Ongoing work: WOCE synthesis

Main objectives

The ongoing work is directed at estimating the general circulation of the Indian Ocean in a dynamically consistent way by quantitative combination of a time-dependent general circulation model (GCM) with World Ocean Circulation Experiment (WOCE) hydrographic data, altimetry, XBTs, wind, heat and freshwater surface flux data. We will analyze the seasonally and intra-seasonally varying dynamics and thermodynamics of the Indian Ocean throughout the water column. We will investigate seasonal changes in mass, heat and freshwater transports including their driving mechanisms, and analyze rectifying effects on the annual mean. Specific objectives are:

To estimate the strength and structure of the meridional overturning circulation.
The analysis of hydrographic sections suggested that the zonally integrated deep mass inflow was moderate (10-15 Sv, 1 Sv is 10⁶ m³/s, Macdonald and Wunsch 1996; Robbins and Toole 1997) or large (27 Sv, Toole and Warren 1993). In contrast, GCMs persistently did not produce significant deep inflow (Wacongne and Pacanowski 1996; Lee and Marotzke 1997, 1998; Garternicht and Schott 1997; Zhang and Marotzke 1998, Fig. 2).
These differing results underscore the large uncertainty in current estimates of Indian Ocean meridional transports. On the one hand, one must question the representativeness of a single or very few hydrographic sections. On the other hand, models, while reasonably consistent among themselves, all used climatological hydrography for initialization, forcing, or as explicit constraints, and hence might all be negatively affected by the lack of spatial resolution in the climatology.
We will analyze the effect of using hydrographic station data, which have a better resolution than the climatology, on the estimated meridional overturning circulation.
To estimate the strength of the Indonesian throughflow and its time-variability.
The Indonesian throughflow (ITF) is widely believed to play an important role in the global thermohaline circulation (e.g., Gordon 1986) and the heat and freshwater budgets of the Indian Ocean (e.g., Godfrey 1996). The results of Zhang and Marotzke (1998), who estimate the ITF, do not support these statements. The estimated ITF is 2.7 Sv westward, near the lower end of but well within the range of previous estimates (which go from 2.6 Sv eastward to 18.6 Sv westward, according to Table 1 in Godfrey's 1996 review). Moreover, the ITF influence on meridional heat and freshwater transports (open boundaries section, Fig. 4) is negligible in Zhang and Marotzke (1998). It is smaller than the difference between surface flux and ocean transport divergence, that is, smaller than the residual indicating deviation from steady state. While these results are intriguing and provocative, we feel that the issue must be revisited with a superior data base to ensure the conclusions' robustness.

Model configuration

This work uses the MIT GCM (Marshall {\et} 1997) and its adjoint with the following configuration:

Domain : Latitude 35°S - 26°N, longitude 24°E - 122°E.
Resolution: Horizontal 1° x 1° (99 x 62 grid points), 23 vertical levels.
Assumptions: Boussinesq and hydrostatic approximations.
Bathymetry: Gridded from the ETOPO5 atlas.
Forcing fields: NCEP wind stress, heat and freshwater fluxes.
Boundary conditions: open boundaries at the Indonesian throughflow and near 35°S; free surface.

Figure bathy

Data sources

Surface forcing: daily wind stress, heat and freshwater fluxes from NCEP.
Assimilated data: sea surface height from the French-US TOPEX/Poseidon altimeter + WOCE hydrography (map above shows the model domain and grid + Indian WOCE sections) obtained from the WOCE Hydrographic Program Office (WHPO). Some XBT data have been made available from CSIRO.

Planned experiments

Two assimilation experiments will be run. In the first experiment using the adjoint, the hydrographic sections will be considered as local, i.e. the hydrographic stations located over one horizontal model grid cell will only directly constrain this grid cell (diagonal weight matrices of the model-data misfit). In the second experiment, the influence of the hydrographic measurements at one horizontal grid cell will be extended to neighboring grid points, i.e. we will consider that the hydrography is not only representative of the local oceanic state but also contains some large-scale information (nondiagonal weight matrices).

Summary

We are now ready to perform a regional model-data synthesis by using the most sophisticated version of our GCM and its adjoint with high-frequency surface forcing, altimetry and hydrographic data from the WOCE Indian Ocean experiment. Especially, we will be able to answer whether the discrepancies between the various estimates of meridional circulation and ITF influence, mentioned above, are due to a lack of resolution in the climatology.

This work is a prototype for regional (limited-domain) ocean modeling including the use of data to estimate open boundary conditions.

Mail

People interested in sharing the output of our numerical model runs are welcome to contact us (Jochem Marotzke or Bruno Ferron)

Back to the index