Identification of blowing snow particles in images from a Multi-Angle Snowflake Camera

Abstract

A new method to automatically discriminate between hydrometeors and blowing snow particles on Multi-Angle Snowflake Camera (MASC) images is introduced. The method uses four selected descriptors related to the image frequency, the number of particles detected per image, and their size and geometry to classify each individual image. The classification task is achieved with a two-component Gaussian mixture model fitted on a subset of representative images of each class from field campaigns in Antarctica and Davos, Switzerland. The performance is evaluated by labeling the subset of images on which the model was fitted. An overall accuracy and a Cohen kappa score of 99.4 % and 98.8 %, respectively, are achieved. In a second step, the probabilistic information is used to flag images composed of a mix of blowing snow particles and hydrometeors, which turns out to occur frequently. The percentage of images belonging to each class from an entire austral summer in Antarctica and during a winter in Davos, respectively, is presented. The capability to distinguish precipitation, blowing snow and a mix of those in MASC images is highly relevant to disentangle the complex interactions between wind, snowflakes and snowpack close to the surface.

Microphysics and dynamics of snowfall associated with a warm conveyor belt over Korea

“…1). MXPol operates at 9.41 GHz with a typical angular resolution of 1 • , range resolution of 75 m, nonambiguous range of 28 km and a Nyquist velocity of 39 m s −1 in dual-pulse pair (DPP) mode or 11 m s −1 in fast Fourier transform (FFT) mode (see Schneebeli et al, 2013, for more details). The scan cycle was composed of three hemispherical range height indicators (RHIs) at 225.8 • (in FFT), 233 • (in DPP) and 325.7 • (in DPP) azimuth.…”

“…From 6000 to 4800 m the crystals grow by vapour deposition, leading to an increase in both Z H and Z DR (Fig. 10) as particles grow mainly along their longest dimension, which results in larger and more oblate crystals (Schneebeli et al, 2013;Andrić et al, 2013;Grazioli et al, 2015). The median of K dp increases to only 0.4 • km −1 , suggesting that the number concentration of oblate particles is small.…”

“…On the other hand, aggregates are much larger than crystals and tend to have higher Z H values. Consequently, decreasing Z DR together with increasing Z H towards the ground is a consistent signature of the aggregation process (Schneebeli et al, 2013;Kumjian et al, 2014;Grazioli et al, 2015). Furthermore, the specific differential phase shift (K dp ; in degrees per kilometre), which is the range derivative of the total differential phase shift on propagation (i.e.…”

“…Being a lower-order moment than Z H , it is more influenced by the number concentration such that a high number concentration of small oblate crystals can lead to an increase in K dp , while Z DR will barely be affected. Owing to this wealth of information, polarimetric variables have been extensively used for snowfall microphysical studies (Bader et al, 1987;Andrić et al, 2013;Schneebeli et al, 2013;Moisseev et al, 2015;Grazioli et al, 2015). The pioneering work of Bader et al (1987) showed that high Z DR values are associated with large dendritic crystals.…”

Unsupervised classification of snowflake images using a generative adversarial network and <i>K</i>-medoids classification

“…Measuring the properties of individual falling snowflakes is fairly challenging, as some important properties such as the snowflake mass are not readily observable using visual techniques due to the variation of the internal structure of snowflakes. Moreover, the widespread optical disdrometers, such as the Parsivel (Löffler-Mang and Joss, 2000) and the 2D Video Disdrometer (2DVD; Schönhuber et al, 2007), can only discern a silhouette of the falling particle, unable to provide information about the surface texture. To address this issue, snowflake-imaging instruments have been actively developed in recent years.…”