URL: http://www.mercator-ocean.fr/html/actualites/news/actu_newpsy_en.html

NewsChronicles

Two major updates on global 1/4° and Atlantic and Mediterranean 1/12°, a flight test for the global 1/12°

In collaboration with Romain Bourdallé, Marie Drévillon, Yann Drillet, Gilles Garric, Jean-Michel Lellouche, Charles-Emmanuel Testut, Benoît Tranchant, Mercator Ocean, July 2008

Two-front improvement of the operational production at Mercator Ocean: global 1/4° and Regional 1/12° on Atlantic and Mediterranean. Global 1/12° feasability is also demonstrated.

The two Mercator flagships systems are presenting major evolutions. The global 1/4° with sea ice modelling at the two poles and in situ data assimilation, stopped in September 2007 because of too strong velocities around the islands, in relation with assimilation adjustments, is coming back more robust and reliable, qualified by a 1 year 1/2 simulation. It will allow a 15 years reanalaysis (1992-2008) and then further in the past.

Sea ice around the Antarctic, as seen by the new PSY3V2, on 23 April 2008: sea ice thickness and ocean temperature all around (black line: sea ice frontier). Click to enlarge image.
Sea ice in the Arctic Ocean, as seen by the new PSY3V2, on 23 April 2008: sea ice thickness and ocean temperature all around (black line: sea ice frontier). Click to enlarge image.
Zoom on the Atlantic and Mediterranean

Concerning the high-resolution Atlantic and Mediterranean system, improvement is concerning both quality and quantity: the new model is crossing the Equator down to 20° South. This system is just like a zoom of the global 1/4°: model, assimialtion and qualification period are the same as in gloabl 1/4°. Both systems look very similar, with high resolution in extra for the 1/12° regional...


Salinity at 1000 metres in the North Atlantic, as seen by the new 1/12° Atlantic and Mediterranean. Salted water is entering Atlantic at Gibraltar Strait. Forecast for 23 April 2008.
Global 1/12°

Mercator Ocean has chosen the Mersea European Programme final meeting, held at Ifremer, Paris, 28 and 29 April 2008, to present analysis and forecast first results of the global eddy-resolving 1/12°. The aim was to demonstrate the feasability of such a challenge.

ORCA model grid used respectively in PSY3V2 (global 1/4°), PSY4 (global 1/12°) and PSY2V3 (Atlantic and Mediterranean 1/12°). Credit : LOCEAN

Such a resolution at a global scale is a worldwide performance since only the US Navy global system is presenting a better resolution (1/32°)... but with a lesser vertical resolution and without in situ data assimilation.

A few figures allow to realize the technological performance deployed: the mesh is formed by more than 13 millions of points on 50 vertical levels. For this flight test, 2 computers have been operated (one for the analysis - 159 processors of the Mercator Ocean SGI), the other one for the forecast (64 processors of the Meteo France NEC).

Mercator Ocean team however will have to face other challenges to make this system operated routinely. Problems still to resolve deal particularly with the post-processing operations to make possible the very big outputs volume diffusion towards the user community.

Around the world in "3 PSY"

Following images (sea surface temperature maps) allow to visualize the benefits of the high resoluton, throughout some "star" ocean areas.

Gulf Stream eddies
Global 1/12°
Atlantic 1/12°
Global 1/4°
Gulf of Mexico loop current
Global 1/12°
Atlantic 1/12°
Global 1/4°
Mauritania upwelling
Global 1/12°
Atlantic 1/12°
Global 1/4°
La Niña in the Eastern Pacific
Global 1/12°
Global 1/4°
Peru-Chili upwelling
Global 1/12°
Global 1/4°
Falklands convergence
Global 1/12°
Global 1/4°
Indian Ocean
Global 1/12°
Global 1/4°
Western Pacific
Global 1/12°
Global 1/4°

For the initiates

The well-advised reader will remark that the Atlantic and Mediterranean new system resolution is lower (1/12°) than the previous one (1/15°). In the Atlantic, the difference is very small. Improvements in shemes and parameterizations in the new model NEMO are compensating, at a lesser extent, at Gibraltar. In the Gulf Stream, resolution is almost the same. In the tropics, ocean structures are presenting larger scale features than in the high latitudes, and 1/12° is assumed good enough. In the Labrador Sea, resolution is increasing but it is not still enough to reproduce very small scale features at 70°N. This is also the case in Mediterranean, where, even 1/16° is not enough. To conclude, this new configuration allows to simulate the same ocean dynamics than the previous one and is better in the Atlantic.

Furthermore, this resolution is resulting from a practical issue: to have a common grid with the global ORCA025 (1/4°) and ORCA12 (1/12°) grids.

We precise finally a new correction technique in the global 1/12° assimilation: Incremental Analysis Update (IAU) allowing the analysis increment to be well distributed in the time and no more sequentially as it was before.

More information

Copyright © 2001-2007 Mercator Ocean. All rights reserved.