Optimization Study on Dual-Polarization Radar Hydrometeor Classification Method for Winter

During winter precipitation processes, the melting layer exhibits low altitude, rapid variation, and spatial heterogeneity, leading to significantly decreased identification accuracy of traditional dual-polarization radar hydrometeor classification algorithms in rain-snow transition regions. Data and Methods Using observational data from 87 dual-polarization radars and vertical profile data from 116 sounding stations across China from July 2023 to July 2024, an improved scheme combining sounding spatiotemporal matching optimization, QVP (Quasi-Vertical Profile) real-time average melting layer monitoring, and three-dimensional MLDA (Melting Layer Detection Algorithm) spatially refined identification was adopted. Objective Research was conducted from two aspects: improving melting layer spatiotemporal positioning accuracy and enhancing the regional applicability of the algorithm, to improve the discriminability of different phase types. Results The improved scheme effectively addressed the problems of rapid variation and spatial heterogeneity of winter melting layer, improving the temporal resolution of melting layer monitoring from 6-12 hours to 6 minutes and achieving three-dimensional spatially refined identification with resolutions of 1 km in range, 0.1 km in height, and 1°in azimuth. Sensitivity experiments using winter data from 7 radars around Nanjing demonstrated that the improved scheme accurately identified the rain-snow boundary, with overall phase identification accuracy improved by 11.91% and rain-snow mixed-phase identification accuracy improved by over 50%. Conclusions Statistics of winter algorithm application duration from 87 radars nationwide show that the annual switching ratio of winter algorithm at plain sites and mountain sites in Northeast China, North China, and Central China reaches 18%-51%, indicating that the improved scheme provides an effective technical approach for winter dual-polarization radar hydrometeor phase identification.