Inquiry About Sea Surface Height Differences in HYCOM 2.3 vs 2.2.98

[Achref OTHMANI]

Dear HYCOM community,

I'm currently investigating why the sea surface height (SSH) appears significantly stronger in HYCOM version 2.3 compared to version 2.2.98. I’ve reviewed the release notes but haven’t found clear documentation on any changes in how SSH is computed between these versions.

If anyone is aware of specific updates or modifications in HYCOM 2.3 that could affect sea level or SSH calculations, I would greatly appreciate your insights.

Thank you in advance for your support.

Best regards,

Achref

[Alan Wallcraft]

The calculation of SSH has not changed, but it includes a contribution from bottom pressure anomaly and we now have a more careful treatment of 1-eta that might change this term.  In addition, epmass now correctly treats E-P as a water flux and that is possibility what you are seeing:

   2      'epmass' = E-P mass exchange flag (0=no,1=yes,2=river)

We have traditionally used epmass=0, which treats E-P (and rivers) as a salt flux.  Since HYCOM is non-Boussinesq the salt flux has some effect on SSH, but not as much as water flux.

This is most important for river plumes, and epmass=2 was relatively recently added to only treat rivers as water flux.  The problem with epmass=1 is that any annual bias in water flux will cause a trend is SSH, and epmass=2 is one way around this.

The figure is from an idealized river-only case.

[Achref OTHMANI]

Dear Alan,

Thank you for your helpful response and explanation regarding the changes in the treatment of epmass and its potential impact on SSH calculations.

In my case, I checked that epmass is still set to 0 in both HYCOM 2.2.98 and 2.3 runs, which should mean that E-P is being treated as a salt flux in both cases.

However, I’m still observing a noticeably stronger SSH signal in version 2.3 compared to 2.2.98.

Given that epmass hasn’t changed between my configurations, could the updated treatment of 1-eta or any other subtle changes in the vertical coordinate or pressure calculation be contributing significantly to this difference in SSH?

I’d appreciate any further clarification on what might be driving this discrepancy.

Thanks again for your time and support.

Best regards,
Achref

[Alan Wallcraft]

Please quantify "noticeably stronger"?  Are you looking at region-wide means (e.g. SSH in the .log file) or mean plots or snapshots?  Provide example plots or .log files.

Alan.

[Achref OTHMANI]

Hi Alan, 

We noticed that the winter mixing is much stronger (1st plot, average of jan-mar 2003-2008), that the top layer EKE is also stronger particularly on the shelf breaks at both boundaries (2nd plot, average of the year 2005) 

and that the area averaged SSH (open ocean only) is much more variable (3rd plot, NEMO line indicates GLORYS, which we use as lateral boundary condition for both runs).

Otherwise there is not much change in the sea ice (it's the same version of CICE and ESMF) and some changes of the river fluxes (probably not that important). 

Were there other changes from v2.2.98 to v2.3 (in the mixing or lateral boundary conditions, ...)? 

Thanks in advance

Achref

[Alan Wallcraft]

Is this your entire region?

If you have epmass=0, all the HYCOM basin wide SSH variation (except flow through the open boundaries) is steric, i.e. due to changes in the depth averaged potential density.  So far as I know NEMO is Bousinesq and so all its basin-wide SSH variation (except flow through the open boundaries) is due to water flux through the surface.

The attached PDF shows basin wide mean S and T and SSH, for a low resolution Arctic-only domain with no flow through its boundaries,  taken from the .log files:

   409121 (1915/001 00) mean      SSH (mm):  116.28  (-5.7E+03 to  2.6E+03)

   409121 (1915/001 00) region-wide mean Temperature:        2.4277687933
   409121 (1915/001 00) region-wide mean Salinity:          34.5939397493
   409121 (1915/001 00) region-wide mean Density Dev:        2.7568112429

I have not plotted Density Dev (w.r.t.1000+thbase) but it should give the cleanest steric signal.

In this case T is increasing and S is decreasing, both of which reduce density and so increase SSH.

I suggest plotting these time series.  My guess is that the 2.3 vs 2.2 SSH difference is due to mean T, and this might be from the deeper winter mixed layer.

Alan.