AbstractA large data set of volume element models of aggregate snowflakes was created, building the snowflakes from various models of ice crystals found in the atmosphere: dendrites, needles, plates, and bullet rosettes, as well as spheroidal crystals for comparison. Several different sizes for the constituent crystals were also used. The radar backscattering cross sections of the snowflakes were computed from the models using the discrete dipole approximation (DDA) at 13.6 GHz (Ku band), 35.6 GHz (Ka band) and 94.0 GHz (W band), and the effects of the choice of crystal model and size on the Ku/Ka band and Ka/W band dual‐wavelength ratios (DWR) was investigated. It was found that the aggregate DWRs were very similar for all naturally occurring ice crystal types investigated in this study. This implies that the choice of crystal type is at most of secondary importance in the forward model of scattering used for snowfall retrievals but also, conversely, that the identification of the crystal type from triple‐frequency observations is likely to be difficult. In contrast, the size of the constituent ice crystals does have a nonnegligible impact on the triple‐frequency signatures. Additionally, it was found that the triple‐frequency signatures found in some experimental data, resembling those resulting from spheroidal model snowflakes, cannot be reproduced using the aggregates with any of the crystal types that were investigated. This suggests that besides aggregation, other mechanisms of snowflake formation from ice crystals must be considered in snowfall retrieval algorithms.
Read full abstract