Limitations (things to be done)

Slow retrieval of data with MARS

For reasons not very clear to me, the retrieval of data using ECMWF's MARS stuff is sometimes very, very, very slow ...

Solution might be to sort the fields to be retrieved for a certain day according to the storage order in the archive. A first attempt to implement something like this is implemented in SCRIPTS/ECmeteo_retr, but not tested yet.

Slow interpolation from spectral to lat/lon fields

Need some improvement in SOURCE/grid_*.f90. Possible improvements:

• evaluation of oposite latitudes might save some time
• instead of interpolation to fine grid and numerical quadrature, evaluate at some smart chosen points only (Gaussian integration?)
• truncate spectral fields according to grid resolution

Reduced Gaussian grid vs zoom area

Some of the TM5 data is first processed at ECMWF grid points (reduced Gaussian grid), and averaged over TM5 cells afterwards. For zoom grid, only those grid points are used in the latitude band around the zoom area. In future, the grid points in the latitude band but outside the zoom area should be skipped too.