SSS damping (nn_sssr=1 or 2) when using the non-linear free surface?

Configurations Regional
1

Hi,
Until recently I was applying a SSS damping using the option nn_sssr=2 combined to the linear free surface. I’m using a regional configuration in which ocean and sea-ice components are coupled.

Now I’ve switched to the non-linear free surface, I am questioning about the good choice to apply a SSS damping.

As a reminder for the 2 SSS damping options in sbcssr.F90:
with nn_sssr=2 . both emp & qns terms are changed :
emp(ji,jj) = emp (ji,jj) + zerp
qns(ji,jj) = qns(ji,jj) - zerp
zerp being the correction term

with nn_sssr=1 only the sfx term is changed :
sfx(ji,jj) = sfx(ji,jj) + zerp

The surface temperature/salinity boundary condition in handled in trasbc.F90 :
! !== Now sbc tracer content fields ==!
DO_2D_OVR( nn_hls, nn_hls, nn_hls, nn_hls )
sbc_tsc(ji,jj,jp_tem) = r1_rho0_rcp * qns(ji,jj)
sbc_tsc(ji,jj,jp_sal) = r1_rho0 * sfx(ji,jj)
END_2D

In the case nn_sssr=2, the modified term emp is used to compute the SSH and also implies changes in the vertical scale factors e3t , so that « salt conservation » is ensured.
While in the case nn_sssr=1, only the term sfx is modified so that we directly add/remove salinity to/from the ocean, meaning no salt conservation …

From there my guess would be to still use the option nn_sssr=2 …. but not really sure about that.

In which case the option nn_sssr=1 would be useful ? It will help me to better understand.

What would be the « best way » to apply a SSS damping when using the non linear free surface ?

Thanks for your feedback.

Claude

2

I would say nn_ssr = 1 as you effectively want to correct the SSS i.e. you want to remove/add salt where needed. I am afraid that nn_ssr = 2 could have some side effects? but I don’t really know…
Try both :grin:

3

Ok, thanks Sébastien. I’ll try.
But does it mean that the option nn_sssr=2 should be only used when using the linear free surface ? If it’s the case, in the sbcssr.F90 the following condition should be used IF ( ln_linssh .AND. nn_sssr == 2 ) … But not sure at all.

4

I catch the “Boss” to get more details.

  • option 1: you directly modify the salinity through a “virtual” salt flux. You add/remove salt without changing the volume.
  • option 2: you modify the volume (changing emp) therefore you change the salinity (and the heat contain). You do not change the total salt mass. This solution corresponds to what happen in real life : you have precipitation/evaporation that change the volume and globally the total salt mass in the ocean is almost constant in time.

Having say that, in regional configuration, you are not supposed the conserve the total salt mass…

The Boss said: I don’t know what is the best solution for your configuration. maybe 1? you should try at lower resolution to do several tests… In your regional configuration, you should avoid using the fresh water budget control on top of the SSS restoring… Do you have an idea of the realism of your “observed” SSS under sea ice? Maybe you should not activate SSS restoring under sea ice?

In one word : no clear or good answer. Good luck… :stuck_out_tongue_winking_eye:

5

Thanks for the reply ! At least I know now that there is not an “easy” neither a good answer.
We do not apply any SSS damping under ice, just in free ice areas such as North Atlantic and GIN seas.
I’ve launched a test to see how looks like the differences between these 2 options in a regional configuration. To follow.

6

Result of the test: for a regional Arctic configuration the option nn_sssr=1, i.e. a virtual salt flux, does not allow to have a good Arctic dynamics over a 20-year long simulation. Especially the Beaufort Gyre which is just disappearing associated to an excessive warm and salty Atlantic water.
And to make it short: the option nn_sssr=2 is definitely the right choice for this configuration type. Thanks for the help !

7

Hello,

My answer comes way too late, since you’ve already performed the tests, but I still wanted to comment.

For ages, we’ve been using nn_sssr = 2 in global ocean-sea ice simulations with a non-linear free surface. I even believe that this value was recommended by the Boss (but I cannot remember for which physical reasons).

It is nice to see a validation of this choice with actual tests :slight_smile: