The case under consideration occurred on March 4, 1993, and was observed as part of the Radiation and Eddy Flux Experiment (REFLEX II) 1993 observational campaign northwest of Spitsbergen. The off-ice flow on this day brought very cold surface air temperatures (-35 ¿C) over a relatively warm ocean surface. The resultant latent and sensible surface heat fluxes produced intense convection and a thermal internal boundary layer (TIBL) which deepened with distance from the ice edge. Two-dimensional cloud-resolving model (CRM) simulations were performed to determine the impact of various cloud parameterizations on the structure and evolution of the TIBL. The model was able to reproduce the observed thermal structure of the boundary layer to within the acknowledged limitations of the CRM approach. Sensitivity studies of cloud type showed that inclusion of mixed-phase microphysics had a large impact of BL depth and structure. Radiative heating of the cloud near cloud base and cooling near cloud top along with latent heat release were found to be significant sources of turbulence kinetic energy even in the present case where very strong surface heat fluxes occur. Ice-phase precipitation processes rapidly depleted the BL of condensate, weakening the radiative thermal forcing. A further consequence of condensate depletion in the mixed-phase cloud was a less humid boundary layer that was able to maintain a larger surface latent heat flux and continuously extract heat through condensation and deposition. Not surprisingly, the presence of clouds had a profound impact on the radiative budget at the surface, with the cloudy BL reducing surface radiative losses more that 60% over clear-sky values. Inclusion of the ice phase significantly affected the radiative budget as compared to purely liquid clouds, illustrating the importance of ice-phase-radiative couplings for accurate simulations of arctic clouds and boundary layer dynamics. ¿ 2000 American Geophysical Union