Impact of the natural variability in snowflake diameter, aspect ratio, and orientation on modeled snowfall radar reflectivity

Significant progress has been made in approximating snowflakes with increasingly realistic shape models and then simulating their microwave scattering properties for snowfall remote sensing. This study aims to complement prior findings by analyzing the natural variability within snowstorms. The approach consists of (i) deriving the variability of snowflake diameters, aspect ratios, and orientation angles from automated snowflake observations during individual snowstorms, (ii) calculating the corresponding variability of snowflake backscatter cross sections at 24, 35, 94, and 183GHz, and (iii) evaluating the impact of the analyzed snowflake variability on modeled snowfall equivalent radar reflectivity factors (Z(e)). For seven snowstorms, the observed natural variability in snowflake diameter, aspect ratio, and orientation leads to distributions of snowflake backscatter cross sections that span up to 6 orders of magnitude. The impact on modeled radar reflectivity is smaller with distributions of Z(e) values spanning less than 1 order of magnitude and with median absolute deviations and standard deviations of up to 25%, depending on the analyzed frequency and the applied snowflake parameterization. An assessment of different characterizations of snowflake aspect ratio and orientation indicates that a simplistic description of aspect ratios and orientation angles through constant values which do not correspond with the observed snowflake distributions for the analyzed snowstorm can lead to relative differences in modeled Z(e) of |Z(e)|60%. These differences in Z(e) are generally smaller than uncertainties in Z(e) associated with the parameterization of snowflake mass but still translate into significant uncertainties for the retrieval of snowfall rates from radar reflectivity measurements.