双线偏振雷达差分传播相移的质量控制

Quality Control of Differential Propagation Phase Shift for Dual Linear Polarization Radar

利用中国气象科学研究院灾害天气国家重点实验室的车载C波段双线偏振多普勒雷达(C-band PolarimetricDoppler Radar on Wheel,CPDRW)的外场试验,在统计分析降水、地物回波差分传播相移φ_(DP)数据的差别与信噪比关系等基础上,提出了一套数据分析和处理的方法。该方法通过φ_(DP)的异常波动并结合回波的强度Z_H和速度V_r信息将地物回波信号分离出来,在降水估测或衰减订正等定量应用时将其剔除。对于气象回波则根据信噪比及零滞后互相关系数ρ_(HV)(0)将φ_(DP)资料分为较好、较差和差3类。对于较好数据直接进行后续的预处理,对于较差数据先订正后处理,而对于差数据将其剔除以保证φ_(DP)资料的整体质量。经过大量资料的验证,该方法在最大程度上保留气象信息的同时也保证了φ_(DP)资料的质量,并能估算出质量较高的差分传播相移率K_(DP)资料。

Data processing and quality control is the foundation of the application of dual-linear polarization Doppler radar. Based on the observation in field experiments by a C-band Polarization Doppler Radar on Wheel （CPDRW）, the difference of differential propagation phase shift φDPe between precipitation and ground clutter and its relationship with signal-to-noise ratio SNR are analyzed and a new data analyzing and processing methodology is suggested. According to this new method, the useless φDP data can be given up and the KDp data with higher accuracy can be acquired. Analysis indicates that φDP data are vulnerable to the influence of the non-meteorological target like ground clutter and usually appears large fluctuations. φDP data are also sensitive to the variability of SNR and cross-correlation coefficient pHV（0）, especially the latter. It appears abnormal fluctuations with the quality of related SNR and pHv（0） becomes poor and that will affect the quality of the estimation of KDp data if no appropriate quality control scheme is adopted. U- sing this kind of KDp data, obvious errors in the quantitative application of precipitation estimation and pre- cipitation particle morphology recognition can be obtained. In this new method, the abnormal volatility of φDP data combining with reflectivity factor ZH and radial velocity Vr information is used to isolate the ground clutter, and then improper data are eliminated in the quantitative application such as quantitative precipitation estimation or attenuation correction. According to SNR and PHv （0）, the meteorological data is divided into good, poor and bad categories. For the good data, the fluctuation is smaller, the increasing trend with distances which accords with theoretical expectations is evident, so the preprocessing algo- rithms and estimate KDp data can be used directly; for the poor data, although the fluctuation is more pro- nounced than the good data, the data continuity begins to become poor and there are some φDP data ＂pile＂ and ＂depression＂, however, much weather information remains and the variation trend is also obvious, so the data correction algorithm are applied so as not to affect the estimated K^p data; and for the bad data, it not only has the large fluctuation, the overall variation trend is also difficult to identify, sometimes even negative growth phenomenon appears which is contrary to the theory, so they are eliminated to ensure the overall quality of φDP data. After a large number of actual data validation, it reveals that the suggested method can keep the meteorological information to the greatest extent and ensure the overall quality of φDP data at the same time, and it can also estimate the high quality of KDp data.