Physical characteristics of winter precipitation in two ice weather cases that lasted for more than five hours during 7 Feb to 15 Feb 2014 are observed using PARSIVEL disdrometer, Automatic Weather Station (AWS), Meteorological Information Comprehensive Analysis And Process System (MICAPS) and other instruments. In this paper, spectral feature of precipitation and particle size distribution for the two cases (occurred on Feb 10 and Feb 13) are analyzed. Both cases were mixed-phase precipitation with freezing rain as the main body. The distribution of terminal velocity and particle size deviated from the G-K curve in the two cases. The mixed phase precipitation is different to conventional liquid-phase precipitation, and frozen rain drops gradually convert to ice and dry snow particles during the precipitation process. The results are as follows. (1) The spectral width of size distribution for total precipitation particles in Case 1 was greater than that in Case 2, whereas the peak number density in Case 2 was greater than that in Case 1. In Case 1, the spectral width of particle size distribution was 10 mm and the peak diameter was 0.5 mm; in Case 2, the spectral width of particle size distribution was 4.25 mm and the peak diameter was 0.5 mm; the spectral widths of size distribution of different precipitation particles are sorted from large to small as dry snow > freezing raindrops > ice pellets in Case 1 and freezing raindrops > dry snow > ice pellets in Case 2. (2) The Gamma distribution is more appropriate for describing and parametrizing the size distribution of freezing rain and mixed-phase precipitation. The Gamma distribution of total precipitation particles can be expressed by N(D)=20D^-3.61exp(-0.08D) in Case 1 and N(D)=30D^-4.68exp(-0.75D) in Case 2. The Gamma distributions for frozen raindrops are N(D)=76D^-2.18exp(-1.11D) and N(D)=30D^-4.67exp(-0.75D) for Case 1 and Case 2, respectively. (3) Mixed-phase precipitation includes dry snow and ice particles, and thus demonstrates an obvious tendency to different and non-empirical direction. This feature provides a new idea for distinguishing rainfall types in synchronous precipitation in the future.