Implementations of Csu Hydrometeor Classification Scheme for C-band Polarimetric Radars

Dual-polarization radar measurements are sensitive to hydrometeor properties such as shape, phase state, and fall behaviour (Bringi and Chandrasekar 2001) so that they can be used for hydrometeor classification purposes. The differential reflectivity (Zdr) is sensitive to the shape and orientation of precipitation particles: it can discriminate rain from spherical hail. The linear depolarization ratio (LDR) is sensitive to shape and orientation, canting, and dielectric constant of precipitation particles, so that tumbling wet nonspherical particles result in large LDR, while drizzle and dry ice particles are associated with low LDR. The magnitude of the co-polar correlation coefficient (ρco) can be useful to identify melting particles or mixed precipitation. The specific differential phase (Kdp) can isolate anisotropic hydrometeors from isotropic ones. The reflectivity factor (Zh) is used to further enhance classification. However, the mapping from polarimetric radar measurement space to hydrometeor classes is not one to one so that Boolean classification techniques cannot be successfully applied. Fuzzy logic represents an attractive approach, since it can a) combine objective knowledge and subjective knowledge, b) manage classification in the presence of imprecisely defined class output, c) cope with overlapping conditions, and d) deal with approximate reasoning. Fuzzy logic based hydrometeor classification systems (HCS) have been successfully applied and evaluated mainly with S-band polarimetric radars (Vivekanandan et al. 1999, Liu and Chandrasekar 2000, Zrnić et al. 2001, Ryzhkov et al. 2005, Lim and Chandrasekar 2005). Adaptation of such classification schemes to C-band is becoming increasingly important, especially because of the extensive use of C-band dual polarization radar systems in Europe. For example, in Italy, the new national radar network will include several C-band radars with the ability to measure Zh, Zdr and differential phase shift. Consequently, HCSs working at C-band may also be used for deployment in operational weather services. Examples of HCSs used at C band are reported in the literature. Höller et al. (1994) applied a decision tree classification to study the evolution of a hail-generating event. Keenan (1999) introduced a fuzzy logic classification scheme based on an additive inference engine which is used in a quasi-operational mode in Australia. Cremonini et al. (2004) studied the sensitivity of two different fuzzy logic classification approaches using Zh and Zdr measurements from different C-band radars in Northern Italy. This paper transforms the CSU algorithm for hydrometeor classification (Liu and Chandrasekar 2001, Lim and Chandrasekar 2005) for C-band applications. Modifying the CSU classification scheme to C-band requires some adaptations, which include both the adjustment of the membership functions and the adoption of implementations for the reduced set of measurements available in current C-band radars. Implementations of the C-band version of the CSU hydrometeor classification scheme to two different Cband radars, namely the Polar 55C radar (Rome, Italy) and the ARMOR (Huntsville, AL) are presented and discussed in the paper.