Strange outputs when using T_PASSIVE

[drivas]

Hello,
I just setup the "T_PASSIVE" option in my model (version 3.0) and is running. I'm using three different tracers, the three initial coordinates. In this example the boundary conditions are set to zero, but the results is the same when I set the values to the initial values.

Here I attach the initial condition of "dye_01" (Dye01_init.jpg) which is OK.


However, after 12 hours of running (Dye01_12h.jpg) the dye field has evolved to something strange: the contours look OK but the values are weird! Now the values are lower than they should be!


Do you have any idea about this problem?

Thanks,

DAVID

[kate]

How odd! Can you plot a vertical cross-section before and after? How about a timeseries at a station? I have recently tried a dye set to a uniform value and it doesn't change like what you are seeing, though it doesn't remain absolutely uniform either.

[drivas]

Thank you very much for your kind reply!

Here is a meridional section (at the center of the domain) at the initial time:


And now 12 hours later:


Here is a zonal section (again at the center of the domain) at the initial time:


And now 12 hours later:


What do you think?

[wilkin]

What are your open boundary conditions?

[drivas]

In this example, the open boundaries are set to zero, but in a previous experiment the boundaries were set to the initial boundaries (for example, at the western boundary the value was -132 always). Actually, I'm still confused about what I should use as boundary condition of a passive tracer! What do you think?

[wilkin]

By "open boundaries set to zero" do you mean they are clamped to be zero? That might account for the appearance of a weak boundary layer on the north boundary in your plot Dye01_12h.jpg. It would also drive some loss of initial tracer, but I don't think really accounts for the large scale loss you see.

Have you inadvertently activated CLIMATOLOGY and TCLM_NUDGING and ended up globally nudging your passive tracers toward zero?

By the way, the boundary condition options you choose for active tracers (temp,salt) are also applied to the passive tracers. So if you are clamping passive tracers you are also clamping temp/salt. Is that what you want to do for a physically well-posed problem?

[drivas]

Indeed, for active tracers I'm using options *_TNUDGING at the boundaries and also TCLM_NUDGING/TCLIMATOLOGY. So, how should I define the boundary conditions for my passive tracers?

[kate]

Now you know the power of TCLM_NUDGING. Are you sure you want to do that to your temperature?

Since all tracer boundary conditions are going to be the same, what do you want to do with T and S? Do you have a file with values? Then add the passive tracers to that file. I would set them to match the initial conditions. I've got an NCL script to read the temperature arrays and add one passive tracer to the boundary condition. However, I only provide two times for the dye and use bio_time for its time dimension (requires a change to varinfo.dat).

Hmm, I'm not seeing the attach a file button right now...

[drivas]

I'm including the Columbia River's plume in my simulations for the Oregon Coast. After numerous experiments, using TCLM_NUDGING was the only way I got the river leaving completely the domain (and thus not interacting with the boundaries and affecting the interior).
I'm not sure if I understand your suggestion, let me think...

[kate]

What exactly are you doing about the open boundaries? Ours started behaving better when we could provide relatively frequent boundary conditions from a larger model. Monthly Levitus just wasn't doing the job.

My comment about TCLM_NUDGING was to say that you are nudging pretty strongly as can be seen by the rapid loss of dye in just twelve hours. That strong a nudging to a T/S climatology is frowned upon in some circles and hard to defend in seminars.

[wilkin]

For what you are trying to do - using passive tracers to indicate the i,j,z source locations of your water masses - the rational values on the boundary (in the boundary conditions netcdf file) to be imposed by the (NORTH,SOUTH*)_NUDGING option (that augments the open boundary TS_RADIATION option radiation scheme) is clearly the same i,j,z values as on initialization.

The 3D passive tracer values provided either in the climatology netcdf file or functional ana_tclima.h should also have the appropriate i,j,z values. The actual nudging is activated by the TCLM_NUDGING flag, and the nudging time scale is set either by Tnudg values in ocean.in, or customized in the functional ana_nudgcoef.h.

You still haven't said what you want to do on the open boundaries for temp/salt, or given any details of the TCLM_NUDGING implementation if you have customized.

I'm afraid that at present you are locked in to having the same scheme for active (temp/salt) and passive tracers. So you should choose based on what the physics needs, and then figure out how to make the passive tracers behave under the same option.

Since you are rapidly losing passive tracer, it seems to me you must be nudging them to zero somewhere - I suspect in all 3 dimensions throughout the domain interior.

John.