Observational evidence of an eddy overturning circulation at the Antarctic margins

Transcription

1 Observational evidence of an eddy overturning circulation at the Antarctic margins Andrew F. Thompson Caltech Karen J. Heywood U. East Anglia Sunke Schmitdtko GEOMAR Last revised: 19 March, 2013 For consideration in Nature Geoscience 1

2 The exchange of water masses across the Antarctic continental shelf break regulates the transport of warm Circumpolar Deep Water (CDW) towards ice shelves and glacial grounding lines, as well as the export of dense shelf waters. 1 The penetration of CDW past the shelf break has been implicated in the dramatic loss of ice shelf mass over much of west Antarctica. 2 4 The Antarctic shelf break also hosts an energetic mesoscale eddy field in high-resolution, regional circulation models, 5, 6 but observations that capture this spatial and temporal variability have been limited. Here we present hydrographic data collected in the northwestern Weddell Sea by ocean gliders, which captures the complex marginal circulation. The mean structure of the stratification in the upper 1000 m, as well as the variability in the frontal currents, support the hypothesis that an eddy-induced transport is a primary contributor to the on-shelf flux of heat and mass. The identification of an eddy overturning suggests an enhanced dependence of cross-slope transport on wind and buoyancy forcing at the shelf break. Lateral eddy fluxes will also ventilate a wide range of density classes at the Antarctic margins, a process not directly resolved and difficult to parameterize in general circulation models. 7 Residual-mean theories, which propose a leading order balance between wind- and eddyinduced overturning cells, have been key in illuminating the dynamics of material exchange across the Southern Ocean s Antarctic Circumpolar Current (ACC). 8 The Antarctic margins (AM) share similarities with the ACC: the circulation is composed of strong, narrow frontal currents; the flow is unblocked by continental boundaries (although the AM span a greater range of latitudes); and surface forcing is largely due to zonal, down-front winds (easterlies as opposed to midlatitude westerlies). Evidence of the generation and persistence of eddy variability in both idealized and realistic 2

3 numerical models, 5, 6, 9 11 as well as mooring data, 12 has led to the proposal that a combination of wind-driven and eddy overturning cells may be active at the AM. 10, 11, 13 Confirming this picture requires observations of both variations in the along-stream velocity and cross-stream buoyancy fields. We present a novel hydrographic data set that achieves this coverage, and supports the presence of an eddy overturning at intermediate depths at the shelf break. Figure 1a shows a map of the northwestern Weddell Sea, including the path of three ocean gliders deployed between January 20, 2012 and March 13, This region is a key gateway for the formation and delivery of Antarctic Bottom Water (AABW) to the Scotia Sea, and eventually to the global circulation. 14 Furthermore, the export of iron-rich shelf waters gives rise to elevated chlorophyll levels in the Scotia Sea, 15, 16 distinguishing this region from most of the Southern Ocean where nutrients are replete, but chlorophyll levels are low. A fundamental understanding of crossslope exchange processes active in this region of the Weddell Sea is critical for both the global overturning circulation and Southern Ocean biogeochemical and ecosystem dynamics. Properties of AABW depend critically on modification near the Antarctic coast due to entrainment or mixing with Circumpolar Deep Water (CDW, or modified CDW). 17 These interactions are enhanced at the shelf break, where MCDW penetrates on-shore, establishing strong lateral water mass gradients. Figure 1b shows temperature along section A in panel (a). MCDW intrudes on the shelf, separating cold Winter Water (WW) above and fresh Weddell Sea Deep Water (WSDW) below; other sections are similar. The injection of warm MCDW occurs via intermittent pulses, consistent with an energetic mesoscale field. This suggests that the modification of dense water, typically attributed to turbulent entrainment during descent along the continental slope, may also be influenced by lateral stirring with interior water masses. While the relative importance of these 3

4 two remains to be quantified, the temperature/salinity diagram in panels (c) and (d) of Figure 1 show that the space enclosed by the three water mass end members is populated by many observations, indicative of strong mixing both along and potentially across density surfaces. Lateral stirring is likely to expose a broad range of density classes to mixing, which will influence the closure of the meridional overturning circulation s lower cell at the southern boundary. 18, 19 The high resolutions sections also provide a striking view of the cross-slope structure of the Antarctic Slope Front (ASF) system. 20 Traditional descriptions of the ASF identify a single, persistent shelf-break current, 1 whereas comparison of multiple sections reveals a more complex picture (supplementary Figure 1). A narrow, bottom-intensified current is nearly always found at, or slightly off-shore, of the shelf break, but as many as three distinct velocity cores may span a distance as short as 50 km. Nearly all of these small-scale fronts have a baroclinic signature and are not simply a signature of a variable barotropic transport. This front or jet separation remains larger than the first baroclinic Rossby deformation radius, which we estimate as λ = NH/f 10 km, where N is the buoyancy frequency, H is the water column depth and f is the Coriolis frequency. In a turbulent flow, potential vorticity gradients, due to a change in water column depth for instance, can give rise to a banded flow, or jets. A standard scaling for the separation of these jets depends on an eddy velocity U e and the background potential vorticity gradient, in this case a topographic beta β T, yielding the Rhines scale l R U e /β T. Using the root mean square value of velocities over the continental slope, U e 0.1 m s 1 and the scaling β T fh 1 H/ y, we arrive at l R 20 km, consistent with the observed spacing of the velocity cores. Across multiple sections, the frontal currents are found at different water column depths, suggesting either a sinuous frontal structure or the migration of the fronts due to tidal variability and topographic waves. 4

5 While both hydrographic properties and the velocity structure support an energetic mesoscale field, we present maps of potential vorticity (PV, defined in Methods) to diagnose the structure of eddy mass transport. PV is a conservative tracer along density surfaces, if diabatic processes are small, and mesoscale stirring is assumed to relax PV gradients. Similar to the ACC 8 or the atmospheric circulation, 21 a down-gradient eddy flux of potential vorticity by eddies in a circumpolar flow acts to relax density layer thickness gradients. In other words, eddies flux mass down thickness gradients via a residual velocity v res. = v h /h, where h the thickness of a density layer and ( ) implies a time and along-stream average. Figure 2 provides an example of the PV distribution found along one of the glider sections; these are similar across all sections. In all cases, PV is enhanced at the shelf break, consistent with the presence of the classic V -shaped structure of the ASF. 1 Over the continental slope, PV develops layers originating from the slope itself, and is consistent with PV anomalies formed in frictional boundary layers associated with the frontal currents. 22 Mapping the PV distribution into a density, as opposed to depth, coordinate reveals that the northwestern Weddell Sea supports substantial along-isopycnal PV (or thickness) gradients, in some cases exceeding an order of magnitude change in PV over 50 km. Density surfaces shoal moving towards the shelf break, and they also thin. Panel (c) shows the thickness of three density classes, superimposed on the background velocity profile. In each case, all three layers show a large-scale thinning towards the shelf break. Interestingly, in the lower-most layer, the offshore edge of the bottom-intensified ASF is associated with a positive thickness anomaly. This is a common feature in the different sections that may arise from the injection of fluid where the bottom Ekman transport is convergent (see Figure 3b). Furthermore, this thickness anomaly would support an enhancement of the eddy overturning across the core of the ASF to close the overturning and maintain the condition of no net (vertically-integrated) 5

6 cross-slope flow across a given depth contour. 13 Following residual-mean theories Figure 2 suggests there is a strong on-shore eddy flux of CDW outside of top and bottom Ekman layers. Figure 3 summarizes the PV distributions from all the sections. The sections are normalized by distance from the shelf break and the abscissa is given in logarithmic units of potential vorticity. Each section is displaced to display trends clearly; the color indicates the water column depth. The values plotted are the mean PV over the density class < γ n < Again, it is clear that a sustained onshore PV gradient is found all along all sections crossing the shelf-break in the northwestern Weddell Sea. Taking an along-slope average, this conforms with the view of an on-shore eddy transport at intermediate depths. Finally, the gray blocks are indicative of locations where there are strong, bottom-intensified velocities. In sections C, D, F, G, H and I these frontal positions are associated with a break in the PV gradient, associated with the thickness gradient anomaly discussed in Figure 2c. Figure 3b summarizes the key elements of the Antarctic meridional overturning circulation implied by the results in panel (a). The shelf break is typically found at latitudes where the wind stress produces an on-shore Ekman transport and a sea surface height gradient consistent with a westward geostrophic flow, namely the ASC. 1, 20 The continental slope focuses the ASC into narrow fronts due to a preference for flow along contours of f/h, or conservation of PV. In regions where dense water is found over the continental shelf, the ASF is bottom intensified. This produces a substantial off-shore (down-slope) frictional Ekman transport that is proportional to the current speed above the Ekman layer. 23 Thus, lateral shear in the ASF produces regions of Ekman convergence (off-shore) and divergence (on-shore). In the interior, the flow is in geostrophic balance, but if along-slope pressure gradients are weak, or zero in the case of a circumpolar flow, 6

7 then a mean cross-slope flow can not be supported. In this case, the mean overturning is similar to the Deacon cell of the ACC, 8 where vertical velocities appear to connect Ekman layers at the surface and seafloor. Vertical velocities of this magnitude are inconsistent with observations in this region. Then, a plausible explanation is that the overturning is closed in the interior by an eddy transport, related to barotropic or baroclinic instabilities of the ASF, directed along density surfaces and mediated by correlations in cross-slope velocity and density layer thickness anomalies. Direct resolution of these dynamics requires at least a sub-5 km grid spacing in numerical models. Parameterization of eddy fluxes in this region is challenging due to the sloped topography. Application of the Gent McWilliam diffusive parameterization 24 encourages the relaxation of tilted density surfaces towards a horizontal orientation. However, the extraction of potential energy from the background stratification is minimized not when the density surfaces are flat, but rather when they are parallel to the bottom bathymetry. 25 A PV-based parameterization is more likely to provide an accurate representation of the eddy fluxes in this region. Characteristics of the circulation in the northwestern Weddell Sea will likely apply along most regions of the AM where the primary mechanical forcing comes from surface winds and the shelf break is dominated by the ASF current. 23 The identification of an eddy overturning departs from traditional descriptions of cross-shelf transport involving friction or along-stream pressure gradients. 17, 26 Furthermore, in the presence of an active eddy field, observational evidence of changes in AABW properties 27 may not stem solely from AABW production changes, but rather modification to mixing properties during export may also play a key role. Earlier work 11 has shown that in a diffusive representation of eddy thickness fluxes, the eddy diffusivity is sensitive to the steepness of the continental slope. Local bathymetry, e.g. troughs and ridges, will also locally 7

8 enhance eddy generation. Thus, there may be significant along-stream variability as compared to sections in the northwestern Weddell Sea. Nonetheless, the observational evidence in convincing that variability at meso- and submesoscales are critical for understanding cross-shelf transport and ultimately the stability of Antarctica s ice sheets. How these mesoscale processes respond to longterm changes in surface wind and buoyancy forcing, and their potential impact on large-scale ocean circulation and climate, 28 remains an open questions. 8

9 Methods Three ocean gliders were deployed in the Weddell Sea between 20 January, 2013 and 13 March, In total the gliders completed over 750 dives, collecting vertical profiles of temperature and salinity, dissolved oxygen, optical backscatter and fluorescence; here we focus on the hydrographic data. Typical glider station spacing was a few kilometers, improving upon standard ship-based surveys by an order of magnitude or more. The gliders completed a V-shaped dive approximately every four hours (depending on water column depth), making measurements roughly every 5 seconds, or every 50 cm. The potential vorticity sections shown in this study are based on optimallyinterpolated sections with a vertical resolution of 5 m and a horizontal resolution of 2 km. The geostrophic shear is obtained by first filtering the data using a 9-point boxcar filter. The barotropic component of the flow is included by using the depth-averaged velocity, as measured from the glider surfacing positions. These are shown to produce good agreement with surface drifter velocities obtained during the same cruise. 29 The Ertel potential vorticity (PV) is defined by Q full = ( ) fˆk + u b, (1) where u and b are relative vorticity and the buoyancy, b = gρ/ρ 0. Along a single section, Q full is not attainable, but PV can be calculated to the extent that Q = fb z + u z b y u y b z, where subscripts represent partial derivatives (z increases upward and y increases shoreward). In the limit that the Rossby number is small, Q fb z, i.e. the stretching component of the PV is dominant. From observations, we find that along the flanks of the strongest shelf break jets, the lateral shear can be sufficiently large to make a contribution to Q, however, over most of the domain the assumption that Q is controlled by density layer fluctuations is valid. 9

10 References 1 A.E. Gill. Circulation and bottom water production in the weddell sea. Deep Sea Research and Oceanographic Abstracts, 20(2): , Andrew Shepherd, Duncan Wingham, and Eric Rignot. Warm ocean is eroding west antarctic ice sheet. Geophysical Research Letters, 31(23):n/a n/a, 4. 3 H. D. Pritchard, S. R. M. Ligtenberg, H. A. Fricker, D. G. Vaughan, M. R. van den Broeke, and L. Padman. Antarctic ice-sheet loss driven by basal melting of ice shelves. Nature, 484(7395): , E. Rignot, S. Jacobs, J. Mouginot, and B. Scheuchl. Ice-shelf melting around antarctica. Science, 341(6143): , Malte Thoma, Adrian Jenkins, David Holland, and Stan Jacobs. Modelling circumpolar deep water intrusions on the amundsen sea continental shelf, antarctica. Geophysical Research Letters, 35(18):n/a n/a, 8. 6 Michael P. Schodlok, Dimitris Menemenlis, Eric Rignot, and Michael Studinger. Sensitivity of the ice-shelf/ocean system to the sub-ice-shelf cavity shape measured by nasa icebridge in pine island glacier, west antarctica. Annals of Glaciology, 53(60): , T00:00:00. 7 Raffaele Ferrari, James C. McWilliams, Vittorio M. Canuto, and Mikhail Dubovikov. Parameterization of eddy fluxes near oceanic boundaries. Journal of Climate, 21(12): , 2014/03/ John Marshall and Kevin Speer. Closure of the meridional overturning circulation through southern ocean upwelling. Nature Geosci, 5(3): ,

11 9 M. S. Dinniman, J. M. Klinck, and W. O. Smith Jr. A model study of Circumpolar Deep Water on the West Antar. Deep Sea Research Part II: Topical Studies in Oceanography, 58: , O. A. Nøst, M. Biuw, V. Tverberg, C. Lydersen, T. Hattermann, Q. Zhou, L. H. Smedsrud, and K. M. Kovacs. Eddy overturning of the antarctic slope front controls glacial melting in the eastern weddell sea. Journal of Geophysical Research: Oceans, 116(C11):n/a n/a, A. L. Stewart and A. F. Thompson. Connecting Antarctic cross-slope exchange with Southern Ocean overturning. J. Phys. Oceanogr., 43: , Arnold L. Gordon, Bruce Huber, Darren McKee, and Martin Visbeck. A seasonal cycle in the export of bottom water from the weddell sea. Nature Geosci, 3(8): , Michael A. Spall. Dense water formation around islands. Journal of Geophysical Research: Oceans, 118(5): , A.H. Orsi, G.C. Johnson, and J.L. Bullister. Circulation, mixing, and production of antarctic bottom water. Progress in Oceanography, 43(1):55 109, Rebecca E. Korb and Mick Whitehouse. Contrasting primary production regimes around south georgia, southern ocean: large blooms versus high nutrient, low chlorophyll waters. Deep Sea Research Part I: Oceanographic Research Papers, 51(5): , Alessandro Tagliabue, Laurent Bopp, and Olivier Aumont. Evaluating the importance of atmospheric and sedimentary iron sources to southern ocean biogeochemistry. Geophysical Research Letters, 36(13):n/a n/a, 9. 11

12 17 Peter G. Baines and Scott Condie. Observations and modelling of Antarctic downslope flows: a review, volume 75 of Antarctic Research Series, pages AGU, ocean, ice and atmosphere: interactions at the antarctic continental margin edition, Kevin Speer, Stephen R. Rintoul, and Bernadette Sloyan. The diabatic deacon cell*. Journal of Physical Oceanography, 30(12): , 2014/03/ Loïc Jullion, Alberto C. Naveira Garabato, Michael P. Meredith, Paul R. Holland, Peggy Courtois, and Brian A. King. Decadal freshening of the antarctic bottom water exported from the weddell sea. Journal of Climate, 26(20): , 2014/03/ S. S. Jacobs. On the nature and significance of the antarctic slope front. Mar. Chem., 35:9 24, R. Alan Plumb. Three-dimensional propagation of transient quasi-geostrophic eddies and its relationship with the eddy forcing of the time mean flow. Journal of the Atmospheric Sciences, 43(16): , 2014/03/ Michael A. Spall, Robert S. Pickart, Paula S. Fratantoni, and Albert J. Plueddemann. Western arctic shelfbreak eddies: Formation and transport. Journal of Physical Oceanography, 38(8): , 2014/03/ A. K. Wåhlin, X. Yuan, G. Björk, and C. Nohr. Inflow of warm circumpolar deep water in the central amundsen shelf*. Journal of Physical Oceanography, 40(6): , 2014/03/ Peter R. Gent and James C. Mcwilliams. Isopycnal mixing in ocean circulation models. Journal of Physical Oceanography, 20(1): , 2014/03/

13 25 Pål Erik Isachsen. Baroclinic instability and eddy tracer transport across sloping bottom topography: How well does a modified eady model do in primitive equation simulations? Ocean Modelling, 39(1 2): , Modelling and Understanding the Ocean Mesoscale and Submesoscale. 26 Hsien-Wang Ou. Watermass properties of the antarctic slope front: A simple model*. Journal of Physical Oceanography, 37(1):50 59, 2014/03/ Sarah G. Purkey and Gregory C. Johnson. Global contraction of antarctic bottom water between the 1980s and 0s*. Journal of Climate, 25(17): , 2014/03/ M. D. Miller, J. F. Adkins, D. Menemenlis, and M. P. Schodlok. The role of ocean cooling in setting glacial southern source bottom water salinity. Paleoceanography, 27(3):n/a n/a, Andrew F. Thompson and Madeleine K. Youngs. Surface exchange between the weddell and scotia seas. Geophysical Research Letters, 40(22): ,

14 (a) Drake Passage Depth (m) Section A 52.6W 52.4W 52.2W 52.0W 51.8W Longitude Salinity Potential Temperature ( o C) H I F G J E A D C B Powell Basin Yearday Joinville Ridge (b) (c) (d) Salinity Figure 1: (a) Map of the northwestern Weddell Sea showing the position of hydrographic profiles form three ocean gliders. The background color shows depth (m); the symbol color provides an indication of the temporal evolution of the glider positions. (b) Temperature distribution along section A; water masses include Winter Water (WW), modified Circumpolar Deep Water (MCDW) and Weddell Sea Bottom Water (WSBW). The dashed line shows the location of the glider observations. (c) Temperature, salinity diagram for section A. (d) Expanded view of the boxed region in panel (c). Add WW, MCDW, WSBW to panels (b and c). Add neutral density curves to panel (d). 14

15 0 8.5 Depth (m) Section A (a) W 52.4W 52.2W 52.0W 51.8W Potential Density (b) 52.6W 52.4W 52.2W 52.0W 51.8W Depth (m) Thickness (m) (c) (d) W 52.4W 52.2W 52.0W 51.8W Longitude h 1 h 2 h 3 Figure 2: (a) Potential vorticity (PV, logarithmic axis, e.g. 9 = 10 9 s 3 ) along section A form Figure 1a; PV is defined in Methods. The solid black line indicates the continental shelf/slope. (b) PV in panel (a) mapped into neutral density coordinates. (c) The bold curves show neutral density contours along section A, which the thin contours are the along-slope geostrophic velocities. Contours are every 4 cm s 1 with black (cyan) indicating cyclonic (anticyclonic) flow. (d) Thickness of the neutral density classes (described in the legend, kg m 3 ) from panel (c). 15

16 (a) Section I Section J (b) On-shore Eddy transport X X X X Surface Ekman transport Cyclonic boundary current Potential vorticity units (s 3, log scale) Section H Section G Section F Section D Section C Continental Shelf Divergent Ekman transport 0.5 Section A 0 PV gradient / Thickness flux Distance from shelf break (km) Figure 3: (a) Summary of the PV structure for all sections. The cross-slope position is normalized to the shelf break and sections are offset vertically for clarity. Color indicates the water column depth and gray blocks indicate the position of fronts with a bottom velocity greater than 0.15 m s 1. The black arrow indicates the direction of the PV gradient and the subsequent direction of the eddy max flux. (b) Schematic diagram of key components of the Antarctic marginal meridional overturning. See text for description. 16

17 W 52.4W 52.2W 52.0W 51.8W S 62.9S 62.8S 62.7S W 53.8W 53.6W 53.4W 53.2W 53.0W Figure 4: Supplementary Figure caption to be updated. Referenced velocity sections along three of the glider transects. Panels (a) and (c) are zonal sections, while (b) is a meridional section; in each case, positive values indicate cyclonic flow. Color gives the cross-section velocity in m s 1. The black ticks indicate the position of the glider surfacings. 17

18 8.5 Potential density W 52.4W 52.2W 52W 51.8W 54W 53.8W 53.6W 53.4W W 53.8W 53.6W Depth (m) Thickness (m) W 52.4W 52.2W 52W 51.8W Longitude W 53.8W 53.6W 53.4W Longitude W 53.8W 53.6W Longitude Figure 5: Supplementary Figure caption to be updated. Summary of potential vorticity structure along sections (a) A, (b) H and (c) I. Upper panels: distribution of PV, Q = fb z, along density surfaces; the axis is logarithmic. White regions indicate density outcropping at the seafloor. The white contour shows the Q = s 3 for reference. Middle panels: Thin contours show the cross-section velocity structure; contour intervals are??? m s 1. Thick contours show the thickness of the following density classes:. Lower panels: Thickness in meters of the density classes shown in the middle panels. 18