Effect of the accretion by water drops on the melting of snowflakes

Abstract

Melting of solid hydrometeors plays an important role in the formation of precipitation. A two-moment bin microphysics scheme was developed to simulate the melting process corresponding with laboratory observations. The following physical processes are taken into consideration: collision—coalescence between the solid and liquid precipitation elements; melting; condensation/deposition; freezing of supercooled drops and re-freezing of partly melted ice particles. Completely melted particles transfer to the water drop category and shedding occurs only when the diameter of the melted solid hydrometeors exceeds 0.8cm. Numerical experiments were made by using a one-dimensional kinematic model to investigate how the accretion by water drops affects melting.Results of the numerical experiments show that an increase of the mass of snowflakes and change of their shape due to riming significantly increase the depth of melting layer. In the presence of water drops within a melting layer, its depth is controlled by two competing processes: increased melted ice particle fall speed acts to increase melting layer depth while vapor depositional growth combined with an increase of liquid fraction due to collision–coalescence growth decreases the depth of the melting layer. If the liquid water is relatively small (0.1gkg−1 or less), the effects of collision are almost negligible within the melting layer. If the liquid water content is relatively large (about 1gkg−1), roughly half of the liquid water content on the surface of the snowflakes forms due to the collision coalescence process, the other half due to melting.An earlier published relation between melting fraction and the mass of snowflake (Szyrmer and Zawadzki (1999) was investigated but our results showed that the melting fraction hardly depends on the mass of the snowflakes in a relatively wide size range. With relatively modest to high amounts of water present, our results show an even larger spread and discrepancy compared to the previous relationship that used spherical rather than oblate spheroid shaped snow. Finally, precipitation formation of a real case with partial snow melting and particle re-freezing was simulated. The type of the surface precipitation produced by the model agreed well with surface observations, especially compared to a similar bulk microphysics scheme that produced far more freezing rain that contradicted the surface observations.