• Bals-Elsholz, T. M., , E. H. Atallah, , L. F. Bosart, , T. A. Wasula, , M. J. Cempa, , and A. R. Lupo, 2001: The wintertime Southern Hemisphere split jet: Structure, variability, and evolution. J. Climate, 14, 41914215, doi:10.1175/1520-0442(2001)014<4191:TWSHSJ>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Baumgartner, A., , and E. Reichel, 1975: World Water Balance: Mean Annual Global, Continental and Maritime Precipitation, Evaporation and Runoff.Elsevier Scientific, 379 pp.

  • Berbery, E. H., , and V. R. Barros, 2002: The hydrological cycle of the La Plata basin in South America. J. Hydrometeor., 3, 630645, doi:10.1175/1525-7541(2002)003<0630:THCOTL>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Berry, G., , M. J. Reeder, , and C. Jakob, 2011: A global climatology of atmospheric fronts. Geophys. Res. Lett.,38, L04809, doi:10.1029/2010GL046451.

  • Bosilovich, M., , F. Robertson, , and J. Chen, 2011: Global energy and water budgets in MERRA. J. Climate, 24, 57215739, doi:10.1175/2011JCLI4175.1.

    • Search Google Scholar
    • Export Citation
  • Bracegirdle, T., , and E. Kolstad, 2010: Climatology and variability of Southern Hemisphere marine cold-air outbreaks. Tellus, 62A, 202208, doi:10.1111/j.1600-0870.2009.00431.x.

    • Search Google Scholar
    • Export Citation
  • Bromwich, D. H., , J. P. Nicolas, , and A. J. Monaghan, 2011: An assessment of precipitation changes over Antarctica and the Southern Ocean since 1989 in contemporary global reanalyses. J. Climate, 24, 41894209, doi:10.1175/2011JCLI4074.1.

    • Search Google Scholar
    • Export Citation
  • Catto, J. L., , and S. Pfahl, 2013: The importance of fronts for extreme precipitation. J. Geophys. Res., 118, 10 791–10 801, doi:10.1002/jgrd.50852.

    • Search Google Scholar
    • Export Citation
  • Catto, J. L., , C. Jakob, , G. Berry, , and N. Nicholls, 2012: Relating global precipitation to atmospheric fronts. Geophys. Res. Lett.,39, L10805, doi:10.1029/2012GL051736.

  • Cerovečki, I., , L. D. Talley, , and M. R. Mazloff, 2011: A comparison of Southern Ocean air–sea buoyancy flux from an ocean state estimate with five other products. J. Climate, 24, 62836306, doi:10.1175/2011JCLI3858.1.

    • Search Google Scholar
    • Export Citation
  • Cerovečki, I., , L. D. Talley, , M. R. Mazloff, , and G. Maze, 2013: Subantarctic Mode Water formation, destruction, and export in the eddy-permitting Southern Ocean State Estimate. J. Phys. Oceanogr., 43, 14851511, doi:10.1175/JPO-D-12-0121.1.

    • Search Google Scholar
    • Export Citation
  • Charnock, H., 1955: Wind stress on a water surface. Quart. J. Roy. Meteor. Soc., 81, 639640, doi:10.1002/qj.49708135027.

  • Condron, A., , G. R. Bigg, , and I. A. Renfrew, 2006: Polar mesoscale cyclones in the northeast Atlantic: Comparing climatologies from ERA-40 and satellite imagery. Mon. Wea. Rev., 134, 15181533, doi:10.1175/MWR3136.1.

    • Search Google Scholar
    • Export Citation
  • Connolley, W. M., 1997: Variability in annual mean circulation in southern high latitudes. Climate Dyn., 13, 745756, doi:10.1007/s003820050195.

    • Search Google Scholar
    • Export Citation
  • Dee, D., and et al. , 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553597, doi:10.1002/qj.828.

    • Search Google Scholar
    • Export Citation
  • Fairall, C., , E. Bradley, , J. Hare, , A. Grachev, , and J. Edson, 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
  • Gruber, N., and et al. , 2009: Oceanic sources, sinks, and transport of atmospheric CO2. Global Biogeochem. Cycles, 23, GB1005, doi:10.1029/2008GB003349.

    • Search Google Scholar
    • Export Citation
  • Hawcroft, M. K., , L. C. Shaffrey, , K. I. Hodges, , and H. F. Dacre, 2012: How much Northern Hemisphere precipitation is associated with extratropical cyclones? Geophys. Res. Lett.,39, L24809, doi:10.1029/2012GL053866.

  • Ho, M., , A. S. Kiem, , and D. C. Verdon-Kidd, 2012: The southern annular mode: A comparison of indices. Hydrol. Earth Syst. Sci., 16, 967982, doi:10.5194/hess-16-967-2012.

    • Search Google Scholar
    • Export Citation
  • Hoskins, B. J., , and K. I. Hodges, 2005: A new perspective on Southern Hemisphere storm tracks. J. Climate, 18, 41084129, doi:10.1175/JCLI3570.1.

    • Search Google Scholar
    • Export Citation
  • Irving, D., , I. Simmonds, , and K. Keay, 2010: Mesoscale cyclone activity over the ice-free Southern Ocean: 1999–2008. J. Climate, 23, 54045420, doi:10.1175/2010JCLI3628.1.

    • Search Google Scholar
    • Export Citation
  • Keeling, R. F., , and M. Visbeck, 2011: On the linkage between Antarctic surface water stratification and global deep-water temperature. J. Climate, 24, 35453557, doi:10.1175/2011JCLI3642.1.

    • Search Google Scholar
    • Export Citation
  • Koch, P., , H. Wernli, , and H. C. Davies, 2006: An event-based jetstream climatology and typology. Int. J. Climatol., 26, 283301, doi:10.1002/joc.1255.

    • Search Google Scholar
    • Export Citation
  • Krinner, G., , O. Magand, , I. Simmonds, , C. Genthon, , and J. L. Dufresne, 2007: Simulated Antarctic precipitation and surface mass balance at the end of the twentieth and twenty-first centuries. Climate Dyn., 28, 215230, doi:10.1007/s00382-006-0177-x.

    • Search Google Scholar
    • Export Citation
  • Lee, S., , and H.-K. Kim, 2003: The dynamical relationship between subtropical and eddy-driven jets. J. Atmos. Sci., 60, 14901503, doi:10.1175/1520-0469(2003)060<1490:TDRBSA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Li, Z. X., , and H. Le Treut, 1999: Transient behavior of the meridional moisture transport across South America and its relation to atmospheric circulation patterns. Geophys. Res. Lett., 26, 14091412, doi:10.1029/1999GL900274.

    • Search Google Scholar
    • Export Citation
  • Lorenz, C., , and H. Kunstmann, 2012: The hydrological cycle in three state-of-the-art reanalyses: Intercomparison and performance analysis. J. Hydrometeor., 13, 13971420, doi:10.1175/JHM-D-11-088.1.

    • Search Google Scholar
    • Export Citation
  • Lovenduski, N., , and T. Ito, 2009: The future evolution of the Southern Ocean CO2 sink. J. Mar. Res., 67, 597617, doi:10.1357/002224009791218832.

    • Search Google Scholar
    • Export Citation
  • Marshall, G. J., 2003: Trends in the southern annular mode from observations and reanalyses. J. Climate, 16, 41344143, doi:10.1175/1520-0442(2003)016<4134:TITSAM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Marsland, S., , and J. Wolff, 2001: On the sensitivity of Southern Ocean sea ice to the surface freshwater flux: A model study. J. Geophys. Res., 106, 27232741, doi:10.1029/2000JC900086.

    • Search Google Scholar
    • Export Citation
  • Meneghini, B., , I. Simmonds, , and I. N. Smith, 2007: Association between Australian rainfall and the southern annular mode. Int. J. Climatol., 27, 109121, doi:10.1002/joc.1370.

    • Search Google Scholar
    • Export Citation
  • Neu, U., and et al. , 2013: IMILAST: A community effort to intercompare extratropical cyclone detection and tracking algorithms. Bull. Amer. Meteor. Soc., 94, 529547, doi:10.1175/BAMS-D-11-00154.1.

    • Search Google Scholar
    • Export Citation
  • Nicolas, J. P., , and D. H. Bromwich, 2011: Precipitation changes in high southern latitudes from global reanalyses: A cautionary tale. Surv. Geophys., 32, 475494, doi:10.1007/s10712-011-9114-6.

    • Search Google Scholar
    • Export Citation
  • Noone, D., , and I. Simmonds, 2002: Annular variations in moisture transport mechanisms and the abundance of δ18O in Antarctic snow. J. Geophys. Res., 107, 4742, doi:10.1029/2002JD002262.

    • Search Google Scholar
    • Export Citation
  • Persson, P. O. G., , J. E. Hare, , C. W. Fairall, , and W. D. Otto, 2005: Air–sea interaction processes in warm and cold sectors of extratropical cyclonic storms observed during FASTEX. Quart. J. Roy. Meteor. Soc., 131, 877912, doi:10.1256/qj.03.181.

    • Search Google Scholar
    • Export Citation
  • Pezza, A. B., , H. A. Rashid, , and I. Simmonds, 2012: Climate links and recent extremes in Antarctic sea ice, high-latitude cyclones, southern annular mode and ENSO. Climate Dyn., 38, 5773, doi:10.1007/s00382-011-1044-y.

    • Search Google Scholar
    • Export Citation
  • Pfahl, S., , and H. Wernli, 2012: Quantifying the relevance of cyclones for precipitation extremes. J. Climate, 25, 67706780, doi:10.1175/JCLI-D-11-00705.1.

    • Search Google Scholar
    • Export Citation
  • Rudeva, I., , and S. K. Gulev, 2011: Composite analysis of North Atlantic extratropical cyclones in NCEP–NCAR reanalysis data. Mon. Wea. Rev., 139, 14191446, doi:10.1175/2010MWR3294.1.

    • Search Google Scholar
    • Export Citation
  • Schanze, J. J., , R. W. Schmitt, , and L. L. Yu, 2010: The global oceanic freshwater cycle: A state-of-the-art quantification. J. Mar. Res., 68, 569595, doi:10.1357/002224010794657164.

    • Search Google Scholar
    • Export Citation
  • Silvestri, G., , and C. Vera, 2009: Nonstationary impacts of the southern annular mode on Southern Hemisphere climate. J. Climate, 22, 61426148, doi:10.1175/2009JCLI3036.1.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., 2003: Modes of atmospheric variability over the Southern Ocean. J. Geophys. Res., 108, 8078, doi:10.1029/2000JC000542.

  • Simmonds, I., , and M. Dix, 1989: The use of mean atmospheric parameters in the calculation of modeled mean surface heat fluxes over the world’s oceans. J. Phys. Oceanogr., 19, 205215, doi:10.1175/1520-0485(1989)019<0205:TUOMAP>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., , and J. C. King, 2004: Global and hemispheric climate variations affecting the Southern Ocean. Antarct. Sci., 16, 401413, doi:10.1017/S0954102004002226.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., , K. Keay, , and E. Lim, 2003: Synoptic activity in the seas around Antarctica. Mon. Wea. Rev., 131, 272288, doi:10.1175/1520-0493(2003)131<0272:SAITSA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., , A. Rafter, , T. Cowan, , A. Watkins, , and K. Keay, 2005: Large-scale vertical momentum, kinetic energy and moisture fluxes in the Antarctic sea-ice region. Bound.-Layer Meteor., 117, 149177, doi:10.1007/s10546-004-5939-6.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., , K. Keay, , and J. A. T. Bye, 2012: Identification and climatology of Southern Hemisphere mobile fronts in a modern reanalysis. J. Climate, 25, 19451962, doi:10.1175/JCLI-D-11-00100.1.

    • Search Google Scholar
    • Export Citation
  • Sodemann, H., , and A. Stohl, 2009: Asymmetries in the moisture origin of Antarctic precipitation. Geophys. Res. Lett.,36, L22803, doi:10.1029/2009GL040242.

  • Talley, L. D., 2008: Freshwater transport estimates and the global overturning circulation: Shallow, deep and throughflow components. Prog. Oceanogr., 78, 257303, doi:10.1016/j.pocean.2008.05.001.

    • Search Google Scholar
    • Export Citation
  • Tietäväinen, H., , and T. Vihma, 2008: Atmospheric moisture budget over Antarctica and the Southern Ocean based on the ERA-40 reanalysis. Int. J. Climatol., 28, 19771995, doi:10.1002/joc.1684.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., , and C. J. Guillemot, 1998: Evaluation of the atmospheric moisture and hydrological cycle in the NCEP/NCAR reanalyses. Climate Dyn., 14, 213231, doi:10.1007/s003820050219.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., , L. Smith, , T. Qian, , A. Dai, , and J. Fasullo, 2007: Estimates of the global water budget and its annual cycle using observational and model data. J. Hydrometeor., 8, 758769, doi:10.1175/JHM600.1.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., , J. T. Fasullo, , and J. Mackaro, 2011: Atmospheric moisture transports from ocean to land and global energy flows in reanalyses. J. Climate, 24, 49074924, doi:10.1175/2011JCLI4171.1.

    • Search Google Scholar
    • Export Citation
  • Tsukernik, M., , and A. H. Lynch, 2013: Atmospheric meridional moisture flux over the Southern Ocean: A story of the Amundsen Sea. J. Climate, 26, 80558064, doi:10.1175/JCLI-D-12-00381.1.

    • Search Google Scholar
    • Export Citation
  • Turner, J., 2004: The El Niño–Southern Oscillation and Antarctica. Int. J. Climatol., 24, 131, doi:10.1002/joc.965.

  • Turner, J., , T. Lachlan-Cope, , J. Thomas, , and S. Colwell, 1995: The synoptic origins of precipitation over the Antarctic Peninsula. Antarct. Sci., 7, 327337, doi:10.1017/S0954102095000447.

    • Search Google Scholar
    • Export Citation
  • Uotila, P., , T. Vihma, , A. B. Pezza, , I. Simmonds, , K. Keay, , and A. H. Lynch, 2011: Relationships between Antarctic cyclones and surface conditions as derived from high-resolution numerical weather prediction data. J. Geophys. Res.,116, D07109, doi:10.1029/2010JD015358.

  • Uotila, P., , T. Vihma, , and M. Tsukernik, 2013: Close interactions between the Antarctic cyclone budget and large-scale atmospheric circulation. Geophys. Res. Lett., 40, 32373241, doi:10.1002/grl.50560.

    • Search Google Scholar
    • Export Citation
  • Vincent, D., 1994: The South Pacific convergence zone (SPCZ): A review. Mon. Wea. Rev., 122, 19491970, doi:10.1175/1520-0493(1994)122<1949:TSPCZA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Wernli, H., , and C. Schwierz, 2006: Surface cyclones in the ERA-40 dataset (1958–2001). Part I: Novel identification method and global climatology. J. Atmos. Sci., 63, 24862507, doi:10.1175/JAS3766.1.

    • Search Google Scholar
    • Export Citation
  • Yu, L., 2011: A global relationship between the ocean water cycle and near-surface salinity. J. Geophys. Res.,116, C10025, doi:10.1029/2010JC006937.

  • Yuan, X., 2004: ENSO-related impacts on Antarctic sea ice: Synthesis of phenomenon and mechanisms. Antarct. Sci., 16, 415425, doi:10.1017/S0954102004002238.

    • Search Google Scholar
    • Export Citation
  • Yuan, X., , J. Patoux, , and C. Li, 2009: Satellite-based midlatitude cyclone statistics over the Southern Ocean: 2. Tracks and surface fluxes. J. Geophys. Res.,114, D04106, doi:10.1029/2008JD010874.

  • View in gallery

    ERA-Interim freshwater flux, EP, at 0600 UTC 24 Jun 2008 (averaged during ±3 h; in units of mm day−1). Contours show SLP in intervals of 5 hPa.

  • View in gallery

    Schematic illustration for the procedure of the attribution of total precipitation to cyclones (label 1) and of strong precipitation to cyclones (label 2) and fronts (label 3) (see text for details).

  • View in gallery

    Climatological cyclone frequency (color; %) and front frequency (dark gray contours; %) for (a) DJF and (b) JJA averaged over the period 1980–2010. The framed areas PAN and PAS in (b) are referred to in section 6 on variability. The green contour highlights the 60°S circle.

  • View in gallery

    Seasonal mean freshwater fluxes (〈EP〉; red shading) and freshwater fluxes associated with cyclones [〈(EP)c〉; blue colors] averaged over the period 1980–2010 for (a) DJF and (b) JJA.

  • View in gallery

    Wintertime (JJA) freshwater fluxes (color) and vertically integrated precipitable water in intervals of 4 kg m−2 (red/blue contours) per cyclone (〈EP〉|cyclone) and under no-cyclone conditions (〈EP〉|no–cyclone).

  • View in gallery

    Wintertime (JJA) ratio of evaporation in the presence of a cyclone to evaporation in no-cyclone conditions (〈E〉|cyclone/〈E〉|no–cyclone). Areas with evaporation below 0.5 mm day−1 in cyclone or no-cyclone conditions, as well as land and sea ice covered regions, are shaded.

  • View in gallery

    The mean rates of (a),(d) strong precipitation (P75); (b),(e) cyclone precipitation; and (c),(f) front precipitation in (left) DJF and (right) JJA. In (a),(d), regions where strong precipitation exceeds 75% of the total precipitation are dotted. Also shown in (b),(e) are the 8%, 12%, 21%, and 38% contours of cyclone frequency (dark gray contours), while in (c),(f) the 2, 3, 4, and 5 m s−1 (1000 km)−1 contours of frontal intensity are shown (dark gray contours). Note that the rates are averaged over all time steps and do not represent the mean rate when P75 is occurring.

  • View in gallery

    The percentage of strong precipitation (P75) falling (a),(c) in cyclones and (b),(d) along fronts in (left) DJF and (right) JJA. Areas where P75 is below 0.01 mm day−1 are shaded gray. Also shown in (a),(c) are the 8%, 12%, 21%, and 38% contours of cyclone frequency (dark gray contours). In (b),(d) the 2, 3, 4, and 5 m s−1 (1000 km)−1 contours of frontal intensity are shown (dark gray contours).

  • View in gallery

    (a) Time series of wintertime (JJA) values of EP, P, and total and attributed P75 averaged over ocean grid points poleward of 50°S. (b) Wintertime mean of SAM and the cyclone frequency index I60S.

  • View in gallery

    (a) As in Fig. 9a, but with the average restricted to the Pacific sector poleward of 50°S. (b) Cyclone frequency indices for the PAN and PAS regions of the South Pacific.

  • View in gallery

    Difference in strong total, cyclone, and front precipitation between winters (JJA) with Ii > 0.5 and with Ii < −0.5 for the (top) PAN and (bottom) PAS regions of the South Pacific. The respective region is framed by the green box.

  • View in gallery

    Randomly attributed precipitation vs precipitation attributed to fronts (see text for details). (a) Seasonal averages over all ocean grid points south of 35°S. Different colors indicate the different overlap thresholds. (b) Wintertime average of randomly chosen grid points over the ocean south of 35°S. As a reference, the solid black line indicates where front precipitation equals randomly attributed precipitation.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 157 157 30
PDF Downloads 158 158 46

The Role of Extratropical Cyclones and Fronts for Southern Ocean Freshwater Fluxes

View More View Less
  • 1 Institute for Atmospheric and Climate Science, and Center for Climate Systems Modeling, ETH Zürich, Zürich, Switzerland
  • | 2 Institute for Atmospheric and Climate Science, ETH Zürich, Zürich, Switzerland
  • | 3 School of Earth Sciences, University of Melbourne, Melbourne, Australia
  • | 4 Institute for Atmospheric and Climate Science, ETH Zürich, Zürich, Switzerland
© Get Permissions
Full access

Abstract

In this study, the important role of extratropical cyclones and fronts for the atmospheric freshwater flux over the Southern Ocean is analyzed. Based on the Interim ECMWF Re-Analysis (ERA-Interim), the freshwater flux associated with cyclones is quantified and it is revealed that the structure of the Southern Hemispheric storm track is strongly imprinted on the climatological freshwater flux. In particular, during austral winter the spiraliform shape of the storm track leads to a band of negative freshwater flux bending toward and around Antarctica, complemented by a strong freshwater input into the midlatitude Pacific, associated with the split storm track. The interannual variability of the wintertime high-latitude freshwater flux is shown to be largely determined by the variability of strong precipitation (>75th percentile). Using a novel and comprehensive method to attribute strong precipitation uniquely to cyclones and fronts, it is demonstrated that over the Southern Ocean between 60% and 90% of the strong precipitation events are due to these synoptic systems. Cyclones are the dominant cause of strong precipitation around Antarctica and in the midlatitudes of the Atlantic and the Pacific, while in the south Indian Ocean and the eastern Atlantic fronts bring most of the strong precipitation. A detailed analysis of the spatial variations of intense front and cyclone precipitation associated with the interannual variability of the wintertime frequency of cyclones in the midlatitude and high-latitude branches of the Pacific storm track underpins the importance of considering both fronts and cyclones in the analysis of the interannual variability of freshwater fluxes.

Corresponding author address: Lukas Papritz, Institute for Atmospheric and Climate Science, ETH Zürich, 8092 Zürich, Switzerland. E-mail: lukas.papritz@env.ethz.ch

Abstract

In this study, the important role of extratropical cyclones and fronts for the atmospheric freshwater flux over the Southern Ocean is analyzed. Based on the Interim ECMWF Re-Analysis (ERA-Interim), the freshwater flux associated with cyclones is quantified and it is revealed that the structure of the Southern Hemispheric storm track is strongly imprinted on the climatological freshwater flux. In particular, during austral winter the spiraliform shape of the storm track leads to a band of negative freshwater flux bending toward and around Antarctica, complemented by a strong freshwater input into the midlatitude Pacific, associated with the split storm track. The interannual variability of the wintertime high-latitude freshwater flux is shown to be largely determined by the variability of strong precipitation (>75th percentile). Using a novel and comprehensive method to attribute strong precipitation uniquely to cyclones and fronts, it is demonstrated that over the Southern Ocean between 60% and 90% of the strong precipitation events are due to these synoptic systems. Cyclones are the dominant cause of strong precipitation around Antarctica and in the midlatitudes of the Atlantic and the Pacific, while in the south Indian Ocean and the eastern Atlantic fronts bring most of the strong precipitation. A detailed analysis of the spatial variations of intense front and cyclone precipitation associated with the interannual variability of the wintertime frequency of cyclones in the midlatitude and high-latitude branches of the Pacific storm track underpins the importance of considering both fronts and cyclones in the analysis of the interannual variability of freshwater fluxes.

Corresponding author address: Lukas Papritz, Institute for Atmospheric and Climate Science, ETH Zürich, 8092 Zürich, Switzerland. E-mail: lukas.papritz@env.ethz.ch

1. Introduction

The net surface flux of freshwater into the atmosphere (viz., EP, where E is evaporation and P is precipitation) is arguably the most important hydrological parameter for the atmosphere and the ocean. Early climatologies of the global water cycle and of freshwater fluxes date back to the 1960s and were summarized by Baumgartner and Reichel (1975). In these early studies, E was obtained from mean atmospheric parameters, where temporal covariances were neglected, which over the Southern Ocean (SO) leads to significant biases in the estimation of the mean fluxes (Simmonds and Dix 1989). Nowadays, satellite observations allow us to monitor the water cycle on synoptic time scales with an almost global coverage. Nevertheless, the quantity EP remains very difficult to estimate and one of the best ways to do so is by using atmospheric reanalyses. These datasets are based on all available observations, are constrained by the fundamental physical laws, and specifically represent the atmospheric moisture cycle (Trenberth and Guillemot 1998; Trenberth et al. 2007; Bosilovich et al. 2011) and atmospheric freshwater fluxes (Schanze et al. 2010) on a global scale. Furthermore, in reanalyses mean surface fluxes can be calculated from the instantaneous fields, which over the SO is considerably more accurate than using mean fields (Simmonds and Dix 1989).

a. Freshwater fluxes over the Southern Ocean

Recently, EP over the Southern Hemisphere (SH) middle to high latitudes and the sea ice region around Antarctica has increasingly become a focus of research. The stratification of the Southern Ocean has been shown to be highly sensitive to changes in EP, which has strong implications for ocean convection and the uptake of anthropogenic CO2 by the ocean (Lovenduski and Ito 2009; Gruber et al. 2009). Freshwater input into the Southern Ocean is a key factor for the formation rates of water masses and for the meridional overturning circulation (Talley 2008; Keeling and Visbeck 2011), which affect the climate system globally. The buoyancy gain required for upwelled deep waters to be converted into less dense waters north of the Subantarctic Front to a large extent comes from the input of freshwater (Cerovečki et al. 2013). Furthermore, the weak surface stratification of the SO, which is largely maintained by EP, plays a key role in adjusting the temperature of the deep waters (Keeling and Visbeck 2011). Over the sea ice region surrounding Antarctica and the nearby ocean, P exceeds E in all seasons (Simmonds et al. 2005). The magnitude of EP impacts on the formation rate of sea ice. It has been shown, for example, that a reduced supply of freshwater can cause the formation of a polynya in the Weddell Sea (Marsland and Wolff 2001), which in turn considerably influences the local climate via the exchange of latent and sensible heat between the atmosphere and the open waters. An accurate knowledge of the pathways of moisture to Antarctica and EP over Antarctica is crucial for estimating the surface mass balance of the ice sheet (Krinner et al. 2007; Bromwich et al. 2011). The study by Sodemann and Stohl (2009) showed that the source regions for Antarctic precipitation over the SO vary greatly between the ocean basins, which in part might be linked to the varying frequency and pathways of extratropical cyclones and fronts (Noone and Simmonds 2002).

Measurements of P over the SO are very sparse because of its remoteness. Evaporation can only be measured via eddy covariances, which needs considerable technical effort and cannot be done over a larger area (Persson et al. 2005). Therefore, on the large scale E can only be estimated indirectly from the stability of the atmospheric boundary layer, wind velocity, and the humidity difference between the atmosphere and the ocean surface layer (Charnock 1955; Fairall et al. 2003). The values thus obtained depend strongly on the chosen parameterization. Because of these difficulties, the study of EP in this region necessarily relies on reanalysis data (e.g., Simmonds et al. 2005; Tietäväinen and Vihma 2008; Bromwich et al. 2011). In an intercomparison study between different reanalyses, Trenberth et al. (2011) demonstrated that there are broad inconsistencies in the representation of EP. As pointed out by Schanze et al. (2010), moisture is not conserved because of numerical inconsistencies in the underlying models and analysis increments. Furthermore, changes in the global observing system can produce artificial trends (e.g., Dee et al. 2011) and inhomogeneities in the time series (Schanze et al. 2010). The ocean represents an integrator of EP via surface salinity (Yu 2011) and buoyancy fluxes, adjusted to estimates of the state of the upper ocean from measurements, provide a benchmark for the accuracy of EP estimates (Cerovečki et al. 2011). For the Interim European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-Interim; Dee et al. 2011) these estimates indicate a bias toward too much precipitation along the Antarctic Circumpolar Current and too little at higher latitudes (Cerovečki et al. 2011). However, the decomposition of the adjusted buoyancy fluxes into EP and heat fluxes depends on the accurate representation of the ocean state and internal ocean dynamics, as well as sea ice processes, which themselves are subject to large uncertainties and limited by the sparsity of observations. Thus, at present this approach is able to indicate biases in EP estimates but is not sufficiently precise to provide a more accurate reference.

Analysis of E and P over the southern high latitudes, particularly over Antarctica (Bromwich et al. 2011; Nicolas and Bromwich 2011), showed that interannual trends are not well represented in and disagree between many of the present reanalyses. Nicolas and Bromwich (2011) concluded that the ERA-Interim provides the most reliable trend estimates and shows fewest artificial jumps in latent heat flux and P that can be attributed to model errors or to changes in the observing system. This is underpinned by the recent intercomparison between three modern reanalyses and observation datasets of the global water cycle by Lorenz and Kunstmann (2012).

Despite apparent deficiencies of reanalysis datasets, the basic physical consistency within a particular dataset, the global coverage, and high temporal resolution are three of the major benefits of using reanalyses, which make multidecadal process studies possible in the first place. Furthermore, the consistent treatment of parameterized physical processes gives hope that the relative contribution of cyclones and fronts to P and EP, compared to other weather systems, is relatively independent of the actual dataset used, provided that the cyclones and fronts themselves are well represented. Having said that, because of the different data assimilation systems used in the various reanalyses, there may indeed be differences in their representation, in particular in data-sparse regions (e.g., see Simmonds 2003). In that respect, ERA-Interim performs particularly well, as the introduction of the four-dimensional variational data assimilation (4D-Var) system leads to more efficient use of observations and thus to significant improvements in the SH (Dee et al. 2011): for instance, in terms of RMS errors of sea level pressure (SLP).

In the southern middle to high latitudes, moisture transport as well as E and P are driven and modulated by transient weather systems (Simmonds and King 2004; Tietäväinen and Vihma 2008; Uotila et al. 2011; Tsukernik and Lynch 2013) and therefore show a strong day-to-day variability. Interannual variations of the meridional moisture flux across the Southern Ocean by transient eddies are strong but, as pointed out by Tsukernik and Lynch (2013), they are not significantly correlated with the principal modes of variability in the SH: namely, the southern annular mode (SAM) and El Niño–Southern Oscillation. Of particular interest are the Amundsen and Bellingshausen Seas, where mean meridional moisture fluxes as well as variability peak with respect to the other sectors of the high-latitude Southern Ocean. In previous studies this region was found to represent a “pole of variability” in SLP (Connolley 1997) and therefore it shows large interannual variability of the frequency of synoptic and mesoscale cyclones.

Figure 1 shows the ERA-Interim EP and SLP at 0600 UTC 24 June 2008. The areas of excessive P are closely confined and strong, while areas of excessive E occupy extended areas but have a smaller magnitude. Furthermore, it is apparent that the negative EP values are mostly found in synoptically active regions, such as near extratropical cyclones and frontal zones, as indicated by the closed SLP contours and the kinks in the isobars. The strong cyclone between South Africa and the Weddell Sea brings excessive P in the warm sector and along the occluded front, mostly within closed SLP contours around its pressure minimum. In addition, its elongated cold front extending into the subtropics clearly shows a strongly negative signal in EP. As has been discussed in earlier studies, cyclones and associated fronts are the most common weather systems in the southern middle to high latitudes (Wernli and Schwierz 2006; Yuan et al. 2009; Simmonds et al. 2012). They are especially frequent in a zonally elongated band along the coast of Antarctica. These synoptic systems were found to be responsible for about 80% of P at Rothera station on the Antarctic Peninsula, whereby in 48% of the cases a front related to a cyclone caused the P (Turner et al. 1995). Catto et al. (2012) found that front P accounts for at least 50% of P in the SH’s extratropics. Furthermore, the majority of extreme P events in the storm-track regions of the SO are caused by the passage of an extratropical cyclone (Pfahl and Wernli 2012).

Fig. 1.
Fig. 1.

ERA-Interim freshwater flux, EP, at 0600 UTC 24 Jun 2008 (averaged during ±3 h; in units of mm day−1). Contours show SLP in intervals of 5 hPa.

Citation: Journal of Climate 27, 16; 10.1175/JCLI-D-13-00409.1

b. Objective identification of extratropical cyclones and fronts

Objective identification of extratropical cyclones has a long history and accordingly a vast number of different methods exists and results between them differ considerably. Nevertheless, in the recent comparison of 15 of the most commonly used methods (Neu et al. 2013), it was shown that climatologically the methods agree surprisingly well in terms of geographical distribution, interannual variability, and the cyclones’ key characteristics. Based on the identification of a cyclone, additional atmospheric parameters such as freshwater fluxes can be studied in its environs. To this end, an area influenced by the cyclone—the so-called cyclone area—must be specified, over which the quantities are integrated. An appropriate choice for this area is that enclosed by the outermost closed isobar surrounding the cyclone center (Wernli and Schwierz 2006).

Even more complex than the identification of cyclones is the automated detection of atmospheric fronts. There is only a small number of such algorithms, and very few have been applied to the SH. The detection of fronts based on a thermodynamic field such as wet-bulb potential temperature seems to be straightforward as this field highlights the boundary between two air masses. However, such methods may identify artificial fronts: for instance, in areas near the sea ice edge, where often stationary temperature gradients occur (Berry et al. 2011). This can be avoided when fronts are detected from Eulerian changes of the meridional wind as was done in the recent study by Simmonds et al. (2012). Their algorithm identifies so-called mobile fronts and was designed and extensively tested for the SH.

By identifying fronts and cyclones, P can be linked to these particular synoptic systems. Catto et al. (2012) used a straightforward technique, considering P at a certain grid point to be of frontal origin if a front is present within a 5° × 5° box. Similarly, for cyclones in the Northern Hemisphere, Hawcroft et al. (2012) used a spherical cap of fixed radius centered on an identified cyclone. With these methods based on distance, cyclone, and front, P cannot easily be disentangled, as parts of the P potentially belong to both categories. However, as regions of the SO are predominantly influenced by either cyclones or fronts, it is of interest to quantify the contribution of both types of synoptic weather systems individually.

c. Specific aims and structure of this study

The purpose of this study is to quantify the important role played by extratropical cyclones and fronts for seasonal EP over the SO and its interannual variability. To that end we present a novel and detailed method for separately attributing strong P (>75th percentile) to cyclones and fronts. In this we build on existing feature identification algorithms for cyclones (Wernli and Schwierz 2006) and fronts (Simmonds et al. 2012). The methods and the ERA-Interim dataset used will be discussed in section 2.

Using the cyclone dataset we demonstrate that the structure and shape of the SH’s storm track, which is briefly discussed in section 3, is of key importance for the climatological distribution of EP over the SO (section 4). In section 5, our method of attribution is used to climatologically quantify strong cyclone and front P, and regions are identified where cyclones or fronts are the primary contributors to strong P. Finally, it is shown in section 6 that attributed strong P is the major driver of interannual variations of wintertime EP and a detailed analysis is presented of the spatial variability of strong front and cyclone P in the South Pacific.

2. Data and methodology

a. Dataset

This study is based on the ERA-Interim dataset. ERA-Interim builds on a four-dimensional variational data assimilation system with a 12-hourly assimilation cycle. Furthermore, the underlying model imposes physical constraints on the observations and allows for their flow-dependent weighting. This technique is particularly beneficial in data-sparse regions such as the SO, where ERA-Interim performs much better than its predecessor, the 40-yr ECMWF Re-Analysis (ERA-40; Dee et al. 2011). While many of the problems related to the hydrological cycle that were present in ERA-40 were eliminated in ERA-Interim, an error in the assimilation of Special Sensor Microwave Imager (SSM/I) radiances led to a spurious drying of the atmosphere and a reduction of global P as more SSM/I observations became available over time (Dee et al. 2011). This drying weakened as the number of observations decreased again. However, according to Nicolas and Bromwich (2011), the influence of this assimilation error over the southern middle to high latitudes is minor.

The analysis is performed over the 31-yr period from December 1979 to November 2010. The data are available every 6 h and the fields are interpolated on a 1° × 1° horizontal grid. The P and surface latent heat flux are accumulated forecast fields from twice-daily short-range forecasts. To avoid biases from model spinup, the averages are calculated from forecast steps between hours 9 and 21 and at a particular synoptic time t we use the centered 6-hourly average: that is, from t − 3 h to t + 3 h. Evaporation from the surface (E) is obtained from the latent heat flux LH via the relation E = −LH/(ρlLυ), where Lυ = 2.5 × 106 J kg−1 denotes the latent heat of vaporization and ρl = 999.84 kg m−3 is the reference density of liquid water. The EP is calculated from the difference of E and P.

b. Cyclone identification

Cyclones are identified with a refined version of the algorithm developed by Wernli and Schwierz (2006), which is based on the identification of local SLP minima. To avoid artifacts from the extrapolation of pressure to sea level, minima are removed where the topography exceeds 1500 m. For every remaining SLP minimum, the cyclone area is bounded by the outermost closed SLP contour that encloses the one minimum under consideration or several additional minima within a radius of 1000 km but no local SLP maximum. The interval for searching SLP contours is 0.5 hPa. To avoid unrealistically large cyclone areas, the maximum contour length is restricted to 7500 km. A cyclone field is defined with the value 1 at grid points that lie inside such an outermost closed contour and 0 at grid points outside. This cyclone field is a natural estimate of the area influenced by the presence of a cyclone and will be used to determine for instance the contribution of cyclones to P. From temporal averaging of the cyclone fields fc, the frequencies of cyclones 〈fc〉 are obtained, where angle brackets denote a temporal average.

c. Front identification

For the fronts we use the dataset of mobile fronts by Simmonds et al. (2012), which provides frontal points and various key characteristics such as frontal length and intensity. The dataset has been extended to cover the entire period from 1979 to 2010. The identification of fronts is based on Eulerian changes of the 10-m meridional wind obtained from ERA-Interim. At a time t grid points are flagged where the wind turns from northwesterly to southwesterly between subsequent times t and t + 6 h and the change of the meridional wind exceeds 2 m s−1 during the same 6 h. Subsequently flagged grid points are clustered by connectivity and the eastern edge of each cluster is determined. After smoothing by multiple filters, this eastern edge determines a “mobile front.” This method is particularly suited for the detection of strongly elongated, meridionally oriented moving fronts, which typically extend far equatorward from the cyclone center. A potential weakness is that more zonally oriented fronts (often warm fronts) are only partly identified or even remain undetected, because the Eulerian change of the meridional wind is below the threshold. However, this poses no major problem for the present analysis, as warm fronts typically are shorter and contained within the cyclone areas.

Finally, the characteristics of a front are calculated. According to Simmonds et al. (2012) the intensity of a front is defined by
e1
where N is the number of frontal points and υi is the meridional wind at point i. This measure accounts for the “integrated effect” of a front; it is proportional to its length and the alongfront change of the meridional wind as the front passes by. Therefore, a long front with a small meridional wind change can have the same intensity as a shorter front with stronger shear. Mean intensities are obtained from gridding the intensity of each front at its center of gravity.

From the identification method, the question arises whether a detected front should be attributed to time t or t + 6 h. Manual comparison of a number of cases of identified fronts with SLP, potential temperature, and P fields revealed that the identified fronts on average correspond best to structures in the other fields at time t + 6 h. Therefore, the fronts are assigned to time t + 6 h. Furthermore, short fronts with a meridional extent of less than 4.5° of latitude are rejected, as they often appear as artifacts of the detection method. The remaining fronts are then interpolated on a 1° × 1° grid and expanded by 1° in every direction, such that frontal lines can be associated with “frontal areas.” At every time step this yields a front field, which, similar to the cyclone field, has a value of 1 where an expanded front is found and 0 elsewhere. As for cyclone frequencies, front frequencies are obtained from time averaging the front fields.

d. Attribution to cyclones and fronts

Because of the different morphological characteristics of the two synoptic weather systems under consideration—cyclones cover a wide area, while fronts are narrow and elongated objects—two different approaches are adopted for the attribution to cyclones and fronts.

1) Cyclones

The contribution of cyclones to E, P, and EP is determined based on colocation with a cyclone. More precisely the value of a field Φ (e.g., EP) is retained where a cyclone is present and set to 0 elsewhere, yielding the cyclone contribution Φc. An illustration of this procedure is shown in Fig. 2a, where P attributed to cyclones (label 1 in Fig. 2a) is highlighted in yellow. Temporal averaging of Φc gives the climatological contribution of cyclones 〈Φc〉. It is also of interest to quantify the average value of Φ given that a cyclone is present: that is, 〈Φ〉 per cyclone. In the following, this quantity will be denoted as 〈Φ〉|cyclone, which is obtained from 〈Φc〉 by dividing by the frequency of cyclones 〈fc〉,
e2
Similarly, the expected value of Φ conditioned that no cyclone is present 〈Φ〉|no–cyclone can be calculated as
e3
Grid points where 〈fc〉 or 1 − 〈fc〉 are below 1% are set to missing data.
Fig. 2.
Fig. 2.

Schematic illustration for the procedure of the attribution of total precipitation to cyclones (label 1) and of strong precipitation to cyclones (label 2) and fronts (label 3) (see text for details).

Citation: Journal of Climate 27, 16; 10.1175/JCLI-D-13-00409.1

2) Fronts

Front areas are narrow and hence considerably smaller than cyclone areas. If the same method for attribution of P to fronts was used as for the attribution to cyclones, a large portion of P a meteorologist classifies as front P would remain unattributed. Since front P can be displaced with respect to the identified front and also front P typically occurs in a much wider band than covered by the front area, the attribution of P to fronts is undertaken in the following three steps:

  1. Precipitation objects (P75): Coherent objects of strong precipitation (P75) are defined as connected grid points (precipitation objects) where 6-hourly P exceeds the climatological 75th percentile for the respective month and grid point.
  2. Attribution to front: For every precipitation object, the overlap between the object and the region where the front field has the value 1 is determined. If the overlap exceeds 10% of the size of the precipitation object, the entire precipitation object is retained; otherwise, it is rejected.
  3. Removal of cyclone contribution: The part of the precipitation object collocated with a cyclone is removed.

Finally, the contribution of cyclones to strong precipitation P75 is obtained from cyclone P by removing P below the climatological 75th percentile for the respective month and grid point. This method ensures that P75 is uniquely classified as cyclone, front, or nonattributed P75. Of the three P75 objects depicted in Fig. 2, the one on the bottom left is fully collocated with the cyclone. The second is attributed to the front (label 3a) but, because part of it is collocated with the cyclone as well, only its upper part is front P75 (labels 2 and 3b). The small object to the right remains unattributed.

The percentile value for the identification of strong P objects and the overlap threshold for the attribution of P75 objects to fronts are tunable parameters. We refer the reader to a discussion of the optimum choice of these parameters in the appendix. It is important to note that this method can only be applied to coherent objects, such as the P75 objects, and is not applicable to E and EP, as these do not allow for a meaningful identification of coherent objects. As will become clear in section 6, the sum of front and cyclone P75 represents an excellent index for the interannual variability of seasonal EP over the SO, particularly the high latitudes, and therefore provides a useful basis for the quantification of the processes driving EP over the SO.

3. Frequency of cyclones and fronts

Figure 3 shows the climatological frequency of cyclones and fronts in summer and winter. Throughout this paper, we refer to regions of relatively high cyclone frequency as storm tracks, which corresponds well with track densities shown in Hoskins and Hodges (2005). A ring of high cyclone frequency extends around Antarctica in both seasons, intercepted only by the Antarctic Peninsula. Three maxima of cyclone frequency with year-round peaks of up to 40%–50% are located in the high-latitude south Indian Ocean and in the Amundsen and Bellingshausen Seas. The latter maximum is shifted to the Ross Sea in winter, where many of the cyclones incoming from the south Indian Ocean decay (Simmonds et al. 2003; Wernli and Schwierz 2006), which contributes to the wintertime reduction of cyclone frequency in the Amundsen Sea. In the midlatitudes, cyclone activity is more pronounced during winter. A major region for cyclogenesis is the lee of the Andes roughly at 30°S (Wernli and Schwierz 2006, their Fig. 7), where the channeling of the air by the Andes gives rise to a northerly low-level jet that advects very moist, tropical air into the La Plata River basin (e.g., Berbery and Barros 2002). As shown by Wernli and Schwierz (2006, their Fig. 11c), these systems constitute a major portion of the midlatitude Atlantic storm activity and typically move southeastward toward the coast of Antarctica, where they connect to the high-latitude storm tracks and give rise to the wintertime spiraliform nature of the SH storm tracks. It is an outstanding feature of the midlatitude south Indian Ocean that also in winter cyclone frequency is very low, despite a strong subtropical jet (e.g., Koch et al. 2006) and high baroclinicity (Hoskins and Hodges 2005). The wintertime subtropical Pacific split jet (Bals-Elsholz et al. 2001) is reflected in a maximum of cyclone frequency detached from the high-latitude storm track, starting off the east coasts of Australia and New Zealand. The majority of these systems follows a slightly poleward curved trajectory (Hoskins and Hodges 2005). In summer a number of cyclones develop between 180° and 150°W in the southward extension of the South Pacific convergence zone (SPCZ). A similar feature can also be observed in the Indian Ocean east of Africa, albeit less prominent. As suggested by Wernli and Schwierz (2006), some of these systems may be tropical cyclones.

Fig. 3.
Fig. 3.

Climatological cyclone frequency (color; %) and front frequency (dark gray contours; %) for (a) DJF and (b) JJA averaged over the period 1980–2010. The framed areas PAN and PAS in (b) are referred to in section 6 on variability. The green contour highlights the 60°S circle.

Citation: Journal of Climate 27, 16; 10.1175/JCLI-D-13-00409.1

As fronts are intimately related to cyclones, it comes as no surprise that front frequencies are aligned with the storm tracks. The fronts belonging to cyclones that travel at high latitudes along the coast of Antarctica cause a circular belt with high front frequency equatorward of the associated storm track, with a year-round local maximum in the south Indian Ocean. In winter this belt extends farther into subtropical latitudes, which is most pronounced in the Pacific Ocean because of the separation of the storm tracks. A high number of quasi-stationary fronts develop in the South Pacific (Berry et al. 2011), related to the high baroclinicity associated with the SPCZ (Vincent 1994), but these are not represented in our dataset because of their immobile character. Furthermore, during summer west and east of the Andes local maxima in front frequency are found, which are absent in winter. Otherwise, frontal frequency shows weak seasonality, as opposed to front intensity, which in winter has pronounced peaks in the eastern Atlantic and Indian Oceans (Simmonds et al. 2012). In the latter region at around 40°S the typical length of fronts exceeds 2000 km in some parts of the basin.

4. Seasonal freshwater fluxes associated with cyclones

a. Contribution of cyclones to E − P

Total EP (〈EP〉) and the contribution of cyclones [〈(EP)c〉] are depicted in Fig. 4 for December–February (DJF) and June–August (JJA). Total EP changes sign in a narrow area separating the subtropics and the midlatitudes with a sharp gradient. In the subtropics E exceeds P, and in the middle to high latitudes P dominates. In both seasons 〈(EP)c〉 is negative in almost the entire region considered. In summer (Fig. 4a) the largest absolute values of 〈(EP)c〉 are found in a ring surrounding Antarctica with a bulge that extends into the subtropical Atlantic. In winter (Fig. 4b) the circular structure is broken and 〈(EP)c〉 assumes a spiraliform shape starting between 30° and 38°S off the coast of South America, turning poleward and sloping around Antarctica in a narrowing band that reaches as far as the Antarctic Peninsula. The magnitude of 〈(EP)c〉 in this band is remarkably uniform from the subtropics to polar latitudes, and it reveals the spiral nature of the SH winter storm track more clearly than the cyclone frequency itself. The spiral hosts local minima, one downstream of the estuary of La Plata River (34°S, 59°W) and a second off the coast of Wilkes Land in Antarctica (65°S, 110°E), collocated with the peak of cyclone frequency there. In addition, a strongly negative signal of 〈(EP)c〉 is found along the Antarctic coastline of the Bellingshausen and Amundsen Seas, where decaying cyclones advect moist air against the steep coast and cyclone P becomes orographically enhanced. Furthermore, in winter sea ice inhibits E from the ocean and hence favors more negative values of 〈(EP)c〉. This is most distinct in the Amundsen and Bellingshausen Seas, where cyclone frequency is larger in summer but because of the lower sea ice extent both cyclone and total EP are less negative. In the Pacific the split storm track causes the second, zonally oriented band of strongly negative 〈(EP)c〉 between 30° and 45°S, which spans from eastern Australia to the Andes.

Fig. 4.
Fig. 4.

Seasonal mean freshwater fluxes (〈EP〉; red shading) and freshwater fluxes associated with cyclones [〈(EP)c〉; blue colors] averaged over the period 1980–2010 for (a) DJF and (b) JJA.

Citation: Journal of Climate 27, 16; 10.1175/JCLI-D-13-00409.1

The major seasonal patterns of 〈(EP)c〉 are similar to those of 〈EP〉, suggesting the important role played by extratropical cyclones for total EP. In most regions, 〈EP〉 has a slightly larger magnitude than 〈(EP)c〉 and strongly negative values extend farther north. This is mainly due to P associated with fronts that extend northward from the closed SLP contours. Unlike in the other two basins, the absence of a midlatitude winter storm track in the Indian Ocean is accompanied by a southward shift to about 40°S of the subtropical band where E exceeds P (Fig. 4b). In the other two basins this is only the case at the location of the climatological anticyclones in their eastern part.

Figure 5a shows wintertime EP per cyclone (〈EP〉|cyclone; see section 2d for a precise definition of 〈EP〉|cyclone). Toward higher latitudes it decreases in accord with precipitable water per cyclone. It is most negative in the Atlantic; in a band that stretches downstream of the La Plata River basin; and in the western south Indian Ocean, in the extension of the peak of subtropical moisture inflow east of South Africa, attaining values of less than −10 mm day−1. The values of 〈EP〉|cyclone are less negative in the midlatitude Pacific, which can be ascribed to two processes that reduce precipitable water. First, the rather zonal track of cyclones erodes atmospheric moisture; second, the meridional inflow of moisture from the subtropics takes place in a broader region centered in the middle of the basin at about 130°W (not shown), so that precipitable water is not replenished sufficiently fast after the passage of storms. Closer to Antarctica 〈EP〉|cyclone is zonally more uniform, with a negative peak off the coast of Wilkes and Victoria Lands between 90°E and 180°.

Fig. 5.
Fig. 5.

Wintertime (JJA) freshwater fluxes (color) and vertically integrated precipitable water in intervals of 4 kg m−2 (red/blue contours) per cyclone (〈EP〉|cyclone) and under no-cyclone conditions (〈EP〉|no–cyclone).

Citation: Journal of Climate 27, 16; 10.1175/JCLI-D-13-00409.1

Figure 5b shows the same fields in no-cyclone conditions. Values of precipitable water are typically smaller than those for cyclone conditions. An exception is the eastern Pacific between 60° and 70°S, where precipitable water inside and outside cyclones is similar. These cyclones are typically older and have already lost a considerable amount of precipitable water because of P during the earlier stages of their development. A much dryer atmosphere in the absence of cyclones is observed near the Amery and the Ross ice shelves, most probably because of dry katabatic winds from Antarctica. The effect of the presence of cyclones on EP is rather dramatic. There is a clear transition from the subtropics, where E strongly exceeds P in the absence of cyclones, to the middle and high latitudes, where E in no-cyclone conditions is compensated by P, causing 〈EP〉|no–cyclone to be only weakly negative (Fig. 5b). The slightly more negative values of 〈EP〉|no–cyclone in the eastern Atlantic and south Indian Oceans are related to front P, while along parts of the coast of Antarctica orographic P and the sea ice cover both lead to a more negative signal of 〈EP〉|no–cyclone.

Along the coast of South America, where the warm Brazil Current encounters the colder waters of the Malvinas Current, a northward protruding branch of the Antarctic Circumpolar Current, 〈EP〉 is positive (>1.1 mm day−1) in a narrow tongue that stretches as far south as 47°S (most pronounced in winter; Fig. 4). At the same time extratropical cyclones that develop over the La Plata River basin or just off the coast are associated with a strongly negative signal of 〈(EP)c〉 and provide about 1.1 mm day−1 of freshwater to the ocean. The moisture content per cyclone there is exceptional when compared to other regions in the SH (Fig. 5a). The southward transport of moist air from the subtropics into the La Plata River basin by the low-level jet in the lee of the Andes (Li and Le Treut 1999; Berbery and Barros 2002) causes the amount of precipitable water per cyclone to be 2 times larger than in their absence. Excessive E occurs in the absence of cyclones (Fig. 5b), when the predominantly westerly flow advects relatively dry air first over cold waters of the Malvinas Current and then along the almost zonal gradient of SST over the warm waters of the Brazil Current.

To quantitatively assess the contribution of cyclones to the total fluxes of E, P, and EP over the SO, the DJF and JJA seasonal means of these fluxes integrated over the ocean between 35°S and the coast of Antarctica are shown in Table 1. To complete the picture, the fluxes per cyclone are also tabulated. Interestingly, the area averaged P per cyclone is the same in both seasons (Table 1). Because of the more frequent cyclones in midlatitudes, the overall contribution of cyclones to P is larger in winter and accounts for about 30% of the total P. The contribution of cyclones to the total E is about 10% in JJA (7% in DJF) and supports the estimate by Yuan et al. (2009). Enhanced cyclone activity in JJA in the midlatitude Atlantic and the Pacific (Fig. 3) implies larger midlatitude contributions of cyclones to E. The bottom line is a freshwater input into the ocean associated with extratropical cyclones that is about 80% of the total freshwater flux. This is not surprising, as enhanced E outside cyclones largely compensates for the additional P that occurs outside cyclones (Fig. 5b). Equal P per cyclone in both seasons and the stronger E during winter cause EP per cyclone to be slightly more negative in summer (Table 1).

Table 1.

Evaporation, precipitation, and EP (mm day−1), averaged over ocean grid points between 35°S and the coast of Antarctica.

Table 1.

b. Role of cyclones for evaporation

Despite strong winds associated with cyclones, typically inside an individual cyclone E is lower than in the climatological mean, which is underpinned by Fig. 6, showing the ratio of E in cyclone and no-cyclone conditions (〈E〉|cyclone/〈E〉|no–cyclone) in winter. In the eastern midlatitude South Atlantic, the western Indian Ocean, and the Pacific, E in the presence of a cyclone is less than 75% of E in no-cyclone conditions. This can be attributed to the fact that the humid air in the warm sector of an extratropical cyclone is associated with a weak near-surface vertical humidity gradient and therefore supports only small moisture fluxes from the ocean surface (Persson et al. 2005). In contrast advection of cold dry air at the back of the cyclone and the mixing induced by strong winds along the bent-back front promotes stronger latent heat fluxes, leading to a compensatory effect between the warm sector and the rear of the cyclone. The study by Rudeva and Gulev (2011) found similar results for the North Atlantic. The picture changes considerably toward higher latitudes and especially near the sea ice edge. There, as a consequence of the advection of cold, dry air from the sea ice, the relative contribution of the strong E at the back of cyclones becomes larger compared to the reduced E in the warm sector. In a band spanning from the Ross to the Bellingshausen Seas E per cyclone even exceeds E under no-cyclone conditions. The effect of sea ice is strongly reduced in summer and E per cyclone is slightly lower (not shown).

Fig. 6.
Fig. 6.

Wintertime (JJA) ratio of evaporation in the presence of a cyclone to evaporation in no-cyclone conditions (〈E〉|cyclone/〈E〉|no–cyclone). Areas with evaporation below 0.5 mm day−1 in cyclone or no-cyclone conditions, as well as land and sea ice covered regions, are shaded.

Citation: Journal of Climate 27, 16; 10.1175/JCLI-D-13-00409.1

5. Strong precipitation attributed to cyclones and fronts

The important role of cyclones in shaping negative 〈EP〉 over the SO via strong events of P calls for a closer analysis of the spatial and temporal variability of cyclone P. Even though a large portion of P outside cyclones is compensated by E, when averaged over the entire SO basin, the northward extension of the negative values of 〈EP〉 with respect to the storm tracks (Fig. 4) suggests that fronts locally cause significant freshwater input to the ocean, which in the mean is compensated by excessive E. Therefore, in this section, in addition to cyclone P we also consider P that falls along fronts. As discussed in section 2, an attribution based on colocation, which works well for cyclones, does not yield meaningful results for fronts. To circumvent this difficulty, we restrict the analysis to strong precipitation above the 75th percentile, allowing us to identify coherent objects of precipitation that can be meaningfully attributed to fronts (see section 2).

a. Climatology of strong precipitation

The P75 over the SO (Figs. 7a,d) is largest and shows the most pronounced seasonality in the midlatitude Atlantic and Pacific, with maxima in winter. In the midlatitude Indian Ocean P75 has a winter peak as well but the seasonality is weaker. The areas of high P75 in the Atlantic and Indian Oceans are connected and show a spiral shape similar to that of EP, but with weaker poleward slopes. The spiral ends south of the Great Australian Bight. The zonal band in the Pacific is also shifted northward with respect to the storm track. Also, upwind of the Andes, the Antarctic Peninsula, and New Zealand, P75 adopts relatively large values. The summer peak in the Andes is confined to the southern edge of the mountain range, while in winter it extends farther north. The summer P75 maxima in the subtropical Atlantic and Pacific are the most southward branches of the South Atlantic convergence zone and the SPCZ, respectively. In midlatitudes, the strongest 25% of 6-hourly P events (i.e., the P75 events) contribute more than 75% of the total P (stippled), while at high latitudes (and also everywhere else) the values are lower but well above 60%.

Fig. 7.
Fig. 7.

The mean rates of (a),(d) strong precipitation (P75); (b),(e) cyclone precipitation; and (c),(f) front precipitation in (left) DJF and (right) JJA. In (a),(d), regions where strong precipitation exceeds 75% of the total precipitation are dotted. Also shown in (b),(e) are the 8%, 12%, 21%, and 38% contours of cyclone frequency (dark gray contours), while in (c),(f) the 2, 3, 4, and 5 m s−1 (1000 km)−1 contours of frontal intensity are shown (dark gray contours). Note that the rates are averaged over all time steps and do not represent the mean rate when P75 is occurring.

Citation: Journal of Climate 27, 16; 10.1175/JCLI-D-13-00409.1

b. Attributed precipitation

The summer and winter seasonal distributions of P75 attributed to cyclones and fronts are depicted in Figs. 7b,c,e,f. The gray contours show cyclone frequency and front intensity, respectively. Cyclones and fronts bring more P75 in winter than in summer. Furthermore, cyclone P75 follows a similar ring structure in summer and a spiral shape in winter as 〈(EP)c〉 (see again Fig. 4), with concurrent maxima. The decreasing moisture content with latitude and therefore lower precipitation per cyclone is compensated by the much larger cyclone frequency at high latitudes such that cyclone P75 keeps its magnitude. The Atlantic peak of frontal P75 is displaced farther downstream of the La Plata River basin than the peak in cyclone P75, because fronts are stronger in the eastern part of the Atlantic basin. Frontal P75 is maximum in the Indian Ocean at about 60°E, where also frontal intensity is largest. The fact that over the midlatitude south Indian Ocean the cyclones’ contribution to P75 is weak leads to a lower magnitude of the total P75 compared to the other basins (cf. Figs. 7a,d). The influence of the fronts is strongest in a band spanning as far as Tasmania and, in winter, fronts account for a major portion of strong P75 along the coast of southwestern Australia. Also the west coast of the southern island of New Zealand receives a high amount of frontal P75 in both seasons. In contrast to the Indian Ocean, in the Pacific the front intensity is lower and therefore frontal P75 is weaker. The local maximum on the upwind side of the Andes is clearly attributed to fronts. In summer mainly fronts associated with high-latitude cyclones in the Bellingshausen Sea cause orographically enhanced P75 as they hit the mountain ridge, whereas in winter the active midlatitude storm track steers fronts toward the Andes also at more northerly latitudes, which explains the northward extension of the P75 maximum. The input of freshwater to the ocean associated with frontal P75 contributes to the negative 〈EP〉 equatorward of the cyclone pathways, which was noted in the previous analysis of EP in section 4.

Our analysis reveals that the attributed frontal P75 coincides best with frontal intensity as a measure of the overall effect of fronts, better than with front frequency or length. To underpin this, the relative contribution of the most intense fronts to the total front P75 has been calculated. While the 50% most intense fronts contribute more than 90% of the front P75, the most intense 10% still accounts for about 40%. Also precipitation extremes are more often related to intense fronts (Catto and Pfahl 2013).

c. Relative contributions

To quantitatively distinguish regions where cyclones or fronts are the major cause of P75, we depict in Fig. 8 the relative contributions of the portion of P75 attributed to cyclones or fronts to the total. Around the coast of Antarctica, cyclones contribute more than 60%, while the fronts’ contribution is low, with 10%–20%. The fact that the cyclones’ relative contribution is considerably higher than the cyclone frequency itself reveals the relevance of cyclones for the formation of P75. If the cyclone frequencies and the relative contributions were close to each other, cyclones would not have a major influence. While equatorward of the high-latitude storm track the summertime contribution of cyclones is mostly below 30%, fronts contribute considerably more, mostly in a band around 40°S, with peaks in the eastern Atlantic, in the Indian Ocean southwest of Australia, and in the Pacific until 15° to the west of the Andes. During winter the relative contributions in midlatitudes differ considerably between the basins. In the Atlantic, cyclones are more relevant in the western part of the basin while the relative importance of fronts increases toward the east. Consistent with their low frequency, cyclones play a minor role in the Indian Ocean basin, whereas fronts account for more than 60% of P75. Finally, in the Pacific the importance of fronts for P75 is only slightly greater than that of cyclones. It is interesting to note that in winter (Figs. 8c,d) the peak in relative front P west of the Andes is greatly reduced. On one hand, this is a consequence of the local increase of cyclone P; on the other hand, this is also a consequence of the cyclone activity minimum in the Bellingshausen Sea, causing less cold fronts to pass the area, a feature also found in the climatology of Catto et al. (2012). The spatial distribution of the relative contribution of front P75 and the seasonal variation compare favorably with the relative contribution of cold fronts in the aforementioned study. The major difference is that the values of the relative contributions of all fronts are larger in Catto et al. (2012). We suppose this is because they did not exclude P falling in cyclones and because of their relatively large front areas.

Fig. 8.
Fig. 8.

The percentage of strong precipitation (P75) falling (a),(c) in cyclones and (b),(d) along fronts in (left) DJF and (right) JJA. Areas where P75 is below 0.01 mm day−1 are shaded gray. Also shown in (a),(c) are the 8%, 12%, 21%, and 38% contours of cyclone frequency (dark gray contours). In (b),(d) the 2, 3, 4, and 5 m s−1 (1000 km)−1 contours of frontal intensity are shown (dark gray contours).

Citation: Journal of Climate 27, 16; 10.1175/JCLI-D-13-00409.1

The relative contributions of cyclones and fronts together add up to more than 50% in DJF poleward of 40°S, with values between 60% and 80% in most areas, except for Antarctica. Equatorward the values quickly drop to 20%–30%, because of the low frequencies of occurrence of cyclones and fronts at these latitudes and the more frequent convective systems. In JJA, the relative contributions are higher with values in the range of 60%–90% almost everywhere over the SO south of 30°S. Therefore, we can conclude, based upon a detailed quantitative analysis, that cyclones and fronts are the most important weather systems contributing to P75 in the considered area.

6. Relation of freshwater fluxes to storm-track variability

a. Zonal average

In this section, we quantify how interannual variations of seasonal EP are linked to changes in the circulation of the SH extratropics and, in particular, the frequency of cyclones. The SAM is frequently used to characterize the state of the SH’s extratropical circulation and there is a variety of indices available used to represent the SAM. In the following, we use JJA mean values of the Marshall index (downloaded from http://www.nerc-bas.ac.uk/icd/gjma/sam.html) based on the SLP difference between six stations located around 40°S and six stations located around 65°S, respectively (Marshall 2003). A positive index is associated with higher pressure at midlatitudes and lower pressure at high latitudes. As pointed out by Ho et al. (2012), this index is to a lesser extent subject to artificial trends than indices derived from reanalysis data and therefore it is a good choice for the analysis of multidecadal periods. In the past the SAM was often linked to the variability of P: for instance, in the context of Australia (Meneghini et al. 2007; Ho et al. 2012), but it was noted that correlations of P (and other atmospheric quantities) and the SAM changed locally over decadal time scales (Silvestri and Vera 2009), which indicates that the same phase of the SAM can characterize a variety of different configurations of the atmospheric flow. It is evident that cyclone frequency is more directly related to attributed P75. Therefore, we compare the SAM to the measure of interannual variations of area averaged seasonal cyclone frequency fc, defined as
e4
where σ denotes the standard deviation of fc and angle brackets denote the climatological mean. Evidence indicates that the cyclones are an important driver of the SAM (Pezza et al. 2012), which is supported by Uotila et al. (2013), who found that during a positive SAM the number of cyclones is significantly decreased equatorward of 60°S and increased poleward. Indeed the index I with cyclone frequency averaged over ocean grid points south of 60°S (I60S; see Fig. 9b) strongly covaries with the SAM (r = 0.71).
Fig. 9.
Fig. 9.

(a) Time series of wintertime (JJA) values of EP, P, and total and attributed P75 averaged over ocean grid points poleward of 50°S. (b) Wintertime mean of SAM and the cyclone frequency index I60S.

Citation: Journal of Climate 27, 16; 10.1175/JCLI-D-13-00409.1

Wintertime EP averaged over ocean grid points poleward of 50°S shows interannual variations of ±0.15 mm day−1 about the mean (Fig. 9), which are highly correlated with total P (r = −0.81) and P75 (r = −0.87). Therefore, the variability of high-latitude EP can largely be ascribed to variations of P75. Jointly with the fact that a significant fraction of P75 is attributed to cyclones and fronts, these systems are the major driver of the interannual variability of EP in this region. Indeed EP is significantly correlated with P75 attributed to cyclones and fronts, with r = −0.69, and also to I60S, with r = −0.41, while correlations with the SAM are lower, with r = −0.33. Therefore, it is mainly the cyclones poleward of 60°S that affect high-latitude EP. Their influence extends also to the area between 50° and 60°S because in this region the cyclones’ fronts bring additional P75.

Variations in cyclone frequency typically do not occur simultaneously in all three ocean basins, and the variability is considerably more complex than suggested by the zonally averaged perspective. This is highlighted by the noteworthy peak of P in 1998 (Fig. 9a), which was restricted to the high-latitude Pacific, where P, P75, and P75 attributed to cyclones and fronts were by far the highest in the study period (Fig. 10a). As an illustrative case, we have a closer look at the linkage between variability of the Pacific storm track and P75 below.

Fig. 10.
Fig. 10.

(a) As in Fig. 9a, but with the average restricted to the Pacific sector poleward of 50°S. (b) Cyclone frequency indices for the PAN and PAS regions of the South Pacific.

Citation: Journal of Climate 27, 16; 10.1175/JCLI-D-13-00409.1

b. Pacific basin

The storm track in the South Pacific preferentially shows simultaneous variations of cyclone frequency of opposite sign between its northern (PAN) and its southern (PAS) branches (Fig. 10b), which are clearly associated with corresponding variations in P at low and high latitudes. The PAN and the PAS areas are framed in Fig. 3. These variations of cyclone frequency are mostly controlled by the strength and position of the subtropical jet stream (Lee and Kim 2003): A strong subtropical jet leads to the frequent development of cyclones in the PAN region and a reduced frequency of cyclones in the PAS region simultaneously with a weaker polar front jet. Tropical SST anomalies associated with El Niño and La Niña modify the Hadley circulation and exert a strong control on the strength of the subtropical jet (e.g., Turner 2004; Yuan 2004). In winter 1998 the subtropical jet was particularly weak because of the strong La Niña, while the polar front jet experienced a dramatic strengthening and an equatorward shift in the Ross Sea region (not shown), resulting in favorable conditions for the growth of cyclones there and their propagation into the Amundsen and Bellingshausen Seas. This tropical forcing caused the strongest cyclone frequency anomalies in the PAN and PAS regions in the period 1980–2010. The consequently increased P75 in the PAS region manifests itself also in an elevated zonal mean P75 and a peak in the zonal mean meridional moisture flux (Tsukernik and Lynch 2013, their Fig. 3).

To clarify where changes in the Pacific cyclone frequency affect attributed P75 and by this EP, composites of cyclone and front P75 are compiled for years with Ii > 0.5 and Ii < −0.5, where i denotes the PAS or PAN regions. For a normally distributed random variable, this corresponds to the selection of 31% of the years, giving 9–10 years in each sample for the 30-yr period, which is nearly satisfied in both regions. In Fig. 11 the difference of P75 between years with Ii > 0.5 and Ii < −0.5 is shown separately for the PAN and PAS regions. In years with many cyclones in the northern branch of the Pacific storm track, P75 is much larger north of 40°S and reduced poleward (Fig. 11a). The increase is mostly driven by cyclone P75 (Fig. 11b), while the poleward decrease is partly due to a reduction of frontal P75 and only at high latitudes because of cyclone P75 (Fig. 11c). In addition to the reduced frequency of cyclones and fronts, a smaller amount of precipitable water due to rain at more northerly latitudes contributes to this decrease (not shown). Because of the often simultaneous occurrence of a weakening of the northern and an amplification of the southern storm track, Fig. 11d is very similar to the reverse of Fig. 11a. The P75 in the PAS region is more strongly affected by cyclone variability (Fig. 11e) compared to the PAN region, where additional front P75 compensates for the reduction in cyclone P75 (Fig. 11f), especially close to the Andes, where the greater portion of P75 is caused by fronts. The PAS storm activity also influences interannual variations of P75 falling on the Antarctic Peninsula. Furthermore, in years with low cyclone activity in the PAS region, more systems decay in the Ross Sea, which contributes to the P75 peak over that region.

Fig. 11.
Fig. 11.

Difference in strong total, cyclone, and front precipitation between winters (JJA) with Ii > 0.5 and with Ii < −0.5 for the (top) PAN and (bottom) PAS regions of the South Pacific. The respective region is framed by the green box.

Citation: Journal of Climate 27, 16; 10.1175/JCLI-D-13-00409.1

We conclude that the variations in cyclone activity explain a major portion of the variability of P75 and consequently of EP. Thereby, associated changes of front frequency and corresponding changes of front P75 play a key role in shaping the total P75 anomalies. Given that E at high southern latitudes and, in particular, over the sea ice is weak, P must be balanced by moisture convergence. Thus, the strong relationship between the variability of the frequency of cyclones and high-latitude P75 requires that also meridional moisture transport covaries with cyclone frequency. It was demonstrated by Tsukernik and Lynch (2013) that transient eddies—defined by them as deviations from the monthly mean—indeed account for a major portion of the high-latitude moisture transport and its variability. The SAM and zonally averaged cyclone frequency poleward of 60°S have some skill in characterizing the interannual variability of wintertime high-latitude EP, but its spatial complexity can only be fully appreciated by nonzonally averaged indices such as high- and midlatitude cyclone frequency in the PAN and PAS areas.

7. Conclusions

Based on the objective identification of cyclones and fronts in the ERA-Interim, in this work the fundamental role played by midlatitude storms for the hydrological cycle over the SO was examined. We found that in the midlatitudes of the SH the shape of the storm track is the key factor that determines the spatial and seasonal distribution of EP. As extratropical cyclones and associated fronts are highly relevant for the formation of strong P [>75th percentile (P75)], a novel method for the objective attribution of 6-hourly P75 to cyclones and to fronts was introduced. This method allows quantifying the separate contribution of both types of synoptic systems to the total P75. It was used to generate a climatology of front and cyclone P75 and to study how interannual variations of storm-track activity impact P75. Major regions were identified where either fronts or cyclones are the primary source of P75. Shifts and variations of the intensity of the storm track in the Pacific basin were shown to be accompanied by major changes of P75. The feature-based climatological assessment of EP, E and P75 yielded the following principal findings:

  1. Extratropical storms are key to the spatial distribution, sign, and magnitude of EP over the SO. The annular structure of negative EP during austral summer and the spiraliform shape during winter are closely related to the freshwater input associated with extratropical storms. Seasonal variations of storm activity are stronger over the middle latitudes compared to the high latitudes. In winter a spiral region of strongly negative EP originates over South America, where the highly negative signal of EP by cyclones is amplified by the inflow of moist subtropical air into the basin of La Plata River, and then curves poleward and slopes around the coast of Antarctica. The magnitude of EP is comparable over middle and high latitudes, because cyclone frequency increases whereas precipitable water decreases toward higher latitudes.
  2. In the presence of an extratropical cyclone E is reduced by up to 25% at midlatitudes. In contrast, off the sea ice edge E can be enhanced under cyclone conditions.
  3. The P75 due to cyclones and fronts has a maximum in winter. Interestingly, the relative contributions to the total of P75 remain fairly constant throughout the year. While cyclones are the dominant cause of P75 around Antarctica, at midlatitudes the relative contributions show significant interbasin differences. Cyclones have a considerable share along the midlatitude storm track of the Atlantic and the Pacific, while fronts related to cyclones that move along the coast of Antarctica are the major cause of P75 over the midlatitude south Indian Ocean, where cyclone activity is weak. Front P75 is the cause of the equatorward displacement of the region where total EP is negative with respect to the main storm tracks.
  4. The interannual variability of wintertime EP at high latitudes is largely explained by variations of P75 attributed to cyclones and fronts and therefore is closely linked to the frequency of cyclones close to Antarctica.
  5. The large percentage of attributed P75 gives credibility to the robustness of our findings and attributed values should be seen as a lower limit for the influence of cyclones and fronts for P75. This is in contrast to the work by Catto et al. (2012) and Hawcroft et al. (2012), whose results, because of their large search radius, should be seen as an upper limit. In addition, our method allows us to distinguish precipitation due to cyclones and fronts.

The central importance of extratropical cyclones and associated fronts for EP and its variability over the Southern Ocean is made apparent by our findings. Nevertheless, there are some important caveats to our study, which need to be kept in mind:

  1. Despite the great improvements in the quality of reanalyses in the SH in general and in particular in the representation of the moisture cycle in ERA-Interim in terms of spatial patterns of P (e.g., Nicolas and Bromwich 2011; Bromwich et al. 2011) and closure of the moisture budget (Trenberth et al. 2011), potential biases in ERA-Interim present a limitation to our study. Thus, it is important that our quantitative results be treated with some caution. However, we are convinced that our major qualitative findings can be treated as robust.
  2. A well-known problem associated with relatively coarse resolution datasets is the underrepresentation of mesoscale cyclones (Condron et al. 2006), potentially leading to an underestimation of air–sea fluxes and cyclone P. Irving et al. (2010), using highly resolved Quick Scatterometer (QuikSCAT) data, found that mesoscale cyclones appear frequently over the ice-free SO around 60°S. As these systems have a small size compared to extratropical cyclones, it is unlikely that they would considerably alter our climatological analysis and, as was shown in section 5, the majority of strong P at around 60°S indeed can be explained by synoptic-scale systems.
  3. The area influenced by an extratropical cyclone is not strictly limited to the area within the outermost closed SLP contour surrounding the pressure minimum. This has implications for the attribution of E in cold air outbreaks, which are forced by the equatorward flow in the rear of a cyclone and often extend out of the SLP contour. Such cold air outbreaks are particularly frequent off the ice edge in the South Pacific (e.g., Bracegirdle and Kolstad 2010), where our method likely underestimates E associated with cyclones.

The linkage between the frequency of cyclones and EP implies a strong impact of interannual variations of cyclone frequency on the buoyancy flux into the ocean. An important conclusion from this study is that shifts and changes in the strength of the SO storm track potentially have a global signal in the climate system by modifying stratification and formation rates of water masses. These are crucial for the strength of the meridional overturning circulation and carbon uptake, as well as deep-water temperatures.

Acknowledgments

MeteoSwiss and the ECMWF are acknowledged for providing access to the ERA-Interim data. We thank D. Byrne, N. Gruber, M. Münnich, and I. Frenger for helpful discussions and comments. The thoughtful comments by L. Talley and two anonymous reviewers greatly helped to improve the manuscript. L. Papritz acknowledges support by ETH Research Grant CH2-01 11-1. Parts of this research were made possible by a grant from the Australian Research Council.

APPENDIX

Attribution of Precipitation to Fronts

a. Identification of coherent objects of strong precipitation

The choice of a specific threshold for the identification of P75 objects is dictated by two competitive criteria. On the one hand, the objects should contain a large part of the total P; on the other hand, they should have a modest size, such that they can be meaningfully attributed to a particular synoptic-scale system. Choosing a percentile threshold at every grid point, as opposed to a global fixed threshold, accounts for the regionally varying intensity of P and ensures that the fraction of the considered P is uniformly distributed. To take into account the seasonal cycle of P, we use monthly percentile values centered on the middle of the month and linearly interpolate them in time. With the choice of the 75th-percentile P objects have a size that makes the attribution to synoptic weather systems meaningful and at the same time ensures that between 75.8% (JJA) and 80.6% (DJF) of the total P at ocean grid points south of 35°S are accounted for (Table A1). Even about 84% of grid-scale P is captured in both seasons. Disregarded P is mostly from the convective parameterization scheme and occurs in situations that are only weakly forced by synoptic-scale weather systems.

Table A1.

Fraction of precipitation in 75th percentile over ocean grid points south of 35°S.

Table A1.

b. The choice of an optimum overlap for front precipitation

The overlap between a P75 object and a front area, required for its attribution to a front, is a tunable parameter. The attribution has been made with overlap thresholds of 5%, 10%, 15%, 20%, and 25% with respect to the size of the P75 object. Seasonal means of attributed front P75 have then been averaged over ocean grid points south of 35°S. In addition, for each overlap threshold the attribution has been applied to randomly chosen P75 objects from the set of P75 objects of the same calendar month in the period 1980–2010. Figure A1a shows the attributed front P75 against randomly attributed P75 averaged over the SO for the different overlap thresholds. During all seasons front and randomly attributed P75 decreases as the required threshold increases. For randomly attributed P75 the reduction is larger than for front P75 when the required threshold increases above 10%. In contrast, for an increase from 5% to 10%, the reduction of front and randomly attributed P75 is almost equal. Thus, the amount of front P75, which is no longer attributed when the threshold is increased from 5% to 10%, cannot be distinguished from randomly attributed P75.

Fig. A1.
Fig. A1.

Randomly attributed precipitation vs precipitation attributed to fronts (see text for details). (a) Seasonal averages over all ocean grid points south of 35°S. Different colors indicate the different overlap thresholds. (b) Wintertime average of randomly chosen grid points over the ocean south of 35°S. As a reference, the solid black line indicates where front precipitation equals randomly attributed precipitation.

Citation: Journal of Climate 27, 16; 10.1175/JCLI-D-13-00409.1

While spatially averaged front P75 clearly exceeds randomly attributed P75, this might not be true at every geographical location. In Fig. A1b wintertime front P75 against randomly attributed P75 is plotted for a set of randomly chosen grid points over the SO (roughly 1 out of 10) for thresholds of 5%, 10%, and 15%. In all three cases there is an almost linear relationship between randomly attributed and front P75: also for low P75 intensities. Hence, for a threshold of 5%, 10%, and 15% front P75 is roughly 2, 4, and 5 times as large as randomly attributed P75. The increase of the slope, and therefore the increase of the signal-to-noise ratio, is clearly largest when the threshold is increased from 5% to 10%.

Therefore, an overlap threshold of 10% is a reasonable choice, which guarantees that a maximum of front P75 is captured. A higher threshold would strongly reduce the amount of actual attributed front P75, while a lower value would not add valuable information on front P75.

REFERENCES

  • Bals-Elsholz, T. M., , E. H. Atallah, , L. F. Bosart, , T. A. Wasula, , M. J. Cempa, , and A. R. Lupo, 2001: The wintertime Southern Hemisphere split jet: Structure, variability, and evolution. J. Climate, 14, 41914215, doi:10.1175/1520-0442(2001)014<4191:TWSHSJ>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Baumgartner, A., , and E. Reichel, 1975: World Water Balance: Mean Annual Global, Continental and Maritime Precipitation, Evaporation and Runoff.Elsevier Scientific, 379 pp.

  • Berbery, E. H., , and V. R. Barros, 2002: The hydrological cycle of the La Plata basin in South America. J. Hydrometeor., 3, 630645, doi:10.1175/1525-7541(2002)003<0630:THCOTL>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Berry, G., , M. J. Reeder, , and C. Jakob, 2011: A global climatology of atmospheric fronts. Geophys. Res. Lett.,38, L04809, doi:10.1029/2010GL046451.

  • Bosilovich, M., , F. Robertson, , and J. Chen, 2011: Global energy and water budgets in MERRA. J. Climate, 24, 57215739, doi:10.1175/2011JCLI4175.1.

    • Search Google Scholar
    • Export Citation
  • Bracegirdle, T., , and E. Kolstad, 2010: Climatology and variability of Southern Hemisphere marine cold-air outbreaks. Tellus, 62A, 202208, doi:10.1111/j.1600-0870.2009.00431.x.

    • Search Google Scholar
    • Export Citation
  • Bromwich, D. H., , J. P. Nicolas, , and A. J. Monaghan, 2011: An assessment of precipitation changes over Antarctica and the Southern Ocean since 1989 in contemporary global reanalyses. J. Climate, 24, 41894209, doi:10.1175/2011JCLI4074.1.

    • Search Google Scholar
    • Export Citation
  • Catto, J. L., , and S. Pfahl, 2013: The importance of fronts for extreme precipitation. J. Geophys. Res., 118, 10 791–10 801, doi:10.1002/jgrd.50852.

    • Search Google Scholar
    • Export Citation
  • Catto, J. L., , C. Jakob, , G. Berry, , and N. Nicholls, 2012: Relating global precipitation to atmospheric fronts. Geophys. Res. Lett.,39, L10805, doi:10.1029/2012GL051736.

  • Cerovečki, I., , L. D. Talley, , and M. R. Mazloff, 2011: A comparison of Southern Ocean air–sea buoyancy flux from an ocean state estimate with five other products. J. Climate, 24, 62836306, doi:10.1175/2011JCLI3858.1.

    • Search Google Scholar
    • Export Citation
  • Cerovečki, I., , L. D. Talley, , M. R. Mazloff, , and G. Maze, 2013: Subantarctic Mode Water formation, destruction, and export in the eddy-permitting Southern Ocean State Estimate. J. Phys. Oceanogr., 43, 14851511, doi:10.1175/JPO-D-12-0121.1.

    • Search Google Scholar
    • Export Citation
  • Charnock, H., 1955: Wind stress on a water surface. Quart. J. Roy. Meteor. Soc., 81, 639640, doi:10.1002/qj.49708135027.

  • Condron, A., , G. R. Bigg, , and I. A. Renfrew, 2006: Polar mesoscale cyclones in the northeast Atlantic: Comparing climatologies from ERA-40 and satellite imagery. Mon. Wea. Rev., 134, 15181533, doi:10.1175/MWR3136.1.

    • Search Google Scholar
    • Export Citation
  • Connolley, W. M., 1997: Variability in annual mean circulation in southern high latitudes. Climate Dyn., 13, 745756, doi:10.1007/s003820050195.

    • Search Google Scholar
    • Export Citation
  • Dee, D., and et al. , 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553597, doi:10.1002/qj.828.

    • Search Google Scholar
    • Export Citation
  • Fairall, C., , E. Bradley, , J. Hare, , A. Grachev, , and J. Edson, 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
  • Gruber, N., and et al. , 2009: Oceanic sources, sinks, and transport of atmospheric CO2. Global Biogeochem. Cycles, 23, GB1005, doi:10.1029/2008GB003349.

    • Search Google Scholar
    • Export Citation
  • Hawcroft, M. K., , L. C. Shaffrey, , K. I. Hodges, , and H. F. Dacre, 2012: How much Northern Hemisphere precipitation is associated with extratropical cyclones? Geophys. Res. Lett.,39, L24809, doi:10.1029/2012GL053866.

  • Ho, M., , A. S. Kiem, , and D. C. Verdon-Kidd, 2012: The southern annular mode: A comparison of indices. Hydrol. Earth Syst. Sci., 16, 967982, doi:10.5194/hess-16-967-2012.

    • Search Google Scholar
    • Export Citation
  • Hoskins, B. J., , and K. I. Hodges, 2005: A new perspective on Southern Hemisphere storm tracks. J. Climate, 18, 41084129, doi:10.1175/JCLI3570.1.

    • Search Google Scholar
    • Export Citation
  • Irving, D., , I. Simmonds, , and K. Keay, 2010: Mesoscale cyclone activity over the ice-free Southern Ocean: 1999–2008. J. Climate, 23, 54045420, doi:10.1175/2010JCLI3628.1.

    • Search Google Scholar
    • Export Citation
  • Keeling, R. F., , and M. Visbeck, 2011: On the linkage between Antarctic surface water stratification and global deep-water temperature. J. Climate, 24, 35453557, doi:10.1175/2011JCLI3642.1.

    • Search Google Scholar
    • Export Citation
  • Koch, P., , H. Wernli, , and H. C. Davies, 2006: An event-based jetstream climatology and typology. Int. J. Climatol., 26, 283301, doi:10.1002/joc.1255.

    • Search Google Scholar
    • Export Citation
  • Krinner, G., , O. Magand, , I. Simmonds, , C. Genthon, , and J. L. Dufresne, 2007: Simulated Antarctic precipitation and surface mass balance at the end of the twentieth and twenty-first centuries. Climate Dyn., 28, 215230, doi:10.1007/s00382-006-0177-x.

    • Search Google Scholar
    • Export Citation
  • Lee, S., , and H.-K. Kim, 2003: The dynamical relationship between subtropical and eddy-driven jets. J. Atmos. Sci., 60, 14901503, doi:10.1175/1520-0469(2003)060<1490:TDRBSA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Li, Z. X., , and H. Le Treut, 1999: Transient behavior of the meridional moisture transport across South America and its relation to atmospheric circulation patterns. Geophys. Res. Lett., 26, 14091412, doi:10.1029/1999GL900274.

    • Search Google Scholar
    • Export Citation
  • Lorenz, C., , and H. Kunstmann, 2012: The hydrological cycle in three state-of-the-art reanalyses: Intercomparison and performance analysis. J. Hydrometeor., 13, 13971420, doi:10.1175/JHM-D-11-088.1.

    • Search Google Scholar
    • Export Citation
  • Lovenduski, N., , and T. Ito, 2009: The future evolution of the Southern Ocean CO2 sink. J. Mar. Res., 67, 597617, doi:10.1357/002224009791218832.

    • Search Google Scholar
    • Export Citation
  • Marshall, G. J., 2003: Trends in the southern annular mode from observations and reanalyses. J. Climate, 16, 41344143, doi:10.1175/1520-0442(2003)016<4134:TITSAM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Marsland, S., , and J. Wolff, 2001: On the sensitivity of Southern Ocean sea ice to the surface freshwater flux: A model study. J. Geophys. Res., 106, 27232741, doi:10.1029/2000JC900086.

    • Search Google Scholar
    • Export Citation
  • Meneghini, B., , I. Simmonds, , and I. N. Smith, 2007: Association between Australian rainfall and the southern annular mode. Int. J. Climatol., 27, 109121, doi:10.1002/joc.1370.

    • Search Google Scholar
    • Export Citation
  • Neu, U., and et al. , 2013: IMILAST: A community effort to intercompare extratropical cyclone detection and tracking algorithms. Bull. Amer. Meteor. Soc., 94, 529547, doi:10.1175/BAMS-D-11-00154.1.

    • Search Google Scholar
    • Export Citation
  • Nicolas, J. P., , and D. H. Bromwich, 2011: Precipitation changes in high southern latitudes from global reanalyses: A cautionary tale. Surv. Geophys., 32, 475494, doi:10.1007/s10712-011-9114-6.

    • Search Google Scholar
    • Export Citation
  • Noone, D., , and I. Simmonds, 2002: Annular variations in moisture transport mechanisms and the abundance of δ18O in Antarctic snow. J. Geophys. Res., 107, 4742, doi:10.1029/2002JD002262.

    • Search Google Scholar
    • Export Citation
  • Persson, P. O. G., , J. E. Hare, , C. W. Fairall, , and W. D. Otto, 2005: Air–sea interaction processes in warm and cold sectors of extratropical cyclonic storms observed during FASTEX. Quart. J. Roy. Meteor. Soc., 131, 877912, doi:10.1256/qj.03.181.

    • Search Google Scholar
    • Export Citation
  • Pezza, A. B., , H. A. Rashid, , and I. Simmonds, 2012: Climate links and recent extremes in Antarctic sea ice, high-latitude cyclones, southern annular mode and ENSO. Climate Dyn., 38, 5773, doi:10.1007/s00382-011-1044-y.

    • Search Google Scholar
    • Export Citation
  • Pfahl, S., , and H. Wernli, 2012: Quantifying the relevance of cyclones for precipitation extremes. J. Climate, 25, 67706780, doi:10.1175/JCLI-D-11-00705.1.

    • Search Google Scholar
    • Export Citation
  • Rudeva, I., , and S. K. Gulev, 2011: Composite analysis of North Atlantic extratropical cyclones in NCEP–NCAR reanalysis data. Mon. Wea. Rev., 139, 14191446, doi:10.1175/2010MWR3294.1.

    • Search Google Scholar
    • Export Citation
  • Schanze, J. J., , R. W. Schmitt, , and L. L. Yu, 2010: The global oceanic freshwater cycle: A state-of-the-art quantification. J. Mar. Res., 68, 569595, doi:10.1357/002224010794657164.

    • Search Google Scholar
    • Export Citation
  • Silvestri, G., , and C. Vera, 2009: Nonstationary impacts of the southern annular mode on Southern Hemisphere climate. J. Climate, 22, 61426148, doi:10.1175/2009JCLI3036.1.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., 2003: Modes of atmospheric variability over the Southern Ocean. J. Geophys. Res., 108, 8078, doi:10.1029/2000JC000542.

  • Simmonds, I., , and M. Dix, 1989: The use of mean atmospheric parameters in the calculation of modeled mean surface heat fluxes over the world’s oceans. J. Phys. Oceanogr., 19, 205215, doi:10.1175/1520-0485(1989)019<0205:TUOMAP>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., , and J. C. King, 2004: Global and hemispheric climate variations affecting the Southern Ocean. Antarct. Sci., 16, 401413, doi:10.1017/S0954102004002226.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., , K. Keay, , and E. Lim, 2003: Synoptic activity in the seas around Antarctica. Mon. Wea. Rev., 131, 272288, doi:10.1175/1520-0493(2003)131<0272:SAITSA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., , A. Rafter, , T. Cowan, , A. Watkins, , and K. Keay, 2005: Large-scale vertical momentum, kinetic energy and moisture fluxes in the Antarctic sea-ice region. Bound.-Layer Meteor., 117, 149177, doi:10.1007/s10546-004-5939-6.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., , K. Keay, , and J. A. T. Bye, 2012: Identification and climatology of Southern Hemisphere mobile fronts in a modern reanalysis. J. Climate, 25, 19451962, doi:10.1175/JCLI-D-11-00100.1.

    • Search Google Scholar
    • Export Citation
  • Sodemann, H., , and A. Stohl, 2009: Asymmetries in the moisture origin of Antarctic precipitation. Geophys. Res. Lett.,36, L22803, doi:10.1029/2009GL040242.

  • Talley, L. D., 2008: Freshwater transport estimates and the global overturning circulation: Shallow, deep and throughflow components. Prog. Oceanogr., 78, 257303, doi:10.1016/j.pocean.2008.05.001.

    • Search Google Scholar
    • Export Citation
  • Tietäväinen, H., , and T. Vihma, 2008: Atmospheric moisture budget over Antarctica and the Southern Ocean based on the ERA-40 reanalysis. Int. J. Climatol., 28, 19771995, doi:10.1002/joc.1684.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., , and C. J. Guillemot, 1998: Evaluation of the atmospheric moisture and hydrological cycle in the NCEP/NCAR reanalyses. Climate Dyn., 14, 213231, doi:10.1007/s003820050219.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., , L. Smith, , T. Qian, , A. Dai, , and J. Fasullo, 2007: Estimates of the global water budget and its annual cycle using observational and model data. J. Hydrometeor., 8, 758769, doi:10.1175/JHM600.1.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., , J. T. Fasullo, , and J. Mackaro, 2011: Atmospheric moisture transports from ocean to land and global energy flows in reanalyses. J. Climate, 24, 49074924, doi:10.1175/2011JCLI4171.1.

    • Search Google Scholar
    • Export Citation
  • Tsukernik, M., , and A. H. Lynch, 2013: Atmospheric meridional moisture flux over the Southern Ocean: A story of the Amundsen Sea. J. Climate, 26, 80558064, doi:10.1175/JCLI-D-12-00381.1.

    • Search Google Scholar
    • Export Citation
  • Turner, J., 2004: The El Niño–Southern Oscillation and Antarctica. Int. J. Climatol., 24, 131, doi:10.1002/joc.965.

  • Turner, J., , T. Lachlan-Cope, , J. Thomas, , and S. Colwell, 1995: The synoptic origins of precipitation over the Antarctic Peninsula. Antarct. Sci., 7, 327337, doi:10.1017/S0954102095000447.

    • Search Google Scholar
    • Export Citation
  • Uotila, P., , T. Vihma, , A. B. Pezza, , I. Simmonds, , K. Keay, , and A. H. Lynch, 2011: Relationships between Antarctic cyclones and surface conditions as derived from high-resolution numerical weather prediction data. J. Geophys. Res.,116, D07109, doi:10.1029/2010JD015358.

  • Uotila, P., , T. Vihma, , and M. Tsukernik, 2013: Close interactions between the Antarctic cyclone budget and large-scale atmospheric circulation. Geophys. Res. Lett., 40, 32373241, doi:10.1002/grl.50560.

    • Search Google Scholar
    • Export Citation
  • Vincent, D., 1994: The South Pacific convergence zone (SPCZ): A review. Mon. Wea. Rev., 122, 19491970, doi:10.1175/1520-0493(1994)122<1949:TSPCZA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Wernli, H., , and C. Schwierz, 2006: Surface cyclones in the ERA-40 dataset (1958–2001). Part I: Novel identification method and global climatology. J. Atmos. Sci., 63, 24862507, doi:10.1175/JAS3766.1.

    • Search Google Scholar
    • Export Citation
  • Yu, L., 2011: A global relationship between the ocean water cycle and near-surface salinity. J. Geophys. Res.,116, C10025, doi:10.1029/2010JC006937.

  • Yuan, X., 2004: ENSO-related impacts on Antarctic sea ice: Synthesis of phenomenon and mechanisms. Antarct. Sci., 16, 415425, doi:10.1017/S0954102004002238.

    • Search Google Scholar
    • Export Citation
  • Yuan, X., , J. Patoux, , and C. Li, 2009: Satellite-based midlatitude cyclone statistics over the Southern Ocean: 2. Tracks and surface fluxes. J. Geophys. Res.,114, D04106, doi:10.1029/2008JD010874.

Save