Roms tools and inter-annual simulations

[fragoso]

Hi all. I'm using Romstools to create the initial, boundary and lateral conditions for ROMS 3.0. I have already changed some of the matlab scripts and managed to run ROMS 3.0 with climatological conditions (monthly winds, geostrophic currents as lateral boundary conditions, Levitus temperature and salinity climatologies ...). So far so good. I'm now trying to run an inter-annual simulation with 6-hour NCEP reanalysis winds, 10-day ECCO boundary conditions, tides from TPXO... The thing is, when I run the make_NCEP and make_OGCM scripts it creates one file per month (e.g ECCO_brasil_Y2003M1.nc, ECCO_brasil_Y2003M2.nc ... ECCO_brasil_Y2003M12.nc). As far as I know (and that is not much) ROMS 3.0 can't work with files like that. Is there someone out there who might help me with that??

Thanks in advance!

Regards,
Mauricio.

[kate]

How big are those files? You can concatenate them using NCO, the Netcdf Operators. You might be better off splitting them into one file per field, then concatenating those. To do multi-year runs, we typically make 13-month forcing files then make sure to switch from one file to the next during a January restart.

Then again, you might want to look into letting ROMS do the interpolation internally. It can do that for forcing files such as NCEP which are coarse, on a uniform lat/lon grid. Then the same global files can be used for any simulation you care to do. We've got such files for the CORE 1 forcing.

[fragoso]

Thank you very much for your answer Kate. Well, after some struggle, I managed to create a matlab script to concatenate the files. I´m curious now about what you said in respect to restart. I mean, do you run the model for one year period and do restart for the next. For instance, if I want to run a 10-year period do I have to restart 9 times? I´m also curious on how to make ROMS do the interpolations? Is there an option that I have to turn on? Sorry my ignorance...

Thanks again,
Mauricio
PROOCEANO - Brasil

[kate]

I run on systems which limit the run time for each job to something like 8-16 hours. I therefore need to restart numerous times per year for a large domain. I think I was getting 120 days per run the last time around, though I might have squeaked out a few extra days to make for three runs/year. It generates enough output that I have to deal with that about as often. The joys of shared supercomputers with quotas.

When ROMS reads forcing files, it checks the dimensions of them. If they don't match the dimensions of your ROMS grid, it will assume that you want it to do the interpolation for you. This only works if the forcing grid is strictly aligned with lat/lon lines. Also, if you want ROMS to interpolate and you happen to make a grid of exactly the same dimensions as your forcing file, it won't interpolate for you. Beware!

[fragoso]

Thanks again for the tips Kate. Well, ROMS never stops amazing me. I´m not used to all those fancy features! As for the restarts, I think I´ll try to run a 5-year long simulation and then restart to another 5 years. I don´t have these problems of shared supercomputers wit quotas. I have my own Mac mini sited on top of my desk (Precious, my precioussss). It probably takes 100 times longer than in a supercomputer, but it´s a start!

Regards,
Mauricio
PROOCEANO - Brasil

[magaesco]

Hi,
It's interesting your post, but I got a question, Kate.
Could be possible to add all files (that I got with make_NCEP and make_OGCM) instead of roms_frc.nc, roms_clim.nc, roms_bry.cn and roms_blk.nc?
It means that to edit the file roms_inter.in like this:

title:
ProyectoCMC-ESPOLEcuador
time_stepping: NTIMES dt[sec] NDTFAST NINFO
21600 1800 60 1
S-coord: THETA_S, THETA_B, Hc (m)
6.0d0 0.0d0 10.0d0
grid: filename
ROMS_FILES/roms_grd.nc
forcing: filename
ROMS_FILES/roms_frc_NCEP_Y2003M1.nc
ROMS_FILES/roms_frc_NCEP_Y2003M2.nc
......
ROMS_FILES/roms_frc_NCEP_Y2003M12.nc
bulk_forcing: filename
ROMS_FILES/roms_blk_NCEP_Y2003M1.nc
ROMS_FILES/roms_blk_NCEP_Y2003M2.nc
...
ROMS_FILES/roms_blk_NCEP_Y2003M12.nc
climatology: filename
ROMS_FILES/roms_clm_ECCO_Y2003M1.nc
ROMS_FILES/roms_clm_ECCO_Y2003M2.nc
...
ROMS_FILES/roms_clm_ECCO_Y2003M12.nc
boundary: filename
ROMS_FILES/roms_bry_ECCO_Y2003M1.nc
ROMS_FILES/roms_bry_ECCO_Y2003M2.nc
...
ROMS_FILES/roms_bry_ECCO_Y2003M12.nc
initial: NRREC filename
1
ROMS_FILES/roms_ini_ECCO_Y2003M1.nc
restart: NRST, NRPFRST / filename
21600 -1
ROMS_FILES/roms_rst.nc
history: LDEFHIS, NWRT, NRPFHIS / filename
T 21600 0
ROMS_FILES/roms_his.nc
averages: NTSAVG, NAVG, NRPFAVG / filename
1 21600 0
ROMS_FILES/roms_avg.nc


Thanks for your comments, They are so much appreciate.

Regards,

[magaesco]

Hi,
It's interesting your post, but I have a question with it, Kate
Could be possible to edit the files roms_frc.nc, roms_clim.nc, roms_blk.nc and roms_bry.nc with all the files generated with NCEP and OGCM for my period of interest(2003)?, for example
title:
Proyecto
time_stepping: NTIMES dt[sec] NDTFAST NINFO
21600 1800 60 1
S-coord: THETA_S, THETA_B, Hc (m)
6.0d0 0.0d0 10.0d0
grid: filename
ROMS_FILES/roms_grd.nc
forcing: filename
ROMS_FILES/roms_frc_NCEP_Y2003M1.nc
ROMS_FILES/roms_frc_NCEP_Y2003M2.nc
...
ROMS_FILES/roms_frc_NCEP_Y2003M12.nc


bulk_forcing: filename
ROMS_FILES/roms_blk_NCEP_Y2003M1.nc
ROMS_FILES/roms_blk_NCEP_Y2003M2.nc
...
ROMS_FILES/roms_blk_NCEP_Y2003M12.nc

climatology: filename
ROMS_FILES/roms_clm_ECCO_Y2003M1.nc
ROMS_FILES/roms_clm_ECCO_Y2003M2.nc
....
ROMS_FILES/roms_clm_ECCO_Y2003M12.nc

boundary: filename
ROMS_FILES/roms_bry_ECCO_Y2003M1.nc
ROMS_FILES/roms_bry_ECCO_Y2003M2.nc
....
ROMS_FILES/roms_bry_ECCO_Y2003M12.nc
initial: NRREC filename
1
ROMS_FILES/roms_ini_ECCO_Y2003M1.nc

restart: NRST, NRPFRST / filename
21600 -1
ROMS_FILES/roms_rst.nc

history: LDEFHIS, NWRT, NRPFHIS / filename
T 21600 0
ROMS_FILES/roms_his.nc

averages: NTSAVG, NAVG, NRPFAVG / filename
1 21600 0
ROMS_FILES/roms_avg.nc


All your comments will be so appreciate, thanks for advances

Maga
Ecuador

[kate]

I can see from the style of the input file that we are not using the same version of ROMS. If you are using an older ROMS from myroms.org, then you have to provide one and only one boundary file for each run. If you are using some other version, then I can't say what the limitations are.

Then there's my hacked branch of ROMS which can deal with a series of climatology files. One could see how those are handled and do the same for the boundary, etc. files.

[magaesco]

I can see from the style of the input file that we are not using the same version of ROMS. If you are using an older ROMS from myroms.org, then you have to provide one and only one boundary file for each run. If you are using some other version, then I can't say what the limitations are.

Thanks Kate, I can see, I've commited a mistake. My romstools is an old version.
My question is How can I run inter annual simulations, if when I use make_NCEP and make_OGCM, I don't get one bondary file or one forcing file, instead of them, I get so much files in this way roms_bry_ECCO_Y2000M1.nc roms_bry_ECCO_Y2000M2.nc roms_frc_NCEPxxx.nc roms_clim_ECCOxxxxx.nc roms_bulk_ECCOxxx.nc for my period. How can I get only one file when use NCEP and ECCO?

Or, maybe, I have to run one month and replace the file for example: roms_clim.nc with roms_clim_ECCO_Y2000MX.nc, and after I have to run the next month but replacing roms_ini.nc whith the last roms_rst.nc generated?

Sorry, I'm confused

Thanks Kate for your answer and help.
Maga

[kate]

I use ncrcat from the NCO package to string files together. How many depends on file size, etc.

[magaesco]

I use ncrcat from the NCO package to string files together. How many depends on file size, etc.

Thanks Kate