雷达回波反射率垂直剖面图的冰雹识别方法  

Hail Recognition Using Radar Echo Reflectivity Cross Section

作　　者：路志英[1] 朱俊秀 田硕[1] 贾惠珍[2] 

机构地区：[1]天津大学电气与自动化工程学院,天津300072 [2]天津市气象局,天津300074

出　　处：《天津大学学报（自然科学与工程技术版）》2015年第8期742-749,共8页Journal of Tianjin University：Science and Technology

基　　金：天津市自然科学基金资助项目(14JCYBJC21800)

摘　　要：为了更加准确地识别冰雹天气,对某地区2005—2010年间24个冰雹过程和19个暴雨过程的825个雷达样本数据进行了分析处理,建立了基于雷达回波反射率垂直剖面图的冰雹自动识别的客观模型.在分析冰雹云体形成机理和结构的基础上确定最佳剖线,并用插值法生成雷达回波反射率垂直剖面图,通过图像处理方法,提取特征数据(强回波(45,d BZ以上)与0,℃和-20,℃温度层的高度差、弱回波区和有界弱回波区的宽度和高度).然后采用粗未选集理论数据挖掘方法对相关特征数据进行处理,建立了自动识别冰雹天气的客观模型.测试结果表明:该识别模型的判别规则对28个冰雹天气过程的383个样本的正确识别命中率是82.77%,可有效地识别和预报冰雹,有助于减轻冰雹灾害天气造成的损失.To recognize the hail weather more accurately,852 radar base data were analyzed. These data were col-lected from 24 hail processes and 19 heavy rain processes in a certain area between 2005 and 2010. Automatic identi-fication model for hail was built by using radar reflectivity cross section. The optimal cutting lines were determined based on the analysis on the forming mechanism and structure of hail cloud. Radar reflectivity cross section was ob-tained by using interpolation algorithm and was analyzed with image processing methods. Features data such as the height differences between strong echo（over 45,dBZ）and 0,℃ as well as-20,℃ isothermal layer,and the height and width of weak echo recognition and bounded weak echo recognition were obtained. Rules for hail and heavy rain recognition were acquired by using data mining based on rough set,and automatic identification model was built. 385 hail samples from 28 hail processes were distinguished with this model. The test results show that the accuracy of rec-ognition rate is 82.77%. This provides an effective method for identification and short-time forecast for hail,and it is helpful for reducing the loss caused by hail.

关 键 词：图像处理 弱回波区 有界弱回波区 粗糙集 数据挖掘 

分 类 号：P412.25[天文地球—大气科学及气象学]