how to properly set the value of 'DSTART'?

[youngstor]

Hi ROMS,

I am getting a little confused by some interesting behavior of ROMS.

So I am driving the model with monthly forcing/clim. The 'ocean_time' has the following values:

Code: Select all

ocean_time =

    15
    45
    75
   105
   135
   165
   195
   225
   255
   285
   315
   345

But I noticed some strange behaviors when running the model. When I set in the ocean.in file :

Code: Select all

DT == 75.0d0
DSTART =  0.0d0

This model was able to run OK. But when I set the following values:

Code: Select all

DT == 30.0d0
DSTART =  0.0d0

the model simply skipped all the time steps, and showed the modeling was done:

Code: Select all

 Initial basin volumes: TotVolume =  3.8940977069E+13 m3
                        MinVolume =  1.5106446558E+05 m3
                        MaxVolume =  6.4468620546E+08 m3
                          Max/Min =  4.2676231169E+03

NL ROMS/TOMS: started time-stepping: (Grid: 01 TimeSteps: 00043201 - 00036864)

 Elapsed CPU time (seconds):

 Node   #  0 CPU:       2.108
 Total:                21.055

 Nonlinear model elapsed time profile:

  Initialization ...................................        16.522  (78.4697 %)
  OI data assimilation .............................         0.005  ( 0.0250 %)
  Reading of input data ............................         5.455  (25.9066 %)
  2D/3D coupling, vertical metrics .................         0.171  ( 0.8142 %)
  Omega vertical velocity ..........................         0.202  ( 0.9611 %)
  Equation of state for seawater ...................         0.407  ( 1.9333 %)
                                              Total:        22.763  108.1099

 Nonlinear model message Passage profile:

  Message Passage: 2D halo exchanges ...............         2.570  (12.2069 %)
  Message Passage: 3D halo exchanges ...............         0.913  ( 4.3349 %)
  Message Passage: data broadcast ..................         6.069  (28.8230 %)
  Message Passage: data reduction ..................         0.033  ( 0.1571 %)
  Message Passage: data scattering..................         5.257  (24.9665 %)
                                              Total:        14.841   70.4884

 All percentages are with respect to total time =           21.055

 ROMS/TOMS - Output NetCDF summary for Grid 01:

 Analytical header files used:

     ROMS/Functionals/ana_nudgcoef.h

 ROMS/TOMS: DONE... Thursday - September 4, 2014 -  4:18:20 PM.

But if I changed the setting to the following, the model again ran OK:

Code: Select all

DT == 30.0d0
DSTART =  15.0d0

I guess what I am really confused is how to properly set the value for 'DSTART'. Does it depend on 'DT'? Does it have to match the 'ocean_time' values?

What also confuses me is whether there is a standard for setting 'ocean_time' in ROMS? In some cases, I see 'ocean_time' has values like '56229' which is # of days since a certain time in history. In some cases, I see 'ocean_time' has values that equals the result of matlab function 'datenum' (e.g. 730625). In my case, I was using the matlab code written by a professor and 'ocean_time' simply has values of days in the middle of each month (15, 45, ..., 345).

Is it up to users to choose which date format for 'ocean_time' and 'DSTART'?

Any comments will be appreciated.

Xiufeng

[kate]

You can read up on TIME_REF at the bottom of the ocean.in file. The idea is that toy problems will have time since some zero time since it doesn't really have anything to do with a calendar while real problems with winds/tides from specific dates do need a calendar. Pick a time origin and be consistent.

I don't understand your troubles with ROMS, though. It should run NTIMES from the DSTART to the end. If you restart, you will need to increase NTIMES.