实测潮位异常值判别方法比较

Comparison of discriminating methods on the measured tide level outliers

针对近岸自动测站实测潮位数据质量偏低的问题,分别采用拉伊达准则、肖维勒准则和方国洪准则对其异常值进行自动判别,利用T_TIDE软件对判别后的实测潮位进行调和回报分析,以检验不同准则的判别效果。结果表明,采用拉伊达准则可以对潮位数据序列中异常值密集和连续出现的情况有较好的判别效果,总体上拉伊达准则较肖维勒准则和方国洪准则表现更优。通过异常值判别后,潮位调和分析的回报精度有明显提高。

In view of the low data quality of the actual measured tidal level in the near-shore automatic station,the Rajda criterion,Chauvenet criterion and Fang Guohong criterion were used to automatically determine the abnormal value,and the T_TIDE software was used to perform the tune-in-return analysis on the measured tide level after the discrimination,which could test the distinguishing effects of different criteria.The results showed that the Rajda criterion could be well used to discriminate the occurrence of dense and continuous outliers in tidal level data series.In general,the Rajda criterion performed better than the Chauvenet criterion and Fang Guohong criterion.The return precision of the tidal harmonic analysis was obviously improved after the discrimination of the abnormal value.