Abstract: Based on conventional upper-air and surface weather observations, ground-based GPS/MET water vapor data, NCEP reanalysis data, and Doppler weather radar and dual polarization radar data, we have performed a comparative analysis on the circulation background and the dynamic, thermodynamic and water vapor transport conditions for the two heavy snowstorm events (hereinafter referred to as "0103" event and "0124" event, respectively) occurred in Anhui Province, respectively, from January 3 to 4 and 24 to 28 in 2018, and discussed their similarities and differences in atmospheric temperature variation when precipitation phases turned during the two events. Results are as follows. (1) Both events occurred under the circumfluence background with the trough at 500 hPa moving eastwards and the low-level shear line extending eastwards, and there are southwest jet and inversion layer at 700 hPa. There is an obvious temperature front zone and strong atmospheric baroclinicity in temperature field at 850 hPa. The time period of the high atmospheric precipitation values corresponds to that of heavy snowfall. The warm advection enhances when the snowstorm occurs. The snowstorm areas are located in the superposition of low-level cold advection and high-level warm advection. (2) The difference between the two events is that during "0103" event the warm and wet air flow increases and moves northward firstly and it supplies enough moisture. There is some unstable stratification. The dynamic convergence zone is deep, and the convective characteristics of snowfall are obvious. But during "0124" event, the low-level cold air moves southward and forms a cold cushion firstly. However, the thickness of vertical movement development and the intensity of the warm advection are not as great as those during "0103" event. Strong cold air mass forms high pressure along the coast of East China and forces the westerly trough to move eastward slowly, causing a long snowfall duration. (3) The precipitation particle phases are correctly identified by dual polarization radar during the two events. The precipitation particle classification products have certain reference value for short-term forecasting and nowcasting of snowfall.