The sea ice model can be thought of as consisting of two mass layers made up of an upper layer which has a constant thickness (10cm) ice layer and an arbitrary snow amount. The second layer mass has a minimum thickness of 10 cm, but can grow arbitrarily thick. There are two thermal layers in each mass layer which consist of a fixed ratio of the mass. Salinity and tracer values are also calculated on the same mass ratio, with the exception that salinity is assumed to be zero in the snow. At the surface, heat and mass fluxes calculated in the SURFCE routine and the precipitation are applied first to the snow layer and then to the ice. There is advection of heat, salt and mass as a function of the regridding of the mass layers in order to preserve the fixed ratios, i.e. there is a sigma-coordinate for sea ice properties.
The basal fluxes are calculated in the ocean only for the configurations where ice thickness is prognostic (i.e. not for the fixed SST cases). This is done in routine UNDERICE. The fluxes are then applied to both the ice and ocean modules. Similarly, lateral fluxes are calculated in MELT_SI. This is done prior to any surface fluxes calculation to avoid the problem of changing the surface type fractions in the midst of the calculation.
Frazil ice formation can occur in the ocean and lakes in open water conditions and under ice. This ice mass formation rate (and it's associated salt (for the ocean) and energy flux are calculated in either the lake or ocean component and added to the sea ice by routine FORM_SI. There is a minimum lead fraction which is a function of the ice thickness.
Snow can be compacted into ice by accumulating over a maximum thickness, or by rain or surface melting. Surface melting is kept track of since it is used to estimate melt pond formation for the albedo calculation. No specific snow-ice formation is yet included.