CLS contribution to CANIGO


SPACE OCEANOGRAPHY (in french) CANIGO
Planning Satellite altimetry Data available Maps and animations Analysis of the ocean circulation variability Altimetric data assimilation in QG model Team-work with other CANIGO participants Publications Bibliography

Note that this QG-model experiment is developped as part of the SHOM (Service Hydrographique et Océanographique de la Marine) SOPRANE (Système Océanique de Prévision Régionale en Atlantique Nord Est) program, which aims at developping a real time forecast ocean modelling system (PI: Pierre Bahurel SHOM/CMO). (This work has been done by CLS under contract with SHOM/CMO. Contract number : 96.87.055.00.470.29.45)

The Quasi-Geostrophic model

The assimilating model has been implemented in the CANIGO area is based on the Blayo et al., (1994) quasi-geostrophic (QG) model of the North Atlantic. The model coverage ranges from 35°W to the European and African continental shelf (200m isobath), and from 24°N to 54°N. The Gibraltar Strait is closed. The model features 10 levels on the vertical and has an horizontal resolution of 1/10° of degree (spherical coordinates). The vertical stratification is computed from the Robinson et al., (1979) atlas. The bathymetry is interpolated from the ETOPO5 1/12° bathymetry file.

Vertical description of the 10 layers of the QG model, with 11 layer interfaces. This 10 layers can resolve the barotropic mode, and the 9 first baroclinic modes, which radii are given in the fourth column

Level interfaces
(depth in meters)

Psi Levels
(depth in meters)

Layer thickness
(meters)

Rossby Radii of
deformation (km)

0

200 100 200 R₁ = 27.16

450 325 250 R₂ = 13.32

650 550 200 R₃ = 8.86

850 750 200 R₄ = 6.74

1150 1000 300 R₅ = 5.81

1500 1325 350 R₆ = 4.98

2050 1775 550 R₇ = 4.55

2700 2325 650 R₈ =3.99

3700 3200 1000 R₉ = 3.81

5500 4600 1800

The open boundaries feature radiative conditions including the tangential component of the radiation equation (Raymond and Kuo, 1984) and a relaxation to the climatology with an adjustable time scale whose typical value is one year. The solid boundary condition derived from the Munk analytical formulation is described in Verron and Blayo (1996). The climatology used is derived from the Paillet and Mercier (1996) inverse model of the North Atlantic. We have chosen to restrain the model experiment to the eastern basin. First because a North Atlantic basin QG model should fulfill the parametrization of the Gulf Stream dynamics, which differs from the eastern basin stratification. Second, because the most sophisticated eddy resolving models of the North Atlantic circulation are still unable to represent the Azores Current energetics, mainly because they cannot simulate the Gulf Stream southern branches which overflow the Mid-Atlantic Ridge. Consequently, we would rather specify the inflow at 35°W using the climatology, and keep the mesoscale energetics by assimilating the altimetric dynamic topography.

Assimilation of altimetric data

Altimetric data are assimilated in the QG-model using a scheme based on the optimal interpolation described in De Mey (1998). Along track sea level anomalies from TOPEX/POSEIDON (T/P) and ERS-1 and ERS-2 are assimilated.
The assimilation method is an optimal interpolation scheme in a reduced space specially tuned for the assimilation of altimeter data from several altimeter missions. The current implementation of the assimilation scheme in this model is used for the assimilation of real-time ERS-2 and TOPEX/POSEIDON data in the framework of the SOPRANE project conducted by the CMO (Centre Militaire d’Océanographie, French Navy). The assimilation occurs schematically as follows : the full 3D model space is projected in a reduced 2D space using vertical empirical orthogonal functions (EOFs). Model data misfits are computed for one assimilation cycle over a one week period, the optimal interpolation is performed at a central date during this period providing the correction term and its error variance in the reduced space. This correction is extended to the full space using an extension scheme based on the same EOF decomposition as described in De Mey and Robinson (1987) and Dombrowsky and De Mey (1992). The vertical EOFs were previously computed from historical data in this area. The initial conditions used for the assimilation are obtained from the Paillet and Mercier (1996) inverse model. Once the assimilation is started through the succession of assimilation cycles, the initial conditions are forgotten after a 100-day transition period, during which the turbulent cascade occurs with the creation of the eddy fields. Therefore, the ocean circulation reaches a statistically and dynamically stable regime at the surface and at depth. The assimilation experiment has been performed over the 1993-1997 period, using a mean annual wind field representing the surface forcing.

First results: The 5-year assimilation run

The model provide the full adjusted dynamic topography at the different levels. For instance the map of the surface streamfunction December 10, 1998 is revealing a large scale slope associated with the subtropical gyre recirculation on the eastern boundary of the Atlantic ocean. As already shown by the altimetric SLA maps the eddy fiels is dominating the ocean circulation. There are strong meanders of the Azores Current near 33-34°N and 30°W, and a cyclonic eddy south of the Canary Islands. The North Atlantic Current is also meandering in the north western corner of the domain.

A first description of the ocean circulation can be obtained by averaging seasonally the surface level outputs:

winter 93
spring 93
summer 93
fall 93

winter 94
spring 94
summer 94
fall 94

winter 95
spring 95
summer 95
fall 95

winter 96
spring 96
summer 96
fall 96

winter 97
spring 97
summer 97
fall 97

Ocean variability
In a similar way as we obtained with the SLA maps, the variability provided by the 5-year surface streamfunction of the model has been analyzed.
The Eddy Kinetic Energy (EKE) map (colors) superimposed on the mean currents for the year (dark contours, interval 2000 m2/s) show the classical distribution of EKE and mean currents in this area. Two high EKE tongues (EKE > 150 m2/s2) corresponding to the North Atlantic Drift near 50°N and the Azores Current near 34°N enter the eastern north Atlantic.
The Azores current penetrates into the eastern basin, recirculating southwestward mainly near the coast in the Canary Current which contributes to the general circulation of the sub-tropical anticyclonic gyre. Several secondary southward recirculation branches can be seen between 35°W and the coast as described by Klein and Siedler (1989) and Tychensky et al., (1998).
The North Atlantic Drift enters the eastern basin mainly through the Charly Gibbs fracture zone near 52°N. It exits the domain primarly 17°W through the Rockall trough and secondary west of the Rockall Plateau. It contributes to the general cyclonic circulation of the sub-polar Gyre. Generally speaking, the North Atlantic Current consists of several branches crossing the ridge through Maxwell, Faraday and Charly Gibbs fracture zones as described by Sy (1988) and Gana and Provost (1993). These current branches meander in the eastern basin interacting with the rough bathymetry and exit the domain by the Northern boundary.

Variance in m2

1993 The EKE is also high west of the Atlantic ridge, between 40°N and 50°N. This high energy area corresponds to the highly energetic extension of the Gulf Stream blocked by the ridge, with only part of it flowing over the ridge to the eastern basin

1994 The EKE slows down in the Azores Current region compared to the 1993 situation. The Azores Current eastern extension is mostly zonal until fall. Then it interacts with large cyclonic and anticyclonic eddies. The Canary current features a lot of eddy activity (EKE>100 m2/s2) mainly late in the year. The North Atlantic Current features 3 main branches, associated with large meanders. EKE reaches 200 m2/s2 in some places, especially above the Ridge.

1995 The Azores Current features a very high activity, with large energetic eddies propagating eastward around 33°N. This eddy activity widens and increases the EKE tongue in this area. Some eddies can be followed all year around. The Canary Current activity is less intense than for the 2 previous years. So is the eastward extension of the Azores Current. This later is supposed to at least partly feed the Canary Current by the north. The EKE is also lower in the North Atlantic Current. The 3 branches of this current are still there, but with a lesser eddy activity as compared to the 2 previous years.

1996 Where is the Azores Current ? The mean current can be hardly depicted from isolated eddies and meanders. The EKE is very low (lower than 100 m2/s2) as compared to the previous year. The Canary current is similar to the Azores Current, in the sense that it is low energy, a slower mean current and that it meanders around large, low energy eddies. The North Atlantic Current is also low energy as compared with the previous years except for a southern branch at around 47°N which interacts with a strong cyclonic eddy at 32°W. This interaction lasts until the end of the year.

1997 The Azores current EKE seems to increase again. It is approximately at the same level as for the year 1994. Though the mean current is not as well established eastward as it was in 1993, the strong interaction with large eddies that was observed two years ago seems to be forgotten. Especially, the eastern extension of this current can be seen again in the mean field. The Canary Current is relatively low energy. The eddy activity in this area is higher in the second half of the year. The North Atlantic Current features now a high energy branch which enters the domain at around 48°N, which corresponds to the southernmost branch of this current described by Sy (1988). This branch go then northward at 31°W. It is high energy. High EKE can also be found west of the Atlantic ridge. This might be the signature of a high eddy activity in the western part of the North Atlantic.

1993 1994

1995 The CANIGO area is subject to strong propagations, as was already show by SLA longitude time-plots. We have plotted here the propagating diagram of the model surface outputs at 34°N, where the Azores Current is meandering. Features, with ~500 km wavelength, are propagating to the west from 10-15°W, to the western boundary of the domain. Note that alitmetric assimilation and the open boundary schemes allow to correctly adjust propagations and reflections at this boundary in the QG model. Between 20°W and 30°W we also observe eastward propagations, over periods of 1-3 months. They may be due to meanders or eddies interacting with the main current. Longitude time plot do not allow to answer whether propagating features are waves (e.g. Rossby waves) or propagating eddies (i.e. close vortices propagating to the west and transporting water masses).

1996 1997