Analysis and Processing of Long Sequence and Massive Temperature and Salinity Data of the North Yellow Sea from 2010 to 2019 [2010~2019年北黄海海域长序列海量温盐数据分析与处理方法]

作者：Chen X.; Liu C.-H.^*; Liu Z.-L.; Wang X.; Wang C.-X.; Jia S.-Y.

来源：海洋与湖沼, 2022, 53(1): 49-61.

DOI：10.11693/hyhz20210700170

• Abnormal data
• Imputation processing
• Missing data
• North Yellow Sea
• Salinity
• Temperature

Ensuring the integrity and reliability of long-term continuous buoy data is the primary issue for the application of the data. Five sets of buoys in the Yellow Sea located in the waters near Changhai County, North Yellow Sea deployed by the Chinese Academy of Sciences Offshore Observation and Research Network were used. Data analysis and processing methods of the sea surface temperature and salt data collected by the buoys for 10 years from 2010 to 2019 were studied. To identify the abnormal values in the original temperature and salinity data, the extreme value method, the Laida criterion, and the box plot method were compared to find the best one to treat abnormal data. In the 2σ principle with the box diagram method, the boundary values were adjusted. In addition, to address the data missing, interpolation combining the SoftImpute and IterativeImpute was proposed, by which the standard deviations of the data could be effectively reduced. Results show that the methods are effective and can be used to eliminate anomalies and imputation defects, correct abnormal points, smooth out data curve, and highlight significant interannual variations and trends in the study sea area. This study provided a reference for enhancing marine observation data for future research.