EGU2020-5644
https://doi.org/10.5194/egusphere-egu2020-5644
EGU General Assembly 2020
© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

From small whirls to the global ocean: how eddies affect the Atlantic Meridional Overturning Circulation

Caroline Katsman1, Nils Brüggemann2, Sotiria Georgiou1, Juan-Manuel Sayol Espana1, Stefanie Ypma1, Carine van der Boog1, and Julie Pietrzak1
Caroline Katsman et al.
  • 1Delft University of Technology, Hydraulic Engineering, Delft, Netherlands (c.a.katsman@tudelft.nl)
  • 2University of Hamburg, Institute of Oceanography, Hamburg, Germany

In the North Atlantic Ocean, intense downward motions connect the upper and lower limbs of the Atlantic Meridional Overturning Circulation (AMOC). In addition, the AMOC also displays a pronounced signature in density space, with lighter waters moving northward and denser waters returning southward.

While at first glance it is appealing to associate this sinking of water masses in the North Atlantic Ocean with the occurrence of the formation of dense water masses by deep convection, this is not correct: the net vertical motion over convection areas is small. The downward flow required to connect the upper and lower branches of the AMOC thus has to occur outside the deep convection areas. Indeed, earlier studies have pointed out theoretically that strong sinking can only occur close to continental boundaries, where ageostrophic processes play a role. However, observations clearly indicate that convected water masses formed in marginals seas constitute an important component of the lower limb of the AMOC.

This apparent contradiction is explored in this presentation, by studying the overturning in the AMOC from a perspective in depth space (Eulerian downwelling) and density space (downwelling across isopycnals). Based on analyses of both a high-resolution global ocean model and dedicated process studies using idealized models we analyze the characteristics of the sinking, of diapycnal mixing, and investigate how these are linked. 

It appears that eddies play a crucial role for the overturning, both in depth space and density space. They control the characteristics of the yearly cycle of convection and restratification, the magnitude of the Eulerian sinking near continental boundaries, and steer the export of dense waters formed in the interior of the marginal seas via the boundary current system.

These studies thus reveal a complex three-dimensional view on sinking, diapycnal water mass transformation and overturning in the North Atlantic Ocean, involving the boundary current, the interior and interactions with the eddy field.  This implies that it is essential to resolve these eddies to be able to properly represent the overturning in depth and density space in the North Atlantic Ocean and its response to changing conditions in a future climate.

How to cite: Katsman, C., Brüggemann, N., Georgiou, S., Sayol Espana, J.-M., Ypma, S., van der Boog, C., and Pietrzak, J.: From small whirls to the global ocean: how eddies affect the Atlantic Meridional Overturning Circulation, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-5644, https://doi.org/10.5194/egusphere-egu2020-5644, 2020.

Display materials

Display file

Comments on the display material

AC: Author Comment | CC: Community Comment | Report abuse

Display material version 1 – uploaded on 04 May 2020
  • CC1: Comment on EGU2020-5644, Quentin Jamet, 05 May 2020

    Hello Caroline,

    Thank you again for your presentation, very interesting!
    In case you have further thougths on this, I ask my question again here: In coarse resolution models, are eddy-parameterization (Gent-McWilliams) able to achieve these exchanges?

    Best,

    Quentin

    • AC2: Q: What does this imply for non-eddy resolving models?  Can a GM-parameterization tackle this?, Caroline Katsman, 05 May 2020

      Great and relevant Q, no answer (yet). We just started to investigate this.  My suspicion is the answer is no. With possible implications for reliability of climate model projections ….. First hints in Katsman et al 2018: a 1 degree model simulation does show sinking near the boundaries (note: all you need is some ageostrophic component in the vorticity balance, so  viscosity basically works but again eddy activity has local peaks while viscosity is constant). But the modeled sinking does not follow theoretical predictions. Definitely more research needed here

  • AC1: Q: is the mechanism proposed the same as in Send & Marshall 1995?, Caroline Katsman, 05 May 2020

    A: No it is not. 

    In SM95, restratification and export of convected waters is assumed to be governed by BC eddies that form on the front surrounding the convection region. However, in reality fronts around convection regions do not become that sharp (Pickart et al 2002) that they can accomplish this (see also Katsman et al 2004; Gelderloos et al 2011). Moreover, the influences of such “rim current eddies” does not reach very far from the front.

    In our view, the eddies that arise from the boundary currents (f.ex along the west coast of Greenland) are the ones that play a crucial role. Note 1: since eddy activity is spatially inhomogeneous, so will the return path to the boundary current. Note 2: this also provides a link between SPG strength to AMOC / dense water export

    • AC5: Reply to AC1, Caroline Katsman, 05 May 2020

      No it is not. 

      In SM95, restratification and export of convected waters is assumed to be governed by BC eddies that form on the front surrounding the convection region. However, in reality fronts around convection regions do not become that sharp (Pickart et al 2002) that they can accomplish this (see also Katsman et al 2004; Gelderloos et al 2011). Moreover, the influences of such “rim current eddies” does not reach very far from the front.

      In our view, the eddies that arise from the boundary currents (f.ex along the west coast of Greenland) are the ones that play a crucial role. Note 1: since eddy activity is spatially inhomogeneous, so will the return path to the boundary current. Note 2: this also provides a link between SPG strength to AMOC / dense water export

  • AC3: Q: Can we see this in observations?, Caroline Katsman, 05 May 2020

    A: Yes, some recent hints. The paper by Le Bras et al in GRL seems to indicate it works like this in the Irminger Sea.

    https://agupubs.onlinelibrary.wiley.com/doi/abs/10.1029/2019GL085989?casa_token=jGReC3R9GykAAAAA:SMU57tQDLn7S9TG600TBnOhXb2g2akXKRKZVZnvw7189sVQ__vSzh7vmokND-cTElaU0d5GozvM1S5Cc

     We are also see it in Argo [paper in prep.]

    • AC6: Reply to AC3, Caroline Katsman, 05 May 2020

      A: Yes, some recent hints. The paper by Le Bras et al in GRL seems to indicate it works like this in the Irminger Sea.

      https://agupubs.onlinelibrary.wiley.com/doi/abs/10.1029/2019GL085989?casa_token=jGReC3R9GykAAAAA:SMU57tQDLn7S9TG600TBnOhXb2g2akXKRKZVZnvw7189sVQ__vSzh7vmokND-cTElaU0d5GozvM1S5Cc

       We are also see it in Argo [paper in prep.]

       

  • CC2: Comment on EGU2020-5644, Patricia Handmann, 05 May 2020

    Thank you Caroline for these very interesting slides. How would you estimate the relative importance of the interior eddy driven process versus the diapycnal process for the total water mass transformation?

    • AC8: Diapycnal mixing - interior versus boundary current, Caroline Katsman, 05 May 2020

      In our Lagrangian studies with the idealized LabSea model (Georgiou et al 2020) you can see that the transports of waters that follow the bdy current and those that pass through the interior and end up at the SW exit are about 50-50 (Fig 3) but these are not all modified. If you look at Fig 7b you see that the overturning in depth space is evenly distributed between BC and INTshort, but that the ones that pass through the interior contribute more to the overturning in density space. 

      Note also that water mass transformation in the interior produces  denser water masses than the convection in the boundary current itself (b/c the boundary current is more stratified), so the pathway matters for the "end product"

       

       

      Note: this is only one marginal sea & idealized – this needs not be the same everywhere

      • AC9: Are the processes in the boundary the same (ie from surface forcing)?, Caroline Katsman, 05 May 2020

        In principle yes. Diapycnal mixing in the bdy current itself occurs because of buoyancy loss, to the atmosphere (Q) or laterally via eddy fluxes (locally inhomogenous). Since it is a stratified environment, it produces less dense water masses than interior convection and it does not reach deep.

        In the Lab Sea the huge buoyancy loss due to eddies shed near the west coast of Greenland allows substantial convection in the bdy current near the Canadian coast, as it already lost much buoyancy. This may be quantitatively different from farther upstream in the SPG.

  • AC4: Q: How would strong interannual variability in the surface buoyancy modify the picture? Should we think of the eddy driven component of the circulation being relatively constant from year to year, while the buoyancy loss has a sporadic impact? And, have y, Caroline Katsman, 05 May 2020

    See Georgiou et al 2019 for a discussion on the relation between Q, convection, eddy activity and overturning (based on our idealized Labrador Sea model).

    https://www.sciencedirect.com/science/article/pii/S1463500318303032?casa_token=XwWxqstM8D0AAAAA:XRpIMIHTFB65tSFREkBx0UruXgfmw4v7vblfPxg8XSitLO4_OhoqhspZ9yVJim3_gFcZ3KnbALs

    Quote from the discussion:

    “We showed that substantial downwelling is predominantly appearing in areas with strong eddy activity and the magnitude of the downwelling in these eddy-rich areas is positively correlated with the magnitude of the surface heat flux. This link between the wintertime cooling and the overturning in the North Atlantic has been pointed out in many numerical and observational studies (e.g. Biastoch et al., 2008; Curry et al., 1998), but here we demonstrate that this link is indirect (Fig. 13). As shown in this study, both the convection and the eddy field are affected by the changes in the surface forcing. In response to a stronger (weaker) surface winter heat loss, convection is stronger and the temperature gradient between the interior and the boundary current increases (decreases). This directly impacts the eddy field; as the temperature gradient increases, the baroclinicity of the boundary current increases, and the boundary current becomes more unstable. While the generation of the eddies is known to be governed by local processes, their impacts are not restricted to their generation region since they propagate away towards the interior (Fig. 4). As a result, the associated eddy heat transport from the boundary current towards the interior strengthens (Figs. 9, 10). This increases the heat loss of the boundary current, which in turn governs the magnitude of the downwelling (Spall and Pickart, 2001; Straneo, 2006b; Katsman et al., 2018), and at the same time provides a negative feedback on the convection depth. These idealized simulations thus highlight that complex interactions between the boundary current and interior are established via the eddy activity, and in concert determine the downwelling in the basin as well as the characteristics of convection.”

    • AC7: Reply to AC4, Caroline Katsman, 05 May 2020

      See Georgiou et al 2019 for a discussion on the relation between Q, convection, eddy activity and overturning (based on our idealized Labrador Sea model).

      https://www.sciencedirect.com/science/article/pii/S1463500318303032?casa_token=XwWxqstM8D0AAAAA:XRpIMIHTFB65tSFREkBx0UruXgfmw4v7vblfPxg8XSitLO4_OhoqhspZ9yVJim3_gFcZ3KnbALs

      Quote from the discussion:

      “We showed that substantial downwelling is predominantly appearing in areas with strong eddy activity and the magnitude of the downwelling in these eddy-rich areas is positively correlated with the magnitude of the surface heat flux. This link between the wintertime cooling and the overturning in the North Atlantic has been pointed out in many numerical and observational studies (e.g. Biastoch et al., 2008; Curry et al., 1998), but here we demonstrate that this link is indirect (Fig. 13). As shown in this study, both the convection and the eddy field are affected by the changes in the surface forcing. In response to a stronger (weaker) surface winter heat loss, convection is stronger and the temperature gradient between the interior and the boundary current increases (decreases). This directly impacts the eddy field; as the temperature gradient increases, the baroclinicity of the boundary current increases, and the boundary current becomes more unstable. While the generation of the eddies is known to be governed by local processes, their impacts are not restricted to their generation region since they propagate away towards the interior (Fig. 4). As a result, the associated eddy heat transport from the boundary current towards the interior strengthens (Figs. 9, 10). This increases the heat loss of the boundary current, which in turn governs the magnitude of the downwelling (Spall and Pickart, 2001; Straneo, 2006b; Katsman et al., 2018), and at the same time provides a negative feedback on the convection depth. These idealized simulations thus highlight that complex interactions between the boundary current and interior are established via the eddy activity, and in concert determine the downwelling in the basin as well as the characteristics of convection.”