“…In this study, we analyzed satellite data set of over 3 years (2007–2009) covered by various instruments loaded on the satellites that fly in a constellation orbit “the A‐Train.” We used the following products: (1) collocated CPR and CALIOP measurements, so‐called “CloudSat‐CALIPSO Merged Data Set”, with a uniform resolution of 240 m vertical × 1.1 km horizontal basically along CloudSat profiles (Hagihara et al, ), which consist of radar reflectivity and lidar backscattering coefficient as well as temperature from the European Centre for Medium‐Range Weather Forecasts; (2) the CloudSat‐CALIPSO hydrometeor particle phase and shape (hereafter, hydrometeor particle type) retrieved from the Merged Data Set using the algorithm developed by Kikuchi et al (); (3) MOD06‐5KM‐AUX product, which provides MODerate resolution Imaging Spectroradiometer (MODIS) cloud top temperature (CTT) collocated with CloudSat along‐track sampling; and (4) Aqua Advanced Microwave Scanning Radiometer for EOS (AMSR‐E) precipitation rate (Liu & Curry, , ) provided in Japan Aerospace Exploration Agency (JAXA) L2 Standard Product selected for the nearest pixels per along‐track CloudSat pixels. Of the total 13 hydrometeor particle types given in (2), we selected the following types that account for 96.4% of the total (see Kikuchi et al, for the percentage breakdown): water (combining warm water and supercooled water), 3D‐ice, 2D‐plate, drizzle (including liquid drizzle and mixed‐phase drizzle), rain, and snow. Note that 3D‐ice refers to ice crystals that are randomly oriented in three‐dimensional space including bullet rosettes, columns, randomly oriented plates, and aggregates, whereas 2D‐plate refers to ice plates that are oriented horizontally in two‐dimensional space (Iwasaki & Okamoto, ; Okamoto et al, ; Sato & Okamoto, ), the former and latter being distinguishable with the lidar measurement capability.…”