Temperature Difference from ROMS to HYCOM

Report or discuss software problems and other woes

Moderators: arango, robertson

Post Reply
Message
Author
chysun
Posts: 22
Joined: Sat Jun 04, 2016 1:00 pm
Location: SUN YAT-SEN UNIVERSITY

Temperature Difference from ROMS to HYCOM

#1 Post by chysun » Thu Aug 18, 2016 2:15 am

Hi,
I have a south china sea(SCS) modelling case using HYCOM data for ROMS initial field and boundary field.My temp field has a difference from HYCOM.

Configurations
The domain is 95~134E and 1~35N and we have spatial varying horizontal resolution from 10km to 5km.For vertical parameterization we take Vtrans=2,Vstr=4,theta_s=10,theta_b=2,hc=400,N=30;We also apply MY2.5 vertical turbulence scheme. NCEP reanalysis is provided for atmospheric forcing with BULK_FLUX defined.Please take a look at cpp file and input file attached for more details.

Results
We make some plots about temp difference from ROMS to HYCOM (say Troms-Thycom) at 50m and 100m to see what our vertical mixing is going on. Seeing in attachments,ROMS initial temp field at 50m is with no doubt similar to HYCOM.Then after 50 days ROMS temp field is cooler at some coastal slopes while at some basins it is warmer. The situation is developed 100 days after. Moreover,100m temp field indicates something like "upwelling" at slopes and "downwelling" at basins :roll: .But our SST is fine.

Has anyone got an idea about this situation?Or I just messed up some aspects of ROMS configurations? :?:
Any comments is appreciated!
Attachments
CHY_N30_arti_2013.in
(118.6 KiB) Downloaded 143 times
chy_bio.h
(2.15 KiB) Downloaded 149 times
HYCOM_DIFF_001_050m.png
HYCOM_DIFF_050_050m.png
HYCOM_DIFF_100_050m.png
HYCOM_DIFF_001_100m.png
HYCOM_DIFF_050_100m.png
HYCOM_DIFF_100_100m.png

chysun
Posts: 22
Joined: Sat Jun 04, 2016 1:00 pm
Location: SUN YAT-SEN UNIVERSITY

Re: Temperature Difference from ROMS to HYCOM

#2 Post by chysun » Mon Aug 22, 2016 10:22 am

Hi,
I checked Haney number (rx1) field and found it coincide with slopes where upwelling current appeared.I think my rx1 field has a close relationship to the fake upwelling current,but neither clear how HPGE leads to upwelling nor know how to decrease rx1 factor directly.

Has anyone experienced this before? Look forward to your comments!
Attachments
rx1_ini.png

chysun
Posts: 22
Joined: Sat Jun 04, 2016 1:00 pm
Location: SUN YAT-SEN UNIVERSITY

Re: Temperature Difference from ROMS to HYCOM

#3 Post by chysun » Fri Aug 26, 2016 1:30 am

Two days ago I smoothed the topography then I did a PGE test with no forcing
and made some simple cross-section figure to check if there is any false upwelling along the slope.

See attachments:
There is w velocity cross-section figure with its unit m/day at day30.The figure shows a strong upwelling at the deep slope and it apparently destroys
water stability.At this case my maximum rx0 factor is 0.3 and maximum rx1 factor is 7.7 which I think is reasonable ,
neither can I find much discussion about upwelling due to PGE from ROMS forum.Hence I'm not sure whether I should smooth more.
Has anyone encountered this situation before?

Hope for your participation!


Haiyun
Attachments
section_location.png
section_wvelocity.png
section_density.png
section_density.png (11.5 KiB) Viewed 5431 times
section_temp.png
sigma_error_test.in
(118.1 KiB) Downloaded 97 times
sigma_error_test.h
(1.57 KiB) Downloaded 98 times

rduran
Posts: 139
Joined: Fri Jan 08, 2010 7:22 pm
Location: Theiss Research

Re: Temperature Difference from ROMS to HYCOM

#4 Post by rduran » Fri Aug 26, 2016 5:30 am

What do the horizontal currents from the pge with no forcing look like? ROMS vertical velocity is related to the horizontal velocity, it is computed diagnostically by integrating the continuity equation. It seems that somehow near the slope you are getting divergence/convergence of horizontal currents produced by the pge (which will probably be strongest near the slope as well) which in turns results in vertical velocity. That looks like complicated terrain, is it causing horizontal divergence?

The only real way to know if rx0 and rx1 are reasonable is to test your application i.e. case by case basis.

chysun
Posts: 22
Joined: Sat Jun 04, 2016 1:00 pm
Location: SUN YAT-SEN UNIVERSITY

Re: Temperature Difference from ROMS to HYCOM

#5 Post by chysun » Sat Aug 27, 2016 9:09 am

Hi rduran,
Thank you for your reply.I make a gif of daily averaged velocity at bottom. The currents are developing all the time and they are obviously too strong.Yes I can see negative vorticity regions along the section line which cause upwelling,your notice is helpful!

I think the bottom velocity due to PGE is still too big (say >0.2m/s) and I have to decrease it more.BTW one thing I haven't made it clear yet is the "stuff" that goes with PGE as you mentioned at viewtopic.php?f=14&t=2841&p=10698&hilit ... rx1#p10698.

1.Is it means wave reflection,or momentum diffusion term? :?:
2.Can I set TNU2 == 0 and VISC2 ==0 to minimize momentum diffusion term if it is the "stuff"?

Cheers,
Haiyun
Attachments
bottom_velocity.gif
bottom_velocity.gif (1.42 MiB) Viewed 5333 times

User avatar
susonic
Posts: 160
Joined: Tue Aug 21, 2007 5:44 pm
Location: Jeju National University
Contact:

Re: Temperature Difference from ROMS to HYCOM

#6 Post by susonic » Sat Aug 27, 2016 12:24 pm

For your information, you'd better run the PGE test with homogeneous(no stratification).

User avatar
kate
Posts: 3780
Joined: Wed Jul 02, 2003 5:29 pm
Location: CFOS/UAF, USA

Re: Temperature Difference from ROMS to HYCOM

#7 Post by kate » Sat Aug 27, 2016 3:38 pm

susonic wrote:For your information, you'd better run the PGE test with homogeneous(no stratification).
You should get a nicer result that way, but the standard PGE test has a density structure that's the same everywhere, but more complex than linear. The errors should go away for homogeneous fluid or for linear stratification.

rduran
Posts: 139
Joined: Fri Jan 08, 2010 7:22 pm
Location: Theiss Research

Re: Temperature Difference from ROMS to HYCOM

#8 Post by rduran » Sun Aug 28, 2016 7:19 pm

1.Is it means wave reflection,or momentum diffusion term? :?:
2.Can I set TNU2 == 0 and VISC2 ==0 to minimize momentum diffusion term if it is the "stuff"?
Its been some time since I last checked for a pge, but I do have a plan for next time I check for it and I can share it with you, just keep in mind it is only an idea at this point and I have not actually tested this test; this would be my first next pge test from which I may or may not need to adjust:

In my previous test I closed my open boundaries, the slope in an eastern ocean boundary created false momentum from the pge that then traveled as a coastally trapped wave / Kelvin Wave around my domain (including the closed boundaries) to then add to the momentum being generated over the slope once it completed a full circle. Because of this I will probably close my boundaries AND add a strong sponge layer to kill anything arriving to the boundary.

I would set diffusion of momentum to zero, and even better than setting the diffusion of temperature and salinity to zero, I would fix them by using TS_FIXED suppressing evolution of density. I would use the mean temperature and mean salinity profiles for my region (averaged from the initial conditions or from the model providing boundary conditions). Using the same vertical profile in each grid cell of course, so that the only difference from cell to cell is how it is discretized. As Kate mentioned you need a density profile beyond constant/linear with depth. I would keep an eye on where the pycnocline (here pycnocline=max buoyancy frequency) is: Are the strongest pge currents near the pycnocline depth? if so, I might need to make sure the test will work for my application -- the pge could be dependent to where the pycnocline is, and therefore a different density profile might generate a different pge. The pycnocline depth would interact differently with the topographic features found at different depths.

I would set advection of momentum to zero (linear model).

And obviously wind through time and the initial SSH, u and v all set to zero.

This should hopefully isolate what can generate momentum to just the pge and limit any influence to the pge-generated momentum.


Not sure if a complicated bathymetry like yours may need additional considerations but I guess you will find out. If you can share what you find that would be great.

Hope that helps.


Rodrigo.

chysun
Posts: 22
Joined: Sat Jun 04, 2016 1:00 pm
Location: SUN YAT-SEN UNIVERSITY

Re: Temperature Difference from ROMS to HYCOM

#9 Post by chysun » Tue Sep 13, 2016 8:27 am

Hi Rodrigo,
Thank you for sharing experience about pge test.I get a better understanding of pge and its influence factor. :wink:
Now rx0 is limited to not more than 0.1 and I do the pge test following your idea (sponge,nonlinear temp profile,zero u,v,zeta,etc) The pge test shows improvement compared with the previous which is good news for me! Now I am running the model using this grid file with the hope that it will be the solution.

At this point my topography could be very smooth and slope shape is obviously changed . I worry that the hydrodynamic force has gone far from reality.
Later I will do another test that limit rx0 to not more than 0.2 which make a compromise between pge and model hydrodynamic force.
Attachments
rx0.1_temp_day_30.png
rx0.1_temp_day_30.png (11.32 KiB) Viewed 5080 times
rx0.1_w_velocity_day030.png
rx0.1_bottom_velocity.gif
rx0.1_bottom_velocity.gif (4.18 MiB) Viewed 5080 times

rduran
Posts: 139
Joined: Fri Jan 08, 2010 7:22 pm
Location: Theiss Research

Re: Temperature Difference from ROMS to HYCOM

#10 Post by rduran » Wed Sep 14, 2016 5:05 am

Hi,

Looking much much better (as far as spurious velocity is concerned), glad I could help. Possibly, you have reduced both terms on the right hand side of vel = PGE + stuff so who knows, perhaps a bit of a steeper topography is ok, probably a good idea to test with higher r.

Ocean models are simply not able to simulate the slope right now, that is just the way it is. We have to choose between the actual bathymetry or the actual zero pg with horizontally-constant stratification, can't have both. Z-level models don't have a pge problem but they have other problems notably a poor representation of the topography. Either way you cant have the actual slope without some potentially serious problems. And there is no good third vertical coordinate for a bottom (or surface) boundary layer, just those two.

Give it another 10 years or so and perhaps we can simulate the actual slope.

Post Reply