tracers(temp and salt )distributed noisy

[xiaocongM]

Greetings,
I have a question about the passive tracer(temp,salt):
My model horizontal resolution is 10 km. The distribution of the tracers seems unphysical. I assume it may be due to the visc coefficient. I set the value to 3.0 because i want to give the disturbances more time to fade out. The result is that there are many isolated points which are abnormally higher or lower than surrounded water points. When I set the value of visc larger, say 240, the phenomenon would disappear. However, the bigger value would damp the signal more rapidly which is not what i want. If my assumption is right, then how to set the value, or there is any inappropriate option in my .h file. I use the option as the following:

Code: Select all

# define SOLVE3D
# define UV_COR
# define UV_ADV
# define UV_LOGDRAG
# define UV_C4ADVECTION
# define DJ_GRADPS


/*                       Grid configuration */
# define CURVGRID
# define SPHERICAL
# define MASKING
# define SALINITY
# define NONLIN_EOS
# undef SPLINES
/*                         Surface Forcing */
# define BULK_FLUXES    
/*   turn ON or OFF bulk fluxes computation */
# define LONGWAVE_OUT 
# define EMINUSP         /* use rain and tmp to computing E-P */
# define SOLAR_SOURCE
# define QCORRECTION   
# define ANA_DQDSST
# define SCORRECTION

[jcwarner]

not sure what you mean by the visc values.

Are you using horiz viscosity:
#define UV_VIS2
#define MIX_S_UV
then you can set
VISC2 == 2.0d0 ! m2/s ??? or something like that

Perhaps the tracer horiz diffusivity is more what you need:
# define TS_DIF2
# define MIX_GEO_TS
TNU2 == 0.5d0 ! m2/s ??? or some value like that
-j

[xiaocongM]

not sure what you mean by the visc values.

Are you using horiz viscosity:
#define UV_VIS2
#define MIX_S_UV
then you can set
VISC2 == 2.0d0 ! m2/s ??? or something like that

Perhaps the tracer horiz diffusivity is more what you need:
# define TS_DIF2
# define MIX_GEO_TS
TNU2 == 0.5d0 ! m2/s ??? or some value like that
-j

Thanks for your prompt reply.Yes, I use the options as you mentioned. The grid cell Reynolds number may be too large. if U=1m/s,v=3.0,dx=10km, the grid cell Reynolds number is 3333, I should increase the v. On the other hand, the larger coefficent may damp out the oscillations quickly.
In addition, Ilıcak et al. (2012) suggest that “The use of a grid Reynolds number below 10 is a sensible and universal recommendation to avoid the saturation level of spurious mixing, but this would not guarantee that the remaining spurious mixing would be insignificant.”
I am confused about the suggestions, if the Reynolds number is 10, then the viscosity coefficient is 1000, which is too large, and i never see anyone using that large value in their model configuration. Do i misinterpret the idea?