Ji Yang1, Kun Zhao1, Guifu Zhang2
1 Key Laboratory for Mesoscale Severe Weather/MOE and School of Atmospheric Science, Nanjing University, Nanjing, China
2 School of Meteorology and Advanced Radar Research Center, University of Oklahoma, Norman Oklahoma 73072, USA
A hydrometeor classification algorithm (HCA) based on the Bayes’s theorem is developed for C-band polarimetric radar observation during the Observation, Prediction and Analysis of Severe Convection of China (OPACC) field campaign in 2014. In contrast to the traditional fuzzy logic approach that uses the empirical membership function, the Bayesian hydrometeor classification algorithm takes a prior occurrence frequency of hydrometeor classification and can overcome the effect of radar noise and calibration error. Five radar variables, including radar reflectivity factor at horizontal polarization (ZH), differential reflectivity(ZDR), the cross-correlation coefficient(RHV), standard deviation of ZH(SDZ), standard deviation of differential phase shift(SDP), are used to identify seven different hydrometeor types. The conditional probability distribution function for each class is constructed by analyzing corresponding hydrometeor data, which is selected by manual. The prior probability distribution function is constructed from training sets, which are produced from the fuzzy-logic based HCA. The new HCA is applied to a squall line that occurred 30 July 2014. Results show that the Bayesian HCA can produce the more reasonable hydrometer classification than the traditional fuzzy logic classifier.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner