一次关于龙卷的天气雷达识别技术研究

A Research of Weather Radar Identification Techniques for Tornadoes

摘要: 龙卷是由空气对流运动造成的强烈的小范围涡旋，利用地基单偏振多普勒雷达以及双偏振雷达数据，使用反射率因子ZH和速度V，谱宽W之间的相关性，中气旋的识别算法叠加单体联合识别来对龙卷极端天气进行识别。单体联合识别算法使用反射率因子ZH，差分反射率ZDR和差传播相位常数KDP，利用DBSCAN聚类算法进行单体聚类。针对使用的阜宁龙卷雷达数据以及广州三水龙卷的雷达数据，识别结果表明，单体联合识别能有效提高识别率，该识别方法对龙卷的预警有参考意义。

Abstract: Tornadoes are intense, small-scale vortices caused by convective air motion. Using ground-based single-polarization Doppler radar as well as dual-polarization radar data, a medium-cyclone identification algorithm overlaid with single-unit joint identification is used to identify tornado extremes using the correlation between the reflectivity factor ZH and the velocity V, and the spectral width W. The single-unit joint identification algorithm uses the reflectivity factor ZH, the differential reflectivity ZDR and the differential propagation phase constant KDP to perform single-unit clustering using the DBSCAN clustering algorithm. For the radar data of the Funing tornado used and the radar data of the Sanshui tornado in Guangzhou, the recognition results show that the single-unit joint recognition can effectively improve the recognition rate, and the recognition method has reference significance for the early warning of tornadoes.