FY-3A MWHS资料的质量控制——以四川“8.19”暴雨为例

FY-3A MWHS微波湿度计辐射率资料空间分辨率高,同化后能有效地提高中尺度系统的数值预报的准确率。但如果不进行质量控制,任由误差大的离群微波资料进入同化系统,将会降低分析场精度,影响预报准确率。本文选取2010年8月17~19日的暴雨个例,应用主成分分析(PCA)方法,找出FY-3A微波湿度计数据中的离群值。结果表明:采用PCA方法,3天的资料总共识别出148个离群值样本,占总数的19.3%,其中通道三离群值占5.5%,通道四离群值占8.2%,通道五占5.4%。去除这些离群值后的MWHS数据总体分布,更靠近数据中心,各EOF方差贡献更加平滑。这说明PCA方法是一种可以抵抗少数离群值对总平均值影响的有效质量控制方法。 

FY-3A MWHS Data Quality Control

The data of FY-3A MWHS has high spatial resolution,and after assimilation the accuracy of the numerical forecast can be effectively improved.But the accuracy of analysis would be brought down without quality control before assimilation.In order to study and selected the outliers data of FY-3A MWHS data,the PCA(Principal Component Analysis) methods has been applied in a rainfall storm case on August 17 in 2010.The results indicated that there were 148 outliers data identified and it accounted for 19.3% of the total by the PCA method in the three-day information.It accounted for 5.5% in Channel 3,8.2% in channel 4,and 5.4% in channel 5,respectively.The distribution of MWHS data got much closer to data center and the variance contribution of EOF(Empirical Orthogonal Function) became much smoother after all the outliers data had been removed,which demonstrated the PCA was an effective quality control methods to reduce the influence on the average value by the outliers data.