Quality Control Of C-Band Doppler Weather Radar Reflectivity Factor

Doppler weather radar data are widely used in numerical weather forecasting,radar quantitative precipitation estimation,and short-term nowcasting.However,in addition to meteorological echoes,the echo signals received by the radar also include non-meteorological echoes such as ground clutters and clear-air echoes.These non-meteorological echoes will have a negative impact on the use of radar data.Current research on quality control of radar data is mainly aimed at S-band radars in the east.The terrain in western China is complex,and non-meteorological echoes such as ground object echoes are more serious.The quality control of radar data is more difficult.This paper proposes a radar reflectivity factor quality control hybrid algorithm(con-NBC)based on the Na(?)ve Bayes classifier,which aims to improve the data quality of C-band Doppler weather radar.First,Na(?)ve Bayes classifier is used to identify precipitation echoes and nonprecipitation echoes,and on the basis of the echo recognition results,Sun Spike filter,speckle filter,and hole filling are used to obtain a relatively complete precipitation echo information.This paper uses the detection data of 7 C-band Doppler weather radars(Xi'an,Yan'an,Yulin,Shangluo,Ankang,Hangzhong and Baoji)in Shaanxi Province to study the radar echo characteristics and the effect of the con-NBC algorithm.The main contents and conclusions are as follows:(1)Through the manual classification of the radar echoes of 7 C-band Doppler weather radars in Shaanxi Province in August 2018,the probability distribution map of the radar echoes and the probability density functions of the characteristic quantities of different types of echoes were obtained.The results show that due to the different terrains,radar beam blockage of each radar are different.Radar beam blockage of the four radars in Xi'an,Yan'an,Yulin,and Baoji are relatively weak,while the remaining three radars are seriously occluded.And the distribution characteristics of ground clutters and clear-air echoes are different among different radars.For different types of echoes,the spatial distribution characteristics of the corresponding feature fields(Z,Td BZ,SPIN,ETH,vgd BZ)are also different,and the feature fields can reflect the differences between precipitation echoes,ground clutters and clear-air echoes to a certain extent,but it is difficult to accurately judge the echo type merely by one single feature field.The distribution characteristics of feature fields of precipitation echoes,ground clutters and clear-air echoes obtained by the statistics of 7 radars in Shaanxi Province are not completely consistent.Therefore,it is necessary to build a Na(?)ve Bayes classifier by obtaining the probability density distribution functions of its own feature fields for each radar.(2)Using the precipitation cases in Shaanxi Province in 2019,the quality control effects of the con-NBC algorithm are analyzed,and the results are compared with the quality control results of the radar reflectivity factor of Shaanxi Province business operation.The results show that using the Na(?)ve Bayes classifier established by the probability density distribution functions of the feature fields obtained by statistics,precipitation echoes and non-precipitation echoes can be well identified in stratiform cloud precipitation,scattered convective precipitation and mixed cloud precipitation.The Sun Spike filter can effectively identify and remove the Sun Spike mixed with the precipitation echoes and retain the precipitation echoes.The speckle filter can effectively remove the scattered non-precipitation echoes that have not been identified by the Na(?)ve Bayes classifier,and hole filling algorithm can retrieve the misidentified precipitation echoes and preserve the integrity of the precipitation systems.The con-NBC algorithm can better identify the precipitation echoes in different types of precipitation,and can more completely retain the precipitation information,especially for the convection cell at close range of the radar,which can be more accurate Identify and keep.