Cloud ice crystal classification using a 95-GHz polarimetric radar

Two algorithms are presented for ice crystal classification using 95-GHz polarimetric radar observables and air temperature (T). Both are based on a fuzzy logic scheme. Ice crystals are classified as columnar crystals (CC), planar crystals (PC), mixtures of PC and small- to medium-sized aggregates and/or lightly to moderately rimed PC (PSAR), medium- to large-sized aggregates of PC, or densely rimed PC, or graupel-like snow or small lumpy graupel (PLARG), and graupel larger than about 2 mm (G). The 1D algorithm makes use of Z_h, Z _dr, LDR_hv, and T, while the 2D algorithm incorporates the three radar observables in pairs, (Z_dr, Z _h), (LDR_hv, Z_h), and (Z _dr, LDR_hv), plus the temperature T. The range of values for each observable or pair of observables is derived from extensive modeling studies conducted earlier. The algorithms are tested using side-looking radar measurements from an aircraft, which was also equipped with particle probes producing simultaneous and nearly collocated shadow images of cloud ice crystals. The classification results from both algorithms agreed very well with the particle images. The two algorithms were in agreement by 89% in one case and 97% in the remaining three cases considered here. The most effective observable in the 1D algorithm was Z_dr, and in the 2D algorithm the pair (Z_dr, Z_h). LDR_hv had negligible effect in the 1D classification algorithm for the cases considered here. The temperature T was mainly effective in separating columnar crystals from the rest. The advantage of the 2D algorithm over the 1D algorithm was that it significantly reduced the dependence on T in two out of the four cases.