[MITgcm-devel] negative snow thickness in thsice

[Martin Losch]

Dear thsice-users/developers,

I have come across a situation where I get slightly negative snow thickness (-0.0118 m), at the ice edge, where the ice cover fract = iceMaskMin (0.05), and iceH is also small (0.016 m). I cannot see anything extraordinary in any other variable. In the code I cannot find a place where this is caught (only iceH < iceMin), but with negative snow thickness the albedo computations are upset and the model stops (albedo < 0.2). 
What are the circumstances under which snow thickness can become negative? I can only think of advection as the culprit (I use 77, so a flux limited scheme, pkg/seaice provides the ice drift velocities). If advection is to blame, it is probably safe to limit snow thickness to 0 for albedo computations. In fact when I do that, the model continues and the negative snow thickness disappears again. 

I am just a little worried that I am masking an underlying problem that could be fixed in a more consistent way. Are there any process other that advection that can cause negative snow thickness?

Martin