progressive drift of temp, salinity and zeta

[coccines]

Dear All,

We are modelling circulation in a small domain (1.5 average horizontal resolution, 90*70 grid points) with atm and bry conditions fixed to those of the same day (boundary and initial conditions derive from numerous synoptic observations interpolated onto a regular 3D cube). After several tests, we found that the model performs better (i.e., simulated values remain closer to our 3D cube) by defining TCLIMATOLOGY and TCLM_NUDG. Currently, we set nudging parameters to 1 day (tracer and momentum) except for ZNUDG which is set to zero; the OBCFAC=5. Overall, the model seems to generate the circulation features we expect, however the average temperature calculated over the whole volume shows a progressive decrease (~4 °C in 18days), while the average salinity shows a progressive (but slower) increase (~0.1 PSU increase in 18days). Also zeta increases, and reaches unrealistic values (reaches >50m!). North bry is closed, west and south are open. After few tests, we found best results by setting the eastern bry (where the bathymetry <100m) to wall.

The first issue now, is to understand why there is a progressive drift in temp, salt and zeta despite the rather strong nudging and fixed boundary conditions (at each time step, all forcing and bry are identical). In particular, values drift more rapidly after the 7th modelling day (could there be any option that activates after reaching a certain threshold?). Do you have any suggestion that can help understand where to look for the causes of the progressive drifts in temp,salt and zeta? Are the nudging factors too strong? I have looked for publications presenting a similar approach (using ROMS for diagnostics), and that could be used as reference, but could not find any.

Your suggestions and comments will be greatly appreciated, as at this point it is hard to obtain significant improvements.

Thank you very much in advance for your time and help.

[uranoscopus]

For salinity and temp it could be that you are imposing every day heavy "winter" conditions, i.e. cold winds cooling the sea surface and tends to concentrate salt? The effect could cumulate in time as it is "fixed", despite you relax to climatology.

Zeta values however suggest that you have some trouble with your bondaries.
YOu could try a chapman at surface instead of fixed value (clamped?).
Cheers,
Antonio

[coccines]

Hello Antonio,
thank you very much for your reply. We found that in our settings the *volume conservation* option was not activated. now that it is defined, in one of our test runs, the variable ZETA stabilizes and does not increase as we had observed before. Seems like we are working in the right direction. I will now check if the tracers behave better than before.

As I mentioned in another post, we are now working with the TS_FIXED option, which allows the diagnostic condition we were looking for (that's why we had "fixed" all boundaries, and applied a strong 3D nudging). Are you familiar with the TS_FIXED option and/or could suggest any document/publication that has adopted it?

Again,thanks for your help!
Ines