The idea is to implement a system that monitors the quality of the linear interpolation of the coupling data, in ARPEGE. This could be checked before sending the coupling files to the different ALADIN countries. If the system indicates that the quality is not sufficient during a certain forecast time interval then for that time interval, extra coupling data will be sent.
Practically to implement the above, the following solution was proposed. ALADIN countries should be able to decide for themselves if they want to receive occasionally such extra coupling data. If they do so, then they will be asked to read coupling files at shorter time intervals unconditionally, e.g. read coupling files at an interval of an hour rather than 3 hours. The idea is then to create the coupling files externally by a linear (or an other) interpolation. If the monitoring system decides to send more coupling data to the ALADIN countries then these files will be overwritten by the extra files. Such an organization would facilitate quick implementations of research results in the future to the operational ALADIN versions without make severe changes to the code.

Future research :

The work to improve the temporal interpolation by means the phase-angle approach will be continued. The last tests in the 3d model were disappointing. In order to continue it was proposed to study the behaviour of physical quantities (e.g. potential vorticity) to get hints of how to proceed.
Spectral coupling. Lateral boundary conditions will be needed here also. It was proposed to study the relation between the lateral and the spectral coupling, e.g. one should reconsider the choice of the relaxation coefficients in the intermediate and in the extension zone.
Starting the work on two-way nesting. The problem here is that one will be forced to make the fields periodic at each time step, which would be too expensive. The proposal here is to revisit the role of the extension zone. Can one live without it? Can the Gibbs waves be cured ?
Well-posedness of the boundary conditions: it was proposed to redo the work of the recent paper of McDonald for a spectral model and then see whether the spectral character of the model will yield some advantage.

Extra issues discussed during the "dynamics and coupling" working group in Medulin :

The following issues were raised during the meeting in Medulin. They are considered to be of lower priority than the above issues.

Which variables should be coupled in the NH model?
Study the sensitivity of the coupling mechanism to the difference in vertical resolution between the coupling and the coupled model.