Problem of abnormal cooling

[tara]

Dear ROMS Users,

I am facing a problem of abnormal cooling specially in the coastal region, to inquire about that, I have tried with different set of momentum forcings (daily as ell as 3 hourly)with all 4 side OBC those are relaxed to 3 days realistic observed boundary conditions.

Please suggest me , if anyone can get the cause of this problem.
Here with I am enclosing the all 2 experiments log and results too.

With Regards--
Tara

[kate]

MIX_GEO_TS Mixing of tracers along geopotential surfaces.
MIX_S_TS Mixing of tracers along constant S-surfaces.

You don't want both and you should never use the latter (except over a flat bottom).

TS_DIF2 Harmonic mixing of tracers.
TS_DIF4 Biharmonic mixing of tracers.

You have a zero coefficient for the first of these - you can save cycles by turning it off. You shouldn't need both viscosities either.

I would look at the horizontal velocities to see if they are reasonable. I would also look at the vertical structure - is that odd SST coming from vertical mixing or from surface fluxes? Extra cross-isopycnal horizontal diffusion? (see above)

[tara]

Thanks kate for your kind reply.

As you suggested, I will check with CPP options(MIX_GEO_TS) 1st. I have already plotted the currents and the wind stress forcings too. But I am not able to find any hint for this issue. I am attaching the current plot for the same year. I have tried two different forcing data sets just to cross check the impact of wind forcings on this issue by keeping all other forcings and inputs same.

further, as I understand for the TS_DIF2 Harmonic mixing of trace,TS_DIF4 Biharmonic mixing of tracers and UV_VISC2, UV_VISC4

The values are not taken from the .in file rather it is calculated by the model by using the cpp options provided in the header files either by #define VISC_GRID, #define DIFF_GRID or by #define TS_SMAGORINSKY , #define UV_SMAGORINSKY

If you have noticed in my earlier log files

the values of these diffusion and viscosity coefficients are taken as---

0.0000E+00 nl_tnu2(01) NLM Horizontal, harmonic mixing coefficient
(m2/s) for tracer 01: temp
0.0000E+00 nl_tnu2(02) NLM Horizontal, harmonic mixing coefficient
(m2/s) for tracer 02: salt
5.0000E+02 nl_tnu4(01) NLM Horizontal, biharmonic mixing coefficient
(m4/s) for tracer 01: temp
5.0000E+02 nl_tnu4(02) NLM Horizontal, biharmonic mixing coefficient
(m4/s) for tracer 02: salt
1.5000E+02 nl_visc2 NLM Horizontal, harmonic mixing coefficient
(m2/s) for momentum.
5.0000E+01 nl_visc4 NLM Horizontal, biharmonic mixing coefficient
(m4/s) for momentum.
1.0000E-06 Akt_bak(01) Background vertical mixing coefficient (m2/s)
for tracer 01: temp
1.0000E-06 Akt_bak(02) Background vertical mixing coefficient (m2/s)
for tracer 02: salt
1.0000E-05 Akv_bak Background vertical mixing coefficient (m2/s)
for momentum.



BUT LITTLE down side in log file it is taking the model calculated coefficients.


Metrics information for Grid 01:
===============================

Minimum X-grid spacing, DXmin = 1.34134527E+01 km
Maximum X-grid spacing, DXmax = 1.38725672E+01 km
Minimum Y-grid spacing, DYmin = 1.34248045E+01 km
Maximum Y-grid spacing, DYmax = 1.38717587E+01 km
Minimum Z-grid spacing, DZmin = 2.34393422E-01 m
Maximum Z-grid spacing, DZmax = 6.72160437E+02 m

Minimum barotropic Courant Number = 2.21857219E-03
Maximum barotropic Courant Number = 4.68428982E-02
Maximum Coriolis Courant Number = 2.45901795E-03

Horizontal mixing scaled by grid size, GRDMAX = 1.38721629E+01 km

Minimum horizontal diffusion coefficient = 1.11826062E+02 m2/s
Maximum horizontal diffusion coefficient = 1.15601358E+02 m2/s

Minimum horizontal viscosity coefficient = 1.11826062E+02 m2/s
Maximum horizontal viscosity coefficient = 1.15601358E+02 m2/s




With Regards--
Tara

[ymamoutos]

Greetings,

I suggest first to change the lateral boundary conditions
for zeta and ubar,vbar. Use chapman implicit if you are planning
to use flather for ubar,vbar or Chapman explicit for Shchepetkin.
As Kate already told you at that post viewtopic.php?f=31&t=4072
if your are using Smagorinsky inactivate TS_DIFF4 and UV_VIS4 and also DIFF_GRID and VISC_GRID.
Try to run without QCORRECTION. Also run using BULK_FLUXES activated and again
QCORRECTION inactivated. Another suggestion is to change the boundary conditions for your 3D fields (T,S,U,V)
from clamped to Radiation+Nudged. One last suggestion is for Jerlov water type that you are using. I think is propably wrong. Check here https://www.myroms.org/projects/src/ticket/609 .You use Open Pacific and your region is on Indian.

Giannis

[tara]

Thank You Very much Dear Ymamoutos,

I will try with your valuable suggestions but I have one doubt regarding OBC--,what I have choosen for my case is given below-

zeta-Flather
ubar- chapman Imp
vbar- chapman Imp

but I am little bit confused whether, as per your suggestion I need to alter the zeta boundary condition from flather to chapmam imp or Chapman explicit if I am using lateral boundary conditions for ubar and vbar as flather , but this is not the current case.

Actually I am trying to see the coastal upwelling
and domain is Indian Ocean. Can you suggest me the the compatible option of lateral boundary conditions.


One more doubt I have regarding the cppdef options--
MIX_GEO_TS, and MIX_GEO_UV
As per my discussion with Kate I should not use the MIX_S_TS option except for the flat bottom case but whether in between MIX_GEO_UV and MIX_S_UV also I should always give preference to Geo Potential(mixing along constant Z surfaces) option. and how these MIX_GEO_TS, and MIX_GEO_UV options are affecting or related with the harmonic as well as biharmonic diffusion and viscosity coefficients

With Regards-
Tara

[kate]

This is what I'm using:

Code: Select all

   LBC(isFsur) ==   Che     Che     Che     Che         ! free-surface
   LBC(isUbar) ==   Shc     Shc     Shc     Shc         ! 2D U-momentum
   LBC(isVbar) ==   Shc     Shc     Shc     Shc         ! 2D V-momentum
   LBC(isUvel) ==   RadNud  RadNud  RadNud  RadNud      ! 3D U-momentum
   LBC(isVvel) ==   RadNud  RadNud  RadNud  RadNud      ! 3D V-momentum

For the lateral mixing, it's important not to introduce cross-isopycnal mixing of tracers, so one has the choice of:

Code: Select all

** MIX_GEO_TS          use if mixing on geopotential (constant Z) surfaces   **
** MIX_ISO_TS          use if mixing on epineutral (constant RHO) surfaces   **

I use the former as being "good enough" while the latter is truly along isopycnals. It is less important to keep the viscosity operator on isopycnals - I just let it mix along s-surfaces (because it's cheaper).

[tara]

Thanks for your kind reply Kate,

I tried with your current suggestion, but I am not getting improvement in my results. As u remember in my previous setup I was getting cooling from the May to nov/oct. But I have run for 6 months, but the results are not upto the mark. Please see the attached figure as well as log_file.

Previous setup,what I was trying also I am attaching with log.


Please, suggest something to come out with this trouble

With Regards-
Tara

[kate]

I didn't think lateral mixing was going to be your only problem. It is time to look at your surface heat fluxes. You saved the averages of it - does it look reasonable? Again, what is the vertical structure of your cooling?

The first thing I notice in your log is:

Code: Select all

Resolution, Grid 01: 0079x0097x040,  Parallel Nodes:  36,  Tiling: 006x006

I doubt you need 36 cores for a problem this size. Communications are probably slowing you down, which you could check if you can get a profile report out of it.

Do you really need WET_DRY? I'd turn it off for now.

[ymamoutos]

Change your OBCFAC to a bigger number (ex. 120.0)
As Kate said check your surface heat fluxes and you
should turn off WET_DRY.I suggest to run with
BULK_FLUXES activated. Use TS_U3HADVECTION and
TS_C4VADVECTION for tracers.

Giannis

[kate]

You are applying heat fluxes without the QCORRECTION term (dQ/dSST). If the heat fluxes are too large upward, the model will keep losing heat without any way to compensate. You need to either have a dQdSST field for the QCORRECTION term or use BULK_FLUXES as suggested.