Using multi-source observation and reanalysis data, the rare ice pellet weather in southern Jiangsu from February 22 to 24, 2024 is analysed, and ice grains, freezing rain, and snow are also compared. The results show that the strong southwest warm wet jet, stably maintained at 700 hPa, provided favourable power, water vapour, and temperature stratification conditions for the formation of ice pellet weather. The development of middle warm wet airflow exhibited significant extremes. The 700 hPa height field in southern Jiangsu was 10-30 gpm higher than the annual average climate (standard deviation multiple of 0.5σ-1σ), and the southwest wind was stronger than the annual average climate by 10 m·s^-1, with a 4-5 ℃ higher temperature (standard deviation multiple of 1.5σ). The middle abnormal warming was one of the important conditions for the rare long-term ice pellet weather in southern Jiangsu. The inversion layer was stable, located between 939.8 hPa and 729.9 hPa, with an average thickness of 1540 metres and an average intensity of 10.2 ℃. The stably maintained middle warm layer and low cold layer provided favourable temperature conditions for the occurrence of ice grain weather: the average thickness of the warm layer was 709 metres, the height was between 779.3 hPa and 683 hPa, and the temperature of the warm layer was less than or equal to 2.9 ℃. The cold layer was basically located below 779.3 hPa, and the lowest temperature was between -8.7 ℃ and -7.6 ℃, which was conducive to the ice particles partially melting in the warm layer and being frozen again during the descent process, ultimately falling to the ground as ice particles. The radar characteristics of the three phases (ice particles, freezing rain, and snow) were different, which had certain indicator significance for the identification of precipitation phases and the forecast of phase approaches: the reflectivity factor of snow was below 30 dBz, while the reflectivity factors of ice particles and freezing rain were greater, generally below 45 dBz; the differential reflectance factor (Z_DR) was ice particles > freezing rain > snow, and the maximum Z_DR of ice particle was 4 dB. From the correlation coefficient (C_C), ice particles had the smallest values, while freezing rain and snow were close to 1.