Blowing-up model - tide current+initial state(ocean current)

Report or discuss software problems and other woes

Moderators: arango, robertson

Post Reply
Message
Author
Rintei
Posts: 6
Joined: Mon Jan 21, 2019 2:45 pm
Location: SNR Technology

Blowing-up model - tide current+initial state(ocean current)

#1 Post by Rintei » Fri May 17, 2019 5:55 am

Hi. I want to reproduce the ocean currents in a wide area including Northeast Asia.
However, in the case of the existing algae condition(with harmonic constant as boundary condition), the model which has no problem starts interpreting HYCOM data and input the initial condition data (input ininame, undef ANA_INITIAL) The status is changed.
This situation is not only input HYCOM data, but also the same problem arises(it does not cause a problem when you insert the result(ocean_rst.nc) before the current time into the ininame. but when u and v factor are 0, the same problem arises).
Is this a matter of setting up my header file, or is it making the input file my mistake?
I would like to hear opinion.

Attached is the initial condition used in the model.

Code: Select all

/*
** Application flag:   test04
** Input script:       test04.in
*/
#define UV_ADV
#define UV_COR
#define UV_VIS4
#define MIX_S_UV
#define TS_DIF4
#define MIX_GEO_TS
#define SOLVE3D
#define UV_LOGDRAG
#undef LIMIT_BSTRESS  /* */
#define SALINITY
#define NONLIN_EOS
#define NONLINEAR      /* */
#define TS_U3HADVECTION
#define TS_C4VADVECTION
#define WET_DRY
#define DJ_GRADPS
#define SPLINES_VDIFF
#define SPLINES_VVISC
#define RI_SPLINES
#define CURVGRID
#define MASKING
#define VISC_GRID
#define DIFF_GRID

#define GLS_MIXING
#if defined GLS_MIXING || defined MY25_MIXING
# define KANTHA_CLAYSON
# define N2S2_HORAVG
# define CANUTO_A
# define CRAIG_BANNER
# define CHARNOK
#endif

#define STATIONS
#define BODYFORCE

#define ANA_MASK
#undef ANA_INITIAL      /* ++++ */
#define ANA_SMFLUX       /* surface momentum stress */
#define ANA_SRFLUX       /* surface shortwave radiation flux */
#undef SHORTWAVE        /* define solar radiation source term+++ */
#define ANA_SPFLUX       /* surface passive tracers fluxes */
#define ANA_STFLUX       /* surface temperature flux */
#define ANA_SSFLUX       /* surface salinity flux */
#define ANA_BPFLUX       /* bottom passive tracers fluxes */
#define ANA_BTFLUX       /* bottom temperature flux */
#define ANA_BSFLUX       /* bottom salinity flux */
#define ANA_TOBC         /* tracer boundary condition */
#define ANA_FSOBC        /* */
#define ANA_M2OBC        /* */
#define ANA_DRAG         /* */

#undef RAMP_TIDES
#define FSOBS_REDUCED

#define ADD_FSOBC        /* */

#undef BULK_FLUXES       /*++*/
#ifdef BULK_FLUXES
# define EMINUSP
# define LONGWAVE_OUT
#endif
#undef SOLAR_SOURCE
#undef DIURNAL_SRFLUX

#undef ANA_NUDGCOEF     /* TNUDGE option check+++ */

/*
#define UV_U3HADVECTION
#define UV_C4VADVECTION
*/
Attachments
ocean_his_0001.nc
blowup result file
(11 MiB) Downloaded 21 times
hns_ini200712_rev2.nc
HYCOM(temp, salt, u, v) + tide current
(8.58 MiB) Downloaded 24 times
grid03_re.nc
gridfile
(1.63 MiB) Downloaded 25 times
test04.in
input.in
(70.16 KiB) Downloaded 25 times
build.sh
compile
(19.13 KiB) Downloaded 21 times

User avatar
kate
Posts: 3713
Joined: Wed Jul 02, 2003 5:29 pm
Location: IMS/UAF, USA

Re: Blowing-up model - tide current+initial state(ocean curr

#2 Post by kate » Fri May 17, 2019 4:25 pm

You don't describe the problem you are having. What happens in your run? We don't provide a service of downloading all your files and rerunning your case (at least I don't).

Rintei
Posts: 6
Joined: Mon Jan 21, 2019 2:45 pm
Location: SNR Technology

Re: Blowing-up model - tide current+initial state(ocean curr

#3 Post by Rintei » Mon May 20, 2019 12:24 am

I did not attach the file in that sense, but I apologize if so.
As shown in the attached figure, the initial free surface value is calibrated with HYCOM data. However, as you can see in the results 10 minutes later, the free surface changes strangely. Immediately thereafter, the model will display an error message in the blowing-up state. In this case, I was just wondering if there were other input data issues or a more reliable header setting depending on the experience of others.
ImageImage
Attachments
result_log.txt
(67.78 KiB) Downloaded 21 times
zeta_t1.jpg
zeta_t1.jpg (820.03 KiB) Viewed 545 times
zeta_initial.jpg
zeta_initial.jpg (707.52 KiB) Viewed 545 times

fagundesmo
Posts: 40
Joined: Wed Dec 03, 2014 1:46 am
Location: University of Georgia

Re: Blowing-up model - tide current+initial state(ocean curr

#4 Post by fagundesmo » Tue May 21, 2019 2:26 pm

Hi,

I was checking your zeta for initial condition and I saw values of -1444 m and max value of 10 m. In my humble opinion as ROMS user, I think you might have not masked properly your land and you should check the region where zeta is 10 m, maybe? I am using HYCOM as well and my model is running smooth so far. I attached your figure (left panel) and my domain (right panel). I know we are in two different regions but take a look at how my land is masked and in yours the land doesn't show at all. I would check the other variables as well.

Cheers,
Matheus
Attachments
comparison.png

Rintei
Posts: 6
Joined: Mon Jan 21, 2019 2:45 pm
Location: SNR Technology

Re: Blowing-up model - tide current+initial state(ocean curr

#5 Post by Rintei » Wed May 22, 2019 4:40 am

Thank you very much for your feedback, Matheus.

Unfortunately, I tried to fix and restart zeta, but that did not improve. What I found is that the initial water temperature and salinity input values ​​return to their initial values ​​determined by the model after the run, because of my fault(or setting). Discussing similar issues earlier in forum, so I've been looking for something related to that part, but I have not found any answer with that part.

If you know about this, I would really appreciate it if you could advise.

fagundesmo
Posts: 40
Joined: Wed Dec 03, 2014 1:46 am
Location: University of Georgia

Re: Blowing-up model - tide current+initial state(ocean curr

#6 Post by fagundesmo » Wed May 22, 2019 1:11 pm

Hey Rintei,

Again this is all based on trial and error (Kate said that to me once ahah). I looked at your IC file and salinity and temperature are also kinda messed up. For example, figures 1 and 2 are from your IC file and 3 from my file (I'm kinda person that needs to visualize my data that's why I am plotting it to explain to you), Figure 1 is SSS plotted straight from your nc file. Figure 2, I masked it based on your mask_rho in your grid file. As you can see, when I masked it there's some places where it was supposed to have data but instead it was either 0 or nan values and therefore it was masked. So again based on my previous experience there's something funky with your interpolation.
salinity.png
Now, the funny part is that your u and v velocities are masked and interpolated just fine (take a look at Figs 4 and 5 for u component for surface and bottom, respectively).
u_comp.png
Are you using the matlab codes that they provide to interpolate your fields? Or pyroms? If not, I would check them out. If so, maybe downloading the codes again just to make sure you didn't modify anything crucial for the interpolation of salinity and temperature fields.

Give it a try and let us know,

Cheers,
Mat

Rintei
Posts: 6
Joined: Mon Jan 21, 2019 2:45 pm
Location: SNR Technology

Re: Blowing-up model - tide current+initial state(ocean curr

#7 Post by Rintei » Mon May 27, 2019 5:14 am

Hi, Matheus. I've looked back at the code you used through your answer, but there has not been a big change (I used d_initial.m to get the initial value, d_ecmwf2roms.m to get the forcing factor.).

I gave up trying to reproduce the tide currents and ocean currents at the same time, began to examine whether it could reproduce the ocean currents. At this way, It did not occur initial problem that was seen at the beginning, but this was because the vertical distribution(temp, salt) was moderately modified. However, it still appears to be much larger u,v vector than the real ocean current. The deeper the depth, the greater the impact.
Mean, 1 day(ocean_avg_0001.nc)
Mean, 1 day(ocean_avg_0001.nc)
Mean, 2 day(ocean_avg_0002.nc)
Mean, 2 day(ocean_avg_0002.nc)
(background color : SST)

Is this because my depth interpolation is too rough? Depth interpolation was done through the grid_builder, and interpolation in some areas proceeded directly. If this problem continues to occur, I would be thinking try to increase the number of grids to reproduce in more detail.

It is confusing whether this judgment is correct. If you have any advice on this, I will accept it humbly. Please let me know if there is anything you would like to tell me about this.

Thanks.

fagundesmo
Posts: 40
Joined: Wed Dec 03, 2014 1:46 am
Location: University of Georgia

Re: Blowing-up model - tide current+initial state(ocean curr

#8 Post by fagundesmo » Tue May 28, 2019 12:48 pm

Hey Rintei,

First, I am glad you figured it out the problem with salt and temp fields.

Secondly, about the tides we would normally create a file with tides only and I don't recall you attaching any tide files so I don't know. I use the matlab files create_roms_tides.m and roms_tide_forc_file.m. These two files they come when you download COAWST. I confess I started working with tides in ROMS only after I started working with COAWST so I don't know if those two files can be found in ROMS tools package (I would say probably yes but double check it). Even when I have the tide netcdf file I still have my currents as BC and IC.

For larger u,v currents, that's a little bit tricky just based on two plots and your description. If you could attach for us the log file of your run we would be able to check its Maximum grid stiffness ratios which is rx0 and rx1, for example (if you think your bathymetry is too rough). Also, the length of the run is also important since if you're showing us the first time steps I would say the model is still getting stable. Give the model a little bit of time. Now, if the model is blowing up that could be your time step. Or even, choosing adequately your turbulence closures, advection flags in your .h file, or your Lateral Boundary Condition Type. As you can see, there are many things that could be happening but without at least the log run of your model is kinda hard to figure it out. Again, as you and me other people have had the same or similar problems. At least every time I face a problem in ROMS I am able to find a similar solution just by googling it. I also learn a lot just by checking the log run it's really good at pointing out where the problem is and I normally copy the problem and try to google it.

Anyways, if you can send us the log run that would be useful to try to find some answers together.

Cheers,
Matheus

Rintei
Posts: 6
Joined: Mon Jan 21, 2019 2:45 pm
Location: SNR Technology

Re: Blowing-up model - tide current+initial state(ocean curr

#9 Post by Rintei » Wed May 29, 2019 12:38 am

Thank you for your advice.
I had never tried to create a log file, I would try to find the problem only on the terminal screen. Once I make the log file and look for the problem. I will post this process, soon.

Post Reply