|
Next: Appendix A: Incremental vs. Up: Main page Previous: 4. Summary of actions 5. Further perspectivesThe rationalization planned above allows to use for each scheme more variants as possible in the current implementations, through a better integration into the code. However some possibilities have not been coded at all so far, and will be required at least for validation.
The LPC_FULL and LPC_TRAJ schemes should be made possible to use in 3-TL framework. This not of interest for operational use, but is needed for scientific validation, in the same sense that 3-TL SI is needed for scientific validation of 2-TL SI, in case of problems. For the same reasons, the LPC_FULL scheme should be made possible possible to use in the Eulerian context. As stated above, the interest of merging first- and second-order decentering code in LATTEX could be examined. A possibility of improved last iteration by recomputing some non-linear residual with the latest iterated variables could be coded if judged interesting. Finally, a similar harmonization of the physics call could be investigated. This would allow a complete flexibility in the introduction of diabatic terms into the time-step for both "models" (i.e. IFS and ARPEGE/Aladin) in a similar way. However, as stated above, this is a subject more vast than the scope of this action.
Next: Appendix A: Incremental vs. Up: Main page Previous: 4. Summary of actions Pierre BENARD 2002-06-17
|