<p>Eddy Compensation in the Southern Ocean</p> Gauss Centre for Supercomputing e.V.


Eddy Compensation in the Southern Ocean

Principal Investigator:
Carsten Eden

Institut für Meereskunde, Universität Hamburg

Local Project ID:

HPC Platform used:

Date published:

The Atlantic Meridional Overturning Circulation (AMOC) transports warm tropical surface water towards northern Europe and returns cold water at depth to the world’s ocean. At the same time it plays a significant role in the global carbon cycle through the ocean’s ability to dissolve carbon dioxide. This overturning is thus of great climatic importance, but a complete picture of its driving forces has not yet emerged due to several observational and theoretical challenges.

One of the leading hypotheses on what governs the AMOC strength is the strong westerly winds that overlie the Southern Ocean. In combination with the unique basin geometry, these winds are able to supply the necessary energy to raise the water from great depths to the surface and return it back towards equator. This idea has gained support from past numerical model studies, but which did not include the explicit effect of ocean meso-scale eddies generated by hydrodynamic instabilities of the large-scale mean flow. This project uses therefore realistic coarse and high resolution ocean models to investigate the ocean response to changes in wind stress and the ability of meso-scale eddy parameterisations to simulate that response.

Scientific work accomplished and results obtained

Recent research has shown that the instability of the Antarctic Circumpolar Current in the Southern Ocean fuels a rich eddy field, which compensate the effect of the winds on the overturning in the Southern Ocean. These studies have however been idealized with respect to bottom topography, basin geometry and governing equations, all of which are aspects of the Southern Ocean that is known to be important to the dynamics. Present day state-of-the-art climate models, like the Community Earth System Model (CESM), relax these simplifications by solving the full set of the primitive equations in a more realistic domain, but are too crude in resolution to explicitly resolve ocean eddies, why they need be parameterized. This involves a complication due to the necessary specification of an eddy diffusivity. We investigate the ability of the model’s eddy parameterisation to correctly reproduce the response to changes in AMOC due to wind stress changes over the Southern Ocean in subproject A and explore new ways for improved parameterisations in subproject B.

Subproject A: Response the Southern Ocean wind increase

Since the middle of the last century the Southern Ocean winds are observed to have increased by up to 30% in strength. Given the potential impact on the AMOC and hence the climate, it is crucial to determine the correct ocean response to wind changes. Wind stress perturbation experiments using the global ocean model CESM with resolved and parameterized ocean meso-scale eddies are therefore conducted. The models are forced with climatology through the first 25 model years, after which the wind stress in the Southern Ocean is increased by 50% and the models are run for another 20 years.

In Poulsen et al. (2018) the ability to parameterize Southern Ocean eddy effects in such a forced coarse resolution ocean general circulation model is assessed and compared to identical experiments performed with the same model in 0.1° eddy-resolving resolution. With forcing of present-day wind stress magnitude and a thickness diffusivity formulated in terms of the local stratification, it is shown that the Southern Ocean residual meridional overturning circulation in the two models is different in structure and magnitude. With a decrease of the zonal wind stress by 50%, Fig. 1 shows that the absolute decrease in the overturning circulation is insensitive to model resolution, and that the meridional isopycnal slope is relaxed in both models. The agreement between the models is not reproduced by a 50% wind stress increase, where the high resolution overturning decreases by 20%, but increases by 100% in the coarse resolution model (Fig. 1). It is demonstrated by Poulsen et al. (2018) that this difference is explained by changes in surface buoyancy forcing due to a reduced Antarctic sea ice cover, which strongly modulate the overturning response and ocean stratification. They conclude that the parameterized eddies are able to mimic the transient response to altered wind stress in the high resolution model, but partly misrepresent the unperturbed Southern Ocean meridional overturning circulation and associated heat transports.

Realization of the project

The experiments with the computationally most expensive high resolution model version are conducted on the JSC HPC system JUQUEEN. The numerical model code (CESM, www.cesm.ucar.edu) contains an active sea-ice and ocean components with observed and prescribed meteorological boundary conditions. The discretisation of the primitive equations is based on finite differences with a horizontal resolution of 1/10 of a degree and with 62 layers in the vertical, which amounts to about 100 million grid points. On JUQUEEN we are using 4096 cores. The model is initialized from present day conditions and forced with climatology through the first 25 model years, after which the wind stress in the Southern Ocean is in- or decreased by 50% and the model is run for another 20 years.

During the first 24 years we saved monthly mean output files. For the remaining integration and analysis, the output consists of three-day mean fields of 26 variables. This high-frequency model output is necessary to calculate meaningful long-term means meso-scale eddy fluxes from covariances during the model integration since meso-scale eddy have time scales of days to weeks. One model year in terms of three-day mean fields amounts to 12 TB output. The output is transferred to the Electronic Research Data Archive (ERDA) at the University of Copenhagen on a daily basis.

In total, 100 model years of integration have been completed on JUQEEN between 2016 and 2018 at a rate of some 0.1 years/day which results in about 2.5 PB output to be analyzed.

Subproject B: Meso-scale eddy parameterisation evaluation

Meso-scale eddy parameterisations often rely on the specification of an eddy diffusivity which is poorly known. The model biases due to the unknown eddy diffusivity can lead to the misrepresentation of large-scale ocean flow changes as shown above. In Poulsen et al (2019) an interpretation of eddy form stresses via the geometry described by the Eliassen-Palm flux tensor is explored. The study shows that the eddy form stress is fully described by a vertical ellipse, whose size, shape, and orientation with respect to the mean flow shear determine the strength and direction of vertical momentum transfers. This geometric framework is used to form an eddy diffusivity that depends on eddy energy and a nondimensional geometric parameter α, bounded in magnitude by unity. The parameter αexpresses the efficiency by which eddies exchange energy with baroclinic mean flow.

The eddy-resolving CESM version is used to estimate the spatial structure of α  in the Southern Ocean and assess its potential to form a basis for parameterization. The eddy efficiency α averages to a low but positive value of 0.043 within the Antarctic Circumpolar Current (Fig.2), consistent with an inefficient eddy field extracting energy from the mean flow. It is found that the low eddy efficiency is mainly the result of that eddy buoyancy fluxes are weakly anisotropic on average. The eddy efficiency is subject to pronounced vertical structure and is maximum at ca. 3 km depth, where eddy buoyancy fluxes tend to be directed most downgradient. Since α  partly sets the eddy form stress in the Southern Ocean, a parameterization for α  must reproduce its vertical structure to provide a faithful representation of vertical stress divergence and eddy forcing, forming the future challenge for this work.


The following project publications form also the main part of the PhD thesis of Mads Poulsen:

  • Poulsen, M. B., Jochum, M., and Nuterman, R. (2018). Parameterized and resolved Southern Ocean eddy compensation. Ocean Modelling, 124, 1-15.
  • Poulsen, M. B., Jochum, M., Maddison, J. R., Marshall, D. P., and Nuterman, R. (2019). A geometric interpretation of Southern Ocean eddy form stress. Journal of Physical Oceanography, 49(10), 2553-2570.

Theses completed within the project

Parameterizing Southern Ocean eddy-induced circulation in coarse resolution ocean models, Mads B. Poulsen, PhD Thesis, January 2019

Additional references

Poulsen, M. B., Jochum, M., Nuterman, R., and Eden, C. (2018). Southern Ocean Eddy Compensation Examined with a High-Resolution Ocean Model. In NIC Symposium 2018 (No. FZJ-2018-02954). John von Neumann-Institut for Computing.

Principal investigator

Prof. Dr. Carsten Eden, Institut für Meereskunde, Universität Hamburg

Project contributors

Prof. Dr. Jochum, Dr. Nuterman, and Dr. Poulsen, Niels Bohr Institute, Copenhagen University, Julien Maries Vej 30, 2100 Copenhagen, Denmark

Scientific Contact

Prof. Dr. Carsten Eden
Institut für Meereskunde
Universität Hamburg
Bundesstr. 53, D-20146 Hamburg (Germany)
e-mail: carsten.eden [@] uni-hamburg.de

JSC project ID: chhh28

May 2020

Tags: JSC Climate Science Universität Hamburg