run_roms_da data assimilation script

Discussion about tangent linear and adjoint models, variational data assimilation, and other related issues.

Moderators: arango, robertson

Post Reply
Message
Author
jpm
Posts: 21
Joined: Thu Aug 27, 2009 4:37 pm
Location: UCSC

run_roms_da data assimilation script

#1 Unread post by jpm »

I created a script called run_roms_da that may be useful for those performing ROMS data assimilation and running multiple consecutive 4DVar cycles.

This script has been created a while ago and is now used by several people in our UCSC data assimilation group. It has become more sophisticated over time and can do a lot more than the `submit_is4dvar.bash` script that comes with ROMS.

The highlights:
  • fully automated file renaming
  • starting of ROMS via mpirun or sbatch job submission
  • slicing of observation files to fit each 4DVar cycle
  • fully configurable using a single configuration file for each assimilation run (unlike `submit_is4dvar.bash`, the script itself is not edited by the user)
  • restart from failed runs
  • written fully in bash, it only requires an installation of the NCO toolkit (for ncks, ncap2, etc.)
  • optional compression of ROMS output after each cycle
While it can do a lot, this script does require a working ROMS 4DVar configuration (executable, forcing and observation files etc.). More information can be found on its github page.

Feel free to ask questions about it here, if I see some interest, I'll expand its documentation on github. If you are encountering issues, please post an issue on the github page page.
Jann Paul Mattern, Ocean Sciences Department, UCSC

Post Reply