Question about SST in WRF-ROMS coupling

[windysun02]

Hello everyone,

I'm trying to simulate a super-typhoon in the WRF-ROMS coupling model through COAWST, now I have something unclear, and need your help.

I have compared the results from the WRF-only simulation and the WRF-ROMS simulation, the results from the WRF-ROMS coupling model are inaccurate, the TC intensity is much weaker than the best-track data. By examining the model output, I found that the error came from the SST which forced the WRF model. When WRF is forced by the NCEP analysis SST in the WRF-only simulation, the result is acceptable. However, when WRF is forced by the SST transferred from ROMS in the WRF-ROMS simulation, the result is bad.

I compared the NCEP SST and the ROMS SST (figure1), it is denoted that the SST from the ROMS is about 2-3℃ colder than the NCEP SST, and a comparing with the OSTIA high resolution SST analysis data (http://podaac.jpl.nasa.gov/dataset/UKMO ... GLOB-OSTIA) showing that the ROMS SST is too cool indeed, making the TC intensity much weaker.

Then I compared the SST transferred to WRF (in the wrfout_d02 files) with the ROMS uppermost layer temperature (in the ocean_his.nc), they are entirely identical (figure2). My question is:

1. Does COAWST transfer the uppermost layer temperature of ROMS to WRF as the SST? I learned from the forum that the uppermost layer in the ROMS may be not the surface layer, it may be serveral meters under the surface, if the temperature of this layer is treated as SST, is it inaccurate because temperature of this layer may be several degrees cooler than the surface?

2. If I misunderstood it, and the SST transferred to WRF is treated in other way, is there any way to improve it? For the SST in my WRF-ROMS simulation is not accurate, I have tried turn on the QCORRECTION in the ROMS, but thing didn't goes better.

The project.h file of my case is in the attachments.

Thank you in advance.

Sun

[hpd14thu]

I heard when using ROMS simulated tyhoon process, the over-mixing often happens which results over-cooling. Maybe you can try to adjust parameters in ROMS mixing scheme.

[windysun02]

I heard when using ROMS simulated tyhoon process, the over-mixing often happens which results over-cooling. Maybe you can try to adjust parameters in ROMS mixing scheme.

Thank you for your suggestions, I'll do some adjustments on the mixing scheme to see if things get better.

[yuke4500]

Hello,
Have you solved this problem? I am running a WRF-ROMS coupling experiment too and I meet a similar problem. The SST seems 2-3 °C colder than the observation.

Best Regards
Yu

[jcwarner]

Looking back at the original post, i see that the SST from NCEP and the SST from ROMS are very different. So the 'coupling' is not the issue. It seems that the issue is "how did you set up the initial conditions for ROMS?" If ROMS is starting out 2-3 degrees cooler, then you cant expect similar results from the coupling with a warmer ocean.
For the coupling - I do transfer the temp from the top ocean level. If you want to explore other options, that is certainly possible: i do the transfers in mct_roms_wrf, you could compute a mix depth and avg over that, or whatever you want to explore. But i dont think that is the problem.

-j

[manhnt181]

Hi John,
I've practiced many cases related to simulate the SST value with WRF_CTL, WRF with updating SST data, COAWST model (WRF+ROMS+SWAN) in which the initial of WRF model is updated with SST data. The result at the initial time is quite good, but after 06-12h simulating, the SST of COAWST model is 2-3 degrees cooler than other cases leading the intensity of TCs is weaker than in the reality.
I think this issue may be come from the way I set up my BCs of SWAN model (in my case I use the constant data)
I've tried many times to run COAWST model with TPAR data which is created by using ww3_swan_input with partitions data, but it failed.
Can you kindly give me some suggestions to improve this situation?
Thank you very much!

[jcwarner]

something is making roms get 2-3 degrees cooler in 6 hour is quite extreme. I do not think it is the waves.
what if you ran roms by itself, with surface forcing from wrf or other source?

-j

[Joeailvyou]

Looking back at the original post, i see that the SST from NCEP and the SST from ROMS are very different. So the 'coupling' is not the issue. It seems that the issue is "how did you set up the initial conditions for ROMS?" If ROMS is starting out 2-3 degrees cooler, then you cant expect similar results from the coupling with a warmer ocean.
For the coupling - I do transfer the temp from the top ocean level. If you want to explore other options, that is certainly possible: i do the transfers in mct_roms_wrf, you could compute a mix depth and avg over that, or whatever you want to explore. But i dont think that is the problem.

-j

Dear jcwarner,
I am also confused with SST from ROMS, because I didn't see the option to output SST in the roms.in. The only option I see is Aout(idTvar) == T T ! temp, salt temperature and salinity. So, how to output SST ?
I am not sure what you mean by "top ocean level". If I have output temp(lon,lat,vertical_level) with vertical_level=35, do you mean temp(:,:,35) is SST ?

Looking forward from you soon!
Best wishes.