Three distinct types of heavy snowfall clouds Revealed by spaceborne Dual-Frequency Precipitation Radar observations (2014–2023)

• Global heavy snowfall is classified into three types using Gaussian Mixture Model. • Topographic heavy snowfall type dominates over highland with the smallest D m . • Continental-like heavy snowfall type shows intermediate characteristics. • Oceanic heavy snowfall type dominates over ocean with the largest D m . • All types differ in vertical structure, environment, and temporal variations. This study examines global heavy snowfall (snow rate > 2.5 mm h −1 ) using 10 years (2014–2023) of observations from the Global Precipitation Measurement (GPM) Dual-frequency Precipitation Radar (DPR). Heavy snowfall cases are categorized into three distinct types using an Expectation-Maximization (EM) approach based on liquid-equivalent mass-weighted diameter (D m ) and normalized intercept parameter (logN w ). The first type represents the majority of heavy snowfall over highland regions and is characterized by smaller snow particles and higher particle number concentrations, along with lower hydrometeor content and colder environments, indicating limited growth of snow particles. The second type exhibits slightly larger D m and lower logN w than the first type and predominantly occurs over continental regions. The third type dominates over oceans and explains the largest contribution to global volumetric heavy snowfall, characterized by relatively larger snow particles and lower particle number concentrations, along with higher hydrometeor content. Based on the characteristic occurrence frequency distributions of each type, the three types are referred to as topographic heavy snowfall (THS), continental-like heavy snowfall (CHS), and oceanic heavy snowfall (OHS), respectively. All three types exhibit pronounced seasonality, whereas their diurnal variation is relatively weak. This study enhances understanding of the characteristics and environmental differences among the three heavy snowfall types, which is expected to support improvements in satellite heavy snowfall retrievals and microphysical parameterizations in numerical models.