[MITgcm-support] shelfice_thermodynamics.F crash

[Martin Losch]

Hi Ian,

it would be great if you could review the related PR 968 on github. Basically I copied the stic_solve4fluxes.F code and modified to avoid solving the quadratic equation for salinity S_b twice. I think this works, but it would be great, if you had a look.

Martin

> On 23. Jan 2026, at 20:54, Dimitris Menemenlis <dmenemenlis at gmail.com> wrote:
> 
> Thank you Ian.  Your notes are super helpful.  Kayhan and I did run into an ocean-salinity = 0 case during initialization of ice shelf cavities for an llc4320 set-up.  This is probably excessive initial melting and undershoot, as you suggest.
> 
> Dimitris
> 
> 
>> On Jan 23, 2026, at 10:46 AM, Fenty, Ian G (US 329B) <ian.fenty at jpl.nasa.gov> wrote:
>> 
>> Martin and Dimitris,
>> 
>> I may have missed this if the discussion, but if the ocean salinity is zero and the salinity of the ice is zero, then the assumptions of the three-equation model break down. In that case, there is no boundary layer at the ocean/ice interface. The melting expression simplifies to
>> 
>> Step 1, find the local melting point: 
>>       T_f = c_0 + b_0 p
>> 
>> Step 2, calculate the heat flux to the ice
>>       <w’T’> = \gamma_T (T_o - T_f)
>> 
>> Step 3, calculate the freshwater flux from melting:
>> Case a: no heat conduction through the ice
>>       \rho_w c_{\pw}  <w’T’> = -Lq 
>> 
>> Case b: heat conduction through the ice
>>       \rho_w c_{\pw}  <w’T’> + \rho_i c_{pi}  \kappa_i (T_i - Tf_f) D_i = -Lq 
>> 
>> q is the ice mass flux [kg s-1 m-2], which can be expressed in terms of velocity of water meltwater coming into the ocean (w_b) as
>> q = \rho_w_b
>> 
>> But yea, it’s hard to imagine the case where the model comes up with a solution of S_b = 0 in the original three equation model.  
>> It can only happen if the heat transfer coefficient is infinite, if the salinity diffusion is zero, or the ocean salinity is zero. 
>> If your ocean salinity is going to zero, it suggestions a different problem. Maybe a numerical artifact (undershoot/overshoot).
>> 
>> I attached my three-equation model notes.  From page 3 on the variable names (the epsilons) are from the shellfire code (or at least some older version of it)
>> 
>> Ian
>> 
>> 
>> 
>> From: MITgcm-support <mitgcm-support-bounces at mitgcm.org> on behalf of Martin Losch <martin.losch at awi.de>
>> Date: Thursday, January 22, 2026 at 10:43 PM
>> To: mitgcm-support at mitgcm.org <mitgcm-support at mitgcm.org>
>> Subject: [EXTERNAL] Re: [MITgcm-support] shelfice_thermodynamics.F crash
>> 
>> I don’t know. If both solutions give the same results, I’d go with SHELFICE_USE_HEATBALANCE_FOR_FRESHWATERFLUX defined, as this is in compact form what is implemented in pkg/steep_icecavity. I have no experience with this version. The other option is a hack that just avoids the division by zero, this is only physically correct if sLoc is really zero at the same time.
>> 
>> M.
>> 
>> On 22. Jan 2026, at 21:05, Dimitris Menemenlis <dimitris.menemenlis at sjsu.edu> wrote:
>> 
>> Done.  I left a review.  Both options work.
>> Which one do you recommend that Kayhan and I use?
>> with or without SHELFICE_USE_HEATBALANCE_FOR_FRESHWATERFLUX defined?
>> 
>> Thanks a million for super-quick turnaround on this request.
>> 
>> D.
>> 
>> 
>> On Jan 22, 2026, at 4:58 AM, Martin Losch <Martin.Losch at awi.de> wrote:
>> 
>> Please have a look at PR #968
>> https://github.com/MITgcm/MITgcm/pull/968 <https://urldefense.us/v3/__https://github.com/MITgcm/MITgcm/pull/968__;!!PvBDto6Hs4WbVuu7!NJMtDwoC-xNdirXVV-Z-UZfpysl7_c7oQrii5iuN41ypBQpoyGEGyCxwEQZW9w2OcmeUGO-LgP7acoh3NpYJVPbeiTE$>
>> 
>> @Dimitry, it would be great, if you could try this code in your little test, and also review it in github.
>> 
>> M
>> 
>> _______________________________________________
>> MITgcm-support mailing list
>> MITgcm-support at mitgcm.org
>> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support <https://urldefense.us/v3/__http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support__;!!PvBDto6Hs4WbVuu7!NJMtDwoC-xNdirXVV-Z-UZfpysl7_c7oQrii5iuN41ypBQpoyGEGyCxwEQZW9w2OcmeUGO-LgP7acoh3NpYJuh3-YM0$>
>> _______________________________________________
>> MITgcm-support mailing list
>> MITgcm-support at mitgcm.org
>> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support
> 
> <Three equation model v9.pdf>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.mitgcm.org/pipermail/mitgcm-support/attachments/20260126/99affa2e/attachment.html>