Evaluation and improvement of the French fuzzy logic polarimetric hydrometeor classifier

Two approaches are developed to evaluate the consistency of hydrometeor types retrieved from the French S, C and X-band polarimetric radars. The classification algorithm is based on fuzzy logic and uses data-based, wavelength-dependent membership functions. The algorithms distinguishes between the following hydrometeor types : rain, wet snow, dry snow, ice crystals, small hail / graupel (diameter < 5 mm), medium hail (diameter between 5 and 20 mm) and large hail (diameter > 20 mm).

The first evaluation approach aims at checking the consistency of the retrieved hydrometeor types between neighboring radars in overlapping areas. The comparison is carried out with great care and ensures that the data collected by the neighboring radars are collected exactly at the same place and date and over the same resolution volume. Different kinds of radar couples are considered : C-C, C-S, S-S and S–X. The differences are analyzed in terms of measurement conditions (attenuation, …) and polarimetric variables.

The second evaluation approach focuses on hail detection and quantification (according to the three diameter classes mentioned before). Reports, videos and pictures of hail observations posted online on different web sites (specialized web sites or social networks) are collected and used as a validation data base. This comparison allows updating the membership functions currently used at X, C and S band for the three categories of hail diagnosed by the classification algorithm.