Past experience revealed some problems of the current version of the EGGX module, defining the ALADIN geometry :
and maybe other problems (all possible cases are not yet investigated).
To correct this and ensure a consistency with the new interactive domain maker, MAKDO, Jean-Daniel Gril wrote a new version of EGGX, tested for a very large set of domains.
The number of input arguments is reduced, as shown in the following table :
|
New input arguments |
Old input arguments |
|---|---|
|
Reference point (ELAT0, ELON0) |
Reference point (ELAT0, ELON0) Type of projection (ERPK) Modification of reference point (NGIV0) Rotation of domain (EBETA) |
|
Centre of domain (ELATC, ELONC) Resolution (EDELX, EDELY) Number of points in each direction (C+I zone) |
SW corner (ELAT1, ELON1) NE corner (ELAT2, ELON2) Isotropy and change of corners (NSOTRP) Number of points in each direction (C+I zone) |
|
|
Change of poles and previous location of the new (0°,0°) point ( NROTEQ, ELATR, ELONR) Total number of points in each direction |
The type of projection is derived from the reference (tangent) latitude : polar stereographic if ±90°, Mercator if 0°, and Lambert conformal conic else, chosen so as to minimize the scaling factor. A rotation of the domain may be imposed through the choice of the reference longitude.
The interface to MAKDO illustrates the choice of arguments.
Some unused and problematic options are suppressed :
and simple latitude × longitude domains are now handled outside EGGX.
The old version will be kept in the ARPEGE / ALADIN code, and the switch will be controlled by the sign of NGIV0 or NROTEQ (not yet decided).
To go further with cleaning, Ryad El Khatib, Jean-Daniel Gril, Denis Paradis and Jean-Marc Audoin will now examine how to optimize the "cadre" of ALADIN files.
|
Home |