All related assimilation code is found in the source directory ASSIM.
The interface to the data assimilation is through the routine assim_surf_atmn. Input needed to this routine and the downstream call-tree is either given as arguments or via a normal initalization through the routine init_surf_atmn.
The assimilation is tile-independent. That means the individual tiles are treated separately and different assimilation alghoritms can be used for the different tiles. This documentation will explain the different options for the different tiles.
Inside the assimilation interface and below, no IO should be performed unless the IO is processor dependent and in that case the number of processors must be kept constant.
After the assimilation is performed the updated fields are written out through the normal SURFEX output routines (write_surf_atmn).
The assimilation is enabled in by setting LASSIM=.TRUE. together with the selected schemes and the started in either inline or offline mode.
Relevant settings in namelist NAM_ASSIM:
| Fortran name | Fortran type | values | default value |
| LASSIM | logical | .FALSE. | |
| NPRINTLEV | INTEGER | 0 |
From cy38h1 and cy40t1 it is (should/could) possible to run the assimilation in SURFEX in an inline mode. The interface to assim_surf_atmn is called at the end of CANARI after first initializing SURFEX the same way as for a normal forecast using SURFEX.
The result of the CANARI analysis exists (...)
The SURFEX offline data assimilation (SODA) was the first implementation of a unified assimilation in SURFEX.
The binary is called SODA and is a wrapper to do intialization and read the needed input data to be able to call the assimilation interface assim_surf_atm_n.
Namelist settings in (...)
RSS 2.0
