• Andreas, E. L, Jordan R. E. , and Makshtas A. P. , 2005: Parameterizing turbulent exchange over sea ice: The Ice Station Weddell results. Bound.-Layer Meteor., 114, 439460, doi:10.1007/s10546-004-1414-7.

    • Search Google Scholar
    • Export Citation
  • Andreas, E. L, Horst T. W. , Grachev A. A. , Persson P. O. G. , Fairall C. W. , Guest P. S. , and Jordan R. E. , 2010: Parametrizing turbulent exchange over summer sea ice and the marginal ice zone. Quart. J. Roy. Meteor. Soc., 136, 927943, doi:10.1002/qj.618.

    • Search Google Scholar
    • Export Citation
  • Andreas, E. L, Jordan R. E. , Mahrt L. , and Vickers D. , 2013: Estimating the Bowen ratio over the open and ice-covered ocean. J. Geophys. Res. Oceans, 118, 43344345, doi:10.1002/jgrc.20295.

    • Search Google Scholar
    • Export Citation
  • Bell, T. G., De Bruyn W. J. , Miller S. D. , Ward B. , Christensen K. , and Saltzman E. S. , 2013: Air–sea dimethylsulfide (DMS) gas transfer in the North Atlantic: Evidence for limited interfacial gas exchange at high wind speed. Atmos. Chem. Phys., 13, 11 07311 087, doi:10.5194/acp-13-11073-2013.

    • Search Google Scholar
    • Export Citation
  • Bell, T. G., De Bruyn W. J. , Marandino C. A. , Miller S. D. , Law C. S. , Smith M. J. , and Saltzman E. S. , 2015: Dimethylsulfide gas transfer coefficients from algal blooms in the Southern Ocean. Atmos. Chem. Phys., 15, 17831794, doi:10.5194/acp-15-1783-2015.

    • Search Google Scholar
    • Export Citation
  • Blomquist, B. W., Fairall C. W. , Huebert B. J. , Kieber D. J. , and Westby G. R. , 2006: DMS sea-air transfer velocity: Direct measurements by eddy covariance and parameterization based on the NOAA/COARE gas transfer model. Geophys. Res. Lett., 33, L07601, doi:10.1029/2006GL025735.

    • Search Google Scholar
    • Export Citation
  • Blomquist, B. W., Huebert B. J. , Fairall C. W. , Bariteau L. , Edson J. B. , Hare J. E. , and McGillis W. R. , 2014: Advances in air–sea CO2 flux measurement by eddy correlation. Bound.-Layer Meteor., 152, 245–276, doi:10.1007/s10546-014-9926-2.

    • Search Google Scholar
    • Export Citation
  • Dunckel, M., Hasse L. , Krügermeyer L. , Schriever D. , and Wucknitz J. , 1974: Turbulent fluxes of momentum, heat and water vapor in the atmospheric surface layer at sea during ATEX. Bound.-Layer Meteor., 6, 81106, doi:10.1007/BF00232478.

    • Search Google Scholar
    • Export Citation
  • Edson, J. B., Hinton A. A. , Prada K. E. , Hare J. E. , and Fairall C. W. , 1998: Direct covariance flux estimates from mobile platforms at sea. J. Atmos. Oceanic Technol., 15, 547562, doi:10.1175/1520-0426(1998)015<0547:DCFEFM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Edson, J. B., and Coauthors, 2011: Direct covariance measurement of CO2 gas transfer velocity during the 2008 Southern Ocean Gas Exchange Experiment: Wind speed dependency. J. Geophys. Res., 116, C00F10, doi:10.1029/2011JC007022.

    • Search Google Scholar
    • Export Citation
  • Edson, J. B., and Coauthors, 2013: On the exchange of momentum over the open ocean. J. Phys. Oceanogr., 43, 15891610, doi:10.1175/JPO-D-12-0173.1.

    • Search Google Scholar
    • Export Citation
  • Else, B. G. T., Papakyriakou T. N. , Galley R. J. , Drennan W. M. , Miller L. A. , and Thomas H. , 2011: Wintertime CO2 fluxes in an Arctic polynya using eddy covariance: Evidence for enhanced air-sea gas transfer during ice formation. J. Geophys. Res., 116, C00G03, doi:10.1029/2010JC006760.

    • Search Google Scholar
    • Export Citation
  • Fairall, C. W., Bradley E. F. , Hare J. E. , Grachev A. A. , and Edson J. B. , 2003: Bulk parameterization of air–sea fluxes: Updates and verification for the COARE algorithm. J. Climate, 16, 571591, doi:10.1175/1520-0442(2003)016<0571:BPOASF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Fairall, C. W., and Coauthors, 2011: Implementation of the Coupled Ocean-Atmosphere Response Experiment flux algorithm with CO2, dimethyl sulfide, and O3. J. Geophys. Res., 116, C00F09, doi:10.1029/2010JC006884.

    • Search Google Scholar
    • Export Citation
  • Fujitani, T., 1985: Method of turbulent flux measurement on a ship by using a stable platform system. Pap. Meteor. Geophys., 36, 157170, doi:10.2467/mripapers.36.157.

    • Search Google Scholar
    • Export Citation
  • Goulden, M. L., Munger J. W. , Fan S. M. , Daube B. C. , and Wofsy S. C. , 1996: Measurements of carbon sequestration by long-term eddy covariance: Methods and a critical evaluation of accuracy. Global Change Biol., 2, 169182, doi:10.1111/j.1365-2486.1996.tb00070.x.

    • Search Google Scholar
    • Export Citation
  • Grachev, A. A., Andreas E. L , Fairall C. W. , Guest P. S. , and Persson P. O. G. , 2007: SHEBA flux–profile relationships in the stable atmospheric boundary layer. Bound.-Layer Meteor., 124, 315333, doi:10.1007/s10546-007-9177-6.

    • Search Google Scholar
    • Export Citation
  • Hall, R. J., Hughes N. , and Wadhams P. , 2002: A systematic method of obtaining ice concentration measurements from ship-based observations. Cold Reg. Sci. Technol., 34, 97102, doi:10.1016/S0165-232X(01)00057-X.

    • Search Google Scholar
    • Export Citation
  • Ibrom, A., Dellwik E. , Flyvbjerg H. , Jensen N. O. , and Pilegaard K. , 2007: Strong low-pass filtering effects on water vapour flux measurements with closed-path eddy correlation systems. Agric. For. Meteor., 147, 140156, doi:10.1016/j.agrformet.2007.07.007.

    • Search Google Scholar
    • Export Citation
  • Kaimal, J. C., Wyngaard J. C. , Izumi Y. , and Coté O. R. , 1972: Spectral characteristics of surface-layer turbulence. Quart. J. Roy. Meteor. Soc., 98, 563589, doi:10.1002/qj.49709841707.

    • Search Google Scholar
    • Export Citation
  • Kohsiek, W., 2000: Water vapor cross-sensitivity of open path H2O/CO2 sensors. J. Atmos. Oceanic Technol., 17, 299311, doi:10.1175/1520-0426(2000)017<0299:WVCSOO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Landwehr, S., Miller S. D. , Smith M. J. , Saltzman E. S. , and Ward B. , 2014: Analysis of the PKT correction for direct CO2 flux measurements over the ocean. Atmos. Chem. Phys., 14, 33613372, doi:10.5194/acp-14-3361-2014.

    • Search Google Scholar
    • Export Citation
  • Landwehr, S., O’Sullivan N. , and Ward B. , 2015: Direct flux measurements from mobile platforms at sea: Motion and airflow distortion corrections revisited. J. Atmos. Oceanic Technol., 32, 1163–1178, doi:10.1175/JTECH-D-14-00137.1.

    • Search Google Scholar
    • Export Citation
  • Le Quéré, C., and Coauthors, 2007: Saturation of the Southern Ocean CO2 sink due to recent climate change. Science, 316, 1735–1738, doi:10.1126/science.1136188.

    • Search Google Scholar
    • Export Citation
  • Loose, B., McGillis W. R. , Perovich D. , Zappa C. J. , and Schlosser P. , 2014: A parameter model of gas exchange for the seasonal sea ice zone. Ocean Sci., 10, 1728, doi:10.5194/os-10-17-2014.

    • Search Google Scholar
    • Export Citation
  • Marandino, C. A., De Bruyn W. J. , Miller S. D. , and Saltzman E. S. , 2007: Eddy correlation measurements of the air/sea flux of dimethylsulfide over the North Pacific Ocean. J. Geophys. Res., 112, D03301, doi:10.1029/2006JD007293.

    • Search Google Scholar
    • Export Citation
  • McGillis, W. R., Edson J. B. , Hare J. E. , and Fairall C. W. , 2001: Direct covariance air-sea CO2 fluxes. J. Geophys. Res., 106, 16 72916 745, doi:10.1029/2000JC000506.

    • Search Google Scholar
    • Export Citation
  • McMillen, R. T., 1988: An eddy correlation technique with extended applicability to non-simple terrain. Bound.-Layer Meteor., 43, 231245, doi:10.1007/BF00128405.

    • Search Google Scholar
    • Export Citation
  • Miller, S. D., Goulden M. L. , Menton M. C. , da Rocha H. R. , de Freitas H. C. , Figueira A. M. S. , and de Sousa C. A. D. , 2004: Biometric and micrometeorological measurements of tropical forest carbon balance. Ecol. Appl., 14, 114126, doi:10.1890/02-6005.

    • Search Google Scholar
    • Export Citation
  • Miller, S. D., Hristov T. S. , Edson J. B. , and Friehe C. A. , 2008: Platform motion effects on measurements of turbulence and air–sea exchange over the open ocean. J. Atmos. Oceanic Technol., 25, 16831694, doi:10.1175/2008JTECHO547.1.

    • Search Google Scholar
    • Export Citation
  • Miller, S. D., Marandino C. A. , and Saltzman E. S. , 2010: Ship-based measurement of air-sea CO2 exchange by eddy covariance. J. Geophys. Res., 115, D02304, doi:10.1029/2009JD012193.

    • Search Google Scholar
    • Export Citation
  • Mitsuta, Y., and Fujitani T. , 1974: Direct measurement of turbulent fluxes on a cruising ship. Bound.-Layer Meteor., 6, 203217, doi:10.1007/BF00232485.

    • Search Google Scholar
    • Export Citation
  • Moat, B. I., and Yelland M. J. , 2008: Going with the flow: State of the art marine meteorological measurements on the new NERC research vessel. Weather, 63, 158159, doi:10.1002/wea.184.

    • Search Google Scholar
    • Export Citation
  • O’Sullivan, N., Landwehr S. , and Ward B. , 2013: Mapping flow distortion on oceanographic platforms using computational fluid dynamics. Ocean Sci., 9, 855866, doi:10.5194/os-9-855-2013.

    • Search Google Scholar
    • Export Citation
  • O’Sullivan, N., Landwehr S. , and Ward B. , 2015: Air-flow distortion and wave interactions on research vessels: An experimental and numerical comparison. Methods Oceanogr., 12, 117, doi:10.1016/j.mio.2015.03.001.

    • Search Google Scholar
    • Export Citation
  • Paulson, C. A., 1970: The mathematical representation of wind speed and temperature profiles in the unstable atmospheric surface layer. J. Appl. Meteor., 9, 857861, doi:10.1175/1520-0450(1970)009<0857:TMROWS>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Popinet, S., Smith M. J. , and Stevens C. , 2004: Experimental and numerical study of the turbulence characteristics of airflow around a research vessel. J. Atmos. Oceanic Technol., 21, 15751589, doi:10.1175/1520-0426(2004)021<1575:EANSOT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Prytherch, J., Yelland M. J. , Pascal R. W. , Moat B. I. , Skjelvan I. , and Neill C. C. , 2010: Direct measurements of the CO2 flux over the ocean: Development of a novel method. Geophys. Res. Lett., 37, L03607, doi:10.1029/2009GL041482.

    • Search Google Scholar
    • Export Citation
  • Prytherch, J., Yelland M. J. , Brooks I. M. , Tupman D. J. , Pascal R. W. , Moat B. I. , and Norris S. J. , 2015: Motion-correlated flow distortion and wave-induced biases in air-sea flux measurements from ships. Atmos. Chem. Phys., 15, 10 61910 629, doi:10.5194/acp-15-10619-2015.

    • Search Google Scholar
    • Export Citation
  • Rintoul, S. R., and Coauthors, 2010: Southern Ocean Observing System (SOOS): Rationale and strategy for sustained observations of the Southern Ocean. Proceedings of OceanObs’09: Sustained Ocean Observations and Information for Society, J. Hall, D. E. Harrison, and D. Stammer, Eds., Vol. 2, ESA Publ. WPP-306, doi:10.5270/OceanObs09.cwp.74.

    • Search Google Scholar
    • Export Citation
  • Schotanus, P., Nieuwstadt F. T. M. , and de Bruin H. A. R. , 1983: Temperature measurement with a sonic anemometer and its application to heat and moisture fluxes. Bound.-Layer Meteor., 26, 8193, doi:10.1007/BF00164332.

    • Search Google Scholar
    • Export Citation
  • Smith, S. D., 1980: Wind stress and heat flux over the ocean in gale force winds. J. Phys. Oceanogr., 10, 709726, doi:10.1175/1520-0485(1980)010<0709:WSAHFO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Takahashi, T., Sweeney C. , Hales B. , Chipman D. W. , Newberger T. , Goddard J. , Iannuzzi R. , and Sutherland S. C. , 2012: The changing carbon cycle in the Southern Ocean. Oceanography, 25 (3), 2637, doi:10.5670/oceanog.2012.71.

    • Search Google Scholar
    • Export Citation
  • Takahashi, T., Sweeney C. , and Sutherland S. C. , 2014: Underway pCO2 measurements in surface waters and the atmosphere during the R/V Nathaniel B. Palmer 2013 expeditions. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, accessed 11 March 2016, doi:10.3334/CDIAC/OTG.VOS_PALMER_2013.

  • Takahashi, T., Sweeney C. , and Sutherland S. C. , 2016: Underway pCO2 measurements in surface waters and the atmosphere during the R/V Nathaniel B. Palmer 2014 expeditions. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, accessed 11 March 2016, doi:10.3334/CDIAC/OTG.VOS_PALMER_2014.

  • Wanninkhof, R., Asher W. E. , Ho D. T. , Sweeney C. , and McGillis W. R. , 2009: Advances in quantifying air-sea gas exchange and environmental forcing. Annu. Rev. Mar. Sci., 1, 213244, doi:10.1146/annurev.marine.010908.163742.

    • Search Google Scholar
    • Export Citation
  • Webb, E. K., Pearman G. I. , and Leuning R. , 1980: Correction of flux measurements for density effects due to heat and water vapour transfer. Quart. J. Roy. Meteor. Soc., 106, 85100, doi:10.1002/qj.49710644707.

    • Search Google Scholar
    • Export Citation
  • Wilczak, J. M., Oncley S. P. , and Stage S. A. , 2001: Sonic anemometer tilt correction algorithms. Bound.-Layer Meteor., 99, 127150, doi:10.1023/A:1018966204465.

    • Search Google Scholar
    • Export Citation
  • Yelland, M. J., Moat B. I. , Pascal R. W. , and Berry D. I. , 2002: CFD model estimates of the airflow distortion over research ships and the impact on momentum flux measurements. J. Atmos. Oceanic Technol., 19, 14771499, doi:10.1175/1520-0426(2002)019<1477:CMEOTA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Yelland, M. J., Pascal R. W. , Taylor P. K. , and Moat B. I. , 2009: AutoFlux: An autonomous system for the direct measurement of the air–sea fluxes of CO2, heat and momentum. J. Oper. Oceanogr., 2, 1523, doi:10.1080/1755876X.2009.11020105.

    • Search Google Scholar
    • Export Citation
  • Zemmelink, H. J., Delille B. , Tison J. L. , Hintsa E. J. , Houghton L. , and Dacey J. W. H. , 2006: CO2 deposition over the multi-year ice of the western Weddell Sea. Geophys. Res. Lett., 33, L13606, doi:10.1029/2006GL026320.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Photographs of the air–sea flux package installed on the RV/IB NBP bow mast from 1 Jan 2013 to 28 May 2014: (a) ship bow dimensions, (b) bow mast instrument configuration, (c) additional R. M. Young 3D sonic anemometers added 25 Jun to 13 Nov 2013, and (d) icing event that left the bow mast ice covered from 20 Aug to 8 Sep 2013.

  • View in gallery

    Flow diagram for air–sea flux system installed in the bosun’s locker of the RV/IB NBP from 1 Jan 2013 to 28 May 2014. The 56-slpm base flow from the mast inlet was split into three lines (bypass = 32 slpm, wet = 11.8 slpm, and dry = 11.8 slpm) upon entering the rackmount enclosure and recombined before leaving the enclosure and reaching the pump. MFC = mass flow controller, P = Mensor 6100 pressure sensor, and S = solenoid valve (two-way = S1, S2, S3; three-way = S4, S5, S6). Solenoid valve S1 was used for inlet tests; the remaining solenoid valves were part of the backflush system. MFC purge had a flow of 6 slpm.

  • View in gallery

    Data acquisition and remote monitoring strategy for air–sea flux system installed on the RV/IB NBP from 1 Jan 2013 to 28 May 2014.

  • View in gallery

    Flow distortion analyses for RV/IB NBP. The x axis is the wind direction relative to the ship (°) measured by CSAT3-stbd. (a) Azimuthal wind deflection (°) relative to CSAT3-stbd for CSAT3-port (black squares), CSAT3-stbd (gray circles), RMY-low (black diamonds), RMY-port (gray triangles), and science mast 2D anemometer (white stars). (b) Wind vector angle of attack (°), same line styles as in (a), with planar fit CSAT3-port (black dashed) and CSAT3-stbd (gray) curves close to zero. (c) Difference in wind stress normalized by CSAT3-stbd. (d) Wind speed difference between science mast and bow mast anemometers (with gray scale according to U10n). All curves for cruise NBP-1305 except “planar fit corrected” CSAT3-port and stbd curves in (b), which were for cruises NBP-1210 and NBP-1402.

  • View in gallery

    Median flux cospectra from CSAT3-stbd measured on RV/IB NBP cruises NBP-1210 and NBP-1402, normalized and frequency weighted: (a) measured momentum flux (black dashed) and Prytherch et al. (2015) corrected (solid gray); (b) as in (a), but for σheave > 0.75 m only; (c) as in (a), but for σheave < 0.75 m; (d) sensible heat flux (black dashed); (e) latent heat flux for open-path LI-7500 (black dashed) and closed-path LI7200-wet (solid gray); and (f) CO2 flux for dried airstream (black dashed), undried airstream (thick solid gray), and dried airstream with |ΔpCO2| < 40 ppm (thin gray with dots). Semiempirical cospectra of K72 shown in all subplots as thin solid black curve.

  • View in gallery

    Momentum flux (N m−2) vs 10-m neutral wind speed for the RV/IB NBP cruises NBP-1210 and NBP-1402: U10n calculated using bow mast anemometer (all data, black squares), U10n calculated using science mast anemometer (all data, gray circles), Prytherch et al. (2015)-corrected τ (all data, black x’s), U10n calculated using science mast anemometer for flux intervals when σheave was less than 0.75 m (light gray triangles), NOAA COARE 3.5 algorithm (black dashed line), and Smith (1980) (gray dashed line). Data are restricted to relative wind direction between ±30° of the bow.

  • View in gallery

    Ship tracks of nine RV/IB NBP cruises between 1 Jan 2013 and 28 May 2014 when the University at Albany air–sea flux system was installed. Gray represents regions where |ΔpCO2| < 40 ppm. Black represents regions where |ΔpCO2| > 40 ppm.

  • View in gallery

    Meteorology summary for the RV/IB NBP cruises (left) NBP-1210 and (right) NBP-1402. Shown are 10-min averages: (a),(b) latitude and longitude (°); (c),(d) air temperature and SST (°C); (e),(f) 10-m neutral wind speed (m s−1); and (g),(h) ΔpCO2 (ppm), where black circles represent open water and gray triangles represent sea ice.

  • View in gallery

    Normalized difference between measured and bulk (W09) CO2 flux plotted against ΔpCO2 for RV/IB NBP cruises NBP-1210 and NBP-1402. Shading of points represents 10-m neutral wind speed.

  • View in gallery

    Flux summary for the RV/IB NBP cruises (left) NBP-1210 and (right) NBP-1402. Fluxes are shown as 3-h averages of 10-min fluxes: (a),(b) momentum flux (N m−2); (c),(d) sensible heat flux (W m−2); (e),(f) closed-path latent heat flux (W m−2); and (g),(h) CO2 flux (mol m−2 yr−1). Black circles represent measured open water fluxes, and gray triangles represent measured fluxes in the SIZ. Gray lines represent the NOAA COARE algorithm (version 3.5 for momentum flux and version 3.0 for Hs and HL) and the W09 quadratic parameterization.

  • View in gallery

    The 10-m neutral drag coefficient vs 10-m neutral wind speed (m s−1) measured from the RV/IB NBP cruises NBP-1210 and NBP-1402: open water (black circles), 0%–20% SIC (gray triangles), 20%–80% SIC (x’s), and 80%–100% SIC (white squares). The black dashed line represents the output of the NOAA COARE 3.5 algorithm (Edson et al. 2013). The gray dashed–dotted line represents the parameterization of Smith (1980). On the right-hand side of the plot are the wind speed–independent CD10n values from the SIC-dependent parameterization of Andreas et al. (2010), calculated using the mean SIC for each of the three ice groups.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 262 124 11
PDF Downloads 268 129 9

Automated Underway Eddy Covariance System for Air–Sea Momentum, Heat, and CO2 Fluxes in the Southern Ocean

View More View Less
  • 1 Atmospheric Sciences Research Center, University at Albany, State University of New York, Albany, New York
Full access

Abstract

A ruggedized closed-path eddy covariance (EC) system was designed for unattended direct measurements of air–sea momentum, heat, and CO2 flux, and was deployed on the Research Vessel Icebreaker (RV/IB) Nathaniel B. Palmer (NBP), an Antarctic research and supply vessel. The system operated for nine cruises during 18 months from January 2013 to June 2014 in the Southern Ocean and coastal Antarctica, sampling a wide variety of wind, wave, biological productivity, and ice conditions. The methods are described and the results are shown for two cruises chosen for their latitudinal range, inclusion of both open water and sea ice cover, and relatively large air–water CO2 concentration differences (ΔpCO2). Ship flow distortion was addressed by comparing mean winds, fluxes, and cospectra from an array of 3D anemometers at the NBP bow, comparing measured fluxes with bulk formulas, and implementing and evaluating several recently published data processing techniques. Quality-controlled momentum, heat, and CO2 flux data were obtained for 25% of the periods when NBP was at sea, with most (86%) of the rejected periods due to wind directions relative to the ship >±30° from the bow. In contrast to previous studies, no bias was apparent in measured CO2 fluxes for low |ΔpCO2|. The relationship between momentum flux and wind speed showed a clear dependence on the degree of sea ice cover, a result facilitated by the geographical coverage possible with a ship-based approach. These results indicate that ship-based unattended EC in high latitudes is feasible, and recommendations for deployments of underway systems in such environments are provided.

Corresponding author address: Scott D. Miller, Atmospheric Sciences Research Center, University at Albany, State University of New York, 251 Fuller Road, Albany, NY 12203. E-mail: smiller@albany.edu

Abstract

A ruggedized closed-path eddy covariance (EC) system was designed for unattended direct measurements of air–sea momentum, heat, and CO2 flux, and was deployed on the Research Vessel Icebreaker (RV/IB) Nathaniel B. Palmer (NBP), an Antarctic research and supply vessel. The system operated for nine cruises during 18 months from January 2013 to June 2014 in the Southern Ocean and coastal Antarctica, sampling a wide variety of wind, wave, biological productivity, and ice conditions. The methods are described and the results are shown for two cruises chosen for their latitudinal range, inclusion of both open water and sea ice cover, and relatively large air–water CO2 concentration differences (ΔpCO2). Ship flow distortion was addressed by comparing mean winds, fluxes, and cospectra from an array of 3D anemometers at the NBP bow, comparing measured fluxes with bulk formulas, and implementing and evaluating several recently published data processing techniques. Quality-controlled momentum, heat, and CO2 flux data were obtained for 25% of the periods when NBP was at sea, with most (86%) of the rejected periods due to wind directions relative to the ship >±30° from the bow. In contrast to previous studies, no bias was apparent in measured CO2 fluxes for low |ΔpCO2|. The relationship between momentum flux and wind speed showed a clear dependence on the degree of sea ice cover, a result facilitated by the geographical coverage possible with a ship-based approach. These results indicate that ship-based unattended EC in high latitudes is feasible, and recommendations for deployments of underway systems in such environments are provided.

Corresponding author address: Scott D. Miller, Atmospheric Sciences Research Center, University at Albany, State University of New York, 251 Fuller Road, Albany, NY 12203. E-mail: smiller@albany.edu

1. Introduction

Air–sea exchange at high latitudes in the Southern Hemisphere affects the global climate system. The fluxes of momentum and heat affect cold-water formation and the global oceanic circulation (Rintoul et al. 2010), and the Southern Ocean is responsible for roughly half the CO2 absorbed by the world’s oceans (Takahashi et al. 2012). The Southern Ocean is also changing—warming more rapidly than the global average sea temperature (Rintoul et al. 2010)—and some studies report the strength of the carbon sink is decreasing (Le Quéré et al. 2007). Despite their importance, air–sea fluxes are poorly sampled throughout the Southern Ocean compared with other regions, and there remains a critical need for systems designed to return high-quality measurements in all seasons and sea states.

The micrometeorological technique eddy covariance (EC) provides direct measurements of air–sea fluxes at small spatial (1–10 km) and temporal (1 h) scales. The high precision of EC makes it particularly well suited for examining differences in air–sea fluxes due to temporal changes (e.g., seasonal), geographical area, ecosystem types, and surface conditions (e.g., wind, waves, ice cover). Short-term (1–4 months) ice field stations in the Antarctic have been effectively used to measure turbulent momentum and heat fluxes (Andreas et al. 2005) and carbon dioxide flux (Zemmelink et al. 2006) using EC. A complementary strategy is to deploy EC systems on Antarctic research and supply vessels to sample the wide range of high-latitude Southern Ocean environments, from high wind and wave latitude bands to the cold, productive sea ice zone (SIZ). The feasibility of this approach depends upon developing EC systems sufficiently robust to overcome the technical and logistical challenges of measuring air–sea CO2 fluxes in harsh environments.

Since the 1970s (Dunckel et al. 1974), advances in technology and methodology have improved EC instrument performance and reliability at sea. Robust corrections for the effects of platform motion on the measured turbulent wind vector have been developed (Mitsuta and Fujitani 1974; Fujitani 1985; Edson et al. 1998), modeling results have increased our understanding of the effects of airflow distortion on both mean flow and fluxes (Yelland et al. 2002; Popinet et al. 2004), and techniques for computing turbulent fluxes that address residual flux contamination due to motion and flow distortion continue to evolve (Landwehr et al. 2015; Prytherch et al. 2015). Compared with momentum and heat fluxes, direct covariance air–sea carbon dioxide flux measurements have advanced more slowly (McGillis et al. 2001). Existing ship-based CO2 fluxes at high latitudes were collected using open-path style sensors (Yelland et al. 2009; Prytherch et al. 2010; Else et al. 2011); however, open-path sensors at sea have been shown to be highly sensitive to contamination that can result in measured CO2 fluxes several orders of magnitude larger than expected (Kohsiek 2000). To address this, Prytherch et al. (2010) proposed an empirical [Peter K. Taylor (PKT)] correction, but Landwehr et al. (2014) showed that the PKT correction is arbitrary and cannot return reliable fluxes, and conclude that a closed-path approach and sample airstream drying are required to obtain reasonable air–sea CO2 flux measurements, consistent with the conclusions of Miller et al. (2010). These advanced techniques have been applied to short manned field campaigns at low–midlatitudes but have not been reported for long-term ship deployments at high latitudes.

We designed an underway EC system for long-term measurements of air–sea momentum, heat, and carbon dioxide fluxes in the Southern Ocean and Antarctic coastal waters. The system was based on the design of Miller et al. (2010) with additional considerations for the harsh Southern Ocean environment. The system was installed on the U.S. Research Vessel Icebreaker (RV/IB) Nathaniel B. Palmer (NBP) during nine cruises over 18 months (January 2013–June 2014), and sampled a variety of wind, wave, and sea ice conditions. During most of this period, the system was unattended and was monitored remotely via automated daily system reports and summary data sent by satellite. Here we describe the system design and performance and show summary flux data from two cruises. We conclude that unattended direct flux measurements in harsh environments are challenging yet feasible and provide recommendations for deployments of underway systems in such environments.

2. Methods

a. RV/IB Nathaniel B. Palmer

The Research Vessel Icebreaker (RV/IB) Nathaniel B. Palmer is a research and supply vessel (length = 94 m, width = 18 m, draft = 7 m) that logs roughly 50 000 km yr−1, mostly in the Southern Ocean and coastal Antarctica. The NBP’s onboard meteorological sensors (Table 1) included two 2D sonic anemometers (1390-PK-062, Gill) that were mounted to the science mast (30 m MSL), one port and one starboard (Fig. 1a). To address flow distortion, the NBP wind product was a composite of the two anemometers, with wind speed and direction reported for the anemometer that recorded the highest wind speed. Seawater measurements, which included a pCO2 system operated by Lamont-Doherty Earth Observatory (methods described in Takahashi et al. 2014, 2016), were collected in the ship’s hydrolaboratory using water from the uncontaminated seawater system (with the main intake located near the stern thruster). EC sensors were mounted on the NBP bow mast, a 7.3-m-tall, 28-cm-diameter steel pipe set back 2.5 m aft of the bow (Fig. 1a). The ship’s bridge superstructure was 15 m aft of the bow mast and had a maximum width of 22.5 m and a height of 19 m MSL. A 3-m-tall crane was located 4 m aft and 3 m starboard of the bow mast.

Table 1.

RV/IB Nathaniel B. Palmer underway systems and SUNY Albany air–sea flux instruments.

Table 1.
Fig. 1.
Fig. 1.

Photographs of the air–sea flux package installed on the RV/IB NBP bow mast from 1 Jan 2013 to 28 May 2014: (a) ship bow dimensions, (b) bow mast instrument configuration, (c) additional R. M. Young 3D sonic anemometers added 25 Jun to 13 Nov 2013, and (d) icing event that left the bow mast ice covered from 20 Aug to 8 Sep 2013.

Citation: Journal of Atmospheric and Oceanic Technology 33, 4; 10.1175/JTECH-D-15-0156.1

b. Eddy covariance measurements

The EC system was designed and built at the University at Albany in summer 2012 and installed on NBP in Punta Arenas, Chile, in December 2012. Air–sea flux sensors were attached to a 1.4-m-tall powder-coated steel frame bolted to the bow mast (Fig. 1b, item b). EC sensor electronics were housed in a 61 cm × 51 cm × 25 cm National Electrical Manufacturers Association (NEMA) 4X aluminum enclosure (Fig. 1b, item c) mounted to a welded steel frame just starboard of the bow mast (Fig. 1b, item d). Additional sensors and electronics were installed below the 02 Deck in the bosun’s locker (Fig. 1a). A 6-cm-diameter vertical steel pipe (Fig. 1b, item e) spanned the length of the bow mast, connected on one end to the base of the NEMA 4X electronics enclosure and on the other end to the 02 Deck bulkhead leading into the bosun’s locker, and was used to protect sample air tubing and electronic cables running from the top of the mast to the bosun’s locker.

1) Wind vector and platform motion

Two 3D sonic anemometers (CSAT3, Campbell Scientific) measured the turbulent wind vector relative to the ship and “sonic” temperature at a 20-Hz sample rate (Fig. 1b, items f,g). The anemometers were separated horizontally by 0.6 m, one to the port side of the bow mast (CSAT3-port) and one to the starboard side (CSAT3-stbd), with their measurement volumes 1 m forward of the bow mast and 6 m above the 02 Deck (4.5 m above the edge of the hull and 14 m MSL). In June 2013, two additional 3D sonic anemometers (model 81000, R. M. Young, Inc.) were installed to assess the flow distortion in the immediate vicinity of the mast (Fig. 1c, items a,b). One of the R. M. Young anemometers (RMY-low) was mounted 1.3 m forward of the bow mast and was aligned 1.3 m below the CSAT3-stbd anemometer. The second R. M. Young anemometer (RMY-port) was mounted 0.05 m above, 0.75 m aft, and 1.32 m to the port side of CSAT3-port.

Two independent strapped-down systems measured ship motion. The first system included three sensors: an analog inertial measurement unit (Motionpak II, Systron Donner; Fig. 1b, item h) to measure linear accelerations and angular rates; a digital compass (C100, KVH) to measure heading; and a GPS (16x-HVS, Garmin Inc.) to measured latitude, longitude, and ship speed. The MotionPak II was located 0.63 m aft and 0.15 m starboard of the CSAT3-port anemometer. Instantaneous Euler angles (pitch, roll, and yaw) and platform velocities were calculated from this system according to Miller et al. (2008). The second motion measurement was a single unit (3DM-GX3-35, Microstrain) located 0.76 m directly aft of the CSAT3-port that integrated measurements of acceleration, angle rate, heading, location, and speed to calculate and output the Euler angles. The comparisons of corrected winds between Systron Donner and Microstrain showed good agreement, and the results based on the two motion measurement systems are used interchangeably in the text and figures.

2) H2O and CO2 mixing ratio

Fast-response water vapor was measured using two independent methods. First, an open-path infrared gas analyzer (IRGA; LI-7500, LI-COR Inc.) mounted with its measurement volume 0.2 m aft and 0.2 m to port of CSAT3-stbd (Fig. 1b, item i) was used to measure water vapor molar density at a 20-Hz sample rate. The water vapor mixing ratio was calculated from the measured molar density, fast-response sonic temperature from CSAT3-stbd, and pressure measured inside of the LI-7500 electronics box mounted approximately 1.5 m below and 1.5 m to the port side of the LI-7500 measurement volume (Miller et al. 2004).

The second measurement of water vapor was made using a closed-path IRGA (LI-7200, LI-COR Inc.), referred to as “LI7200-wet,” housed in a 1.1 m × 0.7 m × 1.2 m ruggedized rackmount case (Hardigg Industries) in the bosun’s locker (Figs. 1a, 2). Air was drawn at a base flow rate of 56 standard L min−1 (slpm) from an inlet located between the CSAT3-stbd and CSAT3-port anemometers on the bow mast (Fig. 1b, item m) and into the bosun’s locker through 10 m of 1.3-cm-diameter stainless steel heated tubing (16.4 W m−1, SL-522-B0849A self-regulating Temptrace, Parker-Balston) using a rotary vane pump (model 5HCD-100TA-M550NGX, Gast Manufacturing; Fig. 2). The base flow was routed into the rackmount case, where an 11.8-slpm sample flow was split off and passed through ~1 m of 0.95-cm-diameter tubing (Synflex Type “1300,” Eaton Corp.) and through LI7200-wet. Both the base flow and sample flow were turbulent (Reynolds numbers roughly 8000 and 2800, respectively), and the heated tubing maintained the sample air temperature measured at LI7200-wet approximately 10°C above the ambient temperature at the bow mast air inlet. The mean pressure in the LI7200-wet sample cell was 89.2 kPa, measured by its onboard sensor and by an external fast-response pressure sensor (model 6100, Mensor Corp.) with a broader measurement range.

Fig. 2.
Fig. 2.

Flow diagram for air–sea flux system installed in the bosun’s locker of the RV/IB NBP from 1 Jan 2013 to 28 May 2014. The 56-slpm base flow from the mast inlet was split into three lines (bypass = 32 slpm, wet = 11.8 slpm, and dry = 11.8 slpm) upon entering the rackmount enclosure and recombined before leaving the enclosure and reaching the pump. MFC = mass flow controller, P = Mensor 6100 pressure sensor, and S = solenoid valve (two-way = S1, S2, S3; three-way = S4, S5, S6). Solenoid valve S1 was used for inlet tests; the remaining solenoid valves were part of the backflush system. MFC purge had a flow of 6 slpm.

Citation: Journal of Atmospheric and Oceanic Technology 33, 4; 10.1175/JTECH-D-15-0156.1

The fast-response CO2 mixing ratio was measured using two closed-path IRGAs (LI-7200, LI-COR Inc.) installed in the rackmount case in the bosun’s locker. One of the IRGAs was the same sensor (LI7200-wet) that measured the fast-response water vapor mixing ratio (described above). The second sensor (LI7200-dry) sampled a separate 11.8-slpm airstream split off of the 56-slpm base flow; however, a moisture exchanger (Nafion PD-200T-12MSS, Perma Pure) was installed immediately upstream of this IRGA (Fig. 2). A 6-slpm dry air counter flow for the moisture exchanger was provided by a zero air generator (model 75-45, Parker Inc.). The use of the Nafion reduced the moisture content of the sample airstream and dramatically reduced the water vapor fluctuations and thus the Webb correction (Webb et al. 1980; Miller et al. 2010). For both LI7200-wet and LI7200-dry, the Webb correction was calculated by LI-7200 onboard electronics. A motion correction was also applied to the LI-7200 CO2 mixing ratios by subtracting a multiple linear regression of CO2 mixing ratio against measured accelerations and angle rates to eliminate spurious signals due to the physical movement of the IRGAs (Miller et al. 2010).

3) Data acquisition and processing

The data acquisition and processing scheme is shown in Fig. 3. Turbulence data were collected at 20 Hz using two CR3000 dataloggers (Campbell Scientific). One CR3000 (NEMA) was installed in the NEMA 4X electronics box on the mast (Fig. 1b, item c) and the other CR3000 (BOSUN) was installed in the rackmount case in the bosun’s locker. To minimize data loss, some signals were logged on both NEMA and BOSUN loggers. To facilitate synchronization of NEMA and BOSUN data streams, a GPS pulse-per-second signal was used to adjust their respective clocks. In the event that this method failed, the datalogger clocks were synchronized during postprocessing using the wind and motion measurements collected on both dataloggers. The dataloggers were connected to the ship’s Ethernet, and their data were polled every 5 min by a rackmount server (server 1, Supermicro, CSE-813M) running LoggerNet software (Campbell Scientific) located in the ship’s forward dry laboratory (Fig. 1a).

Fig. 3.
Fig. 3.

Data acquisition and remote monitoring strategy for air–sea flux system installed on the RV/IB NBP from 1 Jan 2013 to 28 May 2014.

Citation: Journal of Atmospheric and Oceanic Technology 33, 4; 10.1175/JTECH-D-15-0156.1

Fluxes of momentum (τ, N m−2), sensible heat (HS, W m−2), latent heat (HL, W m−2), and CO2 (Fc, mol m−2 yr−1) were calculated for 10-min intervals as
e1
e2
e3
e4
where u, υ, and w (m s−1) are the motion-corrected along-wind, crosswind, and vertical wind components, respectively; ρa (mol m−3) is the dry air density; cp (J kg−1 K−1) is specific heat capacity of air; T (K) is dry air temperature; Lυ (J kg−1) is latent heat of vaporization (or latent heat of sublimation when air temperature was below 0°C; Andreas et al. 2013); q (kg kg−1) is specific humidity; c is CO2 mixing ratio; primes indicate fluctuations about the mean; and the overbar corresponds to the time average. The dry air temperature was calculated from the sonic temperature using the correction of Schotanus et al. (1983) for the effect of water vapor on air density and speed of sound.

4) Motion correction and flow distortion

The measured wind vector was corrected for platform motion following Miller et al. (2008), modified to add the ship speed component after rather than before applying the rotation matrix (Landwehr et al. 2015). Before calculating fluxes, the motion-corrected wind vector was rotated into a reference coordinate frame. Traditionally, a gravity-referenced frame has been used (McMillen 1988) based on the assumption that the mean wind vector over the ocean is horizontal. However, because of flow distortion about the ship superstructure, bow mast, and mounting hardware, an alternative reference coordinate frame may be more appropriate (Wilczak et al. 2001). We examined wind vector azimuthal and elevation angle deflections measured by four 3D anemometers at the bow of the ship, compared measured mean winds and fluxes with well-established bulk formulas, and implemented and evaluated two recently published data processing techniques (Landwehr et al. 2015; Prytherch et al. 2015) to quantify and minimize the effects of flow distortion.

The clustered array of 3D sonic anemometers at the bow was used to calculate azimuthal wind deflections and wind elevation angles (Figs. 4a,b). For bow-on winds, there was a slight (1.7°) portward turning of the wind vector measured by RMY-low compared with CSAT3-stbd that was used as the reference sonic (Fig. 4a, black diamonds). Meanwhile, the wind vector measured by CSAT3-port and RMY-port showed larger portward deflections (7.9° and 6.6°, respectively; Fig. 4a, black squares and gray triangles, respectively). The azimuthal deflection of CSAT3-port and RMY-port sonics showed systematic variation and remained between 5° and 10° as the incoming wind direction measured by CSAT3-stbd varied from −30° to +30°. For incoming angles >45°, the deflections relative to CSAT3-stbd were negative (toward starboard). These patterns in the azimuthal wind angle deflections appear consistent with flow distortion about the bow mast, air–sea flux system components, and mounting hardware (Fig. 1b). In the range of ±30° the measured relative wind direction from the science mast anemometer was generally between the values measured by the CSAT3-port and CSAT3-stbd anemometer (Fig. 4a, white stars).

Fig. 4.
Fig. 4.

Flow distortion analyses for RV/IB NBP. The x axis is the wind direction relative to the ship (°) measured by CSAT3-stbd. (a) Azimuthal wind deflection (°) relative to CSAT3-stbd for CSAT3-port (black squares), CSAT3-stbd (gray circles), RMY-low (black diamonds), RMY-port (gray triangles), and science mast 2D anemometer (white stars). (b) Wind vector angle of attack (°), same line styles as in (a), with planar fit CSAT3-port (black dashed) and CSAT3-stbd (gray) curves close to zero. (c) Difference in wind stress normalized by CSAT3-stbd. (d) Wind speed difference between science mast and bow mast anemometers (with gray scale according to U10n). All curves for cruise NBP-1305 except “planar fit corrected” CSAT3-port and stbd curves in (b), which were for cruises NBP-1210 and NBP-1402.

Citation: Journal of Atmospheric and Oceanic Technology 33, 4; 10.1175/JTECH-D-15-0156.1

The elevation angle of the incoming wind vector (i.e., angle of attack, β) relative to the horizontal plane was calculated for each sonic anemometer as , where is the mean vertical wind speed and is the mean horizontal wind speed. For all sonics, the measured angles of attack were large (10°–15° for winds between −30° and +30° of the bow; Fig. 4b). Therefore, instead of using a gravity-referenced coordinate frame, we rotated the motion-corrected turbulent wind vector into the long-term mean streamlines using the planar fit method (Wilczak et al. 2001) before calculating fluxes. Because the deflection angles were dependent upon wind direction relative to the ship (Fig. 4b), a radially dependent method was employed, with the planar fit angles determined for each 10° wind sector (Landwehr et al. 2015). The resulting angles of attack for CSAT3-stbd and CSAT3-port relative to the planar fit coordinate frame were less than 1° (Fig. 4b).

The motion-corrected uw cospectrum showed residual peaks at ~0.1 Hz, presumably due to incomplete motion correction and/or interaction between motion and flow distortion (Fig. 5a, dotted curve). We found that the residual peaks were large during high-heave periods (standard deviation of heave, σheave, >0.75 m; Fig. 5b) and were small or negligible during low-heave periods (Fig. 5c). Recently, Prytherch et al. (2015) attempted to remove this residual motion contamination by calculating the linear least squares regression between (motion corrected) wind components (vertical and along wind) and six measured components of platform motion (three linear accelerations and three angle rates), and subtracting the regression from the (already motion corrected) wind components. After applying the Prytherch et al. (2015) technique to high-heave periods, the spikes in the cospectra were reduced but not eliminated (Fig. 5b). For low-heave periods, there was little difference between the original and the Prytherch et al. (2015)-corrected cospectra (Fig. 5c), and the impact on the momentum fluxes was small (3.1% mean, 0.28% median). The Prytherch et al. (2015) technique had little effect on the scalar fluxes (heat, CO2). Unless explicitly stated, for the remaining text and figures, we eliminated high-heave periods, and fluxes are shown without the Prytherch et al. (2015) correction.

Fig. 5.
Fig. 5.

Median flux cospectra from CSAT3-stbd measured on RV/IB NBP cruises NBP-1210 and NBP-1402, normalized and frequency weighted: (a) measured momentum flux (black dashed) and Prytherch et al. (2015) corrected (solid gray); (b) as in (a), but for σheave > 0.75 m only; (c) as in (a), but for σheave < 0.75 m; (d) sensible heat flux (black dashed); (e) latent heat flux for open-path LI-7500 (black dashed) and closed-path LI7200-wet (solid gray); and (f) CO2 flux for dried airstream (black dashed), undried airstream (thick solid gray), and dried airstream with |ΔpCO2| < 40 ppm (thin gray with dots). Semiempirical cospectra of K72 shown in all subplots as thin solid black curve.

Citation: Journal of Atmospheric and Oceanic Technology 33, 4; 10.1175/JTECH-D-15-0156.1

The mean measured wind speed was adjusted to neutral stability at 10-m height using a semilogarithmic wind profile and assuming a constant flux layer according to U10n = , where = (τ/ρ)1/2 is the friction velocity (m s−1) measured by CSAT3-stbd, κ is the von Kármán constant of 0.4, z is the measurement height, and z0 is the roughness length (m) calculated as , where represents the stability function of Paulson (1970) for unstable stratification and Grachev et al. (2007) for stable stratification. During open-ocean conditions, when U10n was calculated based on mean wind speed measured from the bow mast CSAT3s (adjusted from 14 m MSL; Fig. 6, black curve with squares), the wind stress measured by CSAT3-stbd exhibited a stronger (steeper) dependence on U10n compared with the NOAA Coupled Ocean–Atmosphere Response Experiment (COARE) 3.5 bulk relationship (Edson et al. 2013). This result suggests the mean wind was underestimated by anemometers at the ship’s bow, a result consistent with ship flow distortion studies that showed decelerated winds measured from bow masts (Yelland et al. 2002). For NBP, much closer agreement between our bow-measured momentum flux and COARE 3.5 was found when adjusting the mean wind speed measured by the 2D sonics on the science mast at 30 m MSL (Fig. 1a, 6, gray curve with circles) to neutral 10-m height. Based on these results, we conclude that the science mast anemometers provided a more accurate mean wind speed measurement, and present U10n based on measurements from the science mast anemometers for the remainder of this paper.

Fig. 6.
Fig. 6.

Momentum flux (N m−2) vs 10-m neutral wind speed for the RV/IB NBP cruises NBP-1210 and NBP-1402: U10n calculated using bow mast anemometer (all data, black squares), U10n calculated using science mast anemometer (all data, gray circles), Prytherch et al. (2015)-corrected τ (all data, black x’s), U10n calculated using science mast anemometer for flux intervals when σheave was less than 0.75 m (light gray triangles), NOAA COARE 3.5 algorithm (black dashed line), and Smith (1980) (gray dashed line). Data are restricted to relative wind direction between ±30° of the bow.

Citation: Journal of Atmospheric and Oceanic Technology 33, 4; 10.1175/JTECH-D-15-0156.1

Flux intervals were rejected when the 10-min mean wind direction relative to the NBP bow (based on CSAT3-stbd) was greater than ±30°. This cutoff was based on the small normalized momentum flux differences for the four 3D sonics (though somewhat larger for RMY-port; Fig. 4c) and the lack of directional dependence of the difference in U10n estimated from the science mast and bow mast anemometers (U10n bowU10n science) when the relative wind direction was between −30° and +30° (Fig. 4d). We note that our range of accepted relative wind directions was more restrictive than previous studies that used cutoffs between ±60° and ±120° (McGillis et al. 2001; Blomquist et al. 2006; Marandino et al. 2007; Miller et al. 2010; Bell et al. 2015).

5) CO2 and H2O fluxes

To calculate covariances between the vertical wind component and CO2 and H2O measured in the bosun’s locker [Eqs. (3) and (4)], the LI7200-wet and LI7200-dry time series were shifted to account for the time delay for the air sample to travel from the inlet on the mast to the closed-path analyzers. The time delays were determined using daily inlet tests controlled by the BOSUN datalogger and solenoid S1 in the NEMA box on the mast (Fig. 1b, item c; Fig. 2). For each test, nine 10-s pulses of purge air from the zero gas generator were injected near the inlet on the mast. The delay for CO2 was 1.95 ± 0.40 s and the delay for H2O was 2.24 ± 0.59 s. The inlet tests were also used to estimate first-order linear filter time constants (0.21 s for CO2 and 3.92 s for H2O) to characterize the loss of high-frequency CO2 and H2O fluctuations (and flux) due to air being drawn through the tubing (Goulden et al. 1996). The time constants were used to low-pass filter the fast-response sonic temperature measurement (Ts), and multipliers (Gc and Gh) for the measured and covariances were calculated as the ratio of unfiltered to filtered wT covariance. To reduce variability in CO2 and H2O fluxes due to variations in Gc and Gh, linear regressions between Gc and Gh and wind speed were calculated and the multipliers for each 10-min flux interval were computed based on the wind speed for that interval. The regression for CO2 was Gc = 0.0012 U + 1.005, and the regression for H2O was Gh = 0.052 U + 1.067. The mean high-frequency correction amounted to 1.3% for CO2 flux and 34.5% for H2O flux, similar to values found by Ibrom et al. (2007).

6) Harsh environment considerations

Compared with campaign-mode (manned) cruises, the unattended Southern Ocean NBP air–sea flux system deployment was especially challenging due to the harsh environment. Extra measures were taken to mitigate the effects of cold temperatures, ice and riming, and liquid water on system performance. As discussed above, the heated tubing bundle was used to warm the air sample stream. We also heated the head of the CSAT3-stbd anemometer to reduce icing and riming. While there are commercially available heated 3D sonic anemometers, we chose to wrap the CSAT3-stbd head with 500 W of heat tape (SRM/E20-1CT, Chromalox Inc.), which was then covered with reflective tape (Fig. 1b, item f). The heat tape was turned on when air temperature at the sonic head, measured using a copper–constantan thermocouple (TT-T-20-TWSH-SLE, Omega, Inc.), dropped below 0°C. The addition of 0.1 slpm of dry air from the zero air generator into the NEMA 4X box at the top of the bow mast helped to keep it relatively dry and pressurized.

A backflush system was installed to reduce the likelihood of seawater entering the IRGAs. An inline liquid water sensor (MK108, Velleman Inc.) was installed where the sample tube entered the bosun’s locker (upstream of the IRGAs; Fig. 2) and was monitored by a datalogger (CR1000, Campbell Scientific). When water was detected in the sample tube, a relay (CD16DC, Campbell Scientific) triggered solenoid valves to shut off flow to the IRGAs (solenoid S3; Fig. 2) and force compressed air at ~100 psi (and seawater) back out the inlet on the mast (solenoid S2; Fig. 2). To avoid a vacuum that could damage the LI-7200 differential pressure sensor, solenoids S4, S5, and S6 were also opened to enable room air to be pulled through the IRGAs and the pump. The mixing ratio of CO2 returned to prebackflush values within seconds of the flow returning to normal, while water vapor mixing ratio took about 1–2 min to recover.

c. Sea surface imagery

Two digital cameras (unheated CC5MPX and heated CC5MPXWD, Campbell Scientific) were used to record images of the sea surface at 1 Hz. The cameras were mounted on the railing of the ice tower (above the bridge; Fig. 1a), one to NBP’s starboard and the other to port. Both cameras were oriented 60° off of the ship’s bow with inclination angles of ~70° above nadir. The footprint of each image was roughly 3000 m2. Images were recorded with 640 × 480 pixel resolution and sent via FTP to a rackmount server in the ship’s forward dry laboratory (server 2; Supermicro, CSE-825), where they were stored as JPEG files. For each image, ice fraction was calculated by converting the image to gray scale, assigning a threshold value (between 0 and 255) above which a pixel was considered ice, and finding the percentage of pixels in the image that passed the threshold (Hall et al. 2002). It was not possible to fully automate the process due to variability of ice types, lighting conditions, glare, etc., and thresholds needed to be assigned for each image individually by visual inspection. The manual evaluation of all 1-Hz images was not feasible; rather, a subsample consisting of one image per minute was evaluated and, for each 10-min flux interval, the mean ice fraction was calculated as the average of the 10 analyzed images during that interval.

d. Remote system monitoring

To monitor the system remotely, automated e-mails with summary and diagnostic data and preliminary fluxes were transmitted daily from the ship to our laboratory at the University at Albany (Fig. 3). The summary and diagnostic data included 5-min averaged meteorological data from BOSUN and NEMA, as well as flow rates, pressures, and data from NBP’s underway systems. Preliminary fluxes were calculated and figures plotted by an automated MATLAB (version 7.4.0, R2007a, MathWorks) script running on rackmount server 2. The calculated data were copied to one of NBP’s data servers, compressed, and attached to an e-mail that was bundled with all the ship’s e-mail and transmitted by satellite to the Antarctic Support Contract office in Denver, Colorado, where it was parsed and forwarded to the University at Albany. The daily data were used to determine if adjustments or repairs were necessary. At the end of each cruise, fast-response flux data, sea surface images, and ship meteorological and underway data were saved to an external hard drive that was mailed to the University at Albany. The techniques deployed on NBP enabled the system to run with minimal intervention. Still, we relied regularly on the NBP onboard technicians for troubleshooting and minor equipment repairs, and occasionally to reboot the rackmount servers.

3. Results

Cruise tracks for the nine cruises while the air–sea flux system was installed on NBP are shown in Fig. 7, and cruise details are listed in Table 2. The tracks show a mix of transit and science cruises. During the 490 days when the system was installed on the NBP, the ship was at sea for 364 days (74% of the time), of which 94% (344 days) included periods when the ship was not within a foreign exclusive economic zone (EEZ; within which government clearance is required for sampling) and the underway pCO2 and air–sea flux systems were turned on.

Fig. 7.
Fig. 7.

Ship tracks of nine RV/IB NBP cruises between 1 Jan 2013 and 28 May 2014 when the University at Albany air–sea flux system was installed. Gray represents regions where |ΔpCO2| < 40 ppm. Black represents regions where |ΔpCO2| > 40 ppm.

Citation: Journal of Atmospheric and Oceanic Technology 33, 4; 10.1175/JTECH-D-15-0156.1

Table 2.

Dates, length, distances, and ports of call for the RV/IB NBP cruises when the University at Albany air–sea flux system was installed. Valparaiso and Antofagasta, Chile; and Papeete, French Polynesia.

Table 2.

The air–sea flux system required two maintenance trips during the 18 months it was installed on NBP. The first trip in June 2013 included minor repairs and maintenance, and installation of the RMY-low and RMY-port anemometers. The second trip in November 2013, delayed 6 weeks due to the U.S. federal government shutdown, was to repair damage from a complete ice over of the bow mast that lasted for 20 days (20 August–8 September 2013; Fig. 1d) while NBP was operating near the Antarctic Peninsula. The repairs included replacement of CSAT3-port (damaged by ice loading) and the sample cells of the LI7200-wet and LI7200-dry gas analyzers (due to seawater damage), and flushing of the pump and inlet tube to remove salt deposits. The total data gap was 94 days (20 August–22 November 2013).

To demonstrate system performance, we focused on two cruises (NBP-1210 and NBP-1402), chosen for their latitudinal range, inclusion of both open water and sea ice cover, and relatively large |ΔpCO2|. NBP-1210 was a 39-day (1 January–8 February 2013) transit from Punta Arenas to McMurdo Station (Fig. 7a) that began with relatively warm (~7°C) SST and air temperatures (Fig. 8c). Temperatures dropped steadily through the Drake Passage, with the air slightly cooler than SST. The next 10 days (days of year 16–26) included sea ice conditions (Fig. 8g, triangles) and little air–water temperature difference (both around −0.5°C). Subsequently, air temperature decreased and remained colder than the water temperature, with a mean water–air temperature difference (SST − Tair) of 3.0°C (maximum 9.6°C). NBP-1210 winds were low–moderate (mean U10n of 6.0 m s−1; Fig. 8e), with generally lower wind speeds in sea ice regions (mean 4.8 m s−1) compared with open water (mean 6.8 m s−1). In the SIZ, ΔpCO2 (defined as pCO2 waterpCO2 air) dropped to nearly −300 ppm (Fig. 8g). Measured ΔpCO2 was less than −40 ppm 51% of the time and less than −100 ppm 41% of the time.

Fig. 8.
Fig. 8.

Meteorology summary for the RV/IB NBP cruises (left) NBP-1210 and (right) NBP-1402. Shown are 10-min averages: (a),(b) latitude and longitude (°); (c),(d) air temperature and SST (°C); (e),(f) 10-m neutral wind speed (m s−1); and (g),(h) ΔpCO2 (ppm), where black circles represent open water and gray triangles represent sea ice.

Citation: Journal of Atmospheric and Oceanic Technology 33, 4; 10.1175/JTECH-D-15-0156.1

NBP-1402 was a 45-day (29 January–15 March 2014) round-trip cruise to/from Hobart, Tasmania, Australia, that spent much of the time near the Antarctic coast in the Adélie Land region (Fig. 8b). Sea ice in the form of icebergs, pack ice, lily pads, and grease ice was encountered south of 65°S. Overall, NBP-1402 spent a larger percentage of time in the SIZ (49% compared to 33% for NBP-1210). Between days 400–432, wind speeds were higher (mean 8.3 m s−1, maximum 28.4 m s−1; Fig. 8f) and the water–air temperature difference was larger (3.6°C; Fig. 8d) than during NBP-1210. During NBP-1402, ΔpCO2 was more modest than NBP-1210 with fewer periods less than −100 ppm (17% of time), but more periods less than −40 ppm (59% of time).

a. Flux cospectra

Flux cospectra were calculated for open water periods during NBP-1210 and NBP-1402 (Figs. 5c–f) and compared with semiempirical cospectra (Kaimal et al. 1972, hereafter K72). For momentum and sensible heat during low-heave periods, the median flux cospectra were broadly similar to K72 (Figs. 5c,d), with a small residual hump near the platform motion frequencies that was likely due to motion and/or flow distortion effects not removed by corrections. The uw cospectrum did not approach zero at high frequencies, particularly for the high-heave intervals (Fig. 5b). This high-frequency tail was less apparent for low-heave intervals (Fig. 5c), suggesting that it was related to platform motion. However, we were unable to isolate the cause of this feature in the cospectrum. Also, the cospectra did not go to zero at the lowest frequency (Figs. 5c–f). While this feature may have been mitigated by using a longer averaging time, the 10-min averaging period was chosen to reduce significant low-frequency features in the CO2 flux cospectra that have be reported elsewhere (Marandino et al. 2007; Miller et al. 2010; Blomquist et al. 2014). We estimated the “missing” low-frequency flux by extrapolating the mean curves (log-linear fit from 0.002 to 0.01 Hz) to zero and integrating beneath both the measured and the extrapolated cospectra. The effects on the calculated fluxes were small, accounting for less than 1% of the total flux for momentum, heat, and CO2.

The open-path wq cospectrum (Fig. 5e, dashed) dropped off more steeply from the spectral peak than K72 as frequencies increased above ~1 Hz, possibly due to sensor separation between the LI-7500 and CSAT3-stbd. Compared with the open-path sensor and the K72 cospectrum, the LI7200-wet wq cospectral peak was shifted to lower frequency (Fig. 5e, solid gray curve), likely due to attenuation of water vapor fluctuations in the sample tubing from the mast inlet to the bosun’s locker (Ibrom et al. 2007).

For CO2, the LI7200-dry cospectral shape was similar to K72, with its peak shifted to slightly lower frequency (Fig. 5f, black dashed curve). Meanwhile, the LI7200-wet wc cospectral peak (Fig. 5f, solid gray curve) had more shift toward low frequency, and was more similar to the LI7200-wet wq cospectrum than to the LI7200-dry wc cospectrum. We suspect this was due to cross talk between H2O and CO2 absorption bands for LI7200-wet, with water vapor erroneously contributing to measured CO2 signal even after the Webb correction was applied by the LI-7200 onboard electronics. Similar contamination for an undried airstream was reported by Landwehr et al. (2014). All CO2 fluxes in the remaining text and figures are from LI7200-dry.

b. ΔpCO2 threshold

Several studies have rejected CO2 fluxes when the magnitude of ΔpCO2 was below an empirical threshold (e.g., 40 ppm) due to poor signal-to-noise ratio (Miller et al. 2010; Edson et al. 2011; Blomquist et al. 2014). Figure 9 shows the normalized difference between measured open-ocean CO2 fluxes and the Wanninkhof et al. (2009, hereafter W09) parameterization as a function of ΔpCO2. Each ΔpCO2 bin includes a range of wind speeds and CO2 fluxes, with larger magnitude ΔpCO2 generally associated with lower wind speeds. When the magnitude of ΔpCO2 was less than 40 ppm, there was increased scatter. However, somewhat unexpectedly, there was little bias relative to the W09 parameterization between low- and high-|ΔpCO2| conditions (Fig. 9, solid curve). Similarly, the wc cospectrum for low |ΔpCO2| exhibited a cospectral shape that was similar to, though narrower than, K72 (Fig. 5f; thin gray curve with dots). We interpret these results as empirical indicators that the low-|ΔpCO2| quality-control criterion may not be necessary for the NBP data. Relaxing this criterion resulted in retention of 615 additional air–sea CO2 flux intervals, 30% more intervals than if the ΔpCO2 criterion were applied.

Fig. 9.
Fig. 9.

Normalized difference between measured and bulk (W09) CO2 flux plotted against ΔpCO2 for RV/IB NBP cruises NBP-1210 and NBP-1402. Shading of points represents 10-m neutral wind speed.

Citation: Journal of Atmospheric and Oceanic Technology 33, 4; 10.1175/JTECH-D-15-0156.1

c. Air–sea fluxes

The combined periods during which NBP-1210 and NBP-1402 were at sea and outside of foreign EEZs included 74 days, or 10 681 potential 10-min flux intervals. The effect of quality control on flux data retention is shown in Table 3. For the combined cruises, equipment malfunction caused the loss of 9% (951) of the potential flux intervals, and 65% (6916) of intervals did not pass the requirement of winds within ±30° of the bow. The elimination of intervals where σheave was greater than 0.75 m caused the loss of 20% (2120) of potential flux intervals. Applying equipment malfunction, relative wind direction, and heave criteria resulted in 2682 quality-controlled flux intervals for momentum, sensible heat, closed-path latent heat, and CO2 (25% of potential intervals). For open-path latent heat flux measurements, an additional quality-control criterion was applied based on the automatic gain control (AGC) diagnostic of the LI-7500, an indicator of contamination on the sensor window. Using a rejection criterion of AGC > 62.5 eliminated 1170 additional intervals (11%). The total number of quality-controlled open-path latent heat flux intervals was 1512, or 14% of potential intervals. Regressions of momentum, sensible and latent heat, and CO2 fluxes calculated using CSAT3-stbd and CSAT3-port sonic anemometers showed slopes close to 1 (0.97–1.03) and high r2 values (0.97–0.98).

Table 3.

Effects of quality-control criteria on air–sea fluxes measured from RV/IB NBP NBP-1210 and NBP-1402 cruises.

Table 3.

Time series of 3-h averaged quality-controlled momentum, sensible and closed-path latent heat, and CO2 flux are shown for NBP-1210 and NBP-1402 in Fig. 10 along with open-water bulk flux parameterizations. Periods when ice was detected are shown as gray triangles. Generally, momentum and sensible heat flux magnitudes were larger for NBP-1402 due to higher winds and greater water–air temperature differences. Open-ocean sensible heat flux measurements were found to agree closely with NOAA COARE 3.0 (Fairall et al. 2003) values (slope = 0.96, intercept = −0.02, r2 = 0.85; Figs. 10c,d). Similar latent heat fluxes were found using the water vapor measured by the open-path sensor (LI-7500) and by the closed-path sensor LI7200-wet after correction for tubing attenuation (NBP-1210 regression slope = 0.95, intercept = 2.9 Wm−2, r2 = 0.77). However, the measured open-ocean latent heat fluxes (closed-path results shown in Figs. 10e,f) were much lower than NOAA COARE 3.0 predictions for both NBP-1210 (closed-path regression slope = 0.52, intercept = −2.3 Wm−2, r2 = 0.74; open-path regression slope = 0.56, intercept = −4.5 Wm−2, r2 = 0.77) and NBP-1402 (closed-path regression slope = 0.26, intercept = −0.66 Wm−2, r2 = 0.43; open-path regression slope = 0.55, intercept = −4.3 Wm−2, r2 = 0.78).

Fig. 10.
Fig. 10.

Flux summary for the RV/IB NBP cruises (left) NBP-1210 and (right) NBP-1402. Fluxes are shown as 3-h averages of 10-min fluxes: (a),(b) momentum flux (N m−2); (c),(d) sensible heat flux (W m−2); (e),(f) closed-path latent heat flux (W m−2); and (g),(h) CO2 flux (mol m−2 yr−1). Black circles represent measured open water fluxes, and gray triangles represent measured fluxes in the SIZ. Gray lines represent the NOAA COARE algorithm (version 3.5 for momentum flux and version 3.0 for Hs and HL) and the W09 quadratic parameterization.

Citation: Journal of Atmospheric and Oceanic Technology 33, 4; 10.1175/JTECH-D-15-0156.1

CO2 fluxes were mostly downward (a sink) during NBP-1210 and NBP-1402. The open-ocean measurements exhibited the highest winds and the lowest magnitudes of ΔpCO2, leading to low-to-moderate CO2 fluxes in reasonable agreement with CO2 flux parameterizations (W09). In the SIZ, ΔpCO2 was generally large (−91 ppm), wind speeds were low, and CO2 fluxes were low (mean −2 mol m−2 yr−1). The largest magnitude of ΔpCO2 (−288 ppm) and CO2 flux occurred in the high-latitude open water regions (polynyas and Antarctic seas). The triangles in Fig. 10 indicate the presence of sea ice but not the percentage of sea ice cover. Some days (e.g., 423–425; Fig. 10h) had complete ice cover and showed near-zero CO2 fluxes, while other days (e.g., 425–430) exhibited only partial ice coverage and reasonably matched open-water bulk CO2 fluxes.

4. Discussion

a. Air–sea flux system performance

The retention rate for quality-controlled air–sea fluxes was 25% for the combined NBP-1210 and NBP-1402 cruises. Most data rejection was associated with poor wind directions relative to the ship rather than equipment/system performance. The measures taken to ruggedize the system likely improved the retention rate and reduced sensor damage that would have been difficult (i.e., slow and costly) to repair. For example, the CSAT3-stbd heat tape was on ~15% of the time during NBP-1210 and NBP-1402. This sonic lasted throughout the study and was the most reliable anemometer, whereas the unheated anemometer required replacement due to damage associated with ice loading (Fig. 1d). Similarly, the backflush system to prevent seawater from entering the IRGAs tripped 45 times during NBP-1210 and NBP-1402, and likely prevented corrosion and damage to electronics. For CO2, the most reliable flux measurements were made using LI7200-dry (Figs. 2, 5f), consistent with other recent studies (Miller et al. 2010; Blomquist et al. 2014; Landwehr et al. 2014). This is our first dataset without apparent need for a ΔpCO2 filter, which enabled retention of 30% more CO2 flux intervals.

Bell et al. (2013) found better agreement with COARE 3.1v (Fairall et al. 2011) when restricting to periods when the ship was on station rather than underway. In contrast, we did not find a significant difference in the τ versus U10n relationship between the underway and on-station periods (data not shown). Our results suggest that the contamination was likely due to ship heave rather than whether the ship was underway (Fig. 5b,c). Interestingly, the curves in Fig. 6 representing all data (circles) and low-heave data only (triangles) both showed close agreement with COARE 3.5. This was possibly due to the symmetric peaks above and below the Kaimal curve in Fig. 5b, whose contribution to the flux may have offset one another. For a similar reason, the Prytherch et al. (2015) correction had little impact on the τ versus U10n relationship (Fig. 6, black x’s). We note that restricting to low heave caused the loss of high-wind periods.

A considerable challenge in measuring air–sea fluxes from NBP was the relatively low mounting height of the flux sensors on the bow mast. The height of the anemometers and air inlet was well within the splash zone for waves hitting the bow during high seas, which increased the likelihood of damage from direct seawater impact, electrical short circuits, and drawing seawater from the inlet into the IRGA cells in the bosun’s locker. Further, the height of the flux sensors was just 4.5 m above the top edge of the hull at the NBP bow (Fig. 1a), which likely increased flow distortion effects on the wind vector. Extending the mast vertically by several meters to get sensors well above the splash zone would have mitigated some of the uncertainties associated with flow distortion; however, modification of the NBP bow mast was not feasible due to cost and logistics. We used a range of approaches to increase our understanding of flow distortion at the bow, and to develop a strategy to obtain quality-controlled mean winds and fluxes. Our final momentum fluxes at the bow showed close agreement with NOAA COARE 3.5 when we used mean wind from the science mast. A similar approach was used by Bell et al. (2015) for measurements from R/V Tangaroa. There remains the possibility that our agreement with COARE was fortuitous and the measures taken to address flow distortion were insufficient. We note that measuring winds uncontaminated by motion and/or flow distortion from a ship is not possible using current technologies, and, in the next section, we show that despite the challenges in obtaining accurate fluxes from ships, the precision of EC can be leveraged to demonstrate the effects of sea ice on surface–atmosphere momentum exchange.

The latent heat flux measurement was most problematic. The closed-path latent heat fluxes during open-ocean conditions were most similar to COARE at the beginning of the first cruise (NBP-1210), when the sample lines were relatively clean (prior to day 13; Fig. 10e). Subsequently, salt buildup in the sample tube may have degraded the closed-path fluxes; however, the similarity between the measured open- and closed-path approaches made it difficult to identify salt contamination as the culprit since the open-path configuration did not include sample tubing. Combined with the significant loss of open-path data due to contamination on its windows (i.e., high AGC values), our latent heat fluxes should be viewed with caution. Alternative approaches and additional hardware techniques, such as regular (e.g., daily) flushing of both the closed-path sample lines and open-path analyzer head (Edson et al. 2011), should be tested to improve the quality and continuity of latent heat fluxes measured by our system.

b. Utility of ship-based air–sea fluxes

The in situ air–sea flux measurements and sea surface imagery were used to examine the effect of sea ice on the momentum flux and 10-m neutral drag coefficient (calculated as CD10n = []2). In open water, there was generally good agreement between measured momentum flux and COARE 3.5 (Figs. 10a,b). Term CD10n was between COARE 3.5 and Smith (1980) for wind speeds greater than 8 m s−1 (Fig. 11, black circles). In the SIZ, our measured momentum flux was typically higher than COARE (Figs. 10a,b). The drag coefficient was separated into three groups according to sea ice cover: 0 < SIC < 0.2, 0.2 < SIC < 0.8, and 0.8 < SIC ≤ 1. For the highest SIC group, an additional quality-control criterion was applied, whereby intervals were rejected when NBP alternated between forward and backward motion corresponding to breaking and/or navigating through ice, resulting in the loss of 150 additional intervals. Compared with open water, CD10n was higher for all ice groups (Fig. 11). The lowest and highest SIC groups (0–0.2 and 0.8–1 SIC) showed modest increases, while the intermediate SIC group (0.2–0.8 SIC) showed the largest increase (Fig. 11). Similar results have been reported previously (Andreas et al. 2010; Fig. 11, right-side margin), attributed to increased roughness of the heterogeneous (open water plus ice) surface at intermediate SIC and relatively lower roughness (i.e., smoother) over the more homogeneous open water and high ice cover surfaces. The ability of the ship-based air–sea flux system to measure across a broad range of sea ice cover facilitated the examination of the effects of sea ice on wind stress.

Fig. 11.
Fig. 11.

The 10-m neutral drag coefficient vs 10-m neutral wind speed (m s−1) measured from the RV/IB NBP cruises NBP-1210 and NBP-1402: open water (black circles), 0%–20% SIC (gray triangles), 20%–80% SIC (x’s), and 80%–100% SIC (white squares). The black dashed line represents the output of the NOAA COARE 3.5 algorithm (Edson et al. 2013). The gray dashed–dotted line represents the parameterization of Smith (1980). On the right-hand side of the plot are the wind speed–independent CD10n values from the SIC-dependent parameterization of Andreas et al. (2010), calculated using the mean SIC for each of the three ice groups.

Citation: Journal of Atmospheric and Oceanic Technology 33, 4; 10.1175/JTECH-D-15-0156.1

The air–sea flux data collected by the NBP system during the nine cruises (Fig. 7) enables additional physical and biogeochemical questions to be studied. The data include five crossings of the Drake Passage (NBP-1210, NBP-1302, NBP-1304, NBP-1309, NBP-1310) and three cruises in the Ross Sea (NBP-1210, NBP-1302, NBP-1310). Further, the direct measurements of the piston velocity during open water periods will allow for determination of the dependence of the gas exchange coefficient on wind speed. This functional relationship is important for biogeochemical studies and for assessing the changing importance of the ocean in the global carbon cycle (Rintoul et al. 2010). The cruises within the SIZ (NBP-1210, NBP-1302, NBP-1310, NBP-1402) will be valuable for testing models of the air–sea ice gas exchange (Loose et al. 2014), and the SIZ data can be used to examine exchange coefficients in polynyas and the seasonality of air–sea CO2 flux (e.g., uptake in summer due to biological production and efflux in the winter due to upwelling; Takahashi et al. 2012).

c. Recommendations for ship-based high-latitude air–sea fluxes

The experience gained in the deployment of the NBP flux system allows us to make specific recommendations for unattended flux measurements from ships in high-latitude environments:

  1. locate turbulence sensors as far forward at the bow and as high above the deck and sea surface as possible, ideally above the splash zone, to reduce the likelihood of seawater damaging sensors and electronics, and to reduce contamination by flow distortion (Yelland et al. 2002);

  2. employ active techniques (e.g., instrument heating and automated backflushing) to reduce instrument damage and data loss;

  3. flush sample lines and rinse open-path sensor head daily to remove salt deposits and to increase the quality of closed- and open-path latent heat fluxes;

  4. for CO2 flux, use a closed-path approach and dry the air sample stream to reduce contamination due to water vapor cross talk; and

  5. use computational fluid dynamics to model flow distortion and to correct the measured wind vector (Yelland et al. 2002; Popinet et al. 2004; Moat and Yelland 2008; O’Sullivan et al. 2013, 2015).

5. Conclusions

Southern Ocean direct air–sea flux measurements are critical for calibrating remote sensing algorithms and developing and validating models, yet few datasets exist compared with temperate and tropical oceans. This study demonstrates that some of the challenges in making unattended high-latitude air–sea flux measurements from ships can be addressed by careful system design. The geographical coverage of ships can be combined with the high precision of eddy covariance to investigate differences in air–sea fluxes between environments, such as varying sea ice concentration (SIC). These data provide an opportunity to study exchange processes at high latitudes where environmental conditions are frequently outside the range typically measured at low and midlatitudes.

Acknowledgments

Thanks to Taro Takahashi for the ΔpCO2 data; Barry Bjork, Andy Nunn, and Bob Kluckhohn for help with installation and logistics; NBP Captains John Souza and Sebastian Paoni; University at Albany staff Chris Sager, John Sicker, Jim Albrecht, and Brian Smith for system fabrication; and three anonymous reviewers. Scientific support aboard NBP was provided by Antarctic Support Contract (ASC). This work was supported by NSF Office of Polar Programs Award 1043623.

REFERENCES

  • Andreas, E. L, Jordan R. E. , and Makshtas A. P. , 2005: Parameterizing turbulent exchange over sea ice: The Ice Station Weddell results. Bound.-Layer Meteor., 114, 439460, doi:10.1007/s10546-004-1414-7.

    • Search Google Scholar
    • Export Citation
  • Andreas, E. L, Horst T. W. , Grachev A. A. , Persson P. O. G. , Fairall C. W. , Guest P. S. , and Jordan R. E. , 2010: Parametrizing turbulent exchange over summer sea ice and the marginal ice zone. Quart. J. Roy. Meteor. Soc., 136, 927943, doi:10.1002/qj.618.

    • Search Google Scholar
    • Export Citation
  • Andreas, E. L, Jordan R. E. , Mahrt L. , and Vickers D. , 2013: Estimating the Bowen ratio over the open and ice-covered ocean. J. Geophys. Res. Oceans, 118, 43344345, doi:10.1002/jgrc.20295.

    • Search Google Scholar
    • Export Citation
  • Bell, T. G., De Bruyn W. J. , Miller S. D. , Ward B. , Christensen K. , and Saltzman E. S. , 2013: Air–sea dimethylsulfide (DMS) gas transfer in the North Atlantic: Evidence for limited interfacial gas exchange at high wind speed. Atmos. Chem. Phys., 13, 11 07311 087, doi:10.5194/acp-13-11073-2013.

    • Search Google Scholar
    • Export Citation
  • Bell, T. G., De Bruyn W. J. , Marandino C. A. , Miller S. D. , Law C. S. , Smith M. J. , and Saltzman E. S. , 2015: Dimethylsulfide gas transfer coefficients from algal blooms in the Southern Ocean. Atmos. Chem. Phys., 15, 17831794, doi:10.5194/acp-15-1783-2015.

    • Search Google Scholar
    • Export Citation
  • Blomquist, B. W., Fairall C. W. , Huebert B. J. , Kieber D. J. , and Westby G. R. , 2006: DMS sea-air transfer velocity: Direct measurements by eddy covariance and parameterization based on the NOAA/COARE gas transfer model. Geophys. Res. Lett., 33, L07601, doi:10.1029/2006GL025735.

    • Search Google Scholar
    • Export Citation
  • Blomquist, B. W., Huebert B. J. , Fairall C. W. , Bariteau L. , Edson J. B. , Hare J. E. , and McGillis W. R. , 2014: Advances in air–sea CO2 flux measurement by eddy correlation. Bound.-Layer Meteor., 152, 245–276, doi:10.1007/s10546-014-9926-2.

    • Search Google Scholar
    • Export Citation
  • Dunckel, M., Hasse L. , Krügermeyer L. , Schriever D. , and Wucknitz J. , 1974: Turbulent fluxes of momentum, heat and water vapor in the atmospheric surface layer at sea during ATEX. Bound.-Layer Meteor., 6, 81106, doi:10.1007/BF00232478.

    • Search Google Scholar
    • Export Citation
  • Edson, J. B., Hinton A. A. , Prada K. E. , Hare J. E. , and Fairall C. W. , 1998: Direct covariance flux estimates from mobile platforms at sea. J. Atmos. Oceanic Technol., 15, 547562, doi:10.1175/1520-0426(1998)015<0547:DCFEFM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Edson, J. B., and Coauthors, 2011: Direct covariance measurement of CO2 gas transfer velocity during the 2008 Southern Ocean Gas Exchange Experiment: Wind speed dependency. J. Geophys. Res., 116, C00F10, doi:10.1029/2011JC007022.

    • Search Google Scholar
    • Export Citation
  • Edson, J. B., and Coauthors, 2013: On the exchange of momentum over the open ocean. J. Phys. Oceanogr., 43, 15891610, doi:10.1175/JPO-D-12-0173.1.

    • Search Google Scholar
    • Export Citation
  • Else, B. G. T., Papakyriakou T. N. , Galley R. J. , Drennan W. M. , Miller L. A. , and Thomas H. , 2011: Wintertime CO2 fluxes in an Arctic polynya using eddy covariance: Evidence for enhanced air-sea gas transfer during ice formation. J. Geophys. Res., 116, C00G03, doi:10.1029/2010JC006760.

    • Search Google Scholar
    • Export Citation
  • Fairall, C. W., Bradley E. F. , Hare J. E. , Grachev A. A. , and Edson J. B. , 2003: Bulk parameterization of air–sea fluxes: Updates and verification for the COARE algorithm. J. Climate, 16, 571591, doi:10.1175/1520-0442(2003)016<0571:BPOASF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Fairall, C. W., and Coauthors, 2011: Implementation of the Coupled Ocean-Atmosphere Response Experiment flux algorithm with CO2, dimethyl sulfide, and O3. J. Geophys. Res., 116, C00F09, doi:10.1029/2010JC006884.

    • Search Google Scholar
    • Export Citation
  • Fujitani, T., 1985: Method of turbulent flux measurement on a ship by using a stable platform system. Pap. Meteor. Geophys., 36, 157170, doi:10.2467/mripapers.36.157.

    • Search Google Scholar
    • Export Citation
  • Goulden, M. L., Munger J. W. , Fan S. M. , Daube B. C. , and Wofsy S. C. , 1996: Measurements of carbon sequestration by long-term eddy covariance: Methods and a critical evaluation of accuracy. Global Change Biol., 2, 169182, doi:10.1111/j.1365-2486.1996.tb00070.x.

    • Search Google Scholar
    • Export Citation
  • Grachev, A. A., Andreas E. L , Fairall C. W. , Guest P. S. , and Persson P. O. G. , 2007: SHEBA flux–profile relationships in the stable atmospheric boundary layer. Bound.-Layer Meteor., 124, 315333, doi:10.1007/s10546-007-9177-6.

    • Search Google Scholar
    • Export Citation
  • Hall, R. J., Hughes N. , and Wadhams P. , 2002: A systematic method of obtaining ice concentration measurements from ship-based observations. Cold Reg. Sci. Technol., 34, 97102, doi:10.1016/S0165-232X(01)00057-X.

    • Search Google Scholar
    • Export Citation
  • Ibrom, A., Dellwik E. , Flyvbjerg H. , Jensen N. O. , and Pilegaard K. , 2007: Strong low-pass filtering effects on water vapour flux measurements with closed-path eddy correlation systems. Agric. For. Meteor., 147, 140156, doi:10.1016/j.agrformet.2007.07.007.

    • Search Google Scholar
    • Export Citation
  • Kaimal, J. C., Wyngaard J. C. , Izumi Y. , and Coté O. R. , 1972: Spectral characteristics of surface-layer turbulence. Quart. J. Roy. Meteor. Soc., 98, 563589, doi:10.1002/qj.49709841707.

    • Search Google Scholar
    • Export Citation
  • Kohsiek, W., 2000: Water vapor cross-sensitivity of open path H2O/CO2 sensors. J. Atmos. Oceanic Technol., 17, 299311, doi:10.1175/1520-0426(2000)017<0299:WVCSOO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Landwehr, S., Miller S. D. , Smith M. J. , Saltzman E. S. , and Ward B. , 2014: Analysis of the PKT correction for direct CO2 flux measurements over the ocean. Atmos. Chem. Phys., 14, 33613372, doi:10.5194/acp-14-3361-2014.

    • Search Google Scholar
    • Export Citation
  • Landwehr, S., O’Sullivan N. , and Ward B. , 2015: Direct flux measurements from mobile platforms at sea: Motion and airflow distortion corrections revisited. J. Atmos. Oceanic Technol., 32, 1163–1178, doi:10.1175/JTECH-D-14-00137.1.

    • Search Google Scholar
    • Export Citation
  • Le Quéré, C., and Coauthors, 2007: Saturation of the Southern Ocean CO2 sink due to recent climate change. Science, 316, 1735–1738, doi:10.1126/science.1136188.

    • Search Google Scholar
    • Export Citation
  • Loose, B., McGillis W. R. , Perovich D. , Zappa C. J. , and Schlosser P. , 2014: A parameter model of gas exchange for the seasonal sea ice zone. Ocean Sci., 10, 1728, doi:10.5194/os-10-17-2014.

    • Search Google Scholar
    • Export Citation
  • Marandino, C. A., De Bruyn W. J. , Miller S. D. , and Saltzman E. S. , 2007: Eddy correlation measurements of the air/sea flux of dimethylsulfide over the North Pacific Ocean. J. Geophys. Res., 112, D03301, doi:10.1029/2006JD007293.

    • Search Google Scholar
    • Export Citation
  • McGillis, W. R., Edson J. B. , Hare J. E. , and Fairall C. W. , 2001: Direct covariance air-sea CO2 fluxes. J. Geophys. Res., 106, 16 72916 745, doi:10.1029/2000JC000506.

    • Search Google Scholar
    • Export Citation
  • McMillen, R. T., 1988: An eddy correlation technique with extended applicability to non-simple terrain. Bound.-Layer Meteor., 43, 231245, doi:10.1007/BF00128405.

    • Search Google Scholar
    • Export Citation
  • Miller, S. D., Goulden M. L. , Menton M. C. , da Rocha H. R. , de Freitas H. C. , Figueira A. M. S. , and de Sousa C. A. D. , 2004: Biometric and micrometeorological measurements of tropical forest carbon balance. Ecol. Appl., 14, 114126, doi:10.1890/02-6005.

    • Search Google Scholar
    • Export Citation
  • Miller, S. D., Hristov T. S. , Edson J. B. , and Friehe C. A. , 2008: Platform motion effects on measurements of turbulence and air–sea exchange over the open ocean. J. Atmos. Oceanic Technol., 25, 16831694, doi:10.1175/2008JTECHO547.1.

    • Search Google Scholar
    • Export Citation
  • Miller, S. D., Marandino C. A. , and Saltzman E. S. , 2010: Ship-based measurement of air-sea CO2 exchange by eddy covariance. J. Geophys. Res., 115, D02304, doi:10.1029/2009JD012193.

    • Search Google Scholar
    • Export Citation
  • Mitsuta, Y., and Fujitani T. , 1974: Direct measurement of turbulent fluxes on a cruising ship. Bound.-Layer Meteor., 6, 203217, doi:10.1007/BF00232485.

    • Search Google Scholar
    • Export Citation
  • Moat, B. I., and Yelland M. J. , 2008: Going with the flow: State of the art marine meteorological measurements on the new NERC research vessel. Weather, 63, 158159, doi:10.1002/wea.184.

    • Search Google Scholar
    • Export Citation
  • O’Sullivan, N., Landwehr S. , and Ward B. , 2013: Mapping flow distortion on oceanographic platforms using computational fluid dynamics. Ocean Sci., 9, 855866, doi:10.5194/os-9-855-2013.

    • Search Google Scholar
    • Export Citation
  • O’Sullivan, N., Landwehr S. , and Ward B. , 2015: Air-flow distortion and wave interactions on research vessels: An experimental and numerical comparison. Methods Oceanogr., 12, 117, doi:10.1016/j.mio.2015.03.001.

    • Search Google Scholar
    • Export Citation
  • Paulson, C. A., 1970: The mathematical representation of wind speed and temperature profiles in the unstable atmospheric surface layer. J. Appl. Meteor., 9, 857861, doi:10.1175/1520-0450(1970)009<0857:TMROWS>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Popinet, S., Smith M. J. , and Stevens C. , 2004: Experimental and numerical study of the turbulence characteristics of airflow around a research vessel. J. Atmos. Oceanic Technol., 21, 15751589, doi:10.1175/1520-0426(2004)021<1575:EANSOT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Prytherch, J., Yelland M. J. , Pascal R. W. , Moat B. I. , Skjelvan I. , and Neill C. C. , 2010: Direct measurements of the CO2 flux over the ocean: Development of a novel method. Geophys. Res. Lett., 37, L03607, doi:10.1029/2009GL041482.

    • Search Google Scholar
    • Export Citation
  • Prytherch, J., Yelland M. J. , Brooks I. M. , Tupman D. J. , Pascal R. W. , Moat B. I. , and Norris S. J. , 2015: Motion-correlated flow distortion and wave-induced biases in air-sea flux measurements from ships. Atmos. Chem. Phys., 15, 10 61910 629, doi:10.5194/acp-15-10619-2015.

    • Search Google Scholar
    • Export Citation
  • Rintoul, S. R., and Coauthors, 2010: Southern Ocean Observing System (SOOS): Rationale and strategy for sustained observations of the Southern Ocean. Proceedings of OceanObs’09: Sustained Ocean Observations and Information for Society, J. Hall, D. E. Harrison, and D. Stammer, Eds., Vol. 2, ESA Publ. WPP-306, doi:10.5270/OceanObs09.cwp.74.

    • Search Google Scholar
    • Export Citation
  • Schotanus, P., Nieuwstadt F. T. M. , and de Bruin H. A. R. , 1983: Temperature measurement with a sonic anemometer and its application to heat and moisture fluxes. Bound.-Layer Meteor., 26, 8193, doi:10.1007/BF00164332.

    • Search Google Scholar
    • Export Citation
  • Smith, S. D., 1980: Wind stress and heat flux over the ocean in gale force winds. J. Phys. Oceanogr., 10, 709726, doi:10.1175/1520-0485(1980)010<0709:WSAHFO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Takahashi, T., Sweeney C. , Hales B. , Chipman D. W. , Newberger T. , Goddard J. , Iannuzzi R. , and Sutherland S. C. , 2012: The changing carbon cycle in the Southern Ocean. Oceanography, 25 (3), 2637, doi:10.5670/oceanog.2012.71.

    • Search Google Scholar
    • Export Citation
  • Takahashi, T., Sweeney C. , and Sutherland S. C. , 2014: Underway pCO2 measurements in surface waters and the atmosphere during the R/V Nathaniel B. Palmer 2013 expeditions. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, accessed 11 March 2016, doi:10.3334/CDIAC/OTG.VOS_PALMER_2013.

  • Takahashi, T., Sweeney C. , and Sutherland S. C. , 2016: Underway pCO2 measurements in surface waters and the atmosphere during the R/V Nathaniel B. Palmer 2014 expeditions. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, accessed 11 March 2016, doi:10.3334/CDIAC/OTG.VOS_PALMER_2014.

  • Wanninkhof, R., Asher W. E. , Ho D. T. , Sweeney C. , and McGillis W. R. , 2009: Advances in quantifying air-sea gas exchange and environmental forcing. Annu. Rev. Mar. Sci., 1, 213244, doi:10.1146/annurev.marine.010908.163742.

    • Search Google Scholar
    • Export Citation
  • Webb, E. K., Pearman G. I. , and Leuning R. , 1980: Correction of flux measurements for density effects due to heat and water vapour transfer. Quart. J. Roy. Meteor. Soc., 106, 85100, doi:10.1002/qj.49710644707.

    • Search Google Scholar
    • Export Citation
  • Wilczak, J. M., Oncley S. P. , and Stage S. A. , 2001: Sonic anemometer tilt correction algorithms. Bound.-Layer Meteor., 99, 127150, doi:10.1023/A:1018966204465.

    • Search Google Scholar
    • Export Citation
  • Yelland, M. J., Moat B. I. , Pascal R. W. , and Berry D. I. , 2002: CFD model estimates of the airflow distortion over research ships and the impact on momentum flux measurements. J. Atmos. Oceanic Technol., 19, 14771499, doi:10.1175/1520-0426(2002)019<1477:CMEOTA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Yelland, M. J., Pascal R. W. , Taylor P. K. , and Moat B. I. , 2009: AutoFlux: An autonomous system for the direct measurement of the air–sea fluxes of CO2, heat and momentum. J. Oper. Oceanogr., 2, 1523, doi:10.1080/1755876X.2009.11020105.

    • Search Google Scholar
    • Export Citation
  • Zemmelink, H. J., Delille B. , Tison J. L. , Hintsa E. J. , Houghton L. , and Dacey J. W. H. , 2006: CO2 deposition over the multi-year ice of the western Weddell Sea. Geophys. Res. Lett., 33, L13606, doi:10.1029/2006GL026320.

    • Search Google Scholar
    • Export Citation
Save