基于K means聚类分析的风廓线雷达降水数据判别方法

在降水环境下,风廓线雷达获得的多普勒信息主要是降水质点运动的结果,从而对大气风场的计算造成很大误差,因此,判断雷达回波信号是否受到降水干扰是很有必要的.采用一种基于K-means聚类算法思想的集合分析法对晴空和降水条件下的数据进行了聚类分析,得到随信噪比变化的垂直速度阈值,再根据该速度阈值对风廓线雷达数据是否受到降水干扰进行判别.采用此方法对南京边界层风廓线雷达2011年8-12月及2012年3-4月的部分观测数据进行了计算分析,结果表明,该方法可在不同的高度区分晴空和降水数据,能有效判别回波信号是否受到降水干扰.

Identification Method of Wind Profiler Precipitation Data Based on K Means Clustering Analysis

Wind profiler is a kind of clear air remote sense device. The Doppler radar information obtained under rain conditions is mainly results of moving precipitation particles, which causes great wind field calculation errors. It is of importance to distinguish whether echo signals are contaminated by precipitation. An ensemble analysis algorithm,based on the K Means clustering, is applied to acquire the vertical radial velocity threshold for SNR variation, which is further used to decide whether the radar data is interfered by rain drops.The method is tested with the observational data from August to December 2011 and March to April 2012 of the Nanjing CLC 11 wind profiler. The results show that this method can be used for distinguishing precipitation and clear air data at different heights, as well as decide properly whether echo signals are contaminated.