1. Seasonal Variations of MOC in the South Atlantic from Observations and Numerical Models Shenfu Dong CIMAS, University of Miami, and NOAA/AOML Coauthors: M. Baringer, G. Goni, C. Menine, S. Garzoli (NOAA/AOML) Research funded by NOAA CPO and AOML XBT Science Team Meeting Beijing, China, November 11-13, 2014

2. Motivation: Impact of SAMOC on Atlantic warming (a) ● Lee et al. (2011): 20th century global ocean simulation shows an important role played by SAMOC on the rapid warming of the Atlantic Ocean since the 1950s.

3. Motivation: Agulhas Leakage  Agulhas leakage through eddies thought to play a central role in the AMOC decadal to multi-decadal variations. Biastoch et al. Nature, 2008 Temperature (color) and velocities (vector) at 450m depth for high-resolution (0.1  ) Agulhas nest hosted in a global ocean model of 0.5° resolution.  Increase of Agulhas leakage from experiment with surface forcing fields during 1958-2004. (Biastoch et al., 2008; 2009)

4. High-Density XBT Transect along 34.5°S The overall objective of the AX18 is to monitor the upper limb of the MOC in the South Atlantic.  39 transects, provide the first and only MOC/MHT time series in the South Atlantic.  Examine the MOC and its associated transports of heat and freshwater in the South Atlantic, as well as their correspondence.  Evaluate the performance of numerical models in simulating the South Atlantic MOC and MHT (AX18). MOC

5. Model Validation: meridional velocity Compared to the results from AX18, the performance of the GFDL CDA is greatly improved in terms of representing the observed MOC and MHT structure: the transports of boundary currents are twice as strong as those during pre ‐ Argo period, and the overturning flow in the interior region is reduced. Meridional velocity from GFDL coupled data assimilation before and after Argo period.

6. Model Validation: salinity Biases in salinity are greatly reduced after assimilating Argo, which improves the boundary current transports.

7. Seasonal Variations in the MOC Both geostrophic and Ekman contributions to the MOC experience annual cycles, but they are out of phase. No significant seasonal cycle was found in the MOC. The Annual cycle in the MOC is dominated by Ekman component, and the geostrophic component shows little seasonal variations.

8. Goal: To investigate what causes the differences in the MOC seasonal variations estimated from observations and numerical models. Methodology Two CMIP5 models are used: NCAR CCSM4 and GFDL ESM2M  Monthly climatologies of T/S on a 1° longitude grid along 34°S are constructed both from observations (Argo/WOA13) and numerical models (last 50-year output) to estimate the geostrophic transport.  Argo drifting velocity at 1000 m as reference velocity for the observations study. The mean velocity at 1000 m from corresponding models are used in model studies.  The Scatterometer Climatology of Ocean Winds (SCOW) is used to compute the Ekman transport in the observational study, and the Ekman transport for each model is derived from the zonal wind stress output of the corresponding model. Model-data Comparison

9. Seasonal Variations in the MOC at 34°S  Observational estimates suggest that gesotrophic and Ekman transports contribute equally to the seasonal variations in the MOC. But in the models, Ekman transport dominates.  The modelled Ekman transport show stronger seasonality than observations.

10. West-to-East Cumulative Volume Transport from Observations and Numerical Models GFDL-ESM2M: both boundary currents are two weak. CCSM4: eastern boundary current is weak, but well reproduces the western boundary current.

11. Model-data Comparison: zonally-averaged meridional velocity  Geostrophic velocity from observations shows vertically coherent seasonal variations, resulting strong seasonal variations in the upper-ocean geostrophic transport.  Geostrophic velocity from the model T/S climatologies do not exhibit this type of vertically coherent seasonal cycle.

12. Density at the Eastern and Western Boundaries and Their Differences The observed east-west density difference is largely controlled by the western boundary, whereas in the coupled models, the eastern boundary dominates. The strong baroclinicity at the base of the mixed layer in the model fields on both the western and eastern sides of the basin results in out-of-phase variations above and below this shear layer. This out-of- phase variation, particularly at the eastern boundary, contributes to the weaker seasonality in the modeled geostrophic transport.

13. Model-data Comparison: vertical density gradient The enhanced baroclinicity in the models is possibly due to the strong stratification in the modeled T/S fields. Vertical density gradient shows large model biases in the vertical stratification from 100 m to 500 m depth, particularly toward the eastern boundary. The strong stratification inhibits the models from transferring information from the mixed layer into the lower layers, resulting in the enhanced baroclinicity.

14. Circulation in the South Atlantic

15. Model-data Comparison: temperature at 1000 m depth The isotherms near the coast experience a stronger northward displacement during austral winter, when the northward flowing Malvinas Current is stronger and the southward flowing Brazil Current is weaker, and weaker displacement during austral summer, when the Malvinas Current is weaker and the Brazil Current is stronger. This seasonality of the currents is induced by the seasonal variations of the wind-driven gyre circulation. This is consistent with the observed positive density anomalies at the western boundary along 34°S during austral winter and negative anomalies during austral summer.

16. Upper Ocean Transport

17. Conclusions  Observational estimates suggest that the geostrophic transport plays an equal role to the Ekman transport in the MOC seasonal variations at 34S, whereas in the models, the Ekman transport controls the AMOC seasonality.  The seasonality of the geostrophic transport from observations is largely controlled by the seasonal density variations at the western boundary, but in the models, the eastern boundary dominates.  The observed density seasonality at the western boundary is linked to the intensity of the Malvinas Current, which is poorly reproduced in the models.  The other factor behind the strong seasonal cycle in the geostrophic transport from the observations is the strong vertical coherence in the velocity down to 1200 m. The models lack this vertical coherence due to erroneously strong vertical density shear, which prevents the models from correctly moving information down from the mixed layer to the deeper layers. All those results clearly show the value of the AX18 transects, in terms of advance our knowledge of the MOC in the South Atlantic and to evaluate numerical models.

18. Thank You!