Main menu


Inhibit sea ice formation in a slab ocean through TocnFrz and depressT?

5 posts / 0 new
Last post
Inhibit sea ice formation in a slab ocean through TocnFrz and depressT?

Hi all,

I'd like to run an SOM compset without sea ice. The idea is to let the water supercool and set the freezing point of sea water so low, that the water stays cool but doesn't form sea ice (as done in this paper,

My set up is e.e104.E_1850_CN.f19_g16 and I changed depressT and Tocnfrz, both in cice4/ice_constants.F90 and cpl_share/ice_constants.F90 and set ATM_NCPL -val 96 and xndt_dyn  = 2 or 3.

Depending on how much I change the parameters, I get vertical thermo errors or flux conservation errors earlier or later (within the first year).

My hope was, that the sea ice would slowly melt away (I tried several CO2 levels) and that I subsquently could branch off the simulations I actually want to do. However, even if it runs for a few months, the sea ice doesn't seem to start melting.

Does anybody have a better idea/procedure how to get rid of the sea ice and then prevent it from forming?

Thanks in advance!



Not sure why on Earth you would want to do this, but I would recommend starting the CICE model with no ice. Set ice_ic to 'none' in the namelist. Depending on the version you are using, this will go in user_nl_cice or cice.buildnml.csh.




Thanks for the quick response! It didn't work... The sea ice comes back within a season. Could you point me to the place in the code where the ice actually forms? Might there be a better place/time then depressT and Tocnfrz? Could I enhance e.g. the melting?

I'm doing q-flux sensitivity tests and locally, changes in qdp cause very strong sea ice feedbacks. I also want to bridge the gap between an aquaplanet version without sea ice to the fully coupled version.



This gets into more of a discussion about the SOM model. I would recommed you read the following if you haven't already. The main issue I see is that the way the SOM works it boils down to a simple equation:

rho*cp*hmix*d(SST)/dt = Fnet + Qflux 

So, if you depress the freezing point for the SOM/CICE it will just keep adjusting the SST colder and colder if the atmosphere is colde enough. Eventually it would get cold enough and you would start growing ice again. Also, I'm not sure what it means to use the latent and sensible heat bulk forumulae when the SST drops below -1.8C. Another approach to consider is to artificially add more heat to Qflux to inhibit ice growth. However, this will end up warming the SST to perhaps unrealistic values. Yet another idea is to just put a lower bound on the SST in the SOM. This would not conserve energy of course. I don't really have an easy fix for this.


This equation is used to generate the q flux but then it is in general assumed to be constant. I test this assumption, i.e. my qdp forcing is based on fully coupled run SSTs (using the equation above), but then I change it locally, i.e. force the SST to adjust. Adding heat works fine, but taking out more of course generates a lot of sea ice (also under very high CO2 scenarios). Thanks for the ideas! I'll ask further in the SOM section.

Log in or register to post comments

Who's new

  • m.kliphuis@...
  • ddc3061993@...
  • hui.ding@...
  • zhouc@...
  • arianna.valmass...