We used a state‐of‐the art one‐dimensional snow and ice model (the LIM1D model), to simulate data collected in winter 2015 north of Svalbard with ice mass balance instruments. The quality of the simulations was assessed by comparing simulated temperature profiles and sea ice thicknesses with the data: The root‐mean‐square difference between observed and modeled temperature was 1.06 °C in snow and 0.19 °C in ice, and the root‐mean‐square difference between simulated and observed ice thickness was 2.0 cm (snow depth was prescribed). The long‐wave heat flux from the ERA‐I reanalysis was adequate to perform winter numerical simulations; in contrast, the ERA‐I air temperature induced large errors in the snow and ice temperature. Snow density had a direct impact on heat transfers and, thus, on the simulation. The joint use of the data and the simulations permitted the adjustment of the snow density profiles with a light (240 kg/m3) snow deposited on top of a denser (370 kg/m3) snow. The ice flooding, which occurred after a storm‐induced breakup of floes loaded with snow, was simulated by prescribing the observed lower limit of the snow. The simulations provided insights on the evolution of sea ice bulk salinity, brine fraction, and the amount of snow ice formed during the flooding event.