Atlantictransport. However, changes in the Indo-AtlanticWSC and associated Sverdrup transport cannotexplain these interior changes (intermodel correlationis not significant, table 1). This is unsurprising as thislatitude intersects the location of the Agulhas currentretroflection.

The location of the BMC is projected to shiftsouthward by ∼1.2°. The relationship between theACC andMC transport, which canmodulate the BMCposition, and surface winds is not straightforward(Fetter andMatano 2008, Lumpkin andGarzoli 2011),especially due to the existence of a complex topo-graphy in the Drake Passage region. Despite south-ward shifts in the WSC pattern, we found that inter-model differences in the projected BMC southwardshift location only showed a significant relationshipwith MC strength. This is consistent with findings ofCombes and Matano (2014), who reported anobserved southward shift of 0.62° decade−1 between1993 and 2008 and related this to a weakening of thenorthern branch of the ACC and a consequent reduc-tion of the MC transport. Moreover, results are con-sistent with Matano (1993) who used idealisedexperiments to show that the BMC is displaced

7

Environ. Res. Lett. 11 (2016) 094013

northwards of theWSC zero line by a strongMC, withBMC location modulated by the strength of the MCtransport.

A key uncertainty in this analysis lies on the coarseresolution of the CMIP5 models. The models areunable to simulate the mesoscale processes that areimportant in this highly energetic region. Despite this,the gross changes described here should be con-strained by basin wide wind changes. Indeed a similaranalysis for the East Australian current (Oliver andHolbrook 2014) found very little difference betweenCMIP5 projections and projections from an eddy per-mittingmodel with regard to changes in the long-termtransport of the WBCs. There is also considerableuncertainty with regards to the future evolution of thewinds over the Southern ocean. This is related to com-pensating effects of greenhouse gases and the ozoneforcing (e.g. Gillett and Fyfe 2013).

Our findings may have important consequencesfor regional climate and ecosystems. BC intensifica-tion can potentially increase heat and salt transportalong the BC pathway, modifying the SST gradientthat can affect regional weather in coastal areas (Pezziet al 2005, DeCamargo et al 2013). Additionally, Pradoet al (2016) observed a decrease in the occurrence oftemperate marine mammals between 30° and 34°Ssince 1970s, where we project a robust change in theBC. Thusly, the changes can potentially modify thedistribution of marine species and affect economicallylocal fishery production (e.g. Andrade 2003, Alemanyet al 2014).

Acknowledgments

For their roles in producing, coordinating, andmakingavailable CMIP5 model output, we acknowledge theclimatemodeling groups (table S2), theWorldClimateResearch Programme’s Working Group on CoupledModelling and the Global Organization for EarthSystem Science Portals. This work was supported bythe National Council for Research Development(Brazil-CNPq) and the Australian Research Council(ARC) including the ARC Centre of Excellence inClimate SystemScience, and theNCINational Facility,Canberra.

References

AlemanyD, Acha EM and IribarneOO 2014Marine fronts areimportant fi shing areas for demersal species at theArgentine Sea (Southwest Atlantic ocean) J. Sea Res. 8756–67

AndradeHA2003The relationship between the skipjack tuna(Katsuwonus pelamis)fishery and seasonal temperaturevariability in the South-Western Atlantic FisheriesOcanography 12 10–8

Barnes EA and Polvani L 2013Response of themidlatitude jets, andof their variability, to increased greenhouse gases in theCMIP5models J. Clim. 26 7117–35

BiastochA et al 2009 Increase inAgulhas leakage due to polewardshift of SouthernHemispherewesterliesNature 462 495–9

Bracegirdle T J et al 2013Assessment of surfacewinds over theAtlantic, Indian, and Pacific ocean sectors of the Southernocean inCMIP5models: historical bias, forcing response, andstate dependence J. Geophys. Res.: Atmos. 118 547–62

CaiW et al 2005The response of the Southern annularmode, theEast Australian current, and the southernmid-latitude oceancirculation to global warmingGeophys. Res. Lett. 32 1–4

CombesV andMatanoR2014Trends in the Brazil/Malvinasconfluence regionGeophys. Res. Lett. 41 799–804

DeCamargoR et al 2013Modulationmechanisms ofmarineatmospheric boundary layer at the Brazil–Malvinasconfluence region J. Geophys. Res.: Atmos. 118 6266–80

Dong S et al 2014 Seasonal variations in the SouthAtlanticmeridional overturning circulation fromobservations andnumericalmodelsGeophys. Res. Lett. 41 4611–8

Fetter A FH andMatanoRP 2008On the origins of the variability oftheMalvinas current in a global, eddy-permitting numericalsimulation J. Geophys. Res. 113C11018

FlignerMAand PolicelloGE 1981Robust rank procedures for thebehrens-fisher problemAm. Stat. Assoc. 76 162–8

Gan J,Mysak LA and StraubDN1998 Simulation of the SouthAtlantic ocean circulation and its seasonal variabilityJ. Geophys. Res. 103 241–51

Garzoli S L 1993Geostrophic velocity and transport variability inthe Brazil-Malvinas Confluence Deep-Sea Research I 401379–403

Garzoli S L et al 2015The fate of the deepwestern boundary currentin the SouthAtlanticDeep-Sea Res. I 103 125–36

Gillett NP and Fyfe J C 2013Annularmode changes in theCMIP5simulationsGeophys. Res. Lett. 40 1189–93

GoniG J, Bringas F andDinezio PN2011Observed low frequencyvariability of the Brazil current front J. Geophys. Res.: Oceans116C10037

GordonAL andGreengrove CL 1986Geostrophic circulation of theBrazil–Falkland confluenceDeep Sea Res.A 33 573–85

Harvey B J, Shaffrey LC andWoollings T J 2014 Equator-to-poletemperature differences and the extra-tropical storm trackresponses of the CMIP5 climatemodelsClim.Dyn. 43 1171

Hill K L, Rintoul S R, ColemanR andRidgwayKR 2008Windforced low frequency variability of the East Australia currentGeophys. Res. Lett. 35 L08602

Hummels R et al 2015 Interannual to decadal changes in the westernboundary circulation in theAtlantic at 11ºSGeophys. Res. Lett.42 7615–22

Lentini CAD,GoniG J andOlsonDB 2006 Investigation of Brazilcurrent rings in the confluence region J. Geophys. Res. 111C06013

Lu J, Vecchi GA andReichler T 2007 Expansion of theHadley cellunder global warmingGeophys. Res. Lett. 34 L06805

Lumpkin R andGarzoli S 2011 Interannual to decadal changes intheWestern SouthAtlantic ’ s surface circulation J. Geophys.Res. 116 1–10

MatanoRP 1993On the separation of the Brazil current from thecoast J. Phys. Oceanogr. 23 79–90

OliverOC J andHolbrookN J 2014 Extending our understandingof South Pacific gyre ‘spin-up’: modeling the East Australiancurrent in a future climate J. Geophys. Res.: Oceans 1192788–805

OlsonDB, PodestaGP, Evans RHandBrownOB1988Temporalvariations in the separation of Brazil andMalvinas currentsDeep Sea Res.A 35 1971–90

PetersonRG and StrammaL 1991Upper-level circulation in theSouthAtlantic ocean Prog. Oceanogr. 26 1–73

Pezzi L P, Souza RB,DouradoMS,Garcia CAE,MataMMandSilva-DiasMAF 2005Ocean-atmosphere in situobservations at the Brazil–Malvinas confluence regionGeophys. Res. Lett. 32 L22603

Polvani LM,WaughDW,CorreaG J and Son SW2011Stratospheric ozone depletion: themain driver of twentieth-century atmospheric circulation changes in the SouthernHemisphere J. Clim. 24 795–812

Prado JHF,Matos PH, Silva KG and Secchi ER 2016 Long-termseasonal and interannual patterns ofmarinemammal

8

Environ. Res. Lett. 11 (2016) 094013

strandings in subtropicalWestern SouthAtlanticPLosOne 11e0146339

SaenkoOA, Fyfe J C and EnglandMH2005On the response of theoceanic wind-driven circulation to atmospheric CO2 increaseClim.Dyn. 25 415–26

SenGupta A, Santoso A, Taschetto A,Ummenhorfer C,Travena J and EnglandM2009 Projected changes to theSouthernHemisphere ocean and sea ice in the IPCCAR4climatemodels J. Clim. 22 3047–78

SenGupta A, ShayneM,Van Sebille A, GanachaudA, Brown J andSantosoA 2016 Future changes to the Indonesianthroughflow and Pacific circulation : the differing role ofwind and deep circulation changesGeophys. Res. Lett. 431669–78

StrammaL 1989The Brazil Current transport south of 23ºSDeep-Sea Res. 36 639–46

Taylor K E, Stouffer R J andMeehl GA 2012 An overview of CMIP5and the experiment design Bull. Am.Meteorol. Soc. 93485–98

Wainer I, Gent P andGoni G 2000Annual cylce of the Brazil–Malvinas confluence region in theNational Center forAtmosphere ResearchClimate SystemModel J. Geophys. Res.105 26 167–77

Wainer I, TaschettoA,Otto-Bliesner B andBardy E 2004Anumerical study of the impact of greenhouse gases on theSouthAtlanticOcean climatology 66 163–89

WuL et al 2012 Enhancedwarming over the global subtropicalwestern boundary currentsNat. Clim. Change 2 161–6

9

Environ. Res. Lett. 11 (2016) 094013

Environmental Research Letters

Supplementary Information for

Projected changes to South Atlantic boundary currents and confluence region in the CMIP5 climate models: the role of wind and deep ocean changes

Gabriel Pontes1, Alex Sen Gupta2, Andréa S. Taschetto2

1School of Oceanography, State University of Rio de Janeiro (UERJ), Rio de Janeiro, Brazil

2Climate Change Research Centre and Australian Research Council (ARC) Centre of Excellence for Climate System Science, The University of New South Wales, Australia

Contents of this file

Figures S1 to S4 Tables S1 to S5

Introduction

This document provides: (1) schematic of the South Atlantic wind-driven circulation (Figure S1); (2) example of the methodology used for defining the western boundary currents (Figure S2); (3) a summary of previous works of BC and NBC transports (Table S1); (2) specifications of each numerical model used in this study (Table S2); (3) full results of the analysis that are described but not completely visualized in the manuscript (Table S3, Table S4, Table S5, Figure S3, Figure S4).

!

!

!

!

!

!

!

!

!Figure S1. Schematic of South Atlantic circulation showing major wind-driven currents (black arrows) and mean wind stress curl (Nm-3). NBC: North Brazil Current. EUC: Equator UnderCurrent. BC: Brazil Current. STG: Subtropical Gyre. BeC: Benguela Current. BMC: Brazil-Malvinas Confluence. SAC: South Atlantic Current. ACC: Antarctic Circumpolar Current. MC: Malvinas Current.

ACC

ACC

MC

STG

BeC

NBC

80oW 40oW 0o 40oE

BMC

20oN

0o

20oS

40oS

60oS

BC

EUC

−2 −1.5 −1 −0.5 0 0.5 1 1.5 2 x 10−7

SAC

!Figure S2. Example from two models showing the extent of the BC western boundary current (black dashed lines). This is selected manually (for each model) to include the main western boundary current core and any weak offshore counter flow. Colours indicate depth integrated (0-1000m) meridional transport: units m3s-1. Note: color bars are different for each panel.

!

!

!

!

!

!

!

!

!

!

!

!

60W 55W 50W 45W 40W 35W45S

40S

35S

30S

25S

20S

15SIPSL-CM5A-MR ×106

-8

-6

-4

-2

0

2

4

6

8

60W 55W 50W 45W 40W 35W45S

40S

35S

30S

25S

20S

15SGFDL-ESM2G ×106

-6

-4

-2

0

2

4

6

Figure S3. a) Upper panel: NADW transport for 20th century (black) and 21st century (red). Lower panel: NADW transport change (21st century minus 20th century). Horizontal solid lines indicate multi-model medians and dashed lines indicate interquartile range b) Same as ‘a’ but for AABW.

Figure S4. Multi-model median transport of the BC a) mean 1900-2000 (20th) and 2050-

2100 (21st), black dots indicate observational transport estimates (Table S1); horizontal

lines indicate the maximum transport latitude; black docs indicate estimates from preview

studies b) Multi-model median change (RCP8.5 minus historical), solid line indicates where

the change is significant at the 95% level. Shaded area indicates interquartile range based

on 19 models (Table S2).

Table S1. Transports of the NBC and BC estimated on previous works. Latitude (°S) Transport (Sv) Source Depth of

reference (m)

Reference

2.5°N 18±2 Pressure gauge

(PG) and Inverted

echo sounder

(IES)

300 (Garzoli et al.,

2004)

11°S 23±3 Hydrographic

data and current

meter (ADCP)

26.8 kg m-3

isopycnal

(Hummels et al.,

2015)

11°S 26±1.1 Current meter

(mooring)

26.8 kg m-3

isopycnal

(Hummels et al.,

2015)

20°S -4 Hydrographic

data

500 (Stramma et al.,

1990)

25°S -5.5 Hydrographic

data

400 (Evans &

Signorini, 1985)

28°S -16.2±2.4 Current Meter 744 (Miiller et al.,

1998)

32/33°S -19.2/-12.2 Hydrographic

data

800 (Stramma, 1989)

36/38°S -23/-20 IES 1000 (Garzoli, 1993)

36-38°S -16±3 Altimetry - (Lentini et al.,

2006)

38°S -18 IES 800 (Garzoli &

Garraffo, 1989)

38°S -19 Hydrographic

data

1400 (Gordon &

Greengrove,

1986)

Table S2. Models’ specifications. Institute Ocean horizontal

resolution [°]

(enhanced

equatorial resol)

Ocean

vertical

resolution

(layers)

Atmosphere

horizontal

resolution [°]

ACCESS1-0 CSIRO-BOM,

Australia

1.0 x 1.0 (0.3) 50 1.9 x 1.2

ACCESS1-3 CSIRO-BOM,

Australia

1.0 x 1.0 (0.3) 50 1.9 x 1.2

CanESM2 CCCMA, Canada 1.4 x 0.9 (0.9) 40 2.8 x 2.8

CCSM4 NCAR, USA 1.1 x 0.6 (0.3) 60 1.2 x 0.9

CESM1-CAM5 NSF-DOE-

NCAR, USA

1.1 x 0.6 (0.3) 60 1.2 x 0.9

CNRM-CM5 CNRM-

CERFACS,

France

1.0 x 0.8 (0.3) 42 1.4 x 1.4

CSIRO-MK3-6-0 CSIRO-QCCCE,

Australia

1.9 x 0.9 (0.9) 31 1.9 x 1.9

GFDL-CM3 NOAA-GFDL,

USA

1.0 x 1.0 (0.4) 50 2.5 x 2.0

GFDL-EMS2G NOAA-GFDL,

USA

1.0 x 1.0 (0.4) 50 2.5 x 2.0

GFDL-EMS2M NOAA-GFDL,

USA

1.0 x 1.0 (0.4) 50 2.5 x 2.0

GISS-E2-R NASA/GISS, NY,

USA

2.0 x 2.0 (2.0) 32 2.5 x 2.0

HadGEM2-CC MOHC, UK 1.0 x 1.0 (0.4) 40 1.9 x 1.2

HadGEM2-ES MOHC, UK 1.0 x 1.0 (0.4) 40 1.9 x 1.2

IPSL-CM5A-LR IPSL, France 2.0 x 1.9 (0.6) 31 3.7 x 1.9

IPSL-CM5A-MR IPSL, France 1.6 x 1.4 (0.6) 31 2.5 x 1.3

MIROC-ESM JAMSTEC, Japan 1.4 x 0.9 (0.6) 44 2.8 x 2.8

MPI-ESM-MR MPI-N, Germany 0.4 x 0.4 (0.4) 40 1.9 x 1.9

MRI-CGCM3 MRI, Japan 1.0 x 0.5 (0.5) 51 1.1 x 1.1

NorESM1-M NCC, Norway 1.1 x 0.6 (0.3) 70 2.5 x 1.9

Table S3. Multiple linear regression of the Brazil-Malvinas confluence position. Predictors: BC is the Brazil Current transport 5° north of the model’s BMC position, MC is the Malvinas Current transport 5° south of the model’s BMC position, DP is the Drake Passage transport, and WSC is the position of the wind stress curl zero line. Columns 2 and 3 show the associated correlation coefficient and p-value.

Predictors* BMC*Position* p0value*

BC# 0.51# 0.03#

MC# 0.79# 5e-05#

DP# 0.28# 0.24#

WSC# 0.37# 0.12#

MC,#BC# 0.81# 2e-04#

MC,#DP# 0.80# 3e-04#

MC,#WSC# 0.81# 2e-04#

MC,#BC,#DP# 0.83# 5e-04#

MC,#BC,#WSC# 0.81# 7e-04#

MC,#BC,#DP,#WSC# 0.83# 2e-03#

Table S4. Same as Table S3 but for the change (c; RCP85 – historical) in BMC position and predictors.

Predictors* BMC*Position*(R)* p0value*

BCc# 0.28# 0.24#

MCc# 0.49# 0.03#

DPc# 0.14# 0.56#

WSCc# 0.24# 0.33#

MCc,#BCc# 0.51# 0.09#

MCc,#DPc# 0.49# 0.11#

MCc,#WSCc# 0.52# 0.08#

MCc,#WSCc,#BCc# 0.54# 0.08#

MCc,#WSCc,#DPc# 0.52# 0.18#

MCc,#WSCc,#BCc,#DPc# 0.54# 0.27#

Table S5. All values calculated from Model’s output in this study. T: transport in Sverdrups. H: historical simulation. Δ (change): RCP85 minus historical. BI:

Basin Interior. (-): negative transport (southward). Max: maximum. Lat: latitude. Multi-model medians change in bold mean that this change is significant

at least at the 90% level based on the Robust Rank Test.

AC

CE

SS1-

0

AC

ESS

1-3

Can

ESM

2

CC

SM4

CE

SM1-

CA

M5

CN

RM

-CM

5

CSI

RO

-MK

3-6-

0

GFD

L-C

M3

GFD

L-E

MS2

G

GFD

L-E

MS2

M

GIS

S-E

2-R

Had

GE

M2-

CC

Had

GE

M2-

ES

IPSL

-CM

5A-L

R

IPSL

-CM

5A-

MR

MIR

OC

-ESM

MPI

-ESM

-MR

MR

I-C

GC

M3

Nor

ESM

1-M

Qua

rtile

25

Qua

rtile

75

MM

M

Agr

eem

ent

BC max

T

H -19.7 -22.8 -27.7 -29.1 -23 -20.6 -20.1 -13 -31 -6.5 -8.6 -13.8 -17.1 -20.1 -20.5 -20.4 -29.6 -11.8 -6.2 -23 -13.2 -20.1 n/a

Δ -8.7 -5.5 -9.1 -8.8 -4 -3.9 -1.5 -10.6 -21.1 -7.6 -4.8 -6.2 -5.3 -1.9 -3 -8 -12.6 -3.7 -2.5 -8.8 -3.7 -5.5 19

BC max

T lat [°S]

H 35.7 35.1 36.6 37.4 33.3 33.2 31.6 34.7 36.9 30.5 32.6 33.3 33.7 30.3 31.6 31.6 33.5 32.1 32.4 35 31.8 33.3 n/a

Δ -1 -0.5 -1.1 -0.7 -0.4 -1.5 -1.3 -1.2 0 -2.4 -1.8 -1.3 -1.6 -1.4 -1.2 -2 -1.1 -0.2 -0.7 -1.5 -0.7 -1.2 18

BC T

30-40S

H -13.7 -17.2 -22.3 -23.3 -17.1 -11.7 -8.9 -11.5 -6.7 3.7 -7.2 -7.1 -9.3 -7.7 -10.2 -12.5 -22.8 -6.6 -3.2 -16.2 -7.1 -10.2 n/a

Δ -7.7 -6 -6.4 -5.9 -3.1 -5.6 -6.2 -8.3 -9.9 -11.8 -3.8 -5.6 -6.5 -4.6 -5.6 -9 -10.8 -2.6 -3.4 -8.1 -4.8 -5.9 19

NBC T

9-15S

H -3.6 19.3 23.3 21.5 13.8 19 17.6 30 27.2 26.5 20.4 16.9 16.1 12.9 15.8 18.2 24.7 15.3 30.2 15.9 24.3 19 n/a

Δ -3.6 -3.7 -4.5 -4.8 -2.4 -4.8 -1.3 -5.8 -5.7 -4.2 -2.1 -1.1 -1.3 0.3 -1.6 0.3 -1.5 -2.3 -4.7 -4.6 -1.4 -2.4 17

BI T 30°S 22.5 24.8 25.9 26.6 24.2 27.9 29.7 27.8 25.4 24.8 23.5 21.2 21 28 29.5 28.9 37 20.1 27.4 23.7 28 25.9 n/a

BI max

ch

T -4.6 -3.9 -2.2 -2.5 -2.7 -4 -2.7 -3.8 -3.5 -3 -3.8 -2.9 -2.8 -4.7 -4.8 -3.4 -4 -0.8 -3.5 -3.9 -2.8 -3.5 19

°S 25 25 19 23 23 29 23 9 29 17 25 23 27 21 25 21 27 19 25 21 25 23 n/a

Table S5. Continued.

AC

CE

SS1-

0

AC

ESS

1-3

Can

ESM

2

CC

SM4

CE

SM1-

CA

M5

CN

RM

-CM

5

CSI

RO

-MK

3-6-

0

GFD

L-C

M3

GFD

L-E

MS2

G

GFD

L-E

MS2

M

GIS

S-E

2-R

Had

GE

M2-

CC

Had

GE

M2-

ES

IPSL

-CM

5A-L

R

IPSL

-CM

5A-M

R

MIR

OC

-ESM

MPI

-ESM

-MR

MR

I-C

GC

M3

Nor

ESM

1-M

Qua

rtile

25

Qua

rtile

75

MM

M

Agr

eem

ent

BI T

21-25S

H 12.8 14.6 13.2 14.6 11.4 17.6 16.3 16.1 13.9 16.4 14.4 12.3 11.7 15.2 13.8 17.8 16.8 11.1 12.9 12.8 16.2 14.4 n/a

Δ -3.9 -3.2 -2.1 -2.3 -2.7 -2.9 -2.7 -3.5 -2 -2.2 -3.3 -2.8 -2.6 -4.4 -4.3 -3 -3.4 -0.1 -3.2 -3.4 -2.4 -2.8 19

BI T

21-25S

[%]

Δ 30 22 16 16 24 16 17 22 14 13 23 23 22 29 31 17 20 01 25 16 24 22 n/a

MC T

50-42S

H 43.1 43.2 34.8 30.4 25.9 51.2 61.8 28.3 17.3 31.1 27.4 50.9 48.9 37 37.4 67.5 54.4 29.9 42.9 30 50.4 37.4 n/a

Δ -4.6 -0.2 -10 -7 -0.5 -2.5 -6.8 -2.3 4 2.5 -2.1 -6.2 -5 -0.9 -3.7 -5.7 -10 -1.5 -8.2 -6.7 -1 -3.7 17

MC T

50-42S

[%]

Δ -11 0 -30 -23 -2 -5 -11 -8 23 8 -8 -12 -10 -2 10 -8 -19 -5 -19 -12 -3 -8 16

Max

WSC [°S]

H 39.7 39.5 39 41 40.2 37.2 38.7 37.3 36.5 36.5 36.4 38.7 38.1 34.7 35.8 34.3 36.5 39.9 38.4 36.4 39.4 38.4 n/a

Δ -1.1 -0.7 -1.8 -1.3 -1.2 -0.8 -1.3 -2.1 -2.9 -2.5 -1.4 -1.3 -1.2 -2.5 -3.3 -1.4 -2.9 -0.3 -1.6 -2.4 -1.2 -1.4 19

WSC zero

line [°S]

H 52.3 52.1 51.2 52.2 52 50.8 49.8 51.2 52.9 51.5 50.6 50.7 51.4 45.2 47.6 47.7 50.1 52.3 52.9 52.2 50.2 51.2 n/a

Δ -1.3 -0.8 -2.1 -2.6 -1.5 -0.7 -0.9 -2.2 -1.1 -1.7 -1.7 -2 -1.6 -4.4 -4.1 -1.7 -1.2 -0.3 -1.3 -2.1 -1.1 -1.6 19

Mean

Deep T

H(-) 16.1 17 18.3 18.2 10.6 17.7 13.3 19.3 18.9 19.2 15.5 15 14.1 8.7 11.8 11.9 15.9 12.7 24 18.2 12.9 15.9 n/a

Δ 4.8 4.6 4.2 5.2 3.4 5 2 5.9 6.2 5 1.8 2.2 1.9 0.7 2.1 1.4 3 1.8 4.2 2 5 3.4 19

Table S5. Continued.

AC

CE

SS1-

0

AC

ESS

1-3

Can

ESM

2

CC

SM4

CE

SM1-

CA

M5

CN

RM

-CM

5

CSI

RO

-MK

3-6-

0

GFD

L-C

M3

GFD

L-E

MS2

G

GFD

L-E

MS2

M

GIS

S-E

2-R

Had

GE

M2-

CC

Had

GE

M2-

ES

IPSL

-CM

5A-L

R

IPSL

-CM

5A-M

R

MIR

OC

-ESM

MPI

-ESM

-MR

MR

I-C

GC

M3

Nor

ESM

1-M

Qua

rtile

25

Qua

rtile

75

MM

M

Agr

eem

ent

Mean

Deep T

[%]

Δ -30 -27 -23 -29 -32 -28 -15 -31 -33 -26 -11 -15 -14 -8 -18 -12 -19 -14 -17 -14 -18 -19 19

NADW T H(-) 18.3 19.2 22.6 21.9 20.8 16.4 22.9 23.4 20.4 21.6 25.4 20.1 19.8 17.2 20 21.6 27.4 14.6 31.7 22.8 19.3 20.8 n/a

Δ 1.4 2.1 1.7 2.8 2.7 2.2 3.3 3.6 4.5 3.6 2.9 2.4 2.2 1.4 4.5 1.3 4.9 2 -0.1 1.8 3.5 2.4 18

NADW

(%) Δ -8 -11 -8 -13 -13 -13 -14 -15 -22 -17 -11 -12 -11 -8 -23 -6 -18 -14 0 -14 -8 -13 18

AABW T H 5.5 7.8 7.4 7.3 3.8 1.9 10.8 3.9 3.1 3.7 6 11.5 11.4 4.6 5.2 8.8 9.2 5.5 2.2 3.8 8.5 5.5 n/a

Δ -1.1 -3.4 0 2.1 1.9 -0.1 0.5 0.9 0.2 -1.6 -2 -1.4 -1.6 -0.6 -0.7 -0.4 -1 -1.3 -0.2 -1.4 0.1 -0.6 11

BMC

[°S]

H 41.5 41.4 43.4 44 41.7 39.1 38.7 42.8 38.6 34.8 42.4 39.1 39.4 38.1 38.6 39.6 41 39.7 39.3 41.6 38.8 39.6 n/a

Δ -1.6 -1 -1.2 -1 -0.5 -2.8 -1.6 -0.7 -0.7 -4.5 -0.8 -1.2 -1.5 -0.7 -1.4 -1.6 -0.9 -0.1 -1.3 -1.6 -0.7 -1.2 19

!

!

!

!

!

!

References Evans D L and Signorini S S 1985 Vertical structure of the Brazil Current Nature 315 48–50 Garzoli S L 1993 Geostrophic velocity and transport variability in the Brazil-Malvinas Confluence Deep Sea Research Part I: Oceanographic Research Papers 40(7) 1379–1403 Garzoli S L Field A Johns W E and Yao Q 2004 North Brazil Current retroflection and transports Journal of Geophysical Research 109 1–14 Garzoli S L and Garraffo Z 1989 Transports, frontal motions and eddies at the Brazil-Malvinas currents confluence Deep Sea Research Part A Oceanographic Research Papers 36(5) 681–703 Gordon A L and Greengrove C L 1986 Geostrophic circulation of the Brazil-Falkland confluence Deep Sea Research Part A Oceanographic Research Papers 33(5) 573–585 Hummels R Brandt P Dengler M Fischer J Araujo M Veleda D and Drugadoo J 2015 Interannual to decadal changes in the Western Boundary Circulation in the Atlantic at 11°S Geophysical Research Letters 42 7615–7622 Lentini C A D Goni G J and Olson D B 2006 Investigation of Brazil Current rings in the confluence region Journal of Geophysical Research 111 1–17 Miiller J Ikeda Y Zangenberg N and Nonato L V 1998 Direct mesuarements of the western boundary current off Brazil between 20S and 28S Journal of Geophysical Research 103(97) 5429–5437 Stramma L 1989 The Brazil current transport south of 23°S Deep Sea Research Part A. Oceanographic Research Papers 36(4) 639–646 Stramma L Ikeda Y and Peterson R G 1990 Geostrophic transport in the Brazil Current region north of 20 ° S Deep Sea Research 37(12) 1875–1886