ERA5_wind_problem

[AMIRHOSSEIN]

Hi
The tide performed well in the model and the zeta parameter fluctuated well. But I have a new problem, which is that when
I add wind files to the model and run the model, I encounter errors such as max speed or max density. However, by making the
time step smaller, I get the same error and the error appears again when I change the boundary condition too. In another part of the forum,
there was a topic with a link that I put below, and I used it a lot to give wind files to the model.
(I put this link for more useful for other friends)

viewtopic.php?p=22234&hilit=Found+Error ... 481#p22234

For wind files, I used the link below, and the modified Python file for receiving wind data is attached below.
Can the errors be caused by wind files or is the problem in the input file? Is the problem due to the resolution of the wind data?

https://github.com/trondkr/ERA5-ROMS


Thank you for your valuable answers

[bilge.tutak]

Hey Amir,

The netCDF file you posted only contains the net longwave radiation flux, but looking at your log files, it seems that the files are not the problem.

One thing I try to do especially if you are starting from the zero SSH condition, is to ramp up the winds. If on the first time step of the simulation,
you have a very strong wind case, that might quickly deteriorate your results.

- Your eastern/southern SSH boundary values are rather low 10^-4 (which is on the order of 0.1 mm). I don't think that is a meaningful boundary condition

Code: Select all

  GET_NGFLD_NF90   - free-surface southern boundary condition,            2013-01-01 00:12:00.00
                      (Grid= 01, Rec=3, Index=2, File: persian_bry.nc)
                      (Tmin=          0.0000 Tmax=         36.5958)   t =          0.0083
                      (Min = -8.61355162E-04 Max = -8.37803353E-04)
  GET_NGFLD_NF90   - 2D u-momentum eastern boundary condition,            2013-01-01 00:12:00.00
                      (Grid= 01, Rec=3, Index=2, File: persian_bry.nc)
                      (Tmin=          0.0000 Tmax=         36.5958)   t =          0.0083
                      (Min = -9.25587490E-04 Max = -8.46367620E-04)

- Your both temperature and salinity open boundaries have min/max values of 0

Code: Select all

  GET_NGFLD_NF90   - temperature eastern boundary condition,              2013-01-01 00:12:00.00
                      (Grid= 01, Rec=3, Index=2, File: persian_bry.nc)
                      (Tmin=          0.0000 Tmax=         36.5958)   t =          0.0083
                      (Min =  0.00000000E+00 Max =  0.00000000E+00)
  GET_NGFLD_NF90   - salinity eastern boundary condition,                 2013-01-01 00:12:00.00
                      (Grid= 01, Rec=3, Index=2, File: persian_bry.nc)
                      (Tmin=          0.0000 Tmax=         36.5958)   t =          0.0083
                      (Min =  0.00000000E+00 Max =  0.00000000E+00)
  GET_NGFLD_NF90   - temperature southern boundary condition,             2013-01-01 00:12:00.00
                      (Grid= 01, Rec=3, Index=2, File: persian_bry.nc)
                      (Tmin=          0.0000 Tmax=         36.5958)   t =          0.0083
                      (Min =  0.00000000E+00 Max =  0.00000000E+00)
  GET_NGFLD_NF90   - salinity southern boundary condition,                2013-01-01 00:12:00.00
                      (Grid= 01, Rec=3, Index=2, File: persian_bry.nc)
                      (Tmin=          0.0000 Tmax=         36.5958)   t =          0.0083
                      (Min =  0.00000000E+00 Max =  0.00000000E+00)

It seems to me that you might have a problem with the boundary conditions, rather than the ERA5 files.

Bilge.

[wilkin]

User Bilge Tutak may be correct. Your temp and salt open boundary data are both zero for all times, which is something you need to fix.

Code: Select all

 GET_NGFLD_NF90   - temperature eastern boundary condition,              2013-01-01 00:18:00.00
                      (Grid= 01, Rec=4, Index=1, File: persian_bry.nc)
                      (Tmin=          0.0000 Tmax=         36.5958)   t =          0.0125
                      (Min =  0.00000000E+00 Max =  0.00000000E+00)
  GET_NGFLD_NF90   - salinity eastern boundary condition,                 2013-01-01 00:18:00.00
                      (Grid= 01, Rec=4, Index=1, File: persian_bry.nc)
                      (Tmin=          0.0000 Tmax=         36.5958)   t =          0.0125
                      (Min =  0.00000000E+00 Max =  0.00000000E+00)
  GET_NGFLD_NF90   - temperature southern boundary condition,             2013-01-01 00:18:00.00
                      (Grid= 01, Rec=4, Index=1, File: persian_bry.nc)
                      (Tmin=          0.0000 Tmax=         36.5958)   t =          0.0125
                      (Min =  0.00000000E+00 Max =  0.00000000E+00)
  GET_NGFLD_NF90   - salinity southern boundary condition,                2013-01-01 00:18:00.00
                      (Grid= 01, Rec=4, Index=1, File: persian_bry.nc)
                      (Tmin=          0.0000 Tmax=         36.5958)   t =          0.0125
                      (Min =  0.00000000E+00 Max =  0.00000000E+00)
                      

But in your log file myChaGra.log I see you have open boundary conditions options for temp and salt that are:

Code: Select all

 Variable               Grid  West Edge    South Edge   East Edge    North Edge
 ---------              ----  ----------   ----------   ----------   ----------
 temp                     1   Closed       Flather      Flather      Closed
 salt                     1   Closed       Flather      Flather      Closed

FLATHER is not a valid option for tracers. It is only for 2D momentum. Valid choices for tracers are closed, clamped, radiation, or RadNud (radiation and nudging).

Nevertheless, your log indicates the open boundary data were read, so I'm not sure what option you actually got (but it wasn't Flather).

Notice, also, that ROMS reports the grid indices where the maximum values are occurring.

Showing the header lines that explain the information typed on every time step, and then your myChaGra.log file just before blow-up, there is

Code: Select all

  
  TIME-STEP YYYY-MM-DD hh:mm:ss.ss  KINETIC_ENRG   POTEN_ENRG    TOTAL_ENRG    NET_VOLUME
                     C => (i,j,k)       Cu            Cv            Cw         Max Speed
... 
       800 2013-01-01 02:13:20.00  1.107350E-01  1.266630E+04  1.266641E+04  2.367520E+14
                     (045,299,10)  6.443446E-03  5.474768E-03  9.687098E-02  1.126712E+01
       801 2013-01-01 02:13:30.00  1.109505E-01  1.266631E+04  1.266642E+04  2.367520E+14
                     (045,299,10)  7.317630E-03  5.485810E-03  9.735145E-02  1.127486E+01
       802 2013-01-01 02:13:40.00  1.111849E-01  1.266631E+04  1.266642E+04  2.367520E+14
                     (045,299,10)  6.547019E-02  5.155013E-03  1.051166E-01  5.193000E+01
 Found Error: 01   Line: 321      Source: ROMS/Nonlinear/main3d.F
 Found Error: 01   Line: 298      Source: ROMS/Drivers/nl_roms.h, ROMS_run

So it is blowing up somewhere in the middle in the i-direction at i=045, at j = 299 which is your northern boundary, at the surface (k=10). I would look for issues along the northern boundary with your grid or bathymetry that might be causing problems.

But fix your open boundary conditions, and open boundary data first.

[AMIRHOSSEIN]

Hi bilge.tutak and dear wilkin,

Thank you very much for your valuable guidance. I will check them, and I will announce the result.

[AMIRHOSSEIN]

Hi dear wilkin,
Did you mean to change the temperature and salinity data so that they should not be zero?
I assume that salinity and temperature are closed in the new implementation and I did not
provide any data for them. I also corrected the open boundary condition but got the same error again.
Once again, using MATLAB, I assumed a minimum depth of 20 meters (in the grid file) for
the entire range, and this made the model run more, but again, like the previous model error, it stopped.

Code: Select all

 4653 2013-01-01 12:55:30.00  1.002121E+00  1.202943E+04  1.203043E+04  2.493155E+14
                     (411,097,10)  5.894537E-04  2.799246E-03  9.955804E+00  1.452783E+01
      4654 2013-01-01 12:55:40.00  9.619426E-01  1.202957E+04  1.203053E+04  2.493162E+14
                     (354,059,10)  4.401671E-03  8.777663E-04  9.272842E+00  1.735941E+01
      4655 2013-01-01 12:55:50.00  9.365775E-01  1.202966E+04  1.203060E+04  2.493167E+14
                     (339,070,10)  8.058252E-03  9.768815E-03  1.050440E+01  2.014264E+01
 Found Error: 01   Line: 321      Source: ROMS/Nonlinear/main3d.F
 Found Error: 01   Line: 298      Source: ROMS/Drivers/nl_roms.h, ROMS_run

 Blowing-up: Saving latest model state into  RESTART file
     REASON: MaxSpeed =  2.0E+01

  WRT_RST_NF90     - wrote re-start    fields (Index=1,2) in record = 3

[wilkin]

You have Rad + Nud boundary conditions for u,v but your logfile indicates the nuding time scale is zero. So you have only radiation, but that's probably OK. Boundary velocities will be controlled more by 2D Flather. I would just use Rad for u,v anyway.

But if boundaries are open for u,v I don't think it makes much sense to choose closed for temp,salt on the same boundary.

You have ANA_VMIX ... a simple uniform vertical turbulent viscosity and diffusivity, and the values are small. This may be allowing strong acceleration in the surface layer if momentum is not able to mix down effectively. You may also be getting statically unstable stratification. A more realistic vertical turbulence closure scheme like GLS_MIXING would likely perform more robustly.

MIX_S_TS with sloping s-coordaintes (you have variable topography) will induce spurious vertical mixing. MIX_GEO_TS would be better.

You have quite small values of horizontal viscosity (visc2) of 5 m2/s, which for your grid size (2800 m) gives a grid Reynolds number (assuming U = 1 m/s) of dx.U/visc2 = 280. So the visc2 is not especially large. Look at your solution and if it is very noisy you might consider increasing the viscosity.

[AMIRHOSSEIN]

Dear wilkin,
Thank you for sharing your experiences.
As you said, I changed the settings, except that I have not yet set the salinity and temperature to the boundary. When I increased the visc2, I still did not see any noticeable changes and there are still MaxDensity and MaxSpeed errors in the northwest part of the domain. By changing the header and input many times, I get an error again in the same part.

[AMIRHOSSEIN]

Hi Wilkin,
As you said, I use Rad for u and v, and RadNud for the free surface(but my nudging time scale is zero). Although the boundary file is read,
it does not affect anything inside the domain. It is as if the edge of the border is stuck(It is visible in 2D mode) and does not allow it to be applied properly inside the domain.

RadNud.JPG (26.84 KiB) Viewed 13927 times

In another case, I considered the Cha for free surface and Rad for u and v, in which the boundary file was not recognized in the output.

As I read in this forum, several modes may be considered for OBC and all of them are correct.
My question is how do I select OBC to read my boundary file correctly?

[wilkin]

The boundary condition options are described here: https://www.myroms.org/wiki/Boundary_Conditions

Only the Clamped and Flather conditions use boundary zeta.
Only the Flather condition uses boundary ubar/vbar.

If you choose to continue this discussion, please start another thread. This topic has nothing to do with ERA5 winds.

[AMIRHOSSEIN]

very well. Thanks for your instant response, wilkin.