Strange Oscillation

[mengqingjun]

Recently I am similating a real-time ocean circulation case of East China Sea. I make initial condiations from HYCOM model on 2012.01.01 and the daily boundary condiations are also made from HYCOM model from 2012.01.01 to 2012.12.31. The forcing files are made from ECMWF datasets every 3 hours(using sustr,svstr,shflux,swrad,dQdSST,SST,SSS)(guiding by d_ecmwf2roms.m provided by Hernan G. Arango). The wind stress are on a regular lon/lat coordinate and I let ROMS do the interpolation internally(I #define CURVGRID and ROMS maybe rotate the wind stress to ROMS grid with whe variable angle). I output the results every 3 hours. But when I plot Total Velocity Vectors (cm/s) over the Potential Temperature (C) using ROMS plotting package. There are many strange oscillations over the whole model domain(especially in Japanese Sea...the velocity is going round and round...). I don't think it is because the model is unstable as the strange oscillations are always there and can not disapear all the year round. Maybe is it related to the wind stress...? But I checked the wind stress it seems resonable... Maybe does ROMS do the wrong interpolation of wind stress internally...? I don't know...Could anyone give me some hints ? Many Thanks!

ps.
year3a.gif is a gif animation of the Total Velocity over the Potential Temperature in the third model year(only providing the first seveal months to let the gif file not too big).(The reference time is 2012.01.01 so 2014 is the third model year.(I used cycle_length attribute for the forcing and boundary files))

[kate]

Yes, the winds can excite inertial oscillations.

[mengqingjun]

Thank you Kate for reply ! The wind can excite inertial oscillations, yes, but I don't think the strange oscillations(the velocity is going round and round...) in my application are resonable. The velocity showed dramatic changes....After three years, I think the barotropic adjustments caused by the wind stress have already finished and the model is stable in somehow. So, how do the strange oscillations(the velocity is going round and round...) come from ? Is it because the wind is high frequency(3 hourly) and there are many mutations in the wind ? Do you think the results I got are resonable ?
Many Thanks !

[LiangPeng]

I think it's inertial oscillation too, have you tested the frequency of the oscillation?

[mengqingjun]

Thank you LiangPeng for your reply! I did not test the oscillation because I think the oscillation is a little bit strange...
The inertial oscillation...um...The inertial oscillations fill all my domain? I have no idea...

[mengqingjun]

I have made a test forced only by the wind stress(3 hourly ecmwf) and output the results every 3 hour.
ps.
wind_stress_08th.gif is a gif animation of the wind stess in Aug;
Total_Velocity_Surface_Aug.gif is a gif animation of the surface Total Velocity in Aug.
The current looks like so unstable and there still exist the strange oscillations...Does it because the dramatic changes of wind stress ? I have no idea...
Anyone could help me ? Many thanks!
Best wishes!

[arango]

What do you mean by unstable? I don't see nothing unstable in your solution There are two types of instability: physical and numerical. For physical instabilities, you need to carry a more specific analysis that just plotting. There are various types of physical instabilities (barotropic, baroclinic, mixed, etc) and they are usually associated with eddies. I will check the Rossby radius of deformation for this area. The numerical instability can come from different sources and it usually makes the model to blow-up. I don't think that this is the case. What you see is caused by external forcing winds and/or boundary conditions.

Yes, those 3-hour winds can produce inertial oscillations. You can carried an spectral analysis to see that this is the case.

Notice that you have a lot stuff entering through the southern and eastern boundaries. In particular, look a the southeastern corner. If I was you, I will check your boundary data. HYCOM boundary data needs to be processed correctly to remove bias before can be use in ROMS. Plot your boundary data. In specific, check the T-S relationship for your boundary data. I usually do this over contours of density. You will be surprise of what you will find out...

[mengqingjun]

Thank you Professor Arango for your suggestions !
Firstly,about instability.

physical instabilities (barotropic, baroclinic, mixed, etc) and they are usually associated with eddies

.There are no obvious developing eddies in my domain,the truth is that there exist strange oscillations(the surface total velocity is going round and round...),so maybe they are not related to instablities,thank you for your guidance.
Secondly,about forcing and boundary condiations.

What you see is caused by external forcing winds and/or boundary conditions.

.Yes,I think this is really the problem.You suggested me to check the HYCOM boundary data.I think I will do as you suggested. I made a ideal numerical test, the model is only forced by the uniform north wind (wind_stress.gif), and no surface heat and salt flux, the initial and boundary condiations are not changed(HYCOM). And I plot the surface total velocity(),the strange oscillations gone !! Can I say that maybe there are no problem with my initial and booundary files ? And the problem comes from the 3-hourly ecmwf forcing ?

Yes, those 3-hour winds can produce inertial oscillations. You can carried an spectral analysis to see that this is the case.

I think the wind forcing need to be carefully checked. Could I carry an spectral analysis of the 3-hour wind stress and surface total velocity to this if there exist the same frequency ?
Why the 3-hour winds produce so many inertial oscillations(the surface total velocity is going round and round...) in all my domain ? Does it because the dramatic changes of wind stress cause these inertial oscillations? But this is not the case in the real ocean...
Thank You Arango again for your reply and could you give me any more advices ?