Opened 4 weeks ago

Last modified 13 days ago

#2407 new Defect

ICB+SI3: blow up under specific circumstances in very shallow cell

Reported by: mathiot Owned by: mathiot
Priority: low Milestone:
Component: ICB Version: 4.0-HEAD
Severity: minor Keywords: ICB, SI3,



MetO eORCA12 (NEMO v4) blows up after few years along Antarctic coast in a very shallow location (10m) because of ice thickness larger than 99m.


We figured out that the iceberg melt was very strong despite surface temperature below surface freezing point. This was happening because:

  • a lot of icebergs were advected into this cell (it was a corner) and get stuck there for a while
  • icebergs are allowed to melt even if surface temperature is under the surface freezing point (as long as sst warmer than -4C) via the bottom melt.
             ! Environment
             zdvo = SQRT( (pt%uvel-pt%uo)**2 + (pt%vvel-pt%vo)**2 )
             ! Melt rates in m/s (i.e. division by rday)
             zMb = MAX( 0.58_wp*(zdvo**0.8_wp)*(zSST+4.0_wp)/(zL**0.2_wp) , 0._wp ) * z1_rday   ! Basal turbulent melting     (eqn M.A7 )

The link between iceberg melt and the 99m of ice is the following: iceberg melt is cooling the ocean (latent heat) ⇒ at some point (especially in very shallow cell) sst get below surface freezing point (super cold water) ⇒ frazil ice are formed by SI3 whatever is the ice thickness or ice fraction until we reach 99m of ice.
In our case the melt rate of icebergs was almost exactly the same than the sea ice production rate.

It should affect all the versions of NEMO using icebergs but we need a bad combination of wind, current and sea ice to trigger the issue.


This is what we (MetO) decided to do :

  • ground icebergs with a keel depth larger than the bathymetry to limit the number of icebergs being advected in very shallow cells.
  • set basal melt to 0 if surface temperature below the surface freezing point.

In our test :

  • the first point tend to delay a bit when the trouble happened but it seems that in our case most of the icebergs were small enough to enter in the shallow cell (we didn't restart the run from scratch, we rested the run 3 months before the model start to build the ice pillar). It can have a large impact on the calving input file: we need to be sure there is no calving point where bathy shallower than 250m. This case is likely to happened because there are some iceshelves where bathymetry is shallower than 250m. We can also build a maxclasses variable associated to the calving file to limit the thickness of the calved icebergs. NEMO seems able to eat it but we tried it yet. At longer term, we should consider adding complexity in the grounding law (see Nach Merino paper for some consideration on this topic).
  • The second point is not beautiful but it fixes the issue. At longer term, we should consider thinking on using another formulation for the iceberg basal melt (3 equation formulation as for the ice shelves maybe, other idea ?)

We are open for discussion about what should be included in NEMO ? only point 2 ?, both points ?, going further in the grounding law or in the melt computation (this is not in our priority at the moment) ?

Commit History (0)

(No commits)

Change History (2)

comment:1 Changed 13 days ago by nemo

  • Version changed from release-4.0 to release-4.0-HEAD

comment:2 Changed 13 days ago by clevy

  • Version changed from release-4.0-HEAD to 4.0-HEAD
Note: See TracTickets for help on using tickets.