Temperature significantly near land areas

General scientific issues regarding ROMS

Moderators: arango, robertson

Post Reply
Message
Author
hoangha
Posts: 8
Joined: Fri Nov 17, 2017 5:02 pm
Location: Dongguk university

Temperature significantly near land areas

#1 Post by hoangha » Sun Nov 11, 2018 8:32 am

Hello

I run for real domain in East Sea of Vietnam with resolution of 9 km, 12 vertical layers.
+ Oceanic variables including temperature, salt, zeta, u and v getting from HYCOM.
+ All atmospheric variables including 'Uwind','Vwind','Pair','Tair','Qair','rain','swrad','lwrad','bhflux','bwflux','swflux'..are set 0 (no interaction with atmospheric)

However, just after 0.5 hour of simulation, temperature significantly increase only near land areas (Fig.0.5 hour. jpg). It is quite weird.

(even, I ]provide real atmospheric variables after running from WRF, temperature still significantly increase only near land areas)

Could you give me any comments?

Thanks
Attachments
0_hour.JPG
0_hour.JPG (42.66 KiB) Viewed 1061 times
0.5_hour.JPG
0.5_hour.JPG (39.29 KiB) Viewed 1061 times
proj_mother.h
(2.15 KiB) Downloaded 64 times
ocean_mother.in
(132.06 KiB) Downloaded 69 times

jcwarner
Posts: 855
Joined: Wed Dec 31, 2003 6:16 pm
Location: USGS, USA

Re: Temperature significantly near land areas

#2 Post by jcwarner » Mon Nov 12, 2018 12:23 am

a couple of things jump out here.
- first it is best to put temp scales the same so we can see the differences.
- second it seems a lot of cooling on east side, looks like a BC issue.
- for the temp spikes near land: how is wrf data interpolated to roms grid? we had issues with the coupling at first, but then realized that wrf heat and momentum fluxes on land can not be used to interpolate to the ocean. The wrf and roms land/sea masking is most likely different, so you have to use only wrf ocean points to interpolate to roms.
- when you have atm fluxes to 0, do you nudge to hycom data ? how is hycom data interpolated to roms grid?

-john

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

Re: Temperature significantly near land areas

#3 Post by susonic » Mon Nov 12, 2018 6:27 am

I think your set up for boundary condition is not common set up.

Why don't you give below boundary condition and see if the result change.

For free surface, Chapman
2d momentum(Ubar, Vbar), Flather
3d momentum(Uvel, Vvel), Clamp
MTKE, Gradient
Tvar (TS), Clamp

-JH

hoangha
Posts: 8
Joined: Fri Nov 17, 2017 5:02 pm
Location: Dongguk university

Re: Temperature significantly near land areas

#4 Post by hoangha » Sat Nov 17, 2018 2:53 am

Thank you very much for your useful comments.

I found above problem was caused by "mask" itself. when I created "grid file" with new version of Matlab (2014a) that problem is gone.
I still do not know why it happen because when I check mask value of both "creating by old version of Matlab, 2013a" (proplem one))and "creating by new version of Matlab, 2014a") mask is exactly the same.

The Matlab code bellow I use to write "mask" in "grid file""

nw{'mask_rho'} = ncdouble('eta_rho', 'xi_rho');
nw{'mask_rho'}.long_name = ncchar('mask on RHO-points');
nw{'mask_rho'}.long_name = 'mask on RHO-points';
nw{'mask_rho'}.option_0 = ncchar('land');
nw{'mask_rho'}.option_0 = 'land';
nw{'mask_rho'}.option_1 = ncchar('water');
nw{'mask_rho'}.option_1 = 'water';


Many thanks

User avatar
corvianawatie
Posts: 25
Joined: Thu May 14, 2015 4:50 pm
Location: Indonesia
Contact:

Re: Temperature significantly near land areas

#5 Post by corvianawatie » Sat Apr 27, 2019 7:51 am

hoangha wrote:Thank you very much for your useful comments.

I found above problem was caused by "mask" itself. when I created "grid file" with new version of Matlab (2014a) that problem is gone.
I still do not know why it happen because when I check mask value of both "creating by old version of Matlab, 2013a" (proplem one))and "creating by new version of Matlab, 2014a") mask is exactly the same.

The Matlab code bellow I use to write "mask" in "grid file""

nw{'mask_rho'} = ncdouble('eta_rho', 'xi_rho');
nw{'mask_rho'}.long_name = ncchar('mask on RHO-points');
nw{'mask_rho'}.long_name = 'mask on RHO-points';
nw{'mask_rho'}.option_0 = ncchar('land');
nw{'mask_rho'}.option_0 = 'land';
nw{'mask_rho'}.option_1 = ncchar('water');
nw{'mask_rho'}.option_1 = 'water';


Many thanks
Dear all, I experienced the same problem. I tried to run using analytical temperature and salinity (3D), so the model domain have the same temperature and salinity value (14 degC and 35 PSU). I put no forcing at all. However, there are abnormal increasing & decreasing both on temperature and salinity value. I didnt turn on CPP NONLIN_EOS. Also, i thought that I have used the correct masking grid. I build the grid using GRIDBUILDER.

What might be the problem?

Post Reply