On the utilization of radar reflectivity data for assimilation into storm-scale NWP models

Radar reflectivity data are commonly utilized for assimilation into storm-scale numerical weather prediction (NWP) models of mixed-phase convective clouds using relatively simple forward operators. These forward operators ignore the effects of non-Rayleigh scattering as well as the distribution of both particle density and mass water fraction across the size spectra of different hydrometeor habits such as melting hail, graupel, and snow. In this study, the associated errors in the radar reflectivity factor Z are quantified via a comparison of two different methods of Z estimation: that which was obtained using the advanced forward operator and cloud models with spectral (bin) microphysics, and that which was obtained using the commonly used, simplistic formulas for the computation of Z from mixing ratios of different species and the corresponding estimates from the models with bulk microphysics.

Such a comparison is performed for the cases of melting graupel/hail in convective storms and melting snow in stratiform cases. Simple one-dimensional bin models of melting hail and snow, in conjunction with the advanced forward radar operator described in Ryzhkov et al. (2011), are utilized to retrieve the values of Z and compare them with the ones obtained from the associated bulk schemes yielding either the same mixing ratios q_i (single-moment schemes) or both q_i and total concentration N_i (double-moment schemes) of different hydrometeor habits.

It is demonstrated that the errors in the estimates of Z from the single-moment bulk schemes are usually unacceptably high to be used in the assimilation process. The errors associated with double–moment bulk schemes are much lower. The utilization of the Rayleigh formulas for computing Z for hail also results in very large errors, especially at C and X bands, and hence the utilization of the T-matrix computations is mandatory.