Are there CPP opions for NetCDF input files ?

[cecilia]

I was looking at some files of the source code of ROMS 3.0....and there is a definition called INI_FILE (in initial.F) that calls the routine get_state.F used to read initial conditions of NetCDF files.....Also I found the definition with name FRC_FILE that is not used but there is a comment relating that name for opening NetCDF forcing, climatological and data assimilation input files.....

Those definitions are not found in SCRUM manual, neither in WIKI of ROMS.....

Then, I would like to ask for help to some experienced in ROMS about which cpp options are there for input NetCDF files (initialization, forcing, climatology, boundary) in ROMS 3.0 to build header files ?

Bye and thanks again

Cecilia

[cecilia]

Yes, I have confirmed that in the header file "globaldefs.h" there are several options conditioning the compiling of the model.

[cecilia]

I am having problems with the reading of my input NetCDF files.

I built my input NetCDF files adapting ROMSTOOLS Matlab scripters Make_QSCAT_daily.m, Make_OGCM.m, and all the associated.

I solved the problem of reading my initial condition input file changing the "scrum_time" name to "ocean_time" name.

Now I am having problems with my NetCDF forcing file, indicating the following problem:

GET_2DFLD - error while reading variable: sms_time at TIME index = 0

ROMS / TOMS - Input error...........exit_flag: 2

.
.
.
ERROR: Abnormal termination: NetCDF IINPUT
REASON: NetCDF: Index exceeds dimension bound


I have checked that the name of the variables (in create_forcing.m of ROMSTOOLS) are correct in my forcing file.

Someone could tell me what could be the problem and give me directions to solve it ?

Bye and have a good time

[kate]

It looks like your problem is here:

Code: Select all

GET_2DFLD - error while reading variable: sms_time at TIME index = 0 

Where index=0 is out of bounds. What are the times in your forcing file and what is the initial time in your model? What calendar are you using? Is the forcing file cyclic in time?

[cecilia]

Hi Kate,

a) I set dstart= 36798 in my input file (the first day of october 2000 minus the reference time of 1900-01-01).

b) In the forcing file I put daily winds for each day of october 2000 (31 days of data), with a ciclic = 0. Also, I put in the same file the climatological (monthly) heat flux and other forcing variables of COADS with a ciclic time equal 360. At first, I set each sms_time starting in 36798.5 in my forcing file, but then I changed it to 36798 (as in my input file)....but the model show the same error message in both situations.

In (b) I followed the example of make_QSCAT_daily, but I not sure if the structure of that kind of file forcing is compatible with ROMS 3.0.

Bye

[kate]

Winds or wind stress? sms_time goes with the wind stress. If you are providing winds, you need to compute the wind stress from it through BULK_FLUX or something similar. The model will then try to read winds and a bunch of other variables that go into computing wind stress.

[cecilia]

...windstress Kate

[kate]

Do you have a working debugger?

[cecilia]

I never had to use any....What is it for ?

[kate]

It's for debugging code... You compile with USE_DEBUG turned on, you fire up the debugger, and if all goes well, it will let you step through the code line by line and let you view the values of all the variables. It's best to try serial code first and you may have to fuss with getting it to read the ocean_xx.in file.

Alternately, you can always use print statements. The troubles are probably coming from get_cycle which figures out which record to read next.

[cecilia]

Ok. I do not know if there is a debugger for Intel Fortran of Linux (ifort)... I will ask for about that to our system administrator.

[cecilia]

Kate,

If I wish run the model for one month with mean daily windstress (from day 1 to 31) how I should define the cycle ? I mean what is the value for that variable in that case ?

[kate]

You don't want to have the cycle attribute on at all.

[cecilia]

Kate

Then I should delete it in my forcing file ? mmmmm now I understand....

In the scripters I used, the length cycle was defined as zero for windstress and 360 for COADS fluxes (others than momentum). And that scripters put the daily windstress data (31 days) and monthly Coads data (12 days, centered in each month of year) in the same forcing file. It maybe that format and attributes are not functioning with ROMS 3.0 also ...

[cecilia]

I was reading the files of the forcing package I downloaded from ROMS page.

There I found a condition telling that if Fcycle is greater than zero then define length cycle attribute for that variable.....

What is Fcycle ? It is a user switch ? or a data of the input file ?

As I have no examples of ascii and Netcdf input /output files it is being a mistery to try with this package.

[kate]

I don't have the package you are using. Fcycle is probably a user switch - you have to tell it if it's cyclic or not. COADS is and the daily winds are not.

[cecilia]

I dowloaded that forcing package (by H. Arango) from

http://www.myroms.org/index.php?page=forcing

but we could not compile it for my PC under linux (I asked for help to the system administrator) , it is maybe a problem with the architecture .

[prasanth]

hai,
I have a while running the model

Node #123 CPU: 15.552
Node #108 CPU: 15.561
Node #233 CPU: 15.563
ROMS/Functionals/ana_nudgcoef.h

ROMS/TOMS - Input error ............. exit_flag: 2


ERROR: Abnormal termination: NetCDF INPUT.
REASON: No error
Node # 25 CPU: 15.529
Node #259 CPU: 15.563
could you please help me.

[kate]

There might be more of a reason given above those CPU numbers. I can't tell from what you gave us.