基于双线偏振参量对北京X波段雷达非降水回波识别方法的研究

Study of Recognition Algorithm of Non-Precipitation Echo for Beijing X-Band Radar Based on Dual Polarization Parameter

为了提高中国X波段双线偏振雷达(简称双偏振雷达)的数据质量,提升应用水平,基于美国多雷达多传感器定量降水估测业务系统质量控制(简称dpQC)方案,在此基础上增加了差分反射率阈值判别并对方案中的计算参数进行本地化调整,设计了针对X波段双偏振雷达的质量控制(简称dpXQC)方案,原理主要包括:将双偏振参量相关系数(CC)区分降水回波与非降水回波的阈值调整为0.9,小于0.9的回波判别为非降水回波;通过计算回波顶高与强回波区位置对冰雹与非均一性波束填充区域进行标识;通过统计探空资料0℃高度上下相关系数的特征,调整了融化层的识别参数;优化算法将双偏振参量差分反射率绝对值大于5 dB判别为非降水回波;加入了条状杂波识别、连续性检验与斑块杂波识别算法。利用dpXQC方案对北京五部X波段多普勒双偏振业务雷达进行试验,检验评估其对非降水回波识别效果,获得以下结果与结论:大部分非降水回波可以通过CC阈值判别被滤除;对于CC阈值判别无法滤除的条幅状、孤立点、块状非降水回波,dpXQC方案利用条幅状杂波识别、连续性检验与差分反射率阈值判别去滤除;dpXQC方案有效保护了冰雹、非均一性波束填充与融化层区域中CC<0.9的降水回波不被滤除;检验结果显示,dpXQC方案对非降水回波正确识别率为91.8%,错误识别率为20.6%,均要优于dpQC方案,但dpXQC方案的错误识别率依然很高,主要有两方面原因,一是由于X波段雷达反射率衰减严重,方案会对一些离雷达较远的非均一性填充区域出现遗漏识别,二是融化层识别算法在所有方位角的位置与厚度是固定的,与实际观测不符,这均会导致被遗漏标识区域中CC<0.9的降水回波被误判为非降水回波。总体来看,dpXQC方案在滤除非降水回波的同时,最大程度保护了降水回波不被滤除,为雷达资料定量化应用等相关业务提供保障。

In order to improve the quality and application level of X-band dual polarization radar data,based on the dual polarization radar quality control(dpQC)scheme from National Multi-Radar and Multi-Sensor(MRMS)Quantitative Precipitation Estimation(QPE)System,a new quality control scheme(dpXQC)is designed for X-band dual polarization radar data.In this scheme,threshold discrimination of differential reflectivity is integrated and some algorithm parameters of the original scheme are adjusted locally.The main principles of the dpXQC scheme include the threshold distinguishing precipitation echo from non-precipitation echo is adjusted to 0.9 when correlation coefficient(CC)of radar echo is used and those less than 0.9 are judged as non-precipitation echoes based on the statistical data;hail and the non-uniform beam filling region are identified by calculating the radar echo top and the location of the powerful echo area;melting layer is distinguished through the feature of CC around 0℃height of sounding data;radar echo whose absolute value of differential reflectivity is larger than 5 dB is judged as non-precipitation echoes.The scheme also includes the algorithm of spike filter,continuous check and speckle filter.The dpXQC quality control scheme is used to do the test on the 5 Beijing X-band dual polarization radars,and the identification effect of non-precipitation echo is evaluated.The results show that most non-precipitation echoes can be filtered by the CC threshold.The residual non-precipitation echoes with CC larger than 0.9 including the stripe clutter,scattered clutter and the small-area clutter,can be removed by spike filter,continuous check and differential reflectivity threshold discrimination.The precipitation echo with CC less than 0.9 in the hail,the non-uniform beam filling and the melting layer region can be protected from CC filter by dpXQC scheme;the statistical result reveals that the hit rate of recognition of non-precipitation echoes of the dpXQC scheme is 91.8%while the false alarm rate is 20.8%,so it is better than dpQC scheme.There are two main reasons for the high false alarm rate:one is that some non-uniform beam filling regions far away from the radar are not identified by algorithm because of severe attenuation of X-band radar,and the other is that the location and thickness of melting layer are fixed in every azimuth in algorithm,which does not accord with the actual observation.These cause some precipitation echoes with CC less than 0.9 to be identified as non-precipitation echoes in the unrecognized region.Overall,the non-precipitation echoes can be filtered while most precipitation echoes could be returned so that some professional work about the quantitative application of X-band dual polarization radar can be ensured by the dpXQC quality control algorithm.