A Case Event Analysis Using Multi-Angle-Snowflake-Camera and CSU-CHILL X-Band Observations in Greeley, Colorado: Degree of Riming and Particle Classification

The Multi-Angle Snowflake Camera and Radar (MASCRAD) field project was centered on an instrument that could record digital photographs of freely-falling snow particles viewed simultaneously from five different spatial directions. The images obtained with this camera system (the multi-angle snowflake camera (MASC)) allow the characterization of several physical attributes of the snow particles, including habit, structural complexity, and surface riming. The modified and customized five-camera MASC system, precipitation occurrence sensor system (POSS), and several other advanced surface optical instruments were housed within a double wind fence at the MASCRAD in-situ surface instrumentation field site, near Greeley, Colorado. The goal of the project is to improve the characterization of microphysical properties of winter precipitation through the synergy between polarimetric radar observations, optical measurements and image processing, and electromagnetic scattering computations.

This presentation focuses on a light snow event (26 – 27 November 2015) during which variations in hydrometeor riming and ice particle habit occurred. In the early portion of this event, the MASC images recorded considerable particle riming. The resultant more spherical particle shapes gave near 0 dB differential reflectivity (Zdr) values at X-band as observed by the dual-polarization CSU-CHILL radar (located 13 km north-northwest of the MASCRAD surface instrumentation site). Later in the event, much more pristine, very lightly rimed single crystals became more common in the MASC observations. Zdr levels increased to +2 to +3 dB during the period of minimal riming. Excellent correlation between the particle classification based on the POSS modal analysis and the CHILL radar X-band Zdr has been observed as well.

In addition to riming analysis, the MASC images were also processed to classify hydrometeor types. This classification method evaluates a set of 25 descriptors of particle texture and geometry from each of the camera images. A multinomial logistic regression model is used to associate the particle characteristics extracted from the camera images with a reference data set developed from a large group of manually-classified images.

An example of the results of these MASC image processing techniques applied to the 26 – 27 November 2015 case is shown in Figure 1. During the first period, riming (indicated by the thickness of the plotted symbols) affected most of the particle types. In the later period, graupel was not present and riming levels were generally less. This paper also demonstrates that these automated MASC image processing techniques can provide useful diagnostic links between the physical properties of frozen hydrometeors and dual-polarization radar observations.