The Effects of Ice Habit Models on Passive Microwave Snowfall Rate Retrievals

This work examines the impact of ice habit models on snowfall rate (SFR) derived from space-borne passive microwave observations. SFR retrieval is highly sensitive to ice habit assumptions. Comparisons of the SFRs based on Cloud Profiling Radar, ERA5, and National Oceanic and Atmospheric Administration Stage IV indicate that dense and sphere-like particles tend to overestimate SFR, whereas most non-spherical particles underestimate it. SFR biases differ by more than 200% between the most extreme cases. Although several ice habits perform well globally, an optimal choice of ice habit is environmentally dependent: A hollow bullet rosette works well in moist and warm conditions, whereas a solid ice sphere excels in cold and dry conditions. A machine learning model integrates multiple ice habits into the SFR algorithm by adjusting their contributions based on environmental conditions. This multi-ice habit approach improves statistical metrics by ∼10% overall (by 40% in deep clouds) compared to a single ice habit method.