Measurements of hydrometeor microphysics and fallspeed in the Arctic with the Multi-Angle Snowflake Camera

Description

Between November 2014 and August 2017, millions of images of frozen precipitation particles in freefall have been captured by the surface-based Multi-Angle Snowflake Camera (MASC) at the Department of Energy’s Atmospheric Radiation Measurement (ARM) Mobile Facility at Oliktok Point, Alaska. For classification of particle type, we have developed a new tool that employs machine learning algorithms such as decision trees, multinomial logistic regression, and neural networking. The accuracy of the algorithm for identifying hydrometeor type compared to identification by eye by a trained atmospheric scientist is between 93% and 95%. For a subset of more than 15,000 frozen particles collected between January and April 2017, we find that the size hydrometeors, on average, is inversely proportional to the degree of riming. However, while the modal maximum dimension of ~1 mm is similar to previous measurements obtained in the Wasatch Front in Utah, the slope of the size distributions is approximately twice as steep. Two distinct modes emerged for fall speeds as related to maximum dimension: one size grouping centered on 1.2 mm with a modal fall speed of ~1 m/s, and another centered on 0.9 mm with a modal fall speed approaching 0 m/s, although numerical simulations suggest this second mode may be an artifact of interactions of winds with the MASC instrument body. Overall, fall speed has a surprisingly weak dependence on size and shape. In particular, there is both an increase and decrease in the settling velocity of the flakes relative to expectations. We present a hypothesis that in a turbulent atmosphere, the terminal fall velocity of a snowflake is not the same as the temporal mean fall velocity. We present a governing equation of motion for settling of a snowflake in an oscillating fluid. It is shown that the temporal mean fall velocity depends on a) the unsteady flow features i.e., amplitude and the angular velocity the oscillating fluid and b) the particle terminal velocity. Overall, the temporal mean fall velocity is lower than the terminal fall velocity, and the ratio is a function of the frequency and amplitude of the fluid oscillation.