*2.3. Clasi*ffi*cation of Hydrometeor Species—Hclass*

Hclass performed in this paper is similar to Sokol et al. [33] and stands on the idea that the TV of diverse hydrometeor species differs. It is natural that the existence of a hydrometeor species depends on ambient air temperature and its shape can be identified by LDR (e.g., shape of ice crystals differs from that of rain). In this paper, we define five hydrometeor species; cloud water (i.e., cloud droplets), rain, graupel, hail, and ice and snow particles together. We merged ice and snow particles into one group following the recommendations provided by Sokol et al. [33], where the algorithm was not able to efficiently distinguish ice particles from snow.

Terminal velocity range of the 5 hydrometeor species between minimum terminal velocity (TVmin) and maximum terminal velocity (TVmax) is provided in Table 2. Terminal velocity ranges of hydrometeor species do not overlap and stem from values provided in COSMO numerical weather prediction model. We use these values as "standard" values for hydrometeors whenever we make model simulations over Czechia (e.g., Sokol et al. [34]). The actual TV of a target for any discrete interval of Doppler spectra corresponds to the result of subtraction of AV from DV for a given peak of the Doppler spectra.


**Table 2.** Terminal velocity range of hydrometeor species classified by Hclass.

We define the ambient air temperature (T [◦C]), which influences the presence of hydrometeor species, from ERA5 reanalyses (www.ecmwf.int). Specifically, we take temperature profiles of a grid point closest to the Milešovka observatory. ERA5 reanalyses provide us with hourly data at a horizontal resolution of 0.25◦ (geographical latitude) × 0.25◦ (geographical longitude). The use of ERA5 reanalyses differs from Sokol et al. [33], which used temperature profiles based on sounding measurements from Praha/Libuš station situated 60 km southeast from the Milešovka observatory. These sounding data are not only distant from the Milešovka observatory, which is a problem especially when investigating thunderstorms, but they are also available only at 00, 06 and 12 UTC. On the other hand, ERA5 reanalyses provide us with vertical profiles at a higher (i.e., hourly) temporal resolution and from a location much closer to the radar site (12 km). Thus, we consider ERA5 reanalyses more

suitable for this study. On the other hand, we are aware that the used temperature profiles are not accurate because the ERA5 data have a low resolution to describe temperature profiles in convective storms that differ from those of the surrounding air.

Similar to Sokol et al. [33], we use 0 ◦C as a temperature threshold in the Hclass algorithm and as in Sokol et al. [33], if T > 0 ◦C, then the Hclass provides cloud water, rain, graupel or hail. However, if T ≤ 0 ◦C, we determine the existence of supercooled water in the higher atmospheric (i.e., tropospheric) layers differently than Sokol et al. [33]. Sokol et al. [33] used a fixed threshold of −20 ◦C below which the supercooled water could not exist although in the case of convective storms, it can happen that the supercooled water is observed at much lower temperatures (even at −50 ◦C in some cases) due to strong updrafts and lack of time to freeze. Therefore, we modified the Hclass provided by Sokol et al. [33] and set that the supercooled cloud water can be found from 0 ◦C up to −40 ◦C (instead of previous −20 ◦C), which corresponds to generally accepted temperature range for the existence of supercooled water in mid-latitude summer thunderstorms. We define the existence of supercooled cloud water within 0 to −40 ◦C only if a condition of AV is met: (i) AV > 1 m/s if T is between 0 ◦C and −20 ◦C or (ii) AV(T) =1 − (T + 20)/5 if T is between −20 ◦C and −40 ◦C. The threshold for AV and relationships determining supercooled cloud water were obtained empirically based on subjective evaluation of their performance at various thunderstorms recorded by the cloud radar.

A limitation of our Hclass is that we did not have the means to objectively verify its results. Since any Hclass depends on selected terminal velocity range of individual hydrometeors and real measurements show ambiguity in the terminal velocity ranges, any change in terms of terminal velocity range of any hydrometeor species will affect the results of all Hclasses. This is the key source of uncertainty of Hclasses in general. Moreover, while fixing the values of given parameters (e.g., terminal velocity range), it is almost impossible to avoid subjectivity. Thus, we tested several values of parameters, compared obtained results and fixed the parameters to values that provided results closest to reality based on our experience and/or literature.
