Due to its high surface albedo, snow plays an important role in the air-ice-ocean interaction in high-latitude regions. Accurate snow mass balance calculations are needed to understand the evolution of sea ice and interaction between snow-ice and atmosphere better. One of the factors affecting snow mass balance is snow density. Constant mean snow bulk density is used to convert snow water equivalent to snow depth in the present 1-D high-resolution thermodynamic snow-ice model (such as HIGHTSI). Simplified to 2 snow layers, being fresh and old, algorithm reference to Lagrangian snow-evolution model (SnowModel-LG) used to treat layered snow compaction is introduced into HIGHTSI to reproduce the physical process of compacting in both the fresh and old layer and affecting the snow depth following the principle of mass conservation. Forced by ERA-Interim reanalysis data, modified HIGHTSI was applied to investigate the impact of snow density on snow depth along drift trajectories of 15 sea ice mass balance buoys (IMB) during snow accumulation period and assess the model results against observation. In contrast to the previous bulk snow density setting, with a constant density of 330 kg/m³ (T1) or 200 kg/m³ (T2), our new algorithm calculates snow depth by considering both the fresh and old snow densifying over time (T3). The simulations indicate that the improved algorithm is more reasonable to deal with the density evolution, and can reproduced the snow depth well. The overaccumulation caused by heaping continuously at the lower density of new snowfall can be avoided by considering the response of both the fresh and old snow depth to compaction. The absolute error calculated by layered snow compaction is reduced by 5 cm by setting the observation as a reference of both the fresh and old snow depth to compaction. The absolute error calculated by layered snow compaction in T2 is reduced by 5 cm by setting the observation as a reference.