Correlation of Snow D/V Relationships to Mean Snow-Water Equivalent (SWE) Using a Gamma-based Mixture Model

Algorithm developers for the Global Precipitation Measurement Mission need parameterizations of microphysical properties of falling snow such as density to estimate the snow-rate associated within a given radar pixel. Our approach uses the observed diameter-velocity (D/V) information provided by video disdrometers like the Particle Imaging Package (PIP) and matches it with observed Snow-Water Equivalent (SWE) rates from a weighing bucket gauge. The D/V relationships are parameterized using physical hypotheses about the size and shape of falling snow, building off the work of previous authors (Huang et. al., 2014). In this work we develop a Gamma/Gamma model (gamma fit to observed snow particle diameter and gamma fit for snow particle fall velocity) for describing D/V distributions of falling precipitation within a small timeframe. Using a general mixture model we automatically generate a small number of distinct parameterizations of the Gamma/Gamma model which describe subpopulations of our samples with different D/V distributions. We show that these subpopulations are correlated to precipitation type, which means this algorithm can be used to automatically detect rain, snow, and mixed-phase precipitation. Finally, we demonstrate how such a model can be correlated to SWE rates to develop an empirical density model based on a particular minute’s classification, and the diameter and velocity of a particle from that minute.