庄硕, 王萍, 侯洁
AuthorsHTML:庄硕, 王萍, 侯洁
AuthorsListE:Zhuang Shuo, Wang Ping, Hou Jie
AuthorsHTMLE:Zhuang Shuo, Wang Ping, Hou Jie
Unit:天津大学电气自动化与信息工程学院，天津 300072
Unit_EngLish:School of Electrical and Information Engineering, Tianjin University, Tianjin 300072, China
Abstract_Chinese:气旋和反气旋系统在理想条件下具有中心对称的特点, 利用来自欧洲中心数值预报产品中的风场格点数据可以实现对这类风场的检测．通过构建多尺度模板匹配规则, 实现气旋和反气旋的初步识别, 同时得到风场的多维向量描述子．通过描述子的运算, 得到基于不对称度及有效风向空缺率的实际气旋的形变指数及隶属度函数, 实现对初检结果的去伪存真．然后对检测点做聚类处理, 得到气旋和反气旋集合, 在每个集合内部根据涡度和风速零点实现气旋和反气旋中心定位．实验表明, 提出的气旋和反气旋自动识别方法的检出率(POD)达到91% 以上, 临界成功指数(CSI)约为77.4% , 中心定准率(LAR)达到88.2% , 实现了气旋和反气旋的自动识别, 同时可以克服高低压中心与气旋中心不一致时, 靠高低压中心识别气旋和反气旋并定位其中心的问题．
Abstract_English:Generally，cyclones and anticyclones are centrally symmetrical under ideal conditions. Wind field grid data from European Centre Medium-Range Weather Forecasts(ECMWF)was used to identify these centrally symmetrical wind fields. Firstly，in order to realize the preliminary identification of central symmetrical wind fields，a multi-scale matching template and formula were built，which contained rules of simplifying and matching wind direction. Meanwhile，a multi-dimensional vector descriptor of wind field was gained. With the operation of the descriptor，the deformation exponent and membership degree function based on asymmetry degree and wind direction vacancy rate were presented to describe the actual cyclones and anticyclones. Thus the real cyclone and anticyclone of preliminary results were obtained. Secondly，the cyclone and anticyclone set was pursued by clustering the wind field testing points. Using each clustering set，the center location was detected according to the vorticity and zero-speed point. Experimental results showed that the probability of detection(POD)of cyclones and anticyclones reached 91% . The critical success index(CSI)was about 77.4% and the location accuracy ratio(LAR)was 88.2% . The presented method could realize the cyclone identification and its center recognition when local minima under sea level pressure(SLP)did not agree with the cyclone center.
Keyword_Chinese:气旋; 隶属度函数; 聚类; 涡度
Keywords_English:cyclone; membership degree function; clustering; vorticity

PDF全文下载地址:http://xbzrb.tju.edu.cn/#/digest?ArticleID=5951