Brine is known to migrate in sea ice, influenced by gravity, by internal pressure changes and by temperature gradient (Kingery and Goodnow, 1963; Jones, 1973; Hoekstra et al., 1965). Brine pockets move slowly towards higher temperatures, and brine moves downwards. Both directions are the same for most of our data.
Brine movement, excluding brine expulsion by pressure changes and stress
cracking, is associated with latent heat of freezing and melting
effects, as ice remains where brine once was. The contribution to heat
flux, of pockets of brine moving with velocity W is 
,
where 
is the volume fraction of sea ice occupied by brine, and
and L are the density and latent heat respectively of pure
water. Pure water values are used because most of the salts remain in
the brine as it migrates.
Then the heat equation is modified to
In the following, we consider what movement of brine is needed to
contribute significantly to heat flow, say by the amount 2 W m 
when 
Cm 
. This is 5%
of the 40 Wm conducted by ice with a thermal conductivity
2 Wm 
C at this temperature gradient. Using 
kg m 
, L=350 kJ kg , and 
corresponding
to 
C gives
so that the peak
velocity of brine pockets (when temperature gradients are high) would
need to be about 
m s , to contribute this
heat flow. This velocity
is reasonable in its implications for desalination of sea ice. It
corresponds to a peak salt flux of 
kg m s , comparable to the gravity drainage
salt fluxes observed by Kingery and Goodnow (1963). So total salt
balance arguments are consistent with the idea that brine movement may
contribute significantly to heat flux. What is less clear is the
exact nature of the mechanism that is responsible for the brine
movement. Possible mechanisms are explored in the following sections.