Shape Recognition of DMT Airborne Cloud Particle Images and Its Application

Abstract

Currently, the most direct and effective way to obtain cloud precipitation microphysical characteristics is from in-situ measurements acquired by airborne imaging probes. There are many studies based on CIP (cloud imaging probe) and PIP (precipitation imaging probe) detection data, which are mostly based on the limited output from the software PADS (Particle Analysis and Display System) provided by DMT (Droplet Measurement Technologies). Since PADS only outputs the second-by-second statistical results rather than the detailed particle-by-particle information, it greatly limits the deep mining and analyzing of cloud particle image data. Besides, particle shapes in previous studies are mainly classified through naked-eye observations, which is time consuming, subjective, and unreliable to conduct statistical analysis on thousands of cloud particle images. Therefore, it is impossible to calculate the hydrometeor content based on mass-dimension relationships for particles of different shapes in ice or mixed cloud observations.The operation principle of airborne two-dimensional optical array probes is introduced. Then techniques of recognition and elimination of shattering particles and fake particles are illustrated in detail. Particle shapes are divided into 8 types (tiny, linear, aggregated, graupel, spherical, plate, dendritic and irregular) based on geometric characteristics of particle shapes. Statistical characteristics of different cloud particle shapes and their areas are analyzed by using gray CIP data detected in three wintertime stratiform clouds in Henan Province. The recognition of particle shapes are basically consistent with results through naked-eye observations, and also consistent with dominant particle shapes in each temperature range obtained by previous studies. The hydrometeor content obtained using the mass-dimension relationship for particles of different shapes is compared with that from treating all particles as spherical liquid particles (i.e., the algorithm used by PADS). It is found that when all particles are treated as spherical liquid particles, the hydrometeor content is roughly one magnitude higher than that from considering different particle shapes, indicating the technique of particle shape classification can improve the accuracy of hydrometeor content in ice or mixed clouds. In addition, some matters needing attention in the use of two-dimensional particle image data are pointed out to ensure proper use of two-dimensional particle image data.

Keywords

• cloud particle
• particle image
• particle shape
• hydrometeor content