file - Biogeosciences

Biogeosciences, 10, 5079–5093, 2013www.biogeosciences.net/10/5079/2013/doi:10.5194/bg-10-5079-2013© Author(s) 2013. CC Attribution 3.0 License.

EGU Journal Logos (RGB)

Advances in Geosciences

Open A

ccess

Natural Hazards and Earth System

Sciences

Open A

ccess

Annales Geophysicae

Open A

ccess

Nonlinear Processes in Geophysics

Open A

ccess

Atmospheric Chemistry

and Physics

Open A

ccess

Atmospheric Chemistry

and Physics

Open A

ccess

Discussions

Atmospheric Measurement

Techniques

Open A

ccess

Atmospheric Measurement

Techniques

Open A

ccess

Discussions

Biogeosciences

Open A

ccess

Open A

ccess

BiogeosciencesDiscussions

Climate of the Past

Open A

ccess

Open A

ccess

Climate of the Past

Discussions

Earth System Dynamics

Open A

ccess

Open A

ccess

Earth System Dynamics

Discussions

GeoscientificInstrumentation

Methods andData Systems

Open A

ccess

GeoscientificInstrumentation

Methods andData Systems

Open A

ccess

Discussions

GeoscientificModel Development

Open A

ccess

Open A

ccess

GeoscientificModel Development

Discussions

Hydrology and Earth System

Sciences

Open A

ccess

Hydrology and Earth System

Sciences

Open A

ccess

Discussions

Ocean Science

Open A

ccess

Open A

ccess

Ocean ScienceDiscussions

Solid Earth

Open A

ccess

Open A

ccess

Solid EarthDiscussions

The Cryosphere

Open A

ccess

Open A

ccess

The CryosphereDiscussions

Natural Hazards and Earth System

SciencesO

pen Access

Discussions

Diapycnal oxygen supply to the tropical North Atlantic oxygenminimum zone

T. Fischer, D. Banyte, P. Brandt, M. Dengler, G. Krahmann, T. Tanhua, and M. Visbeck

GEOMAR Helmholtz Centre for Ocean Research, Kiel, Germany

Correspondence to:T. Fischer ([email protected])

Received: 1 October 2012 – Published in Biogeosciences Discuss.: 17 October 2012Revised: 21 May 2013 – Accepted: 3 June 2013 – Published: 26 July 2013

Abstract. The replenishment of consumed oxygen in theopen ocean oxygen minimum zone (OMZ) off northwestAfrica is accomplished by oxygen transport across and alongdensity surfaces, i.e. diapycnal and isopycnal oxygen sup-ply. Here the diapycnal oxygen supply is investigated usinga large observational set of oxygen profiles and diapycnalmixing data from years 2008 to 2010. Diapycnal mixing isinferred from different sources: (i) a large-scale tracer re-lease experiment, (ii) microstructure profiles, and (iii) ship-board acoustic current measurements plus density profiles.From these measurements, the average diapycnal diffusiv-ity in the studied depth interval from 150 to 500 m is esti-mated to be 1×10−5 m2 s−1, with lower and upper 95 % con-fidence limits of 0.8× 10−5 m2 s−1 and 1.4× 10−5 m2 s−1.Diapycnal diffusivity in this depth range is predominantlycaused by turbulence, and shows no significant vertical gra-dient. Diapycnal mixing is found to contribute substantiallyto the oxygen supply of the OMZ. Within the OMZ core,1.5 µmol kg−1 yr−1 of oxygen is supplied via diapycnal mix-ing, contributing about one-third of the total demand. Thisoxygen which is supplied via diapycnal mixing originatesfrom oxygen that has been laterally supplied within the upperCentral Water layer above the OMZ, and within the AntarcticIntermediate Water layer below the OMZ. Due to the exis-tence of a separate shallow oxygen minimum at about 100 mdepth throughout most of the study area, there is no net ver-tical oxygen flux from the surface layer into the Central Wa-ter layer. Thus all oxygen supply of the OMZ is associatedwith remote pathways.

1 Introduction

The open oceans host distinct permanent regions of low oxy-gen concentration (oxygen minimum zones, OMZs), manyof them situated near the eastern boundaries of the tropicaloceans outside the equatorial belt. Tropical OMZs have somecommon features. They coincide with weak mean circula-tion, and their core is found between 200 and 700 m depth.However, their spatial extent and oxygen levels are diverse.While moderate oxygen levels are encountered in the At-lantic OMZs, very low to vanishing oxygen levels are foundin the Indian and Pacific OMZs (Karstensen et al., 2008;Paulmier and Ruiz-Pino, 2009). OMZs are regions of di-verse microbial and biochemical activity (Lam and Kuypers,2011; Wright et al., 2012), but pose a challenge or evenlethal threat to certain marine animal species, with increas-ing numbers of threatened species the lower the oxygen level(Vaquer-Sunyer and Duarte, 2008; Ekau et al., 2010; Princeet al., 2010). With anthropogenic climate change projectionsof a warming ocean and increased stratification, a net de-cline of the global ocean oxygen content is expected, andalso an expansion and intensification of OMZs is anticipated(Keeling et al., 2010). These expectations are in line with ob-servations over the past decades, which indicate a recent de-cline in global subsurface ocean oxygen content (Strammaet al., 2010; Helm et al., 2011), and also indicate a recent ex-pansion of individual OMZs (Whitney et al., 2007; Bogradet al., 2008; Stramma et al., 2008b). Particularly the tropi-cal North Atlantic OMZ off northwest Africa has shown asignificant oxygen decline and expansion over the past 50years (Stramma et al., 2008b, 2009). Consequences of oxy-gen decline for biogeochemical cycles, marine organismsand ecosystems are expected to be profound, particularly in

Published by Copernicus Publications on behalf of the European Geosciences Union.

5080 Fischer et al.: Oxygen supply to the tropical Atlantic OMZ

regions that are already low in oxygen (Ekau et al., 2010;Gruber, 2011; Wright et al., 2012).

The existence and position of oxygen minimum layers canbasically be explained by a balance between oxygen loss dueto respiration of organic matter (i.e. oxygen consumptionrate), and oxygen supply via ocean circulation (i.e. ventila-tion) (Wyrtki, 1962). Ventilation can be achieved by advec-tive and diffusive supply. Here, supply is defined as excess ofoxygen flux into a fixed volume over oxygen flux out of thevolume. A suitable coordinate system to describe ocean cir-culation is density coordinates, which divide ventilation intoprocesses that act along surfaces of constant density (isopy-cnal) and across surfaces of constant density (diapycnal).This splits ventilation processes into four contributing terms:diapycnal diffusion (molecular, turbulent eddy, and doublediffusive), diapycnal advection, isopycnal diffusion (mainlyeddy), and isopycnal advection. Interpreting the oxygen bal-ance, the consumption rate may be seen as an external forc-ing exerted by sinking organic matter that originates in thesurface layer, while parameters of ocean circulation such asmean flow and eddy diffusivities set the system’s reaction tothe forcing. This reaction is then manifested in the resultingshape of the oxygen concentration field, and in the magnitudeof associated diapycnal and isopycnal oxygen fluxes whichreplenish the oxygen consumed. In this view, feedbacks onorganic matter supply caused by long-term changes of thecirculation, or the effect of very low oxygen levels close tozero on consumption, are ignored.

However, quantitative knowledge about processes control-ling OMZs is limited. Neither oxygen consumption rates northe oxygen supply by the different ventilation processes men-tioned above are well constrained by observations (Keelinget al., 2010). Furthermore, numerical model simulations donot well reproduce the observed global oxygen distribution(Duteil and Oschlies, 2011; Najjar et al., 2007; Meissneret al., 2005). Despite this weakness, simulations under fu-ture climate scenarios consistently predict a global oxygenloss of the ocean (Keeling et al., 2010). But how the extentand intensity of individual tropical OMZs will change in thefuture, is not clear: oxygen solubility, stratification, circula-tion, and organic export production are all affected by climatechange (Bopp et al., 2001, 2002), and each of them, by theireffect on ventilation and consumption, sensitively influencesthe extent and intensity of OMZs (Gruber, 2011). Simulationefforts have resulted in diverse projections of the future vol-ume of OMZs, reaching from expansion (Matear and Hirst,2003; Oschlies et al., 2008) to neutral (Gruber, 2011) to con-traction (Duteil and Oschlies, 2011). The latter study foundeven the sign of the predicted trend being sensitive to theamount of background diapycnal mixing, and thus indicatesthat quantifying diapycnal processes may be important forthe understanding of OMZs.

Recent observational progress concerning ventilation ofOMZs has been achieved in the tropical North AtlanticOMZ using two different approaches. In an idealized frame-

work of an advection–diffusion model fit to observations,Brandt et al.(2010) quantified the contributions of physi-cal processes replenishing oxygen at a given isopycnal sur-face within the OMZ. The diapycnal oxygen supply was esti-mated by assuming a diapycnal diffusivity and calculating adiapycnal oxygen curvature from the observed oxygen distri-bution and the simulated oxygen concentration at that isopy-cnal surface. In another approach,Banyte et al.(2012) deter-mined the diapycnal diffusivity by a large-scale tracer releaseexperiment in the oxycline above the core of the tropicalNorth Atlantic OMZ to be(1.19±0.18)×10−5 m2 s−1. Thisvalue is of the same order as determined in previous studiesin the thermocline of the open ocean, and much lower thanin boundary regimes and other hotspots of diapycnal mix-ing. Nevertheless, we will show below that diapycnal mixingseems to be an important process for the ventilation of thisopen ocean OMZ, and one of the major controls on concen-tration level and shape of the oxygen field. This is in line withthe model-based finding of diapycnal diffusivity controllingthe future evolution of OMZs (Duteil and Oschlies, 2011).

The main objective of the present study is to quantify di-apycnal oxygen supply and its vertical structure in the tropi-cal North Atlantic OMZ. Diapycnal supply is the sum of sup-plies by diapycnal diffusion and by diapycnal advection, butoxygen supply by diapycnal advection is assumed negligiblein the depth range associated with the OMZ (cf. Sect.4.5).Therefore in the following, we focus on estimating the di-apycnal supply by diapycnal diffusion. In order to quantifythe diapycnal diffusive supply within the OMZ, we use anextensive ship-based observational dataset of simultaneousoxygen and mixing measurements. To date, diapycnal oxy-gen supply in the open ocean has not been determined fromappropriate measurements elsewhere. Nevertheless, the ba-sic method to evaluate Fick’s first law of diffusion, using si-multaneous profiles of diffusivities and concentrations, hasbeen used in other contexts to estimate fluxes of solutes andis well established. Turbulence profiles and nutrient profileshave been used in various shelf regions to estimate nutrientsupply to surface waters (Sharples et al., 2001, 2007; Rippethet al., 2009; Hales et al., 2005, 2009; Schafstall et al., 2010).Similarly, Kock et al.(2012) used microstructure turbulenceand nitrous oxide gradients on the shelf and in the open oceanof the tropical Atlantic to estimate gas flux to surface waters.

The dataset was compiled from several cruises to the OMZsince 2008, when observation activities in the tropical NorthAtlantic intensified in the framework of the German researchprogrammes SFB754, SOPRAN, and NORTH ATLANTIC,and the EU project EUROSITES. Diapycnal diffusivity wasobtained by microstructure profiling and shipboard acousticcurrent profiling. The mentioned tracer release experiment– GUTRE (Guinea Upwelling Tracer Release Experiment,Banyte et al., 2012) – happened simultaneously and pro-vides an independent integrative estimate of diapycnal dif-fusivity for the depth range which the tracer patch occupied.The available diffusivity data overlap in the depth range 150

Biogeosciences, 10, 5079–5093, 2013 www.biogeosciences.net/10/5079/2013/

Fischer et al.: Oxygen supply to the tropical Atlantic OMZ 5081

40oW 35oW 30oW 25oW 20oW 15oW 10oW 0o

5oN

10oN

15oN

20oN

25oN

40

50

60

70

80

77

77

67

67 57

µmol/kg

77

NEC

NECC / NEUC

Fig. 1. Oxygen concentrations (coloured dots) in the core of thedeep oxygen minimum, which is situated at depths between 350and 500 m. Oxygen data were collected during the time period of2008 through 2010 of the simultaneous tracer release experimentGUTRE, and stem from seven ship cruises in total. Black labelledlines are oxygen core concentration isolines, the minimum concen-trations derived from World Ocean Atlas WOA09 (Garcia et al.,2010) by vertical spline interpolation. The WOA 77 µmol kg−1 and67 µmol kg−1 isolines roughly enclose the observed 60 µmol kg−1

and 50 µmol kg−1 core concentrations in the period 2008 to 2010,while the WOA 57 µmol kg−1 isoline does not well describe the40 µmol kg−1 observations. The WOA 77 line is used here to de-fine the extent of the OMZ. The dashed black line defines the anal-ysis box (in depth interval 150 to 500 m) that comprises most ofthe available mixing data. NEC: North Equatorial Current, NECC:North Equatorial Countercurrent, NEUC: North Equatorial Under-current.

to 500 m, which comprises the OMZ core and the oxyclineabove. Based on the microstructure, and acoustic and tracerdata, we provide an estimate of the average diapycnal diffu-sivity for the tropical North Atlantic OMZ at 150 to 500 mdepth (Sect.4.1), which may also serve for future modelstudies. We further estimate the average diapycnal oxygensupply profile (Sect.4.2 to 4.5), which makes it possibleto draw some conclusions on this OMZ’s supply pathwaysand their importance. The extensive database further allowsthe deduction of some statistical information, which may be-come useful for similar future observational studies, e.g. theobserved horizontal independence of vertical oxygen gradi-ents and according diapycnal mixing (Sect.3.7), or the es-timated size of confidence limits for the average diapycnaldiffusivity if using smaller datasets (Sect.4.1).


5oN

10oN

15oN

20oN

25oN

400

400

450 500

350

350

27.1

27.05

27.0

27.0

26.95

26.95

Fig. 2.Vertical position of the deep oxygen minimum core in depthcoordinates (black isolines, metres) and in potential density coordi-nates (red isolines, kg m−3). Isoline shapes are mostly derived fromobservations in the period 2008 to 2010. Where observational gapswere too large, isolines were completed by following the shapes ofisolines based on World Ocean Atlas WOA09 (Garcia et al., 2010).

2 Study site

2.1 Tropical North Atlantic oxygen minimum zone

The OMZ in the tropical North Atlantic Ocean (Fig.1) ap-proximately matches the Guinea Dome region (Siedler et al.,1992). This coincidence is most likely caused by both, OMZand Guinea Dome, being linked to the zone of weak cir-culation between North Equatorial Current (NEC at 20◦ Nto 15◦ N) and North Equatorial Countercurrent/Undercurrent(NECC/NEUC at 5◦ N) (Karstensen et al., 2008). Verticaloxygen profiles in this region typically exhibit two oxygenminimum layers at core depths of about 100 m and 450 m(Fig. 4). The shallow minimum with a thickness of about50 m is pronounced in 80 to 90 % of the oxygen profiles.Profiles lacking a shallow oxygen minimum were found tobe scattered throughout the study area, with a tendency to be-come more frequent towards the southern part of the OMZ.The deep oxygen minimum at a depth range of about 300 mto 700 m exists in all profiles in the region, and shows coreconcentrations of typically 40 to 60 µmol kg−1. These mini-mum concentrations are on average 40 µmol kg−1 lower thanthe core concentrations of the shallow minimum. In the fol-lowing we will only study the deeper, more voluminous andmore intense oxygen minimum. Its core is at the interfacebetween Antarctic Intermediate Water (AAIW) and the over-lying Central Water (Stramma et al., 2008a) and has showna noticeable oxygen decline during the last five decades(Stramma et al., 2008b, 2009). The vertical position of thecore in the OMZ region neither perfectly follows isobaths

www.biogeosciences.net/10/5079/2013/ Biogeosciences, 10, 5079–5093, 2013



5oN

10oN

15oN

20oN

25oN

Fig. 3.Minimum covered area of tracer patch after 30 months (ma-genta filled contour) as well as locations of microstructure mea-surements (black squares) and stations with shear data from 75 kHzvessel-mounted ADCP (blue squares) during cruises in the period2008 to 2010. The chosen analysis box is marked with a dashedblack line.

nor isopycnals (Fig.2). However it is better aligned to isopy-cnals: the standard deviation of the core vertical position inthe OMZ is 60 m with respect to the mean core depth of425 m, and 40 m with respect to the mean core isopycnal27.03 kg m−3. The position of the maximum oxygen gradientabove the OMZ core (at about 300 m depth and the isopyc-nal 26.85 kg m−3) will be called deep oxycline throughoutthis study in order to distinguish it from the oxycline just be-low the mixed layer (Fig.4). Approximately, the deep oxy-cline divides the Central Water into upper and lower Cen-tral Water, in accordance with the definition used byKirch-ner et al.(2009). We define horizontal OMZ extent by the60 µmol kg−1 isoline of oxygen core concentration as ob-served between 2008 and 2010 (Fig.1). The reason for thischoice is that the region encompassed by the 60 µmol kg−1

isoline includes as much as possible of the low oxygen re-gion of the tropical Northeast Atlantic, while it excludes ad-jacent regimes which are better ventilated, particularly thoseassociated with the eastward flow of the NECC/NEUC in thesouth and with the subtropical gyre at Cape Verde front inthe north. Furthermore 60 µmol kg−1 can be considered asa limit to hypoxia for marine animals (Diaz, 2001; Vaquer-Sunyer and Duarte, 2008; Ekau et al., 2010; Keeling et al.,2010). Meridional oxygen variations in the OMZ are foundto be associated with a system of alternating zonal currentbands, in which eastward/westward currents coincide withhigher/lower oxygen concentration (Stramma et al., 2008a;Brandt et al., 2010).

2.2 Analysis box

For the analysis of diapycnal oxygen flux, a box is chosenfrom 6 to 15◦ N latitude, 30 to 15◦ W longitude, and 150 to500 m depth. This box covers large parts of the OMZ at its60 µmol kg−1 isoline of core concentration, and it is centredat the location of the lowest observed core concentrationsjust below 40 µmol kg−1 (Fig. 1). Nevertheless it omits someparts of the OMZ: to the northeast a region associated withupwelling, to the west a part where we expect core concen-trations between 50 and 60 µmol kg−1, and to the southeasta part inside the shelf region and exclusive economic zonesof some west African states. The chosen analysis box is theplace where diapycnal mixing can be constrained best be-cause it was horizontally and vertically filled with tracer inthe later phase of tracer release experiment GUTRE (after20 and 30 months), and because most supplemental diapy-cnal mixing measurements that are suitable for our analysiswere conducted during GUTRE surveys inside the analysisbox (Fig.3). The depth interval of interest extends from be-low the shallow oxygen minimum across the deep oxyclinedown to the core of the deep oxygen minimum (Fig.4). It isthus confined to the upper part of the deep oxygen minimumlayer, and diapycnal oxygen flux to the OMZ from the well-oxygenated waters below the OMZ core cannot be quantifiedin this study. Stratification (N2) is near-constant through thechosen depth interval for individual profiles, but varies from1× 10−5 to 2× 10−5s−2 in the region. The higherN2 val-ues are found towards the southeast. For a detailed descrip-tion of theN2 distribution, refer toBanyte et al.(2012).

3 Data and methods

3.1 Diapycnal diffusivity, diapycnal flux, diapycnal fluxdivergence and averaging

In the open ocean, diapycnal mixing that is sustained bymolecular diffusion is often enhanced by turbulence frombreaking internal waves (e.g.Staquet and Sommeria, 2002),and can additionally be enhanced by double diffusive pro-cesses (e.g.St. Laurent and Schmitt, 1999). In this study, di-apycnal diffusivityK will be used to describe diapycnal mix-ing and to estimate the diapycnal flux of water mass proper-ties. Due to the presence of turbulence in the ocean, diapyc-nal property fluxes are greatly enhanced compared to prop-erty fluxes in a non-turbulent fluid. Diapycnal diffusivitiesare thus strongly governed by the turbulent flow characteris-tics of the oceanic region, albeit in principle,K must be dif-ferentiated with respect to the water mass property in focus.As in our case the appropriateKoxygen is inaccessible, othermore accessibleK values were used as approximate esti-mates. If not stated otherwise,K throughout the paper standsfor Koxygen, the one this study aims at. Other meanings ofK

will be indicated by indices, which may specify the technique



40 60 80 100 120 140 160 180 200 220

0

100

200

300

400

500

600

10−6

10−5

10−4

26 26.5 27 27.5

μmol/kg

m2/s

kg/m3

O2 K σθ

m kg/m3

26.426.5

26.6

26.7

26.8

26.9

27.0

27.1

27.2

Deep oxycline

Deep OMZ core

K arithmetic mean and 95% confidence limits

Fig. 4. Profiles of oxygen concentration, diapycnal diffusivityK

from microstructure measurements, and potential density, averagedacross the entire chosen analysis region of the OMZ.K values arearithmetic averages for 20 m bins within their 95 % confidence in-tervals. Dotted lines mark the depth interval where diapycnal oxy-gen flux is evaluated. This is the depth interval where depth intervalsfor K estimates from the different methods overlap.

to obtainK (KTRE, KMSS/ADCP) or the property the specificK is valid for (e.g.K tracer, Kρ) or the mechanism that causedK (K turb from mechanical turbulence,KDD from double dif-fusive enhancement). We use three different measurementstrategies to obtain estimates ofKoxygen. (1) The diapyc-nal spreading of a tracer during GUTRE (Sect.3.4) yieldedKTRE, which is an estimate for bothK tracerandKoxygen, as-suming that dissolved oxygen behaves similar to the tracer.(2) Measurements of the strength of mechanical turbulenceby using a microstructure profiler (Sect.3.5) delivered anestimate of the diapycnal diffusivity of massKρ . Kρ canserve as an estimate forKoxygen if (a) turbulent diffusivity ismuch greater than molecular diffusivities and (b) double dif-fusion is negligible (St. Laurent and Schmitt, 1999; Ferrariand Polzin, 2005). (3) Velocity profiles from vessel-mountedacoustic Doppler current profilers (vmADCP) yielded esti-mates of vertical shear spectra (Sect.3.6) which served toparameterizeKρ . As this parameterization (Appendix A) isbased on the measured microstructure profiles, the use ofvmADCP measurements is not an independent method, butit helps in substantially enhancing the database.

The diapycnal oxygen flux8 can be estimated usingFick’s first law of diffusion if simultaneous profiles of oxy-gen concentration (Sect.3.3) and diapycnal diffusivity areavailable. The diapycnal flux is a down-gradient flux perpen-dicular to density surfaces driven by diapycnal diffusivity.

The necessary diapycnal gradient of oxygen concentration tocalculate8 can be approximated by the vertical gradient ofoxygen concentration so that

8 = −ρ Koxygen∂

∂zc, (1)

with ρ water density in kg m−3, Koxygendiapycnal diffusivityof oxygen in m2 s−1, c oxygen concentration in µmol kg−1

and8 resulting diapycnal flux in µmol m−2 s−1.When regarding the oxygen budget in a volume enclosed

by two isopycnal surfaces, the diapycnal influx across thefirst isopycnal surface does not need to be of the same mag-nitude as the diapycnal outflux across the second isopycnalsurface. The difference of the two diapycnal fluxes is de-scribed by the diapycnal flux divergence∇8, which can beapproximated by the vertical derivative of8, ∂8/∂z. ∇8 isin µmol kg−1 yr−1. A positive value means that the densitylayer loses oxygen via a diapycnal flux imbalance. In orderto quantify the net oxygen supply of the density layer causedby diapycnal processes, it is necessary to observe−∇8.

Ultimately, the goal is to determine averaged profiles of di-apycnal diffusivities, fluxes, and flux divergences for the en-tire OMZ. Thus averaging along isopycnal surfaces is a nec-essary processing step which will be denoted by brackets, asfor example in〈K〉. Unless otherwise specified, the brack-ets refer to arithmetic averaging after objective mapping onisopycnal surfaces. For the objective mapping ofK, thedecorrelation scale was determined by a semivariogram tobe 0.5◦. Confidence limits for isopycnal averages are deter-mined from the combined individual error distributions, tak-ing into account the reduced degrees of freedom caused bythe spatial correlation. Another kind of averaging that willbe used in this study is averaging in diapycnal direction, andwill be denoted by an overbar, as for example inK, so thata total average for the entire analysis box in this examplewould be〈K〉.

3.2 Data overview

The data were collected during four cruises in November2008 (RV Merian MSM10/1), November and December2009 (RVMeteorM80/1 and M80/2), and October 2010 (RVMeteorM83/1) during the tracer surveys for GUTRE. Oxy-gen was measured using a conductivity–temperature–depthprofiler (CTD) with added Clark-type oxygen sensors (CTD-O2). For most of the oxygen profiles, simultaneous estimatesof diapycnal diffusivityKρ from microstructure profiles orvmADCP were available. In total, 400 complete profiles con-sisting of CTD-O2 and diapycnal diffusivity were collectedin the analysis box (Fig.3). The tracer surveys delivered oneadditional independent space–time-averaged estimate of di-apycnal diffusivity for the entire region for the period of 2008through 2010. Another three cruises in February, March, andApril 2008 (RV L’Atalante GEOMAR/3 and GEOMAR/4,and RVMerian MSM08/1) delivered 110 additional oxygen



profiles, but without simultaneousK estimates. That is whythese oxygen measurements only helped in outlining theOMZ (Fig. 1), but did not enter the oxygen flux calculation.

3.3 Oxygen and CTD data calibration

Each ship station that is considered here comprises a CTD-O2 profile down to at least the deep oxygen minimum coreat about 400 to 500 m. On the four cruises evaluated here(MSM10/1, M80/1, M80/2, and M83/1) similar CTD/rosettesystems were used, consisting of a Sea-Bird 911plus CTDequipped with dual Sea-Bird temperature, conductivity, andoxygen sensors. During each of the cruises, several hundredwater samples were collected to calibrate the conductivityand oxygen sensors using a Guildline Autosal 8400B sali-nometer and a Winkler titration stand. The deviations be-tween the in situ conductivity measured by the Sea-BirdCTD and the high-accuracy on-board water sample measure-ments using salinometers were used to derive a correction tothe CTD’s conductivity linear in pressure, temperature, andCTD conductivity itself. After applying the corrections, theresulting absolute accuracy of the CTD salinity values wasestimated to 0.005, 0.002, 0.002, and 0.003 for the cruisesMSM10/1, M80/1, M80/2, and M83/1, respectively. Labo-ratory calibration suggests the accuracy of the temperatureand pressure measurements to be better than 0.002 K and1.5 dbar, respectively. Similar to conductivity, the deviationsbetween the CTD and the on-board Winkler titrations wereused to derive a correction to the CTD’s oxygen measure-ments. The chosen correction was linear in pressure, temper-ature, and oxygen, and we estimate the absolute accuracy ofthe CTD’s oxygen measurements to be better than 3, 1, 1,and 1 µmol kg−1 for the respective four cruises.

While these estimates of the CTD data’s absolute uncer-tainty and their only slowly varying corrections are use-ful for comparisons with other observations, for our anal-ysis of vertical oxygen gradients, an estimate of the oxy-gen sensor noise and precision is more relevant. We foundthat the oxygen sensors had a typical instrument noise ofless than 0.2 µmol kg−1 (95th percentile of the signal resid-uals, after removing the slowly varying signal). The esti-mates of vertical oxygen gradients used here are based on1 dbar averages of the CTD oxygen measurements, whichwere originally recorded at 24 Hz. This results in a noise-induced uncertainty of the vertical oxygen gradient of lessthan 0.03 µmolkg−1m−1 when evaluated at 2 dbar intervals.

3.4 K estimated from GUTRE

The deliberate Guinea Upwelling Tracer Release Experiment(GUTRE) was performed in the deep oxycline in the tropicalNorth Atlantic OMZ in order to obtain a time- and space-integrated estimate of diapycnal diffusivity at the OMZ’s up-per limit. Ninety-two kilograms of the halocarbonic com-pound SF5CF3 (Ho et al., 2008) were released in April 2008

at 8◦ N 23◦ W at the isopycnal surfaceσθ = 26.88 kg m−3.The expanding tracer patch was sampled during three sur-vey campaigns (MSM10/1, M80 and M83/1) 7, 20, and 30months after injection. After 20 and 30 months, the tracer es-sentially covered the analysis box horizontally and vertically.From the increase of the vertical extent of the isopycnally in-tegrated tracer distribution, the time–space-averaged diapyc-nal diffusivity was estimated to be〈K〉TRE = (1.19±0.18)×10−5 m2 s−1 (Banyte et al., 2012). The uncertainty at 95 %confidence level is based on bootstrapping and measurementerrors. During all three surveys the mean tracer concentrationprofile showed no significant skew.

3.5 K estimated from microstructure data

As an independent method from GUTRE, diapycnal diffusiv-ities were also estimated using microscale shear recordingsfrom a loosely tethered microstructure profiler. During thetracer survey cruises, the microstructure profiler MSS90D(termed MSS in the following) of Sea and Sun Technology,Trappenkamp, Germany, was used immediately after certainCTD casts. Usually three profiles in a series down to 500 mat about 0.5 ms−1 sink velocity were collected at these shipstations. In total, MSS data could be obtained on 45 ship sta-tions in the analysis box. On the way to estimate diapycnaldiffusivities from MSS, the signal from two to four airfoilshear sensors at the profiler head was converted to profilesof microscale vertical shear. Then the dissipation rateε ofturbulent kinetic energy could be estimated from the powerspectrum of microscale vertical shear by using a verticalwavenumber range free of instrument vibrational noise andby assuming local isotropy (Oakey, 1982; Stips and Prandke,2000). The diapycnal diffusivity of mass,Kρ , was inferredfrom the Osborn parameterizationKρ = 0 εN−2 (Osborn,1980), with 0 the dimensionless dissipation ratio that is re-lated to mixing efficiency.0 was set at 0.2 because simu-lation data byShih et al.(2005) suggest this value to beappropriate for weak to moderate turbulence, which is es-sentially what was found in the analysis box: the averagevalue of the turbulence activity parameterεν−1N−2 calcu-lated from all data available for this study amounts to 41, andless than 8 % of theεν−1N−2 values exceed 100. When tak-ing into account the eddy diffusivity parameterization sug-gested byShih et al.(2005) for situations ofεν−1N−2 >100,the averageKρ reduces by only 2.5 %. The value 0.2 is fur-ther supported by findings ofSt. Laurent and Schmitt(1999),who report0 between 0.15 and 0.25 in turbulence-dominatedregimes with weak mean shear – conditions that were alsomet in the analysis box. For the analyses of this study, ver-tical averages ofε and N2 for the depth interval 150 mthrough 500 m were used to infer a single averageKρ,MSSper ship station via the Osborn parameterization. Using nar-rower depth bins to calculateKρ,MSS would only marginallychange the results. At the same time, growing measurementerrors inN2, which go hand in hand with narrower depth bins



chosen, would introduce a bias to theKρ estimated from theOsborn parameterization. The uncertainty ofKρ,MSS was es-timated from quantifiable errors to be 60 % on a 95 % con-fidence level. This uncertainty estimate is based on the fol-lowing reasoning: the microstructure probe delivers one es-timate of ε per metre depth, which leaves us with roughly1000ε values in the 150 m to 500 m depth range per stationof three profiles. Theseε values are arithmetically averagedfollowing Davis(1996). To estimate the uncertainty of theεaverage, we follow an approach used byFerrari and Polzin(2005) in a similar situation where they calculated verticallyaveragedKρ values from microstructure measurements inthe eastern North Atlantic. We find a vertical decorrelationscale in ourε profiles of 5 m (comparable to the findings ofFerrari and Polzin(2005) in water depths< 800 m), whichleaves us with roughly 200 degrees of freedom to reduce thespread of theε distribution. Nonetheless, theε uncertaintyremains the main contributor to the total uncertainty of theKρ estimate, together with sensor uncertainties that have tobe accounted for as systematic errors for the duration of anMSS station. There is also a minor contribution from uncer-tainty in N2, and there certainly is some contribution fromthe choice of0, but of unknown amount.

3.6 K estimated from vmADCP

The possibility to obtain additional estimates ofKρ from cur-rent velocity profiles via vmADCP substantially enhancedthe number of “complete” station data comprising CTD plusoxygen plusKρ , which could be used to infer diapycnal oxy-gen flux; the number rising from 45 to 400 inside the analy-sis box (Fig.3). The method is based on the observation thatfinescale vertical shear with vertical wavelengths of the orderof ten to some hundreds of metres and dissipation rateε ofturbulent kinetic energy can be related (Gargett, 1976; Gregg,1989). It had been used before with lowered ADCP data(Polzin et al., 2002; Kunze et al., 2006). The main processingstrategy for the application to vmADCP data during ship sta-tions was as follows: (1) generate velocity profiles with noiselow enough so that vertical shear signals which are of the or-der of the open ocean background level can still be detected;(2) obtain spectra of vertical shear of horizontal velocity inthe finescale vertical wavenumber range; (3) obtain dissipa-tion rateε from shear power spectral level8S by a parame-terization (see Appendix) which uses MSS measurements forcalibration; (4) obtainKρ from ε and CTD-derived stratifica-tion by the Osborn parameterization as in Sect.3.5.

Minutiae of processing were the following: during threesurvey cruises MSM10/1, M80/2 and M83/1, an RDI OceanSurveyor 75 kHz vmADCP continuously recorded currentsdown to 600 m depth at 8 m vertical bin size. The broad-band mode that we used, a high ping frequency (36 min−1),usually calm sea conditions, and a final two-dimensional fil-tering step of velocity fields in depth–time–space, resultedin 1 min-average velocity data of 0.01 ms−1 precision at

95 % level (for technical details refer toFischer, 2011).From the 1 min-averaged and filtered velocity profiles, spec-tra of vertical shear were calculated for 150 to 533 m depthand corrected followingPolzin et al. (2002) for variancelosses caused by ADCP binning, tilting, and ping averag-ing. The resulting 1 min shear power spectra were usable inthe wavenumber band from 1/128 cpm to 1/38.4 cpm. Powerspectra were averaged over the duration of a ship station,and served to estimate a single value of shear power spec-tral level8S for that ship station and the entire depth interval150 to 500 m.8S led to εADCP after using the parameteri-zation Eq. (A5), and toKρ,ADCP after using the Osborn pa-rameterization. The uncertainty ofKρ,ADCP was estimated tobe a factor of 2.7 at 95 % confidence level, resulting frommeasurement noise, discrete spectrum estimation, and errorprogression through, as well as unexplained variance of, theparameterization Eq. (A5).

3.7 Implications of K derived from different sources

Apart from physical differences of the derived diapycnal dif-fusivities, the three applied methods deliveredK estimateswhich also differ in the grade to which they represent spa-tially and temporally averaged quantities. The immediateresult of the MSS measurements were vertical profiles ofKρ,MSS for each ship station at about 0.5 m depth resolution,the result of the vmADCP processing was a single verticalaverageKρ,ADCP for each ship station, and the TRE deliv-ered essentially one time–space-average〈K〉TRE for the en-tire analysis box during the period 2008 to 2010. For the fur-ther processing ofK – i.e. to deliver estimates of〈K〉oxygen,a profile of diapycnal oxygen flux,〈8〉, and a profile of di-apycnal flux divergence – we assume thatK can be treatedas vertically homogeneous through the used depth interval150 to 500 m for timescales larger than several days. Thisassumption is supported by the MSS-derived average pro-file 〈K〉ρ,MSS, which shows no significant gradient at 150 to500 m (Fig.4), as well as by GUTRE tracer profiles, whichshowed no significant deviation from a Gaussian (Banyteet al., 2012). Vertical homogeneity at 150 to 500 m is alsosuggested by the majority of individual density profiles inthe analysis box, which show a near constant stratification.The assumption of vertically homogeneousK implies that(1) vertical averaging ofKρ,MSS for MSS data is equivalentto averaging of longer time series, which we lack, and (2)K

defines the level of the assumed constantK profile at eachship station; in that senseK and K are equivalent for theprocess of isopycnal averaging (e.g. to estimate〈8〉). An-other property of our dataset is the statistical independenceof the oxygen gradient andKρ from MSS/ADCP, as shownin Fig. 5 for all depths between 150 and 500 m. The inde-pendence is expressed in the equivalence of〈Kρ × ∇ c〉 and〈Kρ〉× 〈∇ c〉. This also means that〈8〉 may be calculated as〈〈K〉 ×∇ c〉, i.e.K may also be treated as if it were laterallyhomogeneous (but this time on the long timescale of the TRE



−6 −4 −2 0 2 4 6

x 10−3

100

150

200

250

300

350

400

450

500

550

600

< K ∇c > − < K > < ∇c >

dept

h in

m

µmol/m2/s

Fig. 5. Difference between the lateral average of local diapycnalfluxes and the product of laterally averaged diapycnal diffusivityand laterally averaged oxygen gradient as a function of depth. Thedifference is found to be smaller than the 95 % confidence intervalof 〈K ·∇ c〉 (grey shading) for most depth layers. Dashed lines markthe analysis depth interval whereK observations exist.

duration). The estimates of this effective long-term large-scaleK will be denoted〈K〉TRE and〈K〉MSS/ADCP (MSS andADCP combined because the ADCP method was calibratedwith MSS measurements).

4 Results and discussion

4.1 Diapycnal diffusivities

Estimates of diapycnal diffusivity were derived from the de-liberate tracer release experiment GUTRE as one integrativevalue〈K〉TRE, and from MSS/ADCP methods asKMSS/ADCPfor individual locations in the analysis box. The values are tobe compared and evaluated to what degree they can serve asestimates for〈K〉oxygen.

The KMSS/ADCP were arithmetically averaged after ob-jective mapping on isopycnal surfaces (Section3.1) to al-low comparison to〈K〉TRE. The resulting〈K〉MSS/ADCP

as an estimate for〈K〉ρwas (0.95± 0.15) × 10−5 m2 s−1.〈K〉TRE from GUTRE as an estimate for〈K〉tracer was(1.2± 0.2) × 10−5 m2 s−1. Both uncertainty estimates onlyrepresent the quantifiable errors, i.e. a lower limit to the trueuncertainty. The two methods to estimate〈K〉 further con-tain systematic biases of unknown amount that could notbe accounted for in the methods’ uncertainties. Concerningthe TRE this applies to (1) the fact that for each tracer sur-vey about half of the tracer patch could not be sampled, andthus the deduced diapycnal extent of the patch might be bi-ased; (2)〈K〉TRE being biased towards the early TRE pe-riod in which the tracer only covered a fraction of the entire

region. Concerning MSS measurements this applies to (1)underlying assumptions in the train of calculating turbulentdiffusivity (e.g. the value of dissipation ratio0 in the Os-born parameterization), (2) seasonal bias, and (3) some gapsin the sampling of the analysis box. Thus we conclude thatthe two〈K〉 estimates are not significantly different, even ifthe given minimum confidence limits suggest a barely sta-tistically significant difference. Furthermore, both〈K〉 val-ues can serve as estimates for〈K〉oxygen. 〈K〉 of the orderof 1×10−5 m2 s−1 is much larger than the molecular diffu-sivity of heat (≈10−7 m2 s−1) and the molecular diffusiv-ity of solutes (≈10−9 m2 s−1). Thus, molecular diffusivitiesare negligible, and concerning the one part of eddy diffusiv-ity which is caused by mechanical turbulence,〈K〉oxygen,turb

should be indistinguishable from〈K〉ρ,turb and〈K〉tracer,turb.Double diffusion could in principle enhance the diffusivityto 〈K〉oxygen= 〈K〉oxygen,turb + 〈K〉oxygen,DD. Assuming thatdissolved oxygen behaves as a tracer similar to salt, we es-timate the double diffusive enhancement following the tech-nique ofSt. Laurent and Schmitt(1999) to be of the order of1×10−6 m2 s−1 in the depth range 150 m to 500 m (assum-ing salt-finger thermal diffusivityk(f )

θ to be 6×10−6 m2 s−1).That means〈K〉oxygenmight be 10 % larger than〈K〉ρ,turb asestimated by〈K〉MSS/ADCP , which is an insignificant differ-ence in the range of uncertainties.〈K〉TRE already comprises〈K〉tracer,turb and〈K〉tracer,DD and is thus also a good estimatefor 〈K〉oxygen.

The quintessence so far is that double diffusivity is neg-ligible in the analysis depth range and turbulence is themain mixing driver, so that both〈K〉ρ from MSS/ADCPand 〈K〉tracer from GUTRE can serve as appropriate esti-mates for〈K〉oxygen. The merged confidence interval for〈K〉oxygen, as obtained from the two independent methods,is bounded by 95 % confidence limits of 0.8×10−5 m2 s−1

and 1.4×10−5 m2 s−1.This estimate of 1×10−5 m2 s−1 is distinctly higher than

the diapycnal diffusivity that could be expected for theOMZ’s range of latitudes. An internal wave field with back-ground intensity and with stratificationN2 in the observedrange of 1 to 2×10−5 s−2 should exhibit a〈K〉 of only some10−6 m2 s−1 following the parameterization inGregg et al.(2003). The higher than expected〈K〉 is coincident withrough bottom topography in the analysis box and an inten-sified internal wave field (Fischer, 2011). This aspect alsofits to the finding that elevated diffusivities are found neara seamount chain in the centre of the study area. The spa-tial distribution ofK will be subject to further study and isbeyond the scope of this paper.

Finally, we examine what effect a reduced samplingcould have on the estimate of〈K〉. Comparing the esti-mated diapycnal diffusivity〈K〉MSS/ADCP for Nov 2008,Nov/Dec 2009, and Oct 2010 (from essentially one cruiseper year), we get 0.9×10−5 m2 s−1 ([0.7 1.1]×10−5 m2 s−1

as 95 % confidence interval), 0.95×10−5 m2 s−1 ([0.75



1.15] ×10−5 m2 s−1), and 1.05×10−5 m2 s−1 ([0.851.3]×10−5 m2 s−1), respectively. The three〈K〉MSS/ADCPestimates do not vary significantly between the years.However, each value only represents a month of surveyduring the last quarter of the corresponding year, so aseasonal bias cannot be ruled out. We further use the datasetof 400KMSS/ADCP values to investigate the influence of thenumber of randomly chosen diffusivity data on the resultinguncertainty (Fig.6), after using the same processing as usedfor the entire dataset. One hundred measurements (eachconsisting of either three MSS profiles or an ADCP record-ing during about a 1 h CTD cast) seem to be a reasonablenumber for an efficient estimate of〈K〉MSS/ADCP.

4.2 Diapycnal oxygen flux in the analysis box

Diapycnal diffusivities and oxygen gradients are combinedto an average profile of diapycnal oxygen flux as function ofdensity for the analysis box (Fig.7). Here the depth interval150 to 500 m, for which we haveK estimates, roughly corre-sponds to the potential density range ofσθ = 26.55 kg m−3

to 27.1 kg m−3 in a near-linear relation. Main features of thediapycnal oxygen flux profile are two surfaces of zero aver-age flux which coincide with positions of zero oxygen gra-dients: at the oxygen maximum at about 200 m depth andat the OMZ core at about 450 m depth (Fig.4). In betweenthese two surfaces, there is maximum downward flux of oxy-gen located at the deep oxycline. The depth interval of ourdata only allows oxygen flux calculations for the depth range150 m to 500 m of the water column, encompassing the OMZcore and the deep oxycline. The upper surface of zero diapy-cnal flux atσθ = 26.63 kg m−3 is situated in the upper Cen-tral Water that carries a relatively high oxygen concentration,and it exists because of the widespread presence of a shal-low oxygen minimum above. The existence of the surface ofzero diapycnal flux means that in the analysis box the oxygensupply to the OMZ has no direct link to the surface ocean.The diapycnal downflux of oxygen at the deep oxycline(σθ = 26.85 kg m−3) is about(5± 1) × 10−3 µmol m−2 s−1.This flux may be interpreted as feeding the diapycnal supplyof the OMZ core with about 1.5 µmol kg−1 yr−1 that comevia diapycnal mixing from the upper Central Water. This fluxmay as well be interpreted as a pure internal redistribution ofoxygen inside the water body that is limited by the two sur-faces of zero diapycnal flux at 200 and 450 m.

4.3 Sensitivity to data processing

The processing of the diapycnal flux profile was performedin various ways, with respect to the measurement methodfor K and the sequence of processing steps, in order to ex-plore the sensitivity of the result to processing decisions.The reference method to obtain diapycnal flux, as definedby us, was the calculation of local flux profiles from lo-cal KMSS/ADCP and local oxygen gradients in small (2 m)

0 50 100 150 200 250 300 350 4000.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2x 10

−5

number of stations that contribute to the <K> estimate

<K

> in

m2 /s

_

_

Fig. 6.Average diapycnal diffusivity〈K〉 in the analysis box and its95 % confidence limits when using random subsets of the availabledata. The full dataset comprises 400 stations of mixing data eitherfrom microstructure profiles or vessel-mounted ADCP. Processingof the subsets was identical to the full dataset.

intervals, then averaging local flux profiles isopycnally toan aggregated flux profile, and finally smoothing verti-cally (in a symbolic notation:〈KMSS/ADCP×∇ c〉smoothed).Other ways of processing were tested against this refer-ence. Swapping isopycnal averaging and flux calculation([〈KMSS/ADCP〉 × 〈∇ c〉]smoothed) has, as expected, little ef-fect sinceKMSS/ADCP and ∇ c are statistically indepen-dent in our dataset (Sect.3.7). Thus using 〈K〉TRE tocalculate diapycnal flux by[〈K〉TRE×〈∇ c〉]smoothedessen-tially differs from 〈KMSS/ADCP×∇ c〉smoothedby the factor〈K〉TRE / 〈K〉MSS/ADCP. The attempt to reduce measure-ment error by smoothing local oxygen gradient profiles be-fore both calculating flux and averaging isopycnally (i.e.〈K〉TRE×〈∇ csmoothed〉 vs. [〈K〉TRE×〈∇ c〉]smoothed) not onlyreduces noise but also systematically reduces the gradientand thus the estimated diapycnal flux (Fig.7).

4.4 Sensitivity to coordinate choice

It may be worthwhile asking what the diapycnal flux acrossother surfaces than isopycnals is, even though then diapy-cnal flux is no longer perpendicular to these surfaces andthere also exists an isopycnal flux component across thesesurfaces. Obvious coordinate choices other than isopycnalcould be isobaths or surfaces that are linked to features of theoxygen field. Such a different choice primarily changes thevertical shift with which profiles of oxygen gradient get ag-gregated in the process of averaging laterally (i.e. along thesurfaces appropriate for the chosen coordinate system). Weexamined two cases: (1) diapycnal flux across isobaths, and(2) diapycnal flux across a surface that connects the gradient



−6 −4 −2 0 2 4 6

x 10−3

26.5

26.6

26.7

26.8

26.9

27

27.1

27.2

in mol/m2/yr

θ

150

200

250

300

350

400

450

500

550

dept

h in

m

_

__< K • ∇ c >

< K • ∇ c >, 8th order polynomial

< K • ∇ c >, low pass filtered

_

< K > • < ∇ c >, 8th order polynomial

< K > • < ∇ c >, from smoothed c

_

Deep oxycline

Deep OMZ core

Fig. 7. Diapycnal oxygen flux in density coordinates as determinedby different methods. The given smoothed curves assume that fluc-tuations of smaller vertical scales are an artefact of imperfect sam-pling. Black dots with grey shading: local oxygen gradients over2 m combined with localK from microstructure/ADCP measure-ments to local diapycnal flux, then averaged laterally along isopyc-nals. Red line: corresponding fitted 8th-order polynomial. Blue line:corresponding low-pass-filtered flux profile. Dashed green line: 8th-order polynomial on flux profile that was derived from horizontallyaveraged oxygen gradient and integralK value from the tracer re-lease experiment. Dashed yellow line: flux profile derived from ver-tically smoothed oxygen gradient profiles that have been laterallyaveraged along isopycnals before being combined with the integralK value from the tracer release experiment. Dotted black lines markthe analysis depth interval 150 to 500 m for whichK estimates ex-ist.

maxima of the deep oxycline. For case (1) the resulting di-apycnal flux profile is similar in shape to the diapycnal fluxprofile of Fig.7, but flux values are reduced by about 30 %;the downward diapycnal flux at the deep oxycline is about3× 10−3 to 4× 10−3 µmol m−2 s−1. A coordinate choicelike in case (2) leads to the maximum possible estimate ofdownward diapycnal flux at the deep oxycline; here the re-sulting flux is about 6× 10−3 to7× 10−3 µmol m−2 s−1.

4.5 Diapycnal flux divergence and OMZ oxygen budget

Looking at the oxygen budget of the OMZ for individualisopycnal layers allows further insight into oxygen supplythan just quantifying the oxygen flux across particular sur-faces. The contribution to oxygen supply by diapycnal pro-cesses, i.e. the diapycnal surplus in units of moles oxygenper volume and time, is described by the negative diapyc-nal flux divergence (Sect.3.1). Of the processes contributing

−10 −8 −6 −4 −2 0 2 4 6 8 10

26.5

26.6

26.7

26.8

26.9

27

27.1

27.2

in µmol/kg/yr

Deep OMZ core

Deep oxycline

150

200

250

300

350

400

450

500

550

dept

h in

m

Consumption Diapycnal supplyResidual = Isopycnal supply

UCW

AA

IWLC

W

Fig. 8. Terms of the oxygen budget (diapycnal supply+ isopycnalsupply+ consumption rate = 0) as a function of density. Diapycnalsupply = negative diapycnal flux divergence (black) was determinedin this study (here we chose a smooth line compromising betweendivergences of the polynomial and the low-pass-smoothed fluxesfrom TRE and MSS, as in Fig.7). Oxygen consumption rate (green)was estimated according toKarstensen et al.(2008); the exponen-tial decrease in depth coordinates turns out to be hardly discerniblefrom linear when transformed to isopycnal coordinates in the depthinterval 150 to 500 m. The residual that closes the budget must ac-count for isopycnal supply (red). Uncertainties are 95 % confidencelimits. AAIW: Antarctic Intermediate Water; LCW: lower CentralWater; UCW: upper Central Water.

to diapycnal oxygen supply, we quantified diapycnal flux byturbulent mixing (Sect.4.2), found diapycnal flux by dou-ble diffusive convection to be negligible (Sect.4.1), and as-sume diapycnal flux by diapycnal advection to be negligi-ble in the analysed depth interval 150 m to 500 m. The lat-ter assumption is supported by diapycnal velocity estimatesfrom the measured temperature and salinity fields and theestimated〈K〉 following the conservation equations givenby McDougall (1991). The contribution to diapycnal supplyby diapycnal advection is found this way to be very smallcompared to the other processes (� 1 µmol kg−1 yr−1 at alldepths). Thus for the oxygen budget of the OMZ, we onlyconsider diapycnal oxygen supply by turbulent mixing.

Examining the vertical structure of the negative diapycnaldiffusive flux divergence in Fig.8, the maximum local di-apycnal supply is located at about the OMZ core, with about1.5 to 2 µmol kg−1 yr−1. Other distinguished locations in theflux divergence profile are the surfaces where local diapyc-nal supply is zero, which coincide with the locations of max-imum oxygen gradients above and below the OMZ core. Theupper Central Water above the deep oxycline loses oxygento the OMZ below by diapycnal mixing, but also upwardstowards the shallow oxygen minimum layer. Concerning the



other terms of the oxygen budget – consumption rate andisopycnal supply – a consumption rate estimate may be takenfrom the literature, while isopycnal supply is the differenceof consumption rate and diapycnal supply, assuming steadystate. Reported consumption rate profiles vary throughoutthe global ocean and are uncertain (Keeling et al., 2010),but their principal shape is typically assumed to be expo-nential with depth (Martin et al., 1987; Karstensen et al.,2008). Karstensen et al.(2008) estimated an exponentialconsumption rate profile specific for the Atlantic and Pa-cific OMZ regions from apparent oxygen utilization and esti-mated water ages. This profile is the only one reported specif-ically for the tropical Northeast Atlantic OMZ. It is usedhere with the consumption rate in µmol kg−1 yr−1 given as−0.5+ 12· exp(−0.0021· z) with z depth in metres. The un-certainty of the consumption rate was estimated to be 40 %(95 % confidence level) from the scatter of oxygen utiliza-tion rates reported byKarstensen et al.(2008). Thus the re-sulting uncertainty of the deduced profile of isopycnal supplyis large, but the main pattern seems clear: the profile of isopy-cnal oxygen supply resembles a step function, with largeisopycnal supply in the upper Central Water layer above thedeep oxycline, and little isopycnal supply in the OMZ layerbelow the deep oxycline. For a rough check of plausibility ofthe deduced isopycnal supply below the deep oxycline, onecan use an isopycnal diffusivity of 500 m2 s−1 (e.g.Banyteet al., 2013) and a typical curvature of the oxygen field onan isopycnal surface of 1×10−10 µmol kg−1 m−2, and obtainan estimate of 2 µmol kg−1 yr−1. Brandt et al.(2010) ob-tained consistent estimates for the isopycnal eddy supply, inan idealized framework of a mean latitudinally alternatingeast- and westward current system that connects the west-ern boundary regime with the OMZ. Depending on the as-sumed strength of the zonal currents, they derived an isopy-cnal eddy supply of 0.5 to 0.9 µmol kg−1 yr−1 when usingan isopycnalK of 200 m2 s−1, with the isopycnal advec-tive supply varying between 2 and 0 µmol kg−1 yr−1. All ofthese values are in the range of the expected isopycnal supplybelow the deep oxycline.

The assumption of steady state, which served to estimatethe isopycnal supply profile, is in fact violated, and we expecta small imbalance in the oxygen budget asStramma et al.(2008b) observed a long-term trend of -0.35 µmol kg−1 yr−1

in the depth interval 300 m to 700 m. However, this expectedimbalance is much smaller than the error margins of the bud-get terms in Fig.8, and thus could not be detected by observ-ing the recent oxygen budget.

The diapycnal contribution to the supply of the tropicalNorth Atlantic OMZ is maximum in the water layer be-tween isopycnals 26.95 and 27.05 kg m−3, i.e. surroundingthe OMZ core, and may be estimated to be about one-thirdof the demand, with large uncertainty.

5 Conclusions

The tracer release experiment and the MSS/ADCP mea-surements complement each other and deliver consis-tent estimates of diapycnal diffusivity. The two inde-pendent methods constrain diapycnal diffusivity for theOMZ in the depth range 150 m to 500 m to an inter-val of [0.8 1.4]×10−5 m2 s−1, or a single value of〈K〉 =1× 10−5 m2 s−1. Diapycnal diffusivity in the region is pre-dominantly caused by turbulence, such that〈K〉oxygen ≈

〈K〉solutes≈ 〈K〉ρ ≈ 〈K〉. Using the diapycnal diffusivity of1×10−5 m2 s−1 could serve to constrain models and possi-bly improve the oxygen representation in the tropical NorthAtlantic OMZ.

Diapycnal oxygen supply replenishes one-third of the oxy-gen demand in the OMZ core (between isopycnals 26.95 and27.05 kg m−3), as can be estimated from the vertical struc-ture of diapycnal oxygen supply and consumption (Fig.8).Thus diapycnal mixing is a major controlling process for theformation and maintenance of this OMZ and will have to betaken into account when exploring the controlling processesof other OMZs. The value “one-third” is potentially subjectto some adjustments when better estimates of isopycnal sup-ply and consumption rate become available. It will also besubject to some change in time when consumption and/orcirculation and/or mixing change. Rating the diapycnal oxy-gen supply as a major controlling process may seem debat-able: it might be equally valid to assign the oxygen supplyof the OMZ entirely to isopycnal processes. If choosing alayer bounded by isopycnal surfacesσθ = 26.63 kg m−3 and27.03 kg m−3 (each characterized by zero diapycnal flux), thesupply of the water column in between must be completelyisopycnal, and the role of diapycnal mixing would reduceto just generating internal oxygen redistribution. But even ifone adopts this latter perspective, the controlling effect of di-apycnal mixing on the OMZ structure remains substantial:if diapycnal mixing was less intense than observed, the oxy-gen distribution in the analysis box would certainly be an-other. Given the other parameters kept constant, the verticaloxygen gradient would most likely become sharper, and theoxygen concentration in the OMZ core would become lower.By the same argument we also conclude that the higher thanexpected level of diapycnal mixing in the study region (cf.Sect. 4.1) plays a substantial role in shaping the oxygenconcentration field and supposedly also in causing relativelyhigh levels of oxygen in the core of the tropical North At-lantic OMZ compared to other OMZs. The exact reasons forthe higher than expected level of diapycnal mixing will bethe object of further study.

All available information combined, particularly the verti-cal structure of diapycnal and isopycnal supply (Fig.8), thisstudy also advances insight into the pathways through whichoxygen is transported from abroad to the OMZ core. Besidesdirect isopycnal supply at all depths, there is an importantsupply branch which brings oxygen laterally within the upper



Central Water layer above the deep oxycline; from there oxy-gen spreads diapycnally towards the OMZ core. There is nonet oxygen contribution from the surface layer of the OMZregion because the presence of the shallow oxygen minimumcauses a layer of upward average oxygen flux which virtuallydecouples the OMZ oxygen budget from the surface.

The contribution of diapycnal mixing to supply the lowerpart of the deep OMZ (below the OMZ core) with oxy-gen from AAIW can hardly be quantified with the presentdata because most measurements did not reach the neces-sary depths. But there are indications that diapycnal mixing issubstantial here as well. Although gradient and curvature ofoxygen concentration are found to be smaller below the OMZcore than above the OMZ core, turbulent diffusivity valuesbelow the OMZ core are found equal or larger than above(Fig. 4), and turbulent diffusivity is enhanced by double dif-fusion here (adds 50 % if following the method ofSt. Laurentand Schmitt, 1999). The resulting total diapycnal oxygen fluxfrom below might even reach similar values to those above,while the oxygen consumption rate to be balanced is lowerthan above. Another TRE will elucidate the relative impor-tance of diapycnal and isopycnal supply for the layer at andbelow the OMZ core. This TRE (Oxygen Supply Tracer Re-lease Experiment, OSTRE) was launched in December 2012,and surveys are scheduled from June 2013 until 2015 and in-clude ADCP and microstructure profiles down to 800 m.

The uncertainty of the consumption profile is causing highuncertainty in the estimated profile of isopycnal supply (how-ever, some confidence to the magnitude of isopycnal supplyis inspired by the independent rough calculations of isopy-cnal oxygen supply given in Section4.5). Local profiles ofconsumption rate are still difficult to obtain (e.g.Keelinget al., 2010) but are important for evaluating the oxygen bud-get. Nevertheless, the general tendency of consumption rateto diminish with depth seems to be described well enough bythe fitted exponential ofKarstensen et al.(2008), and mod-erate deviations from that profile would not change the mainfeatures of the deduced isopycnal oxygen supply profile asof Fig. 8. In fact, some deviation from an exponential shapeseems even probable, e.g. caused by active organic mattertransport during diel vertical migration of zooplankton (An-gel, 1985; Steinberg et al., 2002). The local impact of thelatter process on the OMZ is unknown, but collected acous-tic data during the survey cruises showed strong diel mi-grant activity. The acoustic backscatter records suggest en-hanced oxygen consumption at about 300 m and diminishedoxygen consumption at about 200 m depth. This is a shapenot uncommon in reported consumption profiles for the Pa-cific (Feely et al., 2004), but on their reported level the effectwould be small: the jump in the isopycnal supply profile ofFig. 8 would shift slightly downwards, but the profile’s prin-cipal shape would not change.

What is the reason for the found sharp decline of isopyc-nal oxygen supply at the depth of the deep oxycline? Thiscontrast in isopycnal oxygen supply could basically origi-

10−10

10−9

10−8

10−10

10−9

10−8

measured

p

aram

etri

zed

m2/s3

m2/s3

Fig. 9. Predicted versus measured turbulent dissipation rates. Pre-dictions by parameterizationε = 1/100· f 5/9

· N19/9· G4/3 that

resulted from a two-parameter fit in91 = N2· G (vertical shear

power) and92 = f · N−1 (internal wave slope). Measurements bymicrostructure profiler MSS90D in the region of the tropical NorthAtlantic OMZ (black dots) and values reported byPolzin et al.(1995) performed with the High Resolution Profiler in differentmidlatitude locations (squares). Of the total predictions, 95 % liewithin a factor 2.5 from measured values.

nate in isopycnal eddy diffusion and/or isopycnal advection.However, several potential explanations for the sharp con-trast seem improbable: we do not expect lateral diffusivi-ties to experience a jump at the deep oxycline, nor do weexpect a vertical jump of proper sign in the second deriva-tive of isopycnal oxygen distribution (supported by WOA09climatological data;Garcia et al., 2010). This makes eddyprocesses an improbable reason for the sharp decline. Fur-thermore, we do not expect mean current velocities to jumpin magnitude or direction at the depth of the deep oxycline.Thus it is suggested that the main contribution to the shape ofisopycnal oxygen supply is due to advection of water carry-ing specific oxygen profiles when entering and leaving theOMZ region. Water of greater than average vertical oxy-gen gradient would enter the OMZ region, would undergodepth-dependent consumption and a diapycnal redistributionof oxygen inside the region, and would leave the region witha weaker than average vertical oxygen gradient. The OMZ re-gion would just act to reduce the oxygen gradient at the deepoxycline, and the source of water with a strong oxygen gra-dient would have to be sought outside the OMZ region. Thewestern boundary circulation characterized by strong verti-cal shear is a likely candidate for generating sharp verticaloxygen gradients that could be transported within eastwardcurrent bands into the OMZ.



Finally, this study can help planning and data processing insimilar studies by giving hints on the necessary measurementeffort to obtain〈K〉 estimates (Fig.6) and by making use ofthe found lateral independence ofK and vertical oxygen gra-dient. The latter feature allowed us to estimate the average di-apycnal oxygen flux in the OMZ region from the regional av-erage of diapycnal diffusivity and the regional average of theoxygen gradient profile (further assuming that these averagesare representative for the analysis box). If such an indepen-dence ofK and concentration gradients is also valid for otherregions and substances, regional diapycnal fluxes can be esti-mated from datasets which comprise non-simultaneous mea-surements.

Appendix A

Dissipation rate parameterized from vessel-mountedADCP data

There is a widely known parameterization of dissipation rateε for the open ocean (Polzin et al., 1995; Gregg et al., 2003)which is based on the concept of breaking internal wavescausing turbulence, dissipation, and mixing. Important pa-rameters needed to use this parameterization are finescalevertical shear and finescale vertical strain; “finescale” in thiscontext comprises vertical wavenumbers of the order of 10to order 100 m. In this study, we had access to vertical shearvia vmADCP data (Sect.3.6), but for strain we lacked simul-taneous data of satisfying quality. As a remedy we fitted an-other parameterization with fewer parameters, using our owndata from the tropical North Atlantic Ocean plus data frommidlatitudes, which were reported byPolzin et al.(1995) andformed the base of their parameterization. Here two variableswere used which have commonly been identified to influencemixing intensity in the open ocean: (1) the intensity of ve-locity vertical shear in the vertical wavelength band of tensto hundreds of metres as providing the forcing, and (2) thecharacteristic slope of internal waves, which influences theirprobability to break and cause turbulence, dissipation, andmixing. The predictors91 and 92 we defined for the pa-rameterization were derived from these two variables, hav-ing been inspired by the Garrett–Munk (GM) internal wavemodel (Garrett and Munk, 1975; Munk, 1981). 91 was cho-sen proportional to the power spectral level8S of verticalfinescale shear as follows:

91 = N2· G, (A1)

with N buoyancy frequency andG a nondimensional numberdescribing the power spectral level of shear,8S, in relationto its GM background level8S,GM = const. · N2. G was es-timated from vmADCP data by

G =8S

kmin...kmax

8kmin...kmaxS,GM

, (A2)

and both 8Skmin...kmax and 8

kmin...kmaxS,GM were calculated

in the vertical wavenumber rangekmin = 1/128 cpm tokmax = 1/38.4 cpm by weighted averaging as6((9− i)/8 ·

8S,i)/6((9− i)/8) for the appropriate8S,i at wavenumberski = (2+ i)/384 cpm. This way to estimateG is specific tothe used vmADCP configuration and the used depth intervalfor the analysis of 150 to 500 m. ButG is not expected to besensitive to the particular choice ofki in the finescale range.92 was chosen as follows:

92 =f

N, (A3)

with f the Coriolis parameter because the characteristicslope of the internal wave field approximately is a power off · N−1 when assuming the GM model. In order to fit a re-gression model we took the existing 45 stations with simul-taneous microstructure profiles, CTD profile, and vmADCPmeasurements, plus data from the High Resolution Pro-filer collected at several midlatitude locations as reported byPolzin et al.(1995). Predictor91 relied on vmADCP datafor G and CTD data forN2, predictor92 needed latitudeand CTD data forN , and predictandε was provided by mi-crostructure data. In order to avoid a skewed distribution inparameter space, we performed linear regression in logarith-mic space, i.e. the regression model was logε = a0 +a1· log91+a2· log 92. The best model fit (Fig.9) can be written asfollows:

ε =1

100· 9

4/31 · 9

5/92 , (A4)

with multiple correlationR2= 0.80. This relation translates

to

ε =1

100· f 5/9

· N19/9· G4/3. (A5)

Acknowledgements.This study benefitted from cruises, infrastruc-ture, and financial support by the German Federal Ministry of Ed-ucation and Research through the cooperative projects SOPRANand NORTH ATLANTIC, and by the German Science Founda-tion’s Sonderforschungsbereich SFB754 “Climate BiogeochemistryInteractions in the Tropical Ocean”. We acknowledge the supportof the European Commission (FP7-EuroSITES grant agreementno. 202955). The support of captains, crews, and scientific crews ofMeteorcruise 80,Meteorcruise 83/1,Merian cruise 08/1,Meriancruise 10/1, andL’Atalante cruises GEOMAR/3 and GEOMAR/4is highly appreciated. We thank Johannes Karstensen for fruitfuldiscussions. We thank two anonymous reviewers, who helped toimprove the manuscript.

The service charges for this open access publicationhave been covered by a Research Centre of theHelmholtz Association.

Edited by: B. Dewitte



References

Angel, M. V.: Vertical migrations in the oceanic realm: possiblecauses and probable effects, in: Migration: Mechanisms andAdaptive Significance, Contributions in Marine Science, suppl.Vol. 27, edited by: Rankin, M. A., Checkley, D., Cullen, J., Kit-ting, C., and Thomas, P., Marine Science Institute, Austin, Texas,45–70, 1985.

Banyte, D., Tanhua, T., Visbeck, M., Wallace, D. W. R.,Karstensen, J., Krahmann, G., Schneider, A., Stramma, L., andDengler, M.: Diapycnal diffusivity at the upper boundary of thetropical North Atlantic oxygen minimum zone, J. Geophys. Res.,117, C09016, doi:10.1029/2011JC007762, 2012.

Banyte, D., Visbeck, M., Tanhua, T., Fischer, T., Krahmann, G.,and Karstensen, J.: Lateral diffusivity from tracer release exper-iments in the tropical North Atlantic thermocline, J. Geophys.Res. Oceans, 118, doi:10.1002/jgrc.20211, 2013.

Bograd, S. J., Castro, C. G., Di Lorenzo, E., Palacios, D. M., Bai-ley, H., Gilly, W., and Chavez, F. P.: Oxygen declines and theshoaling of the hypoxic boundary in the California Current, Geo-phys. Res. Lett., 35, L12607, doi:10.1029/2008GL034185, 2008.

Bopp, L., Monfray, P., Aumont, O., Dufresne, J.-L., Le Treut, H.,Madec, G., Terray, L., and Orr, J. C.: Potential impact of climatechange on marine export production, Global Biogeochem. Cy.,15, 81–99, 2001.

Bopp, L., Le Quere, C., Heimann, M., and Manning, A. C.:Climate-induced oceanic oxygen fluxes: Implications for thecontemporary carbon budget, Global Biogeochem. Cy., 16, 1022,doi:10.1029/2001GB001445, 2002.

Brandt, P., Hormann, V., Kortzinger, A., Visbeck, M., Krah-mann, G., Stramma, L., Lumpkin, R., and Schmid, C.: Changesin the Ventilation of the Oxygen Minimum Zone of theTropical North Atlantic, J. Phys. Oceanogr., 40, 1784–1801,doi:10.1175/2010JPO4301.1, 2010.

Davis, R. E.: Sampling turbulent dissipation, J. Phys. Oceanogr., 26,341–358, 1996.

Diaz, R. J.: Overview of Hypoxia around the World, J. Environ.Qual., 30, 275–281, 2001.

Duteil, O. and Oschlies, A.: Sensitivity of simulated extent and fu-ture evolution of marine suboxia to mixing intensity, Geophys.Res. Lett., 38, L06607, doi:10.1029/2011GL046877, 2011.

Ekau, W., Auel, H., Portner, H.-O., and Gilbert, D.: Impacts ofhypoxia on the structure and processes in pelagic communities(zooplankton, macro-invertebrates and fish), Biogeosciences, 7,1669–1699, doi:10.5194/bg-7-1669-2010, 2010.

Feely, R. A., Sabine, C. L., Schlitzer, R., Bullister, J. L., Meck-ing, S., and Greeley, D.: Oxygen utilization and organic car-bon remineralization in the upper water column of the PacificOcean, J. Oceanogr., 60, 45–52, 2004.

Ferrari, R. and Polzin, K. L.: Finescale Structure of the T-S Relationin the Eastern North Atlantic, J. Phys. Oceanogr., 35, 1437–1454,2005.

Fischer, T.: Diapycnal diffusivity and transport of matter in theopen ocean estimated from underway acoustic profiling and mi-crostructure profiling, Ph.D. thesis, Leibniz Institute of MarineSciences (GEOMAR), University of Kiel, Germany, 105 pp.,2011.

Garcia, H. E., Locarnini, R. A., Boyer, T. P., Antonov, J. I., Bara-nova, O. K., Zweng, M. M., and Johnson, D. R.: World OceanAtlas 2009 Volume 3 (Dissolved Oxygen, Apparent Oxygen

Utilization, and Oxygen Saturation), NOAA Atlas NESDIS 70,edited by: Levitus, S., US Government Printing Office, Washing-ton, DC, 344 pp., 2010.

Gargett, A. E.: An investigation of the occurrence of oceanic turbu-lence with respect to finestructure, J. Phys. Oceanogr., 6, 139–156, 1976.

Garrett, C. and Munk, W.: Space-time scales of internal waves:A progress report, J. Geophys. Res., 80, 291–297, 1975.

Gregg, M. C.: Scaling turbulent dissipation in the thermocline, J.Geophys. Res., 94, 9686–9698, 1989.

Gregg, M., Sanford, T. B., and Winkel, D. P.: Reduced mixing fromthe breaking of internal waves in equatorial waters, Nature, 422,513–515, 2003.

Gruber, N.: Warming up, turning sour, losing breath: ocean bio-geochemistry under global change, Philos. T. Roy. Soc. A, 369,1980–1996, doi:10.1098/rsta.2011.0003, 2011.

Hales, B., Moum, J. N., Covert, P., and Perlin, A.: Irreversible ni-trate flux due to turbulent mixing in a coastal upwelling system, J.Geophys. Res., 110, C10S11, doi:10.1029/2004JC002685, 2005.

Hales, B., Hebert, D., and Marra, J.: Turbulent supply of nutrients tophytoplankton at the New England shelf break front, J. Geophys.Res., 114, C05010, doi:10.1029/2008JC005011, 2009.

Helm, K. P., Bindoff, N. L., and Church, J. A.: Observed decreasesin oxygen content of the global ocean, Geophys. Res. Lett., 38,L23602, doi:10.1029/2011GL049513, 2011.

Ho, D. T., Ledwell, J. R., and Smethie Jr., W. M.: Use of SF5CF3for ocean tracer release experiments, Geophys. Res. Lett., 35,L04602, doi:10.1029/2007GL032799, 2008.

Karstensen, J., Stramma, L., and Visbeck, M.: Oxygen minimumzones in the eastern tropical Atlantic and Pacific oceans, Prog.Oceanogr., 77, 331–350, doi:10.1016/j.pocean.2007.05.009,2008.

Keeling, R. F., Kortzinger, A., and Gruber, N.: Ocean deoxygena-tion in a warming world, Annu. Rev. Mar. Sci., 2, 199–229,doi:10.1146/annurev.marine.010908.163855, 2010.

Kirchner, K., Rhein, M., Huttl-Kabus, S., and Boning, C. W.: On thespreading of South Atlantic Water into the Northern Hemisphere,J. Geophys. Res., 114, C05019, doi:10.1029/2008JC005165,2009.

Kock, A., Schafstall, J., Dengler, M., Brandt, P., and Bange, H. W.:Sea-to-air and diapycnal nitrous oxide fluxes in the easterntropical North Atlantic Ocean, Biogeosciences, 9, 957–964,doi:10.5194/bg-9-957-2012, 2012.

Kunze, E., Firing, E., Hummon, J. M., Chereskin, T. K., and Thurn-herr, A. M.: Global abyssal mixing inferred from lowered ADCPshear and CTD strain profiles, J. Phys. Oceanogr., 36, 1553–1576, 2006.

Lam, P. and Kuypers, M. M. M.: Microbial nitrogen cycling pro-cesses in oxygen minimum zones, Annu. Rev. Mar. Sci., 3, 317–345, doi:10.1146/annurev-marine-120709-142814, 2011.

Martin, J. H., Knauer, G. A., Karl, D. M., and Broenkow, W. W.:VERTEX: carbon cycling in the northeast Pacific, Deep-SeaRes., 34, 267–285, 1987.

Matear, R. J. and Hirst, A. C.: Long-term changes in dis-solved oxygen concentrations in the ocean caused by pro-tracted global warming, Global Biogeochem. Cy., 17, 1125,doi:10.1029/2002GB001997, 2003.

McDougall, T. J.:Water Mass Analysis with Three ConservativeVariables, J. Geophys. Res., 96, 8687–8693, 1991.


http://dx.doi.org/10.1029/2011JC007762

http://dx.doi.org/10.1002/jgrc.20211

http://dx.doi.org/10.1029/2008GL034185

http://dx.doi.org/10.1029/2001GB001445

http://dx.doi.org/10.1175/2010JPO4301.1

http://dx.doi.org/10.1029/2011GL046877

http://dx.doi.org/10.5194/bg-7-1669-2010

http://dx.doi.org/10.1098/rsta.2011.0003

http://dx.doi.org/10.1029/2004JC002685

http://dx.doi.org/10.1029/2008JC005011

http://dx.doi.org/10.1029/2011GL049513

http://dx.doi.org/10.1029/2007GL032799

http://dx.doi.org/10.1016/j.pocean.2007.05.009

http://dx.doi.org/10.1146/annurev.marine.010908.163855

http://dx.doi.org/10.1029/2008JC005165

http://dx.doi.org/10.5194/bg-9-957-2012

http://dx.doi.org/10.1146/annurev-marine-120709-142814

http://dx.doi.org/10.1029/2002GB001997


Meissner, K. J., Galbraith, E. D., and Volker, C.: Denitrification un-der glacial and interglacial conditions: A physical approach, Pa-leoceanography, 20, PA3001, doi:10.1029/2004PA001083, 2005.

Munk, W. H.: Internal waves and small-scale processes, in: Evolu-tion of Physical Oceanography; Scientific Surveys in Honor ofHenry Stommel, edited by: Warren, B. A. and Wunsch, C., MITpress, Cambridge and London, 264–291, 1981.

Najjar, R. G., Jin, X., Louanchi, F., Aumont, O., Caldeira, K.,Doney, S. C., Dutay, J.-C., Follows, M., Gruber, N., Joos, F.,Lindsay, K., Maier-Reimer, E., Matear, R. J., Matsumoto, K.,Monfray, P., Mouchet, A., Orr, J. C., Plattner, G.-K.,Sarmiento, J. L., Schlitzer, R., Slater, R. D., Weirig, M.-F., Ya-manaka, Y., and Yool, A.: Impact of circulation on export pro-duction, dissolved organic matter, and dissolved oxygen in theocean: results from phase II of the Ocean Carbon-cycle ModelIntercomparison Project (OCMIP-2), Global Biogeochem. Cy.,21, GB3007, doi:10.1029/2006GB002857, 2007.

Oakey, N. S.: Determination of the rate of dissipation of turbu-lent energy from simultaneous temperature and velocity shearmicrostructure measurements, J. Phys. Oceanogr., 12, 256–271,1982.

Osborn, T. R.: Estimates of the local rat of vertical diffusion fromdissipation measurements, J. Phys. Oceanogr., 10, 83–89, 1980.

Oschlies, A., Schulz, K. G., Riebesell, U., and Schmittner, A.:Simulated 21st century’s increase in oceanic suboxia by CO2-enhanced biotic carbon export, Global Biogeochem. Cy., 22,GB4008, doi:10.1029/2007GB003147, 2008.

Paulmier, A. and Ruiz-Pino, D.: Oxygen minimum zones(OMZs) in the modern ocean, Prog. Oceanogr., 80, 113–128,doi:10.1016/j.pocean.2008.08.001, 2009.

Polzin, K. L., Toole, J. M., and Schmitt, R. W.: Finescale parame-terizations of turbulent dissipation, J. Phys. Oceanogr., 25, 306–328, 1995.

Polzin, K., Kunze, E., Hummon, J., and Firing, E.: The finescaleresponse of lowered ADCP velocity profiles, J. Atmos. Ocean.Tech., 19, 205–224, 2002.

Prince, E. D., Luo, J., Goodyear, C. P., Hoolihan, J. P., Snod-grass, D., Orbesen, E. S., Serafy, J. E., Ortiz, M., andSchirripa, M. J.: Ocean scale hypoxia-based habitat compressionof atlantic istiophorid billfishes, Fish. Oceanogr., 19, 448–462,doi:10.1111/j.1365-2419.2010.00556.x, 2010.

Rippeth, T. P., Wiles, P., Palmer, M. R., Sharples, J., and Tweddle, J.:The diapycnal nutrient flux and shear-induced diapycnal mixingin the seasonally stratified western Irish Sea, Continental ShelfRes., 29, 1580–1587, doi:10.1016/j.csr.2009.04.009, 2009.

Schafstall, J., Dengler, M., Brandt, P., and Bange, H. W.: Tidal-induced mixing and diapycnal nutrient fluxes in the Mau-ritanian upwelling region, J. Geophys. Res., 115, C10014,doi:10.1029/2009JC005940, 2010.

Sharples, J., Moore, C. M., and Abraham, E. R.: Internal tide dissi-pation, mixing, and vertical nitrate flux at the shelf edge of NENew Zealand, J. Geophys. Res., 106, 14069–14081, 2001.

Sharples, J., Tweddle, J. F., Green, J. A. M., Palmer, M. R., Kim, Y.-N., Hickman, A. E., Holligan, P. M., Moore, C. M., Rippeth, T. P.,Simpson, J. H., and Krivtsov, V.: Spring-neap modulation of in-ternal tide mixing and vertical nitrate fluxes at a shelf edge insummer, Limnol. Oceanogr., 52, 1735–1747, 2007.

Shih, L. H., Koseff, J. R., Ivey, G. N., and Ferziger, J. H.: Parameter-ization of turbulent fluxes and scales using homogeneous shearedstably stratified turbulence simulations, J. Fluid Mech., 525, 193–214, 2005.

Siedler, G., Zangenberg, N., and Onken, R.: Seasonal changes inthe tropical Atlantic circulation: observation and simulation ofthe guinea dome, J. Geophys. Res., 97, 703–715, 1992.

Staquet, C. and Sommeria, J.: Internal gravity waves: from instabil-ities to turbulence, Annu. Rev. Fluid Mech., 34, 559–593, 2002.

Steinberg, D. K., Goldthwait, S. A., and Hansell, D. A.: Zooplank-ton vertical migration and the active transport of dissolved or-ganic and inorganic nitrogen in the Sargasso Sea, Deep-Sea Res.Pt. I, 49, 1445–1461, 2002.

Stips, A. and Prandke, H.: Recommended Algorithm for DissipationRate calculation within PROVESS, European Commission, JRC,Space Applications Institute, Technical Note No. 1.00.116, avail-able at:www.pol.ac.uk/provess/bodc/doc/dissialgo.pdf(last ac-cess: 10 October 2012), 2000.

St. Laurent, L. and Schmitt, R. W.: The contribution of salt fingersto vertical mixing in the North Atlantic Tracer Release Experi-ment, J. Phys. Oceanogr., 29, 1404–1424, 1999.

Stramma, L., Brandt, P., Schafstall, J., Schott, F., Fischer, J., andKortzinger, A.: Oxygen minimum zone in the North Atlanticsouth and east of the Cape Verde islands, J. Geophys. Res., 113,C04014, doi:10.1029/2007JC004369, 2008a.

Stramma, L., Johnson, G. C., Sprintall, J., and Mohrholz, V.: Ex-panding Oxygen-Minimum Zones in the Tropical Oceans, Sci-ence, 320, 655–658, doi:10.1126/science.1153847, 2008b.

Stramma, L., Visbeck, M., Brandt, P., Tanhua, T., and Wal-lace, D. W. R.: Deoxygenation in the oxygen minimum zoneof the eastern tropical North Atlantic, Geophys. Res. Lett., 36,L20607, doi:10.1029/2009GL039593, 2009.

Stramma, L., Schmidtko, S., Levin, L. A., and Johnson, G. C.:Ocean oxygen minima expansions and their biological impacts,Deep-Sea Res. Pt. I, 57, 587–595, doi:10.1016/j.dsr.2010.01.005,2010.

Vaquer-Sunyer, R. and Duarte, C. M.: Thresholds of hypoxia formarine biodiversity, P. Natl. Acad. Sci. USA, 105, 15452–15457,doi:10.1073/pnas.0803833105, 2008.

Whitney, F. A., Freeland, H. J., and Robert, M.: Persis-tently declining oxygen levels in the interior waters ofthe eastern subarctic Pacific, Prog. Oceanogr., 75, 179–199,doi:10.1016/j.pocean.2007.08.007, 2007.

Wright, J. J., Konwar, K. M., and Hallam, S. J.: Microbial ecologyof expanding oxygen minimum zones, Nat. Rev. Microbiol., Ad-vance Online Publications, doi:10.1038/nrmicro2778, 2012.

Wyrtki, K.: The oxygen minima in relation to ocean circulation,Deep-Sea Res., 9, 11–23, 1962.


http://dx.doi.org/10.1029/2004PA001083

http://dx.doi.org/10.1029/2006GB002857

http://dx.doi.org/10.1029/2007GB003147


http://dx.doi.org/10.1111/j.1365-2419.2010.00556.x

http://dx.doi.org/10.1016/j.csr.2009.04.009

http://dx.doi.org/10.1029/2009JC005940

www.pol.ac.uk/provess/bodc/doc/dissi_algo.pdf

http://dx.doi.org/10.1029/2007JC004369

http://dx.doi.org/10.1126/science.1153847

http://dx.doi.org/10.1029/2009GL039593

http://dx.doi.org/10.1016/j.dsr.2010.01.005

http://dx.doi.org/10.1073/pnas.0803833105


http://dx.doi.org/10.1038/nrmicro2778

Documents

file - Biogeosciences