Polarimetric radar-based methods for evaluation of hydrometeor mixtures in numerical weather prediction models

The Numerical Weather Prediction (NWP) Model ICON-LAM (operational weather predictions model of the German Weather Service) simulates number and mass concentration of six hydrometeor types namely cloud water, cloud ice, snow, graupel, rain and hail. To date the exploitation of radar polarimetry to improve model parametrizations is still at the beginning. Radar-based hydrometeor classification (HMC) schemes in widespread use only classify the dominant hydrometeor type of a hydrometeor distribution within a measurement volume are very sensitive to uncertainties in scattering simulations, and do not provide information about mixtures of types. The different definitions of hydrometeor types and numbers considered in HMCs and models make a direct comparison even more challenging. In this study, a two steps method is used to calculate the mixing ratios from the polarimetric variables, which in turn can be directly compared to the mixing ratios of the model. In the first step, the radar measurements are clustered based on their similarity of their multidimensional polarimetric signatures (Agglomerative Hierarchical Clustering, AHC) and afterwards the hydrometeor class of the resulting clusters are identified with conventional HMC. In the second step, it is shown how the centroids derived from the multidimensional polarimetric clusters and their probability distributions are used to determine the mixing ratios of the individual hydrometeor classes. The comparison introduced is based on a local polarimetric radar composite including the two polarimetric X-band radars in Jü lich (JuXPol) and Bonn (BoXPol).