By using densely-observed snowfall, surface conventional observation, FY-2E satellite TBB and ERA5 reanalysis data, comparative analysis is carried out for two snowfall events with significantly different snow to liquid ratios (SLR) in North China. Based on the analysis of difference of the snowfall characteristics, vertical thermodynamic structure in clouds, verticaldistribution of precipitation particles and surface temperature, the effect of vertical thermodynamic structure and water vapor conditions on snow density is revealed. The results show that the event with higher SLR (hereinafter referred to the 3rd January Case) occurs in a colder condition from surface to upper atmosphere, and the vertical layer with temperature between -18℃ to -12℃ is much thicker, which almost coincides with the level of maximum ascending motion in the cloud. The perfect match of temperature and vertical motion in the cloud leads to the formation of dendritic snowflake and a higher SLR. In addition, the precipitation particles are mainly ice-phase particles in the 3rd January Case. The case with lower SLR (hereinafter referred to the 29th November Case) appears in a warmer condition. The vertical layer with 〖JP2〗temperature between -18℃ to -12℃ is rela-〖JP〗tively thinner, which is located below the level of its maximum ascending motion. This kind of profiles of temperature and vertical motion in the cloud would result in a low SLR snowfall. Besides this, there are large amounts of super-cooled water droplets in the lower level of the cloud, thus the riming process may occur, which would cause a lower SLR further. The 3rd January Case is mainly caused by a short-wave trough, thus its accent motion is relatively weaker with its peak in 〖JP2〗low-level layer. The layer with tempera-〖JP〗ture between -18℃ to -12℃ is located near the frontal zone of a warm front, where there is a strong warm advection and a small temperature lapse rate. This is the main reason why the -18℃ to -12℃ layer for the 3rd January Case is thick. While the 29th November Case is mainly caused by a deep upper trough, which generates a strong updraft with its peak in upper-level layer. The layer with temperature between -18℃ to -12℃ is located over the frontal zone of a cold front, where the temperature lapse rate is much larger, thus -18℃ to -12℃ layer is shallow. Meanwhile, water vapor provided by easterly over boundary layer, which is the primary source for the 3rd January Case, is generally less; while the southwesterly in front of the upper trough brings a plenty of water vapor for the 29th November Case.