1. Beyond Currents – The Next Phase in GOCE Oceanographic Research
Rory Bingham, Chris Hughes and Keith Haines

2. Outline of presentation
What we hoped to achieve
What we have achieved
What is yet to be achieved

3. 1. What we hoped to achieve

4. What we hoped to achieve
Understanding short-spatial-scale dynamical processes:
jets
Boundary/frontal currents
Eddy-mean flow
Topographic controls
Estimating basin scale heat/freshwater fluxes
Model initialisation
Ocean forecasting

5. What we hoped to achieve

6. What we hoped to achieve
Small scale ocean dynamics: eddy-mean flow interaction
Show a figure similar to 3.10 showing what GOCE hoped to deliver relative to GRACE, which was expected to deliver to d/o=80=250 km spatial scales – use OCCAM 1/12th. Consider what GRACE actually delivered before launch and later. Could use my POV vs d/o plot for MDT and currents with expected delivery points marked on them, eg d/o=20, 80 and 200.

7. What we hoped to achieve
Basin scale transports – uncertainty reduction
Legrand, 2001,JGR Oceans
Legrand and Minster, 1999, GRL

8. What we hoped to achieve
Constraining ocean models
See page 38 of GOCINA report for impact on transports

9. 2. What we have achieved
By “have achieved” we mean released data published work

10. What we have achieved Improved MDT and ocean currents – Agulhas region
GTIM1
GTIM2
GTIM3
GTIM4
GTIM5

11. What we have achieved
Time-wise informal MDT errors at d/o=200

12. What we have achieved Small scale ocean dynamics: ACC fronts
The Antarctic Circumpolar Current (ACC) is the largest (in terms of transport) current on Earth transporting about 140 Sv (1 Sv0106 m3/s) eastward across the Drake Passage (Cunningham et al. 2003). The ACC is the only circumpolar current that connects all major oceans. The flow of the ACC is known to be concentrated in a number of jets associated with fronts that are characterized by large horizontal gradients of temperature and salinity (e.g., Deacon 1937; Orsi et al. 1995; Sokolov and Rintoul 2002, 2007, 2009). The processes of water mass transformation as well as the meridional energy and tracer transports in the Southern Ocean are most intense near the fronts, and therefore,
it is important to establish their position and spatial extent.
Volkov and Zlotnicki, 2012, Ocean Dynamics

13. What we have achieved
Small scale ocean dynamics: eddy-mean flow interaction
Figure 2. Eastward acceleration of the mean flow by time-dependent eddies, based on 13 years of satellite altimetry data (a dynamically passive, irrotational component has been removed). Contours representing intervals of 20 centimetres show mean dynamic topography based on the DTU10 [16,17] mean sea surface and the TUM2013C geoid [18], averaged over 0.25◦ blocks and with 25 km Gaussian smoothing applied.
Eddies shape the mean flow:
Figure 2 (adapted from [19]) shows an example of what can be learned about eddy-mean flow interaction from satellite altimetry. The shading represents the eastward acceleration of the mean flow owing to momentum fluxes carried by time-dependent eddies (measured using altimetry alone), whereas the contours of mean dynamic topography illustrate the mean flow (for which a geoid measurement is also needed). It had been thought that eddies radiated out fromjets would tend to exert an eastward acceleration on the jets, but the observations showed a more complicated relationship, indicating that the simple theory is too idealized to apply to the real Southern Ocean, and suggesting that the interaction of eddies with topography is important.
Eastward acceleration of the mean flow by time-dependent eddies, based on 13 years of satellite altimetry data. Contours representing intervals of 20 centimetres show mean dynamic topography based on the DTU10 mean sea surface and the TUM2013C geoid.
Tamisiea et al, 2014, Phil. Trans. R. Soc.

14. What we have achieved Exploiting formal GOCE errors
Mean dynamic topography model MDT_TIM04S and associated standard deviations based on the TIM04 gravity field model with additional smoothness conditions applied to the MDT parameters.
(left) Estimated mean dynamic topography with derived geostrophic velocities based on the different GOCE gravity field models within a section of the North Atlantic Ocean. (right) Corresponding standard deviations of the different mean dynamic topography estimates.
Becker, S., J. M. Brockmann, and W.-D. Schuh (2014), Mean dynamic topography estimates purely based on GOCE gravity field models and altimetry, Geophys. Res. Lett., 41, 2063–2069, doi: /2014GL

15. What we have achieved
Data assim./transport - Inverse Finite Element Ocean Model (IFEOM)
Fig. 4. Atlantic Meridional Overturning Circulation (AMOC) in IFEOM using unfiltered MDT (left) and filtered MDT (right) in the assimilation.
Fig. 5. Atlantic Meridional heat transport in IFEOM
Freiwald, G. (2013) A new filter for the Mean Dynamic Topography of the ocean derived directly from satellite observations, Journal of Geodynamics, (72), pp , doi: /j.jog
Becker, S. , Freiwald, G. , Losch, M. and Schuh, W. D. (2012) Rigorous fusion of gravity field, altimetry and stationary ocean models, Journal of Geodynamics., , pp , doi: /j.jog

16. 3. What is yet to be achieved
By “yet to be achieved” we mean work in progress eg MO data assim.

17. What is yet to be achieved
Heat and freshwater fluxes
(left) Cumulative zonal ACC transports, calculated from MDT gradients and SOSE density gradients for the data products for (a) SR1, (c) SR2 and (e) SR3. (right) Cumulative zonal barotropic transports associated with MDT gradients, for (b) SR1, (d) SR2 and (f) SR3. SOSE (black), CNES‐CLS09 (red), MN05 (blue), GGM02C (green), and EGM08 (magenta).
Mean Dynamic Ocean Topography (MDT) is the difference between the time-averaged sea surface height and the geoid. Combining sea level and geoid measurements, which are both attained primarily by satellite, is complicated by ocean variability and differences in resolved spatial scales. Accurate knowledge of the MDT is particularly difficult in the Southern Ocean as this region is characterized by high temporal variability, relatively short spatial scales, and a lack of in situ gravity observations. In this study, four recent Southern Ocean MDT products are evaluated along with an MDT diagnosed from a Southern Ocean state estimate. MDT products differ in some locations by more than the nominal error bars. Attempts to decrease this discrepancy by accounting for temporal differences in the time period each product represents were unsuccessful, likely due to issues regarding resolved spatial scales. The mean mass transport of the Antarctic Circumpolar Current (ACC) system can be determined by combining the MDT products with climatological ocean density fields. On average, MDT products predict higher ACC transports than inferred from observations. More importantly, the MDT products imply an unrealistic lack of mass conservation that cannot be explained by the a priori uncertainties. MDT estimates can possibly be improved by accounting for an ocean mass balance constraint.
Griesel, A., M. R. Mazloff, and S. T. Gille (2012), Mean dynamic topography in the Southern Ocean: Evaluating Antarctic Circumpolar Current transport, J. Geophys. Res., 117, C01020, doi: /2011JC

18. What is yet to be achieved
Dynamics– what determines the mean circulation
High resolution
Sensitivity to MDT/topography/eddies
Low resolution climate models
Coupled models

19. What is yet to be achieved
Dynamics – what determines the mean circulation

20. What is yet to be achieved
Understanding time-mean coastal sea level gradients
Mean sea level along the Atlantic eastern boundary (OCCAM)
equilibrium position
dynamic displacement
Due to poorly understood ocean dynamic processes the UK mean sea level is almost 0.5 m below equilibrium

21. What is yet to be achieved
Understanding time-mean coastal sea level gradients
Mean sea level along the Atlantic eastern boundary
equilibrium position
1 Sv Med-inflow ≈10 cm sea level

22. What is yet to be achieved
Understanding time-mean coastal sea level gradients

23. What is yet to be achieved
see poster
Operational ocean forecasting
The UK Met Office
GOCE input MDT
Bias-corrected MDT
Bias
TOPAZ – plans to use GOCE in next version
Our ideas for inclusion of errors is as follows: 
- TOPAZ has a bias estimation loop taking advantage from the EnKF dynamic error covariances,
in particular for estimation of sea surface heights biases. 
- The initial uncertainty of the sea surface heights bias is set arbitrarily. 
- The principle and primary results have been described in Sakov et al. (OS, 2012) 
- Since the article has been published, the forward propagation of uncertainties in SSH bias 
(error inflation) has been replaced by a first order auto-regressive process with arbitrary variance. 
- The GOCE error estimates would provide a sounder scientific basis for the initial and forward 
uncertainties of the bias estimates. 
bias correction
Dynamic topography
Sakov et al. (OS, 2012)

24. What is yet to be achieved
Understanding errors – formal vs. informal
Informal MDT (CLS01-GTIMX) error
Informal GTIMX error
Formal GTIMX error

25. What is yet to be achieved
Understanding errors – formal vs. informal
Formal errors (16.9 cm)
Informal errors (16.3 cm)
Cumulative errors for the GTIM3 geoid at d/o=250

26. What is yet to be achieved
Understanding errors – MSS contribution
Timewise
Direct
CLS01: dashed
CLS11: solid MW LW
LW: >250 km
MW: km

27. What is yet to be achieved
Understanding errors – defining length scales
GOCE error
covariance patterns

28. Summary: Beyond currents
Phase 1:
Improved MDT and associated currents
Improved techniques and processing
Validation studies
Phase 2:
Heat and freshwater fluxes: Basin-scale; Across current sections
MDT dynamics: Model representation; Mean-flow eddy interaction; Topographic controls; Coastal MDT
Assimilation/initialisation: Operational forecast systems; Low-high resolution models; Coupled climate models; Understanding errors