In ice microphysical processes within cloud resolving models, ice particles are often representing by separating them into 3 categories: cloud ice, snow and graupel. Conversion processes among these categories are usually tuned to reproduce surface precipitation amounts accurately, but their consistency with real cloud microphysical processes has not been thoroughly validated. Furthermore, it is important to accurately represent the conversion process from cloud ice to snow, which is an initial stage in the growth of ice particles. Polarimetric radar observations provide polarimetric parameters that reflect cloud microphysical properties such as particle number concentration, size, and shape. To improve ice microphysical processes through comparisons with polarimetric radar observations, it is necessary to accumulate knowledge about the sensitivity of microphysical processes to polarimetric parameters. In this study, polarimetric radar parameters (e.g., differential reflectivity Zdr and specific differential phase Kdp) were derived using a polarimetric radar simulator from the outputs of a cloud-resolving model, such as mixing ratios and number concentrations, for comparison with polarimetric radar observations. Both Zdr and Kdp are sensitive to particle aspect ratios; hence, cloud ice, which is typically more oblate than snow, can exhibit higher values. The aim of this clarify the sensitivity of polarimetric parameters to the conversion process from cloud ice to snow in the Cloud Resolving Storm Simulator (CReSS; Tsuboki, 2023).

In CReSS, when the mean diameter of cloud ice exceeds the minimum snow diameter via deposition or aggregation, a portion of the cloud ice is converted into snow. Therefore, increasing the minimum snow diameter suppresses this conversion. We conducted two experiments: a control run (CTL), which used the conventional minimum snow diameter, and a sensitivity experiment with a larger minimum snow diameter. In the CTL experiment, cloud ice didn’t exist below 8 km altitude; snow was the dominant condensate. As a result, both Zdr and Kdp were nearly zero (0 dB and 0°/km) above the 0°C level and remained at small, constant values compared to observations. In the sensitivity experiment, where the minimum particle size of snow was increased, Zdr showed little change, but Kdp increased from the 0°C level upward and exhibited a wider range of possible values, leading to a profile closer to the observations. Because Kdp is proportional to both particle size and number concentration, the increased number concentration of more oblate cloud ice likely contributed to the higher Kdp. Reproducing the vertical variation of Kdp suggests that CReSS is now better capturing the changes in cloud ice number concentration accompanying the conversion of cloud ice to snow via aggregation growth.

On the other hand, Zdr did not change appreciably. Since Zdr is the ratio of radar reflectivities (which scale with approximately the sixth power of particle diameter), the presence of larger snow particles compared to cloud ice likely resulted in a small change in Zdr .

In this study, we used the polarimetric radar simulator POLARRIS to calculate the polarimetric parameter Kdp from the output of CReSS and compared it with observations. This comparison revealed that CReSS underestimated the cloud ice number concentration. Furthermore, such Kdp comparisons provided constraints on the appropriate cloud ice number concentration in the cloud-resolving model, helping to guide improvements in ice-phase microphysical processes.