Interview

How is a bulletin produced?

The first thing to be said is that bulletins are generated by the first prototype version of the Mercator analysis and prediction system, called PSY-1 v-1 (for version 1 of Prototype System 1). This prototype is the first in a series of increasingly complex systems geared towards developing an ocean analysis and prediction capability. We are currently in the first phase of the Mercator project. Bulletins are generated by a model of the North Atlantic at a resolution of one third of a degree.

Mercator bulletins are produced by an analysis and prediction system that runs automatically at the Météo-France weather center in Toulouse, France, every Wednesday starting at 4.00 a.m. precisely. This process is initiated after retrieving forcing data from the European Centre for Medium-range Weather Forecasting (ECMWF) and altimetry data from Collecte Localisation Satellites (CLS). Data analysis takes about 45 minutes. All output data is ready by about 5.30 a.m. The system then generates two-week forecasts and the Mercator analysis and prediction process is complete by about 6.00 a.m. A further two hours or so of post-processing are needed to produce the 700 graphic products for the bulletin. These products are posted on the website. The complete bulletin is validated in house and checked by an expert before being officially released.

What sort of information does a bulletin contain?

Bulletins contain information relating directly to the ocean, plus data of a more technical nature produced by the analysis and prediction system.
Analysis and prediction products describe ocean conditions on a particular day, that is, every Wednesday, and forecast conditions for the following week and the week after that. These products provide a regional picture, vertical cross-sections and time series of moored buoy data.
Maps of the North Atlantic give a synoptic view of ocean conditions. We can then zoom in and take a closer look at any of the regions shown. Obviously, this picture of the ocean is a horizontal one. But vertical cross-sections of the ocean basin reveal in more detail what is happening. For example, we can locate a current on a chart and identify its temperature and salinity in a vertical cross-section. Likewise, we can then use a nearby mooring to establish whether the current is transitory or sustained, oscillatory or propagational. By combining information from charts, cross-sections and moorings, we can thus track certain phenomena temporally and spatially.
The technical bulletin pulls together sea level observations and information on assimilation (i.e., how observations were merged and the model used to generate analyses and predictions).

How are bulletins validated?

Rather than validate or reject a "simulation" generated from the model, it is better to attempt to identify regional discrepancies. We are in the process of building a full-scale validation system. The first phase of this development will seek to describe product quality within the limitations of the current system.
The current process covers:

• qualitative validation, based on our knowledge of ocean currents;
• validation of forecasts with respect to satellite observations;
• validation of analysis, based on in-situ measurements.

Dynamic topography mapped by TOPEX/POSEIDON and ocean currents

A more comprehensive validation process will be put in place in the future. Mercator product quality will be a major focus of validation activities. Christian Le Provost, Director of the LEGOS space geophysics and oceanography research laboratory, will be coordinating product validation, which is also one aspect of the Mercator/CORIOLIS 2001 call for proposals.

PSY1 team:

• Eric Dombrowsky
• Mounir Benkiran
• Eric Greiner
• Lucas Nouel