The Montreal Urban Snow Experiment (MUSE) was dedicated to further understanding of micro-meteorological processes involved in the late winter to early spring transition period in a Canadian city. A surface energy budget site was installed in a dense residential area of Montreal for several weeks in 2005 and 2006. A previous study conducted for the 2005 experiment underlined the impact of the snow-melting process on the urban surface energy budget. Difficulties to accurately simulate the surface energy budget appeared for days when snow had melted and before vegetation became active. Thus, the present study's objectives are to better understand the physical processes involved during this specific period and to examine their importance on the surface energy budget. The study focuses on the last six days of the 2006 experiment (23-28 March 2006), using the Town Energy Balance (TEB) canopy model for the built-up surfaces (78 %) and the Interactions-Soil-Biosphere-Atmosphere (ISBA) surface model for the natural covers (22 %). Both models are used in an off-line mode and are forced with meteorological data from measurements at the top of a 25-m instrumented tower. Preliminary results displayed deficiencies similar to those encountered in the MUSE-2005 modeling study. But sensitivity studies indicate that a large portion of these problems is related to the latent energy involved in natural soil freeze/thaw processes, which has a significant effect on the urban surface energy budget. It was found that the energy in this particular situation was quite sensitive to the thermal roughness length used for local energy exchange over the built-up surfaces.