model time exceeds ending value for var

[uranoscopus]

Hi everybody,
I'm newby to roms.
I'm trying to provide IC and BC from a coarser model to my own implementation, with ana surface forcings.

While initial are passed correctly (I think), when bounds are checked the following error occurrs:

Code: Select all

 SET_NGFLD  - current model time exceeds ending value for variable: zeta_west
              TDAYS     =      15928.0000
              Data Tmin =      15927.0000  Data Tmax =      15938.0000
              Data Tstr =      15938.0000  Data Tend =      15927.0000
              TINTRP1   =      15928.0000  TINTRP2   =      15927.0000
              FAC1      =         -1.0000  FAC2      =          0.0000

My real problem is that I didn't really understand HOW and WHERE to set such a "model time" in my code.
I tried various different combinations of TSTART, TIME_REF and DSTART parameters in *.in file but the error is always the same.

Waiting for suggestions,
Regards,
AO

[kate]

What are the time values in your boundary condition file? Could you possibly have your records reversed, with the last time in the first record? Here is what one of my GET_NGFLD reports looks like:

Code: Select all

    GET_NGFLD   - free-surface western boundary condition,   t = 40939 00:00:00
                   (Rec=0000033, Index=2, File: HYCOM_Bdry_SCS_2012.nc)
                   (Tmin=      40907.0000 Tmax=      41143.0000)
                   (Min =  0.00000000E+00 Max =  1.04377458E+00)

[uranoscopus]

Hi kate,
records are not reversed. I had days but now I turn all the stuff in seconds.

I think I found a first error: I just realized that my "ocean_time" in initial conditions was wrong of 588 days (!!) and didn't correspond to the first record of *_time variables in bry.nc.
I fixed this but still doesn't work. Same error.
Now I have all in seconds (since 1968-5-23).

What about NTIMES? Should it start from 0 or from "my" zero?

these are the print for get_ngfld and set_ngfld:

Code: Select all

 NL ROMS/TOMS: started time-stepping: (Grid: 01 TimeSteps: 00000721 - 00001440)

 GET_NGFLD   - free-surface western boundary condition,   t = ***** 00:00:00
                   (Rec=0001, Index=1, File: roms_bry_mercator008_Y2012M1.nc)
                   (Tmin= 1376179200.0000 Tmax= 1376956800.0000)
                   (Min = -2.82085456E-01 Max = -5.98562482E-02)

 SET_NGFLD  - current model time exceeds ending value for variable: zeta_west
              TDAYS     =      15928.0000
              Data Tmin = 1376179200.0000  Data Tmax = 1376956800.0000
              Data Tstr = 1376956800.0000  Data Tend = 1376179200.0000
              TINTRP1   = 1376956790.0000  TINTRP2   = 1376179200.0000
              FAC1      = 1376163272.0000  FAC2      = ***************

[kate]

It looks like you were better off with days instead of seconds.

You have picked a time origin. That's good. You then set DSTART to correspond to the number of days between your time origin and the starting time of the current run. NTIMES is then the number of times to go from DSTART to your ending time. The idea is that on restart, you leave DSTART alone and increase NTIMES.

[uranoscopus]

OK,
now it's fixed.

There was a disagreement between "model time" (the ocean_time in the IC) and the time in the BC. Further, there was some problem with the old BC netcdf file (I don't know where extactly): I rewrote BC on a simpler manner (less attributes, variables ...etc...less stuff) and fixed the dates and now it ran 2 days.

Thanks,
A.