中文摘要：

基于传统的交叉相关追踪回波（TREC）算法，通过对“匹配元”所含的多普勒雷达反射率进行一定的分类来动态选择符合判定条件的“匹配元”进行交叉相关计算．为了提高反演风场的可靠性和稳定性，通过选取多重不同尺度的“匹配元”、分位法剔除风向奇异点、用椭圆型权重再插值等技术来订正反演的风场，在一定程度上弥补了传统TREC技术对较小尺度回波云团精度判定不足的缺陷，使得“匹配元”中的雷达反射率要素特征覆盖更有效．结果表明：对TREC算法进行修订后得到的反演风场，可以更好地反映雷达回波团的整体流场，同时一些回波单体的个性风场特征得到较好的保留．

英文摘要：

Doppler radar reflectivity which ＂matching cell＂ contained was classified based on the traditional tracking of radar echo with correlations（TREC） algorithm. Then certain qualified matching cell＇was dynamically selected to perform cross-correlation calculation. In order to improve the reliability and stability of retrieval wind, the following corrective techniques were adopted： selected a different multiscale ＂matching cell＂, eliminated singular points with percentile method, re-interpolated by means of elliptic weight. The deficiency of accuracy was made up to a certain extent for selecting small-scale radar echoes with traditional TREC technique. The characteristics of radar echoes were covered by ＂matching cell＂ more efficiently. Results showed that： the retrieval wind can reflect the whole circulation fields of radar echoes after improving the TREC algorithm, while the individual characteristics of the wind field in some radar echoes were well reserved.