Southern Ocean Gas Exchange Experiment: Setting … Ocean Gas Exchange Experiment: Setting the stage...

Southern Ocean Gas Exchange Experiment: Setting the stage

David T. Ho,1 Christopher L. Sabine,2 David Hebert,3,4 David S. Ullman,3

Rik Wanninkhof,5 Roberta C. Hamme,6 Peter G. Strutton,7 Burke Hales,8

James B. Edson,9 and Bruce R. Hargreaves10

Received 1 December 2010; revised 27 April 2011; accepted 14 July 2011; published 14 October 2011.

[1] The Southern Ocean Gas Exchange Experiment (SO GasEx) is the third in a series ofU.S.led open ocean process studies aimed at improving the quantification of gas transfervelocities and airsea CO2 fluxes. Two deliberate

3He/SF6 tracer releases into relativelystable water masses selected for large DpCO2 took place in the southwest Atlantic sectorof the Southern Ocean in austral fall of 2008. The tracer patches were sampled in aLagrangian manner, using observations from discrete CTD/Rosette casts, continuoussurface ocean and atmospheric monitoring, and autonomous drifting instruments to studythe evolution of chemical and biological properties over the course of the experiment. CO2and DMS fluxes were directly measured in the marine air boundary layer withmicrometeorological techniques, and physical, chemical, and biological processescontrolling airsea fluxes were quantified with measurements in the upper ocean andmarine air. Average wind speeds of 9 m s1 to a maximum of 16 m s1 were encounteredduring the tracer patch observations, providing additional data to constrain wind speed/gasexchange parameterizations. In this paper, we set the stage for the experiment by detailingthe hydrographic observations during the site surveys and tracer patch occupations thatform the underpinning of observations presented in the SO GasEx special section.Particular consideration is given to the mixed layer depth as this is a critical variable forestimates of fluxes and biogeochemical transformations based on mixed layer budgets.

Citation: Ho, D. T., C. L. Sabine, D. Hebert, D. S. Ullman, R. Wanninkhof, R. C. Hamme, P. G. Strutton, B. Hales, J. B. Edson,and B. R. Hargreaves (2011), Southern Ocean Gas Exchange Experiment: Setting the stage, J. Geophys. Res., 116, C00F08,doi:10.1029/2010JC006852.

1. Introduction

[2] The Southern Ocean Gas Exchange Experiment(SO GasEx) was an open ocean process study aimed atimproving the quantification of gas transfer velocities and

airsea CO2 fluxes under the high wind and rough seaconditions frequently found in that region. Previous GasExstudies were conducted in the North Atlantic (GasEx98)[Feely et al., 2002] and in the Equatorial Pacific (GasEx2001) [Sabine et al., 2004]. The GasEx98 study was oneof the first dedicated 3He/SF6 dual tracer experiments inthe open ocean, and resulted in the first robust micrometeorological measurements of CO2 fluxes in the openocean [McGillis et al., 2001; McGillis et al., 2004]. SOGasEx sought to build on insights from and techniques honedin the previous GasEx studies to explore the biogeo-chemical and physical controls on airsea gas exchange athigh winds (in excess of 10 m s1) in a globally significantCO2 flux region.[3] One of the goals of these GasEx studies was to be able

to quantify gas transfer velocities on regional scales fromremote sensing such that, when combined with DpCO2,regional airsea CO2 fluxes can be determined. A systematicapproach to accomplish this goal involved the followingsteps: (1) Make micrometeorological (or direct) flux mea-surements in the field to obtain shortterm local CO2 fluxes/gas transfer velocities; (2) reconcile the direct CO2 fluxmeasurements with integrated measurements of gas transfervelocities using the 3He/SF6 dual tracer technique and watercolumn mass balance estimates; (3) understand the mechan-

1Department of Oceanography, University of Hawaii at Manoa,Honolulu, Hawaii, USA.

2NOAA Pacific Marine Environmental Laboratory, Seattle,Washington, USA.

3Graduate School of Oceanography, University of Rhode Island,Narragansett, Rhode Island, USA.

4Now at Fisheries and Oceans Canada, Bedford Institute ofOceanography, Dartmouth, Nova Scotia, Canada.

5NOAA Atlantic Oceanographic and Meteorological Laboratory,Miami, Florida, USA.

6School of Earth and Ocean Sciences, University of Victoria, Victoria,British Columbia, Canada.

7Institute for Marine and Antarctic Studies, University of Tasmania,Hobart, Tasmania, Australia.

8College of Oceanic and Atmospheric Sciences, Oregon StateUniversity, Corvallis, Oregon, USA.

9Department of Marine Sciences, University of Connecticut, Groton,Connecticut, USA.

10Department of Earth and Environmental Sciences, Lehigh University,Bethlehem, Pennsylvania, USA.

Copyright 2011 by the American Geophysical Union.01480227/11/2010JC006852

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 116, C00F08, doi:10.1029/2010JC006852, 2011

C00F08 1 of 19

http://dx.doi.org/10.1029/2010JC006852

isms controlling ocean mixed layer pCO2 on short time andspace scales; (4) elucidate the forcing functions controllinggas transfer; and (5) relate forcing functions to parameters thatcan be retrieved by remote sensing.[4] SO GasEx took place on the National Oceanic and

Atmospheric Administration (NOAA) Ship Ronald H.Brown from 29 February to 12 April 2008 (yeardays 60 to103), with 31 scientists representing 22 academic institu-tions and government laboratories. The experiment wasbased around two deliberate tracer releases. The first tracerpatch was created on March 8 (yearday 68) and studied forapproximately 6 days before the ship was diverted 560 kmto the south of the study site, near South Georgia Island, fora storm to pass. The ship returned to the study area four dayslater where some remnants of the tracer patch were locatednear one of two drifting buoys deployed when the patch wasfirst created. After collecting a final set of water columnsamples and recovering the two buoys, the ship was relocatedto the northwest, close to the area where the first patch wasstarted. A second tracer patch was created on March 21(yearday 81) and studied for approximately 15 days beforebreaking off the experiment and transiting to Montevideo,Uruguay.[5] The scientific work during SO GasEx concentrated on

quantifying gas transfer velocities using deliberately injectedtracers (3He/SF6), measuring CO2 and DMS fluxes directlyin the marine air boundary layer, and elucidating the phys-ical, chemical, and biological processes controlling airseafluxes with measurements in the upper ocean and marine air.The oceanic studies used a Lagrangian approach to study theevolution of chemical and biological properties over thecourse of the experiment using shipboard and autonomousdrifting instruments. The categories of different researchprojects performed during SO GasEx are listed in Table 1.[6] The specific research objectives for SO GasEx were to

answer the following questions:[7] 1. What are the gas transfer velocities at high winds?[8] 2. What is the effect of fetch on the gas transfer?[9] 3. How do effects indirectly related to wind influence

gas transfer?[10] 4. How does variability in pCO2 and in DMS levels

affect the airsea CO2 and DMS fluxes?

[11] 5. What is the near surface horizontal and verticalvariability in turbulence, pCO2, and other relevant bio-chemical and physical parameters, and how are they relatedto one another?[12] 6. How do biological processes influence pCO2 and

gas exchange?[13] 7. Do fluxes estimated by different approaches agree,

and if not why?[14] In this special section, there are contributions focused

on different objectives of SO GasEx. Ho et al. [2011],Edson et al. [2011] and Yang et al. [2011] address questionsof gas transfer, and factors that affect CO2 and DMS fluxes.Balch et al. [2011] and Moore et al. [2011] focus onunderstanding the contribution of various factors to carboncycling in the Lagrangian tracer patches. Del Castillo andMiller [2011] examine the colored dissolved organic mat-ter dynamics during SO GasEx. Dwivedi et al. [2011] use anocean circulation model along with data assimilation of insitu and remote sensing data to simulate the ocean stateduring SO GasEx. V. P. Lance et al. (Primary productivity,new productivity and carbon export during two SouthernOcean Gas Exchange (SO GasEx) tracer experiments,unpublished manuscript, 2011), Lee et al. [2011], and R. C.Hamme et al. (Dissolved O2/Ar and other methods revealrapid changes in productivity during a Lagrangian experi-ment in the Southern Ocean, submitted to Journal of Geo-physical Research, 2011) quantify productivity during SOGasEx.

2. Study Site

2.1. Selection Criteria

[15] SO GasEx took place in the southwest Atlantic sectorof the Southern Ocean (nominally at 50S, 40W), nearSouth Georgia Island (Figure 1). The location was chosen,based on inspection of available satellite and in situ data(Figures 2 to 12) to fit the following criteria: (1) an airwaterpartial pressure gradient of CO2,DpCO2, of at least 40 matmto ensure a large enough signaltonoise for direct covari-ance measurements of CO2 fluxes; (2) area with a relativelystable water mass (i.e., relatively weak currents and lowmesoscale eddy variability) and a mixed layer depth less

Table 1. Categories of Research Projects on SO GasEx

Research Projects Method

Direct Flux Measurements (CO2, ozone and DMS) Airsea CO2 (NDIR), Ozone and DMS (APIMS) flux systemsBulk Meteorology and Turbulent Fluxes (winds, momentum,water vapor, temp, IR, Solar radiation, etc.)

Sonic anemometer, thermometer, pyranometer, pyrgeometer,MicroSAS

Integrated Gas Transfer Velocities with Deliberate Tracers(SF6 and

3He)Continuous and discrete SF6 systems (GCs) and He isotope mass spec

Surface and Subsurface variability (CO2, nutrients, calcite,DMS, chlorophyll)

Shipboard underway systems, NDIR CO2 systems, GC, EcoVSF,ICPOES, fluorometer, ACS, ISUS, SuperSoar/TOMASI

Autonomous Platforms MAPCO2, SAMI, ASIS, surface drifters, SOLO floatsSurface and nearsurface ocean processes (wave spectra,white capping, currents)

Shipboard radar; microwave altimeter, video camera, ADCP

Water column hydrography, carbon and related tracers(DIC, pCO2, TAlk, temp, sal, O2, nutrients, DOC, CDOM,PIC, O2/Ar, DMS, particles, TSM, Chl., POC)

SOMMA, NDIR, titration, CTD, Winkler, nutrient autoanalyzer,spectrophotometer, mass spec., GC, HPLC, Chl andCDOM fluorometers

Productivity rates 14C and 15N incubations, O2/Ar, O2 isotopes, O2 and pCO2 sensors,photosynthesisirradiance experiments

Ocean Optics PAR sensor, FRRF, IOP cage, HTSRB, MVSM

HO ET AL.: SOUTHERN OCEAN GAS EXCHANGE EXPERIMENT C00F08C00F08

2 of 19

than 5070 m to allow the 3He/SF6 patch to be followed forup to 3 weeks; (3) relatively high wind speeds, long fetch,and large waves; and (4) close to the ports of call (PuntaArenas, Chile and Montevideo, Uruguay) to minimizetransit time. While the Pacific sector of the Southern Oceanwould have been better in terms of winds, fetch and swell,the DpCO2 and logistical criteria ultimately led us to theAtlantic sector.

2.2. Site Survey

[16] After arriving at the study site, we conducted anunderway survey for 48 h. The main in situ parametersexamined during the site surveys were: waterside pCO2 toensure a large enough DpCO2, along with productivitymeasurements from O2/Ar, and chlorophyll measurementsfrom a fluorometer; expendable bathythermograph (XBT)temperature profiles (Figures 7 and 9) to determine thedepth and variability of the mixed layer; and shipboardacoustic Doppler current profiler (ADCP) measurements ofthe currents to ensure a region with relatively low currentsand velocity shears. In addition to the in situ measurements,we also tried to ensure that we did not inject the tracer on theedge of an eddy by examining tracks of drifters deployed inthe SO GasEx study site, remote sensing images of oceancolor, sea surface temperature, and sea surface height.

2.3. Surface Underway Surveys

[17] Underway pCO2 was measured using a shower typeequilibrator coupled to an IR analyzer with procedures anddata reduction as described by Pierrot et al. [2009]. Netcommunity production (NCP) was estimated from under-way measurements of O2/Ar made with an equilibrator inletmass spectrometer [Cassar et al., 2009], using a mass balanceapproach and partitioning between airsea gas exchange andbiological O2 production [Reuer et al., 2007]. Underwaychlorophyll fluorescence (Fchl; measured with a TurnerDesigns Cyclops7 sensor with wiper, attached to a C6instrument) and incident photosynthetically active radiation(PAR) (Lance et al., unpublished manuscript, 2011) were

Figure 2. Climatological maps of wind speed, sea surface temperature (SST), chlorophyll (Chl) and theairsea pCO2 difference (DpCO2) for the month of March. The wind climatology is based on QuikSCATdata from 2000 through 2009; SST data are from the Advanced Microwave Scanning RadiometerEOS(AMSRE), 2003 through 2009; Chl data are from the Seaviewing Wide Fieldofview Sensor (Sea-WiFS), 1998 through 2010; and the pCO2 data are from the Takahashi 940 k database (the version thatwas available to us prior to the cruise).

Figure 1. Map showing the cruise tracks at the SO GasExstudy site near South Georgia Island, along with the locationsof the two ports of call, Punta Arenas, Chile, and Montevideo,Uruguay.


3 of 19

Figure 3. Preinjection underway surface survey for Patch 1 showing (a) ocean minus atmosphere pCO2,DpCO2, (b) Net Community Production (NCP) from O2/Ar measurements, (c) chlorophylla concentrationfrom fluorometry corrected for daytime quenching, (d) salinity, (e) temperature, and (f) mixed layer kineticenergy. Black circle shows injection location. The black line denotes the location of the Lamont PumpingSeaSoar survey (Figures 1012). Color scales for NCP and chlorophyll have been truncated to better showvariability at lower values.

Figure 4. Preinjection underway surface survey for Patch 2 showing (a) ocean minus atmosphereDpCO2, (b) NCP from O2/Ar measurements, (c) chlorophylla concentration from fluorometry correctedfor daytime quenching, (d) salinity, (e) temperature, and (f) mixed layer kinetic energy. Black circleshows injection location. Color scales for Figures 4a5f are the same as Figure 3.


4 of 19

collected continuously (110 s intervals for Fchl). IncidentPAR data were used to adjust daytime Fchl (suppressed bynonphotochemical quenching) to match nighttime averagevalues (dark Fchl) each day, and periodic measurements ofchlorophylla concentration (Lance et al., unpublishedmanuscript, 2011) from the ships seawater system were usedto calibrate dark Fchl in units of mg chla m

3. Temperatureand salinity were measured with a SeaBird SBE 21 at theships intake, and by a SeaBird SBE 45 as part of otherunderway systems in the main lab and calibrated againstthe CTD/Rosette surface temperatures and salinities. Sali-nities in this paper are expressed on the practical salinityscale (PSS78).[18] Mixed layer depths were calculated from XBT

deployments during the underway surveys (Figures 6 to 9).The mixed layer was defined as the depth at which temper-ature was 0.1C less than the mean temperature between 10and 20 m. The 10 to 20 m interval was chosen becausetemperatures shallower than 10 m were not accurate due tothe time needed for equilibration. The mean mixed layerfor survey 1 was 54.2 m (range 47.4 to 60.7 m), while forsurvey 2 the mean was 55.0 m (range 41.4 to 68.7 m). Usingthe 0.3C criterion as applied to the drifter data (see below)would deepen the mixed layers by 5.2 m on average.[19] Water velocities were obtained using a hullmounted

ADCP (75 kHz RDI Ocean Surveyor), acquired with Uni-versity of Hawaiis Data Acquisition System (UHDAS) andprocessed and edited using the Common Ocean Data AccessSystem (CODAS) software package. The ADCP was oper-ated with interleaving broadband and narrowband pings. Wefocus only on the broadband data since it provides higherresolution near the surface and we are interested in the

Figure 5. (a and c) NCP versus DpCO2 and (b and d) NCP versus chlorophyll for Patches 1 and 2. Notethat chlorophyll scale in Figure 5d is expanded relative to that in Figure 5b.

Figure 6. Satellite (AMSRE) SST image for 5 and 6March2008 (yeardays 65 and 66) with the Patch 1 survey (blackline), XBT locations (solid circles, labeled 14) and the patchinjection site (red circle).


5 of 19

mixed layer velocities. The ADCP data were used to predictthe advection of the tracers in the mixed layer as an aid intracking and surveying the tracer patch. Since the ADCPvelocity is at fixed depths, the average velocity between25 m (the first bin of the broadband data) and 49 m wasoperationally defined as the mixed layer velocity, withkinetic energy (KE) calculated as 0.5 times the square of thedepthaverage mixed layer velocity.[20] Water masses detected by surface spatial surveys

surrounding the injection sites had wide ranging propertieswith DpCO2 of 100 to 20 matm (aqueous < atmosphere),salinities of 33.7 to 33.9, and temperatures of 3.5 to 6.5C(Figures 3 and 4). The area around Patch 1 showed muchhigher productivity and chlorophyll than around Patch 2(compare Figure 3 with Figure 4). A particularly productivewater mass was present at the southern end of the Patch 1site survey, but was not chosen for injection due to a sub-surface eddy with higher velocity shear in that area. Addi-tionally, areas with high kinetic energy (KE) of the mixedlayer or areas where KE changed rapidly, indicative ofregions with high horizontal velocity shear, were avoided(Figures 3f and 4f).[21] Although we observed wideranging water mass

properties in the larger site surveys, the water masses wherethe tracer was actually injected were much more homoge-neous (Table 2). However, there was a water mass withmuch higher NCP and chlorophyll directly to the southwestof Patch 2 (Figure 4b and more clearly by Figure 8 ofHamme et al. (submitted manuscript, 2011)).[22] Several distinct relationships can be seen between

DpCO2 and NCP during the Patch 1 survey, but essentially norelationship was present for the Patch 2 survey (Figure 5).Because NCP estimates are based on mixed layer O2/Armeasurements, this measure of productivity integrates overthe residence time of O2 in the mixed layer with respect toairsea gas exchange, defined as the ratio of mixed layerdepth (h) to gas transfer velocity (k), of about 10 days underthese conditions, while the residence time of CO2 is tentimes longer. A weak relationship between DpCO2 andNCP indicates that the DpCO2 signal was mainly caused bya previous episode of productivity that was largely finishedby the time of the survey. Except for the highly productiveregion at the southernmost point in the Patch 1 survey, therewas no strong relationship between NCP and chlorophyllconcentration (Figure 5). Hamme et al. (submitted manu-script, 2011) show that productivity rates within the tracerpatches were changing rapidly. If this was true of the widerregion, we might not expect a relationship between biomass

represented by chlorophyll and productivity rates integratedover the previous 2 weeks.

2.4. OSU Towed Vehicle Survey

[23] As part of the preinjection site survey for Patch 1,we resolved the 2dimensional structure (vertical and zonal)of the study site with the Lamont Pumping SeaSoar [Halesand Takahashi, 2002]. A single 60 km EW section30 km SE of the injection site was completed before avehicle failure ended the SeaSoar operations for theremainder of the cruise. Nonetheless, the section provideduseful information regarding the horizontal and verticalstructure of Patch 1. Conditions were fairly uniform, with

Figure 7. XBT profiles 14 to correspond with the station locations in Figure 6.

Figure 8. Satellite (AMSRE) SST image for 19 and20 March 2008 (yeardays 79 and 80) with the Patch 2 survey(black line), XBT locations (solid circles, labeled 5 and 6) andthe patch injection site (red line).


6 of 19

surface T, S, (Figure 10) O2, and pCO2 (Figure 11) varyingby less than 0.5C, 0.05, 6 mmol kg1 and 4 matm, respec-tively, coincident with low and weakly variable beamattenuation, Fchl, and optical backscatter (Figure 12).[24] Despite these smallmagnitude signals, the combi-

nation of T, S, O2, pCO2, and biooptical distributionsidentified three distinct hydrographic conditions for thesurface waters of this section. The first is a large region ofthe warmest (T > 6C), freshest (S < 33.70) water seen at thesurface, clearly expressed centered at a longitude of 37.8Win waters shallower than 40 m. This water carries moder-ately elevated signals of optical backscatter and beamattenuation, and locally depressed pCO2. O2 appears to be ata local minimum, but is still supersaturated with respect tothe atmosphere by about 2.2%. The second is a pool ofwater with essentially the same T as the first, but moderatelyhigh salinity (S 33.73) situated at about 38.2W. Thiswater has elevated biooptical signals and O2 concentra-

tions, the latter corresponding to a supersaturation of 2.6%.Finally, there is a water mass with moderate T (5.56.0C)and very high S (>33.75) at the eastern extreme of the line(37.637.5W) that carries strong signals of biomass and O2enrichment and pCO2 depletion.

Figure 9. XBT profiles 5 and 6 to correspond with the sta-tion locations in Figure 8.

Figure 10. Physical hydrography of the preinjectionpumping SeaSoar survey. (top) Temperature (color map,C) with density (st from 26.526.7, white contours,kg m3) overlain. (middle) Salinity (color map) with vehicleposition (black symbols) overlain. (bottom) BruntVislfrequency squared (N2, s2), with the top of the density tran-sition zone (based on the shallowest position of the N2 = 3 105 s2 contour, white line) overlain.

Figure 11. Dissolved gas distributions during the preinjection pumping SeaSoar survey. (top) O2 concentrations(mmol kg1), measured using a SeaBird SBE43 sensoraboard the SeaSoar vehicle; sensor was factory calibrated,and calibrated sensor output was corrected to match theTO2 relationships determined on the cruise using highaccuracy Winkler titrations of surface water samplesdrawn from the ships surface intake line. (bottom) ThepCO2 measured using a membrane contactor interfacedwith a LICOR infrared analyzer.

Figure 12. Biooptical characteristics of the preinjectionpumping SeaSoar survey. (top) Relative beam attenuation(Cp; m

1), measured using a WetLabs CStar transmissome-ter aboard the pumping SeaSoar. (middle) Chlorophyll fluo-rometer voltage (uncalibrated), measured using a WetLabsWetStar fluorometer plumbed in the shipboard end of theSeaSoar sample stream; positions of samples indicated bythe black symbols. (bottom) Optical backscatter voltage(uncalibrated), measured using a WetLabs EcoBB sensoraboard the SeaSoar.


7 of 19

[25] Beneath all of these waters lies a pycnocline, ordensity transition zone, defined by a band of elevated BruntVisl frequency squared (N2) bound above by the 26.6 stand the 5.5C density and temperature horizons, respec-tively. The stratification maximum represents a barrier tomixing, and the shallow limits of the density transition zonecan be thought of as analogous to the bottom of the mixed,or mixable, layer. The depths of this feature (white contourin Figure 10 (bottom)) are similar to mixed layer depthsdetermined for Patch 1 (discussed below) in the period priorto the South Georgia excursion. The depth and intensity ofthis stratification maximum vary across the section from 40to 60 m and 0.5 to 2 104 s2. There are local minima inthe depths of this feature beneath the expressions of the firsttwo surface water masses, but beneath the third there is littlesignal of stratification within the depth range the SeaSoarwas able to sample.[26] Examination of the surface T, S, and pCO2 distribu-

tions in the patch surveys immediately following the injec-tion suggest that the patch was centered in a water massmost similar to the second water mass identified above, with

a 40 m mixed layer situated above a reasonably stronglystratified density transition zone. The lateral proximity ofwater masses with differing hydrographic properties,including variable mixed layer depths and temporal varia-tions in mixed layer, however, advises caution in the inter-pretation of the tracer patch as a simple onedimensionaltime evolution.

3. Physical Environment

3.1. Winds

[27] Wind speed and direction were measured by 3 sonicanemometer packages deployed from the forward mast at aheight of about 18 m above the mean sea surface by groupsfrom University of Connecticut, NOAAs Earth SystemResearch Laboratory (ERSL), and LamontDoherty EarthObservatory (LDEO). Each of the sonic anemometerpackages included sensors capable of motion correcting thesonic velocities using the approach described by Edson et al.[1998]. The group also deployed a suite of sensors tomeasure air temperature, humidity, pressure, and CO2, alongwith 2 highresolution cameras, a Wave and Surface CurrentMonitoring System (WaMoS) 2D wave radar, and amicrowave altimeter to quantify the wavefield and breakingevents as described by Edson et al. [2011].[28] A time series of wind speed and direction was con-

structed using all 3 sonic anemometer systems. The relativewind direction was used to select the sonic(s) expected to beleast impacted by flow distortion; e.g., all 3 sonics wereaveraged when the relative wind was bowon but only theport sonic was used when the relative wind was from theport side. An empirical correction for flow distortion based

Table 2. Mean Surface Properties From Preinjection UnderwaySurveys Within 0.05 Latitude/Longitude of Injection Site

Patch 1 Patch 2

DpCO2 (matm) 66.8 5.1 77.5 3.5Net Community Production

(mmol C m2 d1)21.3 2.6 5.0 3.2

Chlorophyll a (mg m3) 1.06 0.13 no measurementsSalinity 33.757 0.004 33.754 0.001Temperature (C) 5.52 0.07 5.05 0.07

Figure 13. Time series of 10min averaged wind speed, wind direction, and air and water temperaturesduring SO GasEx. The red and magenta lines on these plots delineate the time periods for Patches 1 and 2,respectively. Passage of a storm center caused the winds to rotate by greater than 180 during the transectto Uruguay at the end of the experiment (yearday 96). The passage of storms centers stayed to our northduring all other periods.


8 of 19

on previous flow distortion studies [Dupuis et al., 2003;Fairall et al., 2003] was applied to the measured winds.This approach increased the usable relative wind direction toapproximately 130 from the bow. Relative wind direc-tions beyond this window were removed and the remainingdata interpolated to generate the time series shown inFigure 13. This procedure limited the number of gaps inexcess of one hour to 24 during the 37day cruise withonly three gaps longer than two hours.[29] The time series of the wind speed and direction is

characteristic of the wind field found in the circumpolar gyreat our latitude of 51S, with highfrequency variabilitysuperimposed on more slowly varying zonal flow. A series ofmidlatitude synoptic systems generally propagated to thenorth of our study region during the experiment. Theobserved high frequency transients are thought to be a resultof interaction between these synoptic scale systems and thelow frequency planetary scale waves associated with thecircumpolar gyre [Cuff and Cai, 1995; Karoly, 1990]. As aresult, the wind direction generally oscillated between SSWand NNW over 24 day cycles.[30] A combination of the barotropic component of the

temperature field associated with the meandering gyre andwarm and cold air advection associated with the north-

westerly and southwesterly wind directions, respectively,drove air temperature variability of approximately 5C(Figure 13). The variable air temperature and relativelystable sea surface temperature caused significant variabilityin atmospheric stability. Conditions of static stability (i.e.,Tair > Tsea) were generally encountered during the first halfof the experiment and unstable conditions during the secondhalf.[31] An example of the synoptic systems to our north and

the meandering atmospheric gyre over our study site isshown in Figure 14. Figure 14 shows the position of ourstudy area relative to the daily averaged sea surface pressureand air temperature fields on 22 March 2008 (yearday 82)and 23 March 2008 (yearday 83). The wind direction shiftedfrom south to northwest over this period and the temperatureincreased by approximately 3C. The resulting atmosphericstratification switched from unstable to slightly stable con-ditions during this period. This variability was typical of theconditions encountered during the experiment.[32] A primary objective of the SO GasEx is to determine

the gas transfer velocities at highwinds between 10 to 20m s1.The number distribution of 10 min averaged wind speedsencountered over the patches and for the entire experiment areshown in Figure 15, with the mean and standard deviations

Figure 14. A synoptic view of the (top) surface pressure and (bottom) temperature fields on (left)22 March 2008 (yearday 82) and (right) 23 March 2008 (yearday 83). The arrow provides the aver-age wind vector at the ships location on those days and indicates cold and warm air advection on22 and 23 March, respectively.


9 of 19

used to determine the Weibull distributions. These distribu-tions have a shape parameter of 3.6 and scaling parameters of10.1 for just the winds encountered during the tracer patchesand 10.7 for the entire cruise. The large value of the shapeparameter is indicative of the peaked distribution shown inFigure 15 due to the fairly steady wind speeds encounteredduring the experiment.[33] The cumulative percentage of 10 min averaged wind

speeds indicate that wind speeds in excess of 10 and 13 ms1 were encountered over the study region for approxi-mately 38% and 9% of the time, respectively. However,winds speeds in excess of 15 m s1 were only encountered1% of the time. Wind speeds in excess of 20 m s1 wereencountered on the transect back to Uruguay and several ofthe papers in this special section take advantage of thesedata. The average wind speeds computed for the two tracerpatches were very similar, but with significantly more var-iability in the wind speed measured during the second,longer duration release. The average wind speeds were9.10 1.71 and 9.11 3.25 m s1 over the first and secondtracer patches, respectively. The maximum wind speed of16.2 m s1 over the study region was observed on March 24(yearday 84) after the second tracer release.

3.2. SF6 Tracer Patch

[34] The SF6 tracer patch was monitored using an auto-mated SF6 analysis system that analyzed samples at ca. 1minintervals [Ho et al., 2002] to quantify the lateral dispersion atthe surface. The main aims of the surface SF6 mapping wereto provide a Lagrangian framework for the biogeochemical

studies, and to constrain the patch boundaries so that the patchcenter could be identified.[35] The continuous SF6 analysis system consists of a gas

extraction unit, which strips SF6 continuously from theships uncontaminated seawater line, and the analytical unit,made up of a gas chromatograph (GC) equipped with anelectron capture detector (ECD), which measures the SF6.Instrument control and data acquisition are handled by acombination of hardware and software (LabVIEW) runningon a personal computer. The system has a detection limit of1 1014 mol L1. At a typical ship speed of 510 knotsduring SO GasEx, a spatial resolution of every 0.150.3 kmwas usually achieved. Concentration data were incorporatedin near realtime into a plot of the areal distribution of thepatch, by integrating the SF6 data with the ships GPS position.This enabled rapid alteration of ship speed and direction inresponse to variation in the SF6 signal and ensures resolutionof patch boundaries for distinguishing stations inside andoutside the patch. The realtime SF6 signal was used todetermine the center of the patch for discrete 3He and SF6sampling.[36] After injection, underway SF6 surveys show that the

tracer patches spread to a width of approximately 0.1 degreelatitude/longitude (11 7 km) and were advected to thesoutheast (Figures 16 and 17).

Figure 15. The number distribution of 10min average windspeed observations during the experiment. The distributionshown in blue presents the data collected over the tracerpatches, while the data distribution shown in red includesthe period near South Georgia Island and the transect to Uru-guay. Weibull distributions with a shape parameter 3.6 andscaling parameters of 10.1 (bottom curve) and 10.7 (topcurve) were determined from fits to the mean and standarddeviation of the wind speed measured over the tracer patchand during the entire cruise, respectively.

Figure 16. Surface distributions of SF6 (fmol L1) during

Patch 1. Colors show optimum interpolation of underway sur-face SF6 concentration measurements on logarithmic scale,derived using the GLOBEC Kriging software (EasyKrig3.0)for MATLAB. Green circles show locations of CTD/Rosettecasts. Smaller gray dots show cruise track. The patch advectedfrom the northwest to the southeast over the time period ofobservation from yearday 70 to 74.


10 of 19

3.3. ADCP Velocities

[37] For Patch 1, the mean velocity was toward thesoutheast with a slight increase over the period of obser-vation (Figure 18). There were significant higher frequencyfluctuations due to tidal and inertial motions that could notbe separately resolved (Figure 19) with approximately 60%more energy in the meridional component. (The peak in theeastward velocity at 7 101 cycles per hour (90 min) is anartifact due to the wellknown Schuler oscillation of thegyroscope while transiting in a north or south direction.)[38] For Patch 2, the mean velocity was initially to the

southeast similar to Patch 1. However, around yearday 88,the patch stalled for a day or so before starting to move tothe southwest (Figure 18). High frequency velocities,including the inertial and tidal motions, were even strongerduring Patch 2 compared to Patch 1 (Figure 20), with moreenergy contained in the zonal component. The highfrequency velocity fluctuations were strongest at the startof Patch 2 observations and decreased over the first 6 days.The storm that forced the end of the Patch 1 survey likelygenerated the large inertial oscillations in the mixed layerobserved at the beginning of Patch 2. Inertial wavespropagate out of the mixed layer over several days[DAsaro, 1985; DAsaro et al., 1995]. The observeddecrease of the magnitude of the inertial oscillationsduring Patch 2 is likely due to this process.

3.4. Mixed Layer Depth

[39] A key parameter for constraining gas transfer veloc-ities based on mass balances is the ocean surface mixedlayer depth (MLD), as it determines the reservoir in contactwith the airsea interface. Conceptually, it can be defined asthe depth at the surface over which chemical and physicalproperties are homogeneous, which implies that mixing isactive throughout the layer. However, homogeneity ofproperties is a function of fluxes and transformations in themixed layer such that its extent can differ for differentproperties and from the operational definition. For gases, itis assumed that all water in this region is in contact with theatmosphere on time scales that are significantly shorter thanthe residence time of the gases. The mixed layer is boundedat depth by a region with a sharp gradient in these properties.[40] However, homogenous properties do not guarantee

the whole layer is mixing. Additionally, at the base of themixed layer, there is some exchange between the mixedlayer and the pynocline, through the density transition zone[Johnston and Rudnick, 2009]. Thus, the definition of theMLD depends on the problem that is being addressed. Inaddition, the depth will vary on time scales from hourly toseasonal due to heating and cooling cycles, wind, andinternal wave activity. Below, we operationally define thedepth of the mixed layer in several different ways dependingon the issue studied.

Figure 17. Surface distributions of SF6 (fmol L1) during Patch 2. Plotted same as Figure 16.


11 of 19

[41] For 3He/SF6 dual tracer experiments in the openocean, the total mass decrease of 3He and SF6 in the upperwater column must be determined. Some studies, becausethey were limited by the number of 3He/SF6 samples taken,used a density or temperaturebased definition of the MLDfrom CTD profiles [e.g., Wanninkhof et al., 2004]. Otherstudies with a sufficient number of 3He and SF6 samples

have used a tracer based definition in which the mixed layerdepth is where the SF6 drops to 50% of its concentration inthe top 20 m or so [e.g., Ho et al., 2006; Nightingale et al.,2000]. Stevens et al. [2011] proposed a buoyancy fre-quencybased definition of MLD that fit the tracer MLDfrom a previous 3He/SF6 dual tracer experiment in theSouthern Ocean, but was not able to reproduce the tracer

Figure 18. Average mixed layer (a) eastward and (b) northward velocities during Patch 1 and Patch 2from the 5min averaged hullmounted ADCP data (black line). Lowpassed (fourthorder Butterworthfilter with 36 h cutoff) velocities are shown in red.

Figure 19. Spectra of the average mixed layer velocities for Patch 1.


12 of 19

MLD during SO GasEx, suggesting that this empirical MLDdefinition is not universal.[42] When determining the depth over which tracers are

ventilating with the atmosphere, a tracerbased MLD is mostappropriate. However, even in SO GasEx where a largenumber of discrete 3He and SF6 samples were taken, thevertical sample resolution was 510 m, so this identificationof the MLD is resolution limited. Here, we provide severalestimates ofMLD related to the scientific questions addressedbased on CTDdensity, a onedimensional numerical model,SF6 concentration, and temperature measurements from theMAPCO2 drifter (Table 3).

3.5. CTD Based

[43] We defined the CTDbased MLD as the shallowestdepth with a density at least 0.01 kg m3 greater than thedensity at 5 dbar. This 5 dbar depth is the shallowest depththat all CTD casts sampled. Only the 1 dbar binaverageddowncasts of the CTD were examined, as the position of thetemperature and conductivity sensors at the bottom of theCTD/Rosette package and turbulence generated by therosette package may bias the upcasts. As will be discussedbelow, the presence of internal wave motion and othervariability does not warrant determining the mixed layerdepth at better than 1 m resolution.[44] Two CTD casts were made each day: near local noon

and local midnight. While downcasts were used to calculatethe CTDbased MLD, the tracerbased MLD was based onwater collected by bottles fired during the upcast. Com-paring the depth of bottle samples with the depth of thesame density horizons on the downcast between the mixedlayer base and 225 m revealed a mean depth difference of0.6 4.8 m. Over the time to make a CTD cast (about anhour), the root mean square vertical displacement of iso-pycnals (and the base of the mixed layer) was on the orderof 5 m. The coarse temporal sampling by the CTD and

aliasing by internal wave displacement make determiningthe highresolution evolution of the MLD from CTD andtracer profiles challenging.

3.6. OneDimensional Model Based

[45] We used the onedimensional (1D) GeneralizedOcean Turbulence Model (GOTM) to examine the evolutionof the mixed layer. GOTM solves the equations for thevertical transport of horizontal momentum, heat, and saltwith vertical turbulent fluxes parameterized using a widerange of turbulence closure schemes [Burchard et al., 2005].For SO GasEx, the fluxes at the ocean surface were esti-mated from the meteorological measurements made aboardthe ship (see above). Because SO GasEx focused on thesurface mixed layer and for computational efficiency, thelower boundary of the model was placed at 500 m, at whichvertical fluxes of momentum, heat, and salt were set to zero.The model was run with 500 vertical levels with grid reso-lution of 0.55 m at the ocean surface and 1.3 m at the 500 mbase of the model grid. Vertical profiles of hydrographyand velocity from CTD casts and the hullmounted ADCPat approximately the times of tracer injection were usedas model initial conditions. Additionally, a separate series ofGOTM runs included nudging of the model velocities, T,and S to the observations on a 1day timescale.[46] The observed velocity profiles, lowpass filtered to

remove tides and inertial oscillations, were used to estimatetimedependent horizontal pressure gradients that wereimposed as additional model forcing. This forced the 1Dmodel to produce more realistic vertical shears, whichthemselves influenced the turbulent mixing. The barotropicand baroclinic pressure gradients, the latter arising fromhorizontal density gradients assumed here to vary linearlywith depth, were estimated from the lowpassed ADCPvertical profiles using the assumption of geostrophy atsubinertial timescales. For each ADCP ensemble, the pro-

Figure 20. Spectra of the average mixed layer velocities for Patch 2.


13 of 19

files of eastward and northward velocities were fit to aquadratic function of depth.[47] The results presented here were obtained using the k"

turbulence closure [Rodi, 1987] to parameterize the turbu-lent vertical momentum, heat, and salt fluxes in the model.Injection of turbulent kinetic energy at the ocean surface dueto wave breaking was specified using the parameterizationof Burchard [2001]. Model results were not sensitive toclosure scheme; using the nonlocal K profile parameteri-zation (KPP) for turbulence within the surface mixed layeryielded very similar results.[48] In general, for both patches, the 1D GOTM pre-

diction of the MLD agrees with the CTD observations(Figure 21), particularly given that the CTD observationshave internal wave displacements on the order of 5 m thatare not present in the model. In the model simulations, thereare instances where surfaces fluxes (mainly rainfall) producea shallow mixed layer that is subsequently mixed into theolder mixed layer. As expected, the model runs with

nudging toward observed mixed layer properties reproducedthe observed MLD slightly better.[49] The evolution of the modeled T, S, and density of the

mixed layer without nudging diverges quickly from theobservations, especially for Patch 2 (Figure 21). Thisdivergence could be due to incorrect surface forcing butmore likely is a result of horizontal advection/diffusion. Ifthe CTD casts really represent the Lagrangian evolution ofthe mixed layer, then colder, saltier, denser waters musthave been mixed into the patch. This water must have comefrom the south. The CTD casts were made in what wasbelieved to be the center of the tracer patch. However, if thetracer preferentially mixed horizontally in one directionmore than the other, a similar result would be seen. In thiscase, the center of the patch would have moved into colder,saltier water (i.e., to the south).

3.7. Tracer Based

[50] The mixed layer depth used to determine the gastransfer velocity with the deliberate tracer SF6 and

3He

Table 3. Surface Properties and Mixed Layer Depths for All CTD Castsa

Station Lat Long Yearday S (5 dB) T (5 dB) s (5 dB) rMLD (dB) GOTMMLD (m) SF6MLD (m) Drifter MLD (m)

1 50.75 38.499 66.64 33.7591 5.8446 26.5916 292 50.713 38.585 70.20 33.7519 5.5639 26.6201 47 43 353 50.731 38.555 70.63 33.7513 5.6429 26.6101 45 46 54 344 50.755 38.476 71.13 33.7540 5.5728 26.6207 47 47 50 355 50.758 38.442 71.64 33.7545 5.5770 26.6206 38 43 52 386 50.79 38.41 72.12 33.7538 5.6267 26.6140 38 40 41 387 50.803 38.355 72.62 33.7546 5.7542 26.5992 15 10 598 50.866 38.306 73.16 33.7545 5.6654 26.6099 27 21 449 50.875 38.234 73.63 33.7491 5.6605 26.6062 30 30 49 4710 50.862 38.239 74.09 33.7532 5.7356 26.6003 49 50 4513 51.04 37.698 78.32 33.7534 5.9713 26.5714 31 77 2414 50.946 37.797 80.62 33.7416 6.3247 26.5173 2816 51.14 38.375 82.12 33.7383 4.9711 26.6784 46 60 4818 51.192 38.304 82.65 33.7498 4.9406 26.6831 52 51 54 4819 51.191 38.264 83.16 33.7418 4.9299 26.6859 56 53 58 4921 51.2 38.135 83.66 33.7386 4.9172 26.6847 52 52 41 5122 51.222 38.019 84.11 33.7388 4.9201 26.6846 50 52 55 5224 51.239 37.997 84.65 33.7438 4.8219 26.6995 63 55 68 5225 51.258 37.868 85.15 33.7464 4.7759 26.7067 66 57 58 5526 51.294 37.687 85.62 33.7471 4.7511 26.7100 56 56 61 5727 51.327 37.563 86.08 33.7461 4.7560 26.7087 56 15 64 5828 51.3 37.503 86.56 33.7159 4.7894 26.6810 12 6 6429 51.292 37.427 87.06 33.7129 4.8409 26.6729 15 9 5730 51.318 37.347 87.55 33.7328 4.7966 26.6936 21 11 5531 51.318 37.314 88.04 33.7312 4.8109 26.6908 22 22 6435 51.31 37.28 89.74 33.7391 4.8039 26.6978 51 51 60 4937 51.312 37.297 90.13 33.7421 4.7520 26.7059 55 36 63 4938 51.297 37.33 90.65 33.7454 4.7153 26.7126 53 54 60 5639 51.303 37.35 91.04 33.7464 4.6814 26.7171 62 54 58 5940 51.303 37.342 91.61 33.7454 4.6851 26.7159 47 52 57 5941 51.794 36.885 91.78 33.7488 4.4680 26.7422 64 50 71 5942 51.28 37.363 92.06 33.7470 4.7019 26.7154 56 52 5843 51.336 37.433 92.57 33.7467 4.6852 26.7169 54 52 5844 51.34 37.463 93.05 33.7479 4.7022 26.7160 53 52 6045 51.367 37.473 93.54 33.7490 4.6910 26.7181 52 53 6146 51.383 37.467 94.04 33.7523 4.7433 26.7150 53 52 6147 51.407 37.517 94.58 33.7528 4.7995 26.7092 47 49 5848 51.437 37.477 95.10 33.7524 4.7270 26.7169 58 49 6349 51.453 37.438 95.58 33.7526 4.6897 26.7211 62 50 7250 51.45 37.25 95.78 33.7583 4.9201 26.7000 71 50 8151 51.465 37.407 96.03 33.7537 4.7277 26.7178 64 72aThe rMLD, GOTMMLD, SF6MLD, and Drifter MLD give the mixed layer depth from the CTD density measurements, the GOTM model results,

the penetration depth of the SF6 tracer, and the MAPCO2 drifter temperature measurements with an 18h filter, respectively.


14 of 19

should represent the average depth of water that ventilateswith the atmosphere on a daily basis for the entire patch [Hoet al., 2011]. This depth was determined from the depthprofiles of SF6 at every sampling station. The tracers wereinjected near the surface and rapidly mixed throughout themixed layer. Penetration downward below the mixed layeroccurs by slow diffusive processes on timescales signifi-cantly longer than airsea gas transfer. Since sampling withthe Niskin bottles was at coarser resolution than the CTDtrace, the SF6 gradient at the bottom of the mixed layer wasinterpolated based on the density gradient. The averagemixed layer depths for gas exchange, for every cast whereSF6 measurements were taken, were determined from thelevel where SF6 reached 50% of the averaged concentrationin the top 20 m.[51] In general the tracer based mixed layer depth is

greater than CTDbased estimate, because the 50% surfaceSF6 concentration criteria is deeper than the start of thepycnocline (Figure 22). In other words, the ventilated mixedlayer is slightly deeper than the mixed layer defined bystability. This is verified by the combined 3He and SF6profiles, with the mixed layer defined by the region wherethe ratios of the two tracers are invariant (Figure 23).

3.8. Drifter Based

[52] A drifting autonomous buoy, the MAPCO2 designedby NOAA/PMEL, was used to make high frequency phys-ical and biogeochemical measurements in the tracer patchduring the experiment. It was drogued to help it stay withthe water mass that contained the tracers throughout theexperiment. Below the buoy was a 118 m string of instru-ments, including 15 HOBO Pro v2 water temperature sensorspositioned no more than 10 m apart, that sampled every30 min with an accuracy of 0.2C. MLDwas calculated as thedepth at which the depthinterpolated temperature was 0.3Ccolder than the temperature at 5 dbar, To remove 12h periodinternal waves present in all of the data with amplitudes ofabout 5 m, the data was 18 h lowpass filtered. Following amajor rain event that began on 26 March (yeardays 8689), alow salinity layer at the surface prevented accurate MLDdeterminations from the drifter temperature data. A detailedanalysis of theMAPCO2 data is given byMoore et al. [2011].[53] The MAPCO2 was deployed on 8 March 2008

(yearday 68), shortly after Patch 1 was created, recoveredon 12 March (yearday 72) and redeployed in Patch 1 on13March, and finally recovered again on 18March (yearday 78)after the ship returned from South Georgia Island. The pres-ence of internal waves, documented in the drifter data,complicated estimation of the MLD. In particular, there was atendency during the first few days of Patch 1 for the CTDcasts to occur during the deepest excursion of the internalwaves, such that the CTD and tracer basedMLD during thesecasts were likely deeper than the average MLD during thisperiod (Figure 22).[54] The drifter based MLD increased from 32 m to a

maximum of 54 m on 17 March (yearday 77), the day beforethe ship returned and the MAPCO2 buoy was recovered.Both the MAPCO2 and CTD data show that the averagemixed layer density decreased slightly over this sameperiod. Neither the CTD nor tracer based estimates show theconsistent deepening of the MLD seen in the MAPCO2estimates (Figure 22 and Table 3), which may be partiallyattributed to the 5 m excursions of the internal tide.MAPCO2 recorded a significant warming event that shoaledthe MLD to 24 m on 18 March (yearday 78), which wascorroborated by the CTD cast performed when the ship re-turned to the MAPCO2 location (Figure 22 and Table 3).Traces of the original patch were found near the drifter whenit was recovered, but the MLD inferred from the SF6 mea-surements was substantially deeper than the CTD or drifterMLDs. This is reasonable, because a thermal shallowing ofthe mixed layer due to surface heat fluxes would not bemanifested in the tracer concentrations.[55] The MAPCO2 was deployed into the second patch on

22 March (yearday 82). The MLD increased from 48 to58 m by 26 March (yearday 86) when there was a rainstorm.The rainstorm did not alter the temperature structure enoughto affect the drifter based mixed layer depth estimates, butthe changes in salinity recorded by the CTD casts indicatedthat the rain caused the mixed layer to shoal to

[56] Around the time of the rainstorm, the tracer patchsheared in two. The largest fraction of the tracer patchessentially stopped moving. However, MAPCO2 and someportion of the tracer patch continued their southeasterlyadvection until the drifter was recovered 31 March (yearday91) approximately 50 km from the primary patch. The CTD,GOTM, and tracer based MLD estimates apply to the pri-mary tracer patch, while the drifter documented MLDchanges in the portion that moved to the southeast. Despitethis separation, the ship and drifter based MLD estimateswere similar within the error imposed by excursions of theinternal tide (Figure 22).

3.9. Vertical Diffusivity From 1D Model

[57] As might be inferred from the variability in MLD(Figure 21), vertical mixing within the upper ocean likelyunderwent large fluctuations during SO GasEx. Maximumvertical turbulent diffusivity for heat within the mixed layerin the 1D GOTM runs was typically in the range 102101 m2 s1, occurring within a broad region in the interiorof the mixed layer (Figure 24a). The maximum diffusivityis, however, 12 orders of magnitude lower during periodswhen the MLD decreases as a result of weak winds com-bined with surface heating (e.g., prior to yearday 72.5 duringPatch 1) or precipitation (e.g., yeardays 8687 duringPatch 2). A timescale for vertical homogenization of themixed layer can be estimated as: t = HML

2 /Kmax, whereHML is the mixed layer thickness and Kmax is the maxi-mum diffusivity within the mixed layer. A typical mixingtime is on the order of 24 h, but can be as short as 45 h duringperiods of rapid deepening of the mixed layer and an order

of magnitude longer during brief periods of weak mixing(Figure 24b).[58] Diffusivity within the mixed layer decreases toward

the surface as eddy length scales are reduced near theboundary. Diffusivity strongly decreases near the bottom ofthe mixed layer; however, the diffusivity at the base of theoperationally defined density based mixed layer varies over

Figure 23. Depth profiles of temperature, SF6 and3He/SF6

normalized to their respective mixed layer values for Station27 (51.32S, 37.563W; yearday 86.08). As shown here andTable 1, the CTDdensity based mixed layer (56 m) depth isshallower than that based on change in SF6 concentration(64 m).

Figure 22. Mixed layer depth based on r(z)r(5 dbar) 0.01 kg m3 (red circles), 0.5 [SF6] = [SF6]20m(blue squares) and T(5 dbar)T(z) 0.3C fromMAPCO2 drifter (black line = 30min frequency, cyan line =18h filter applied).


16 of 19

at least 5 orders of magnitude (Figure 24c). This emphasizesthe importance of distinguishing between the actively mix-ing layer and the well mixed layer, the latter resulting fromthe time history of mixing [Brainerd and Gregg, 1995].High vertical diffusivity at the mixed layer base is associatedwith deepening of the mixed layer, most clearly seen afteryearday 72.5 in the Patch 1 simulation and to a lesser extentduring yeardays 86.689.9 during Patch 2. During the latterperiod, there are intervals when mixing within the mixedlayer and at its base weakened, likely due to strong surfaceheating during daytime hours. Mixed layer shoaling generallyoccurs concurrently with periods of weaker mixing at themixed layer base. However, not all times of weak mixing areaccompanied by shoaling of the mixed layer, because mixed

layer shoaling requires a mechanism of active restratification,such as surface heating or precipitation.

4. Conclusions

[59] We described the site selected for the Southern Ocean(SO) Gas Exchange Experiment (GasEx) and the generalmixed layer dynamics as background to study questionsrelated to gas exchange and CO2 fluxes in the SouthernOcean. Using ship, buoy, and remotesensing based methodsto study changes in Lagrangian tracer patches, the contribu-tions in this volume address research questions ranging fromimproved gas exchange parameterizations to better under-standing of Southern Ocean biological processes.

Figure 24. (a). Vertical turbulent diffusivity from GOTM 1D model runs with nudging versus time anddepth. The magenta line shows the depth of the model mixed layer, defined as the depth where densityvaries by less than 0.01 kg m3 from its value at 5 m depth. (b) Mixing timescale within model mixedlayer estimated from the maximum diffusivity and the mixed layer thickness. (c). Vertical diffusivityat the base of the mixed layer versus time.


17 of 19

[60] Acknowledgments. We thank those members of the SO GasExPlanning group, who helped draft the SO GasEx implementation plan andassisted with initial site selection, but who did not participate in SO GasEx(MaryElena Carr, Paty Matrai, and Janet Sprintall). Several other scientistsprovided essential data that aided initial site selection decisions, includingTaro Takahashi, who provided pCO2 measurements from the LDEO CO2database; Jacqueline Boutin, who provided pCO2 measurements fromCARIOCA buoys; Dorothee Bakker, who provided pCO2 measurementsfrom recent cruises to the SO GasEx study region; Andy Jacobson, whoprovided wind speed data; David Woolf, who provided wave height data;Joaquin Trianes, who provided satellite altimeter and drifter data; andPeter Miller, who provided satellite altimeter, SST, and Chl a data. WilliamBalch, Bruce Bowler, and David Drapeau (Bigelow Laboratory for OceanSciences) provided underway salinity data for the Patch 1 site survey.Kathy Tedesco, former Program Manager of the Global Carbon Cycle Pro-gram in NOAAs Climate Program Office, was an important liaisonbetween the science planning group and the agencies that funded SO GasEx(NSF, NASA, and NOAA). We thank the participants of SO GasEx, whoendured technical and logistical challenges and made SO GasEx a success-ful expedition, and the Captain and the crew of the NOAA Ship Ronald HBrown who, in their various roles, ensured that we were safe, well fed, andto a large extent, able to execute our various scientific projects. This isPMEL contribution 3696.

ReferencesBalch, W. M., D. T. Drapeau, B. C. Bowler, E. Lyczskowski, E. S. Booth,and D. Alley (2011), The contribution of coccolithophores to the opticaland inorganic carbon budgets during the Southern Ocean Gas ExchangeExperiment: New evidence in support of the Great Calcite Belthypothesis, J. Geophys. Res., 116, C00F06, doi:10.1029/2011JC006941.

Brainerd, K. E., and M. C. Gregg (1995), Surface mixed and mixing layerdepths, Deep Sea Res., Part I, 42, 15211543, doi:10.1016/0967-0637(95)00068-H.

Burchard, H. (2001), Simulating the waveenhanced layer under breakingsurface waves with twoequation turbulence models, J. Phys. Oceanogr.,31, 31333145, doi:10.1175/1520-0485(2001)0312.0.CO;2.

Burchard, H., K. Bolding, M. R. Villarreal, T. Rippeth, N. Fisher, and A. Stip(2005), The GOTM modeling system, in Marine Turbulence: Theories,Observations, and Models. Results of the CARTUM Project, edited byH. Z. Baumert, J. Simpson, and J. Sundermann, pp. 213224, CambridgeUniv. Press, New York.

Cassar, N., B. A. Barnett, M. L. Bender, J. Kaiser, R. C. Hamme, andB. Tilbrook (2009), Continuous highfrequency dissolved O2/Ar mea-surements by equilibrator inlet mass spectrometry, Anal. Chem., 81,18551864, doi:10.1021/ac802300u.

Cuff, T. J., and M. Cai (1995), Interaction between the low and highfrequency transients in the Southern Hemisphere winter circulation, Tellus,Ser. A, 47, 331350, doi:10.1034/j.1600-0870.1995.t01-2-00004.x.

DAsaro, E. A. (1985), The energy flux from the wind to nearinertialmotions in the surface mixed layer, J. Phys. Oceanogr., 15, 10431059,doi:10.1175/1520-0485(1985)0152.0.CO;2.

DAsaro, E. A., C. C. Eriksen, M. D. Levine, P. Niiler, C. A. Paulson, andP. Vanmeurs (1995), Upperocean inertial currents forced by a strongstorm. 1. Data and comparisons with lineartheory, J. Phys. Oceanogr.,25, 29092936, doi:10.1175/1520-0485(1995)0252.0.CO;2.

Del Castillo, C. E., and R. L. Miller (2011), Horizontal and vertical distribu-tion of colored dissolved organic matter during the Southern Ocean GasExchange Experiment, J. Geophys. Res., 116, C00F07, doi:10.1029/2010JC006781.

Dupuis, H., C. Guerin, D. Hauser, A. Weill, P. Nacass, W. M. Drennan,S. Cloche, and H. C. Graber (2003), Impact of flow distortion correc-tions on turbulent fluxes estimated by the inertial dissipation methodduring the FETCH experiment on R/V LAtalante, J. Geophys. Res.,108(C3), 8064, doi:10.1029/2001JC001075.

Dwivedi, S., T. W. N. Haine, and C. E. Del Castillo (2011), Upper oceanstate estimation in the Southern Ocean Gas Exchange Experiment regionusing the fourdimensional variational technique, J. Geophys. Res., 116,C00F02, doi:10.1029/2009JC005615.

Edson, J. B., A. A. Hinton, K. E. Prada, J. E. Hare, and C. W. Fairall(1998), Direct covariance flux estimates from mobile platforms at sea,J. Atmos. Oceanic Technol., 15, 547562, doi:10.1175/1520-0426(1998)0152.0.CO;2.

Edson, J. B., C. W. Fairall, L. Bariteau, C. J. Zappa, A. CifuentesLorenzen,W. R. McGillis, S. Pezoa, J. E. Hare, and D. Helmig (2011), Direct covari-ance measurement of CO2 gas transfer velocity during the 2008 Southern

Ocean Gas Exchange Experiment: Wind speed dependency, J. Geophys.Res., doi:10.1029/2011JC007022, in press.

Fairall, C. W., E. F. Bradley, J. E. Hare, A. A. Grachev, and J. B. Edson(2003), Bulk parameterization of airsea fluxes: Updates and verificationfor the COARE algorithm, J. Clim., 16, 571591, doi:10.1175/1520-0442(2003)0162.0.CO;2.

Feely, R. A., R. Wanninkhof, D. A. Hansell, M. F. Lamb, D. Greeley,K. Lee, and J.Z. Zhang (2002), Water column CO2 measurements duringthe Gas Ex98 expedition, in Gas Transfer at Water Surfaces, Geophys.Monogr. Ser., vol. 127, edited by M. Donelan et al., pp. 173181, AGU,Washington, D. C.

Hales, B., and T. Takahashi (2002), The pumping SeaSoar: A highresolution seawater sampling platform, J. Atmos. Oceanic Technol.,19, 10961104, doi:10.1175/1520-0426(2002)0192.0.CO;2.

Ho, D. T., P. Schlosser, and T. Caplow (2002), Determination of longitu-dinal dispersion coefficient and net advection in the tidal Hudson Riverwith a largescale, high resolution SF6 tracer release experiment, Environ.Sci. Technol., 36, 32343241, doi:10.1021/es015814+.

Ho, D. T., C. S. Law, M. J. Smith, P. Schlosser, M. Harvey, and P. Hill(2006), Measurements of airsea gas exchange at high wind speeds inthe Southern Ocean: Implications for global parameterizations, Geophys.Res. Lett., 33, L16611, doi:10.1029/2006GL026817.

Ho, D. T., R. Wanninkhof, P. Schlosser, D. S. Ullman, D. Hebert, and K. F.Sullivan (2011), Toward a universal relationship between wind speed andgas exchange: Gas transfer velocities measured with 3He/SF6 during theSouthern Ocean Gas Exchange Experiment, J. Geophys. Res., 116,C00F04, doi:10.1029/2010JC006854.

Johnston, T. M. S., and D. L. Rudnick (2009), Observations of the transitionlayer, J. Phys. Oceanogr., 39, 780797, doi:10.1175/2008JPO3824.1.

Karoly, D. J. (1990), The role of transient eddies in lowfrequency zonalvariations of the Southern Hemisphere circulation, Tellus, Ser. A, 42,4150, doi:10.1034/j.1600-0870.1990.00005.x.

Lee, Z., et al. (2011), An assessment of optical properties and primary pro-duction derived from remote sensing in the Southern Ocean (SO GasEx),J. Geophys. Res., 116, C00F03, doi:10.1029/2010JC006747.

McGillis, W. R., J. B. Edson, J. D. Ware, J. W. H. Dacey, J. E. Hare, C. W.Fairall, and R. Wanninkhof (2001), Carbon dioxide flux techniques per-formed during GasEx 98, Mar. Chem., 75, 267280, doi:10.1016/S0304-4203(01)00042-1.

McGillis, W. R., et al. (2004), Airsea CO2 exchange in the equatorialPacific, J. Geophys. Res., 109, C08S02, doi:10.1029/2003JC002256.

Moore, T. S., M. D. DeGrandpre, C. L. Sabine, C. J. Zappa, W. R. McGillis,R. C. Hamme, R. A. Feely, and W. M. Drennan (2011), Sea surface pCO2and O2 in the Southern Ocean during the austral fall, 2008, J. Geophys.Res., doi:10.1029/2010JC006560, in press.

Nightingale, P. D., P. S. Liss, and P. Schlosser (2000), Measurements ofairsea gas transfer during an open ocean algal bloom, Geophys. Res.Lett., 27, 21172120, doi:10.1029/2000GL011541.

Pierrot, D., C. Neill, K. Sullivan, R. Castle, R. Wanninkhof, H. Luger,T. Johannessen, A. Olsen, R. A. Feely, and C. E. Cosca (2009),Recommendations for autonomous underway pCO2 measuring systemsand datareduction routines, Deep Sea Res., Part II, 56, 512522,doi:10.1016/j.dsr2.2008.12.005.

Reuer, M. K., B. A. Barnett, M. L. Bender, P. G. Falkowski, and M. B.Hendricks (2007), New estimates of Southern Ocean biological produc-tion rates from O2/Ar ratios and the triple isotope composition of O2,Deep Sea Res., Part I, 54, 951974, doi:10.1016/j.dsr.2007.02.007.

Rodi, W. (1987), Examples of calculation methods for flow and mixing instratified fluids, J. Geophys. Res., 92, 53055328, doi:10.1029/JC092iC05p05305.

Sabine, C. L., R. A. Feely, G. C. Johnson, P. G. Strutton, M. F. Lamb,and K. E. McTaggart (2004), A mixed layer carbon budget for the GasEX2001 experiment, J. Geophys. Res., 109, C08S05, doi:10.1029/2002JC001747.

Stevens, C., B. Ward, C. Law, and M. Walkington (2011), Surface layermixing during the SAGE ocean fertilization experiment, Deep SeaRes., Part II, 58, 776785, doi:10.1016/j.dsr2.2010.10.017.

Wanninkhof, R., K. F. Sullivan, and Z. Top (2004), Airsea gas transfer inthe Southern Ocean, J. Geophys. Res., 109, C08S19, doi:10.1029/2003JC001767.

Yang, M., B. W. Blomquist, C. W. Fairall, S. D. Archer, and B. J. Huebert(2011), Airsea exchange of dimethylsulfide in the Southern Ocean: Mea-surements from SO GasEx compared to temperate and tropical regions,J. Geophys. Res., 116, C00F05, doi:10.1029/2010JC006526.

J. B. Edson, Department of Marine Sciences, University of Connecticut,1080 Shennecossett Rd., Marine Sciences Bldg., Rm. 193, Groton, CT06340, USA.


18 of 19

B. Hales, College of Oceanic and Atmospheric Sciences, Oregon StateUniversity, 104 Ocean Admin. Bldg., Corvallis, OR 97331, USA.R. C. Hamme, School of Earth and Ocean Sciences, University of

Victoria, PO Box 3055, STN CSC, Victoria, BC V8W 3P6, Canada.B. R. Hargreaves, Department of Earth and Environmental Sciences,

Lehigh University, 1 W. Packer Ave., Bethlehem, PA 180153001, USA.D. Hebert, Fisheries and Oceans Canada, Bedford Institute of

Oceanography, PO Box 1006, Dartmouth, NS B2Y 4A2, Canada.D. T. Ho, Department of Oceanography, University of Hawaii at Maoa,

1000 Pope Rd., MSB 517, Honolulu, HI 96822, USA. ([email protected])

C. L. Sabine, NOAA Pacific Marine Environmental Laboratory,7600 Sand Point Way, Seattle, WA 98115, USA.P. G. Strutton, Institute for Marine and Antarctic Studies, University of

Tasmania, Private Bag 129, Hobart, Tas 7001, Australia.D. S. Ullman, Graduate School of Oceanography, University of Rhode

Island, South Ferry Road, Narragansett, RI 02882, USA.R. Wanninkhof, NOAA Atlantic Oceanographic and Meteorological

Laboratory, 4301 Rickenbacker Cswy., Miami, FL 33149, USA.


19 of 19

/ColorImageDict > /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /CropGrayImages false /GrayImageMinResolution 300 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.00000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /CropMonoImages false /MonoImageMinResolution 1200 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 400 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.00000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects true /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile () /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False

/CreateJDFFile false /Description > /Namespace [ (Adobe) (Common) (1.0) ] /OtherNamespaces [ > > /FormElements true /GenerateStructure false /IncludeBookmarks false /IncludeHyperlinks false /IncludeInteractive false /IncludeLayers false /IncludeProfiles true /MarksOffset 6 /MarksWeight 0.250000 /MultimediaHandling /UseObjectSettings /Namespace [ (Adobe) (CreativeSuite) (2.0) ] /PDFXOutputIntentProfileSelector /DocumentCMYK /PageMarksFile /RomanDefault /PreserveEditing true /UntaggedCMYKHandling /UseDocumentProfile /UntaggedRGBHandling /UseDocumentProfile /UseDocumentBleed false >> ]>> setdistillerparams> setpagedevice

Date post:	08-Apr-2018
Category:	Documents
Upload:	phungkien
View:	222 times
Download:	3 times

Southern Ocean Gas Exchange Experiment: Setting … Ocean Gas Exchange Experiment: Setting the stage...

Documents