摘要
利用1960—2013年渤海冰情等级资料和500 h Pa、1000 h Pa NCEP高度场再分析格点资料,采用滑动平均、相关分析、主成分分析等统计方法,找出了与渤海冰情密切相关的大气环流因子,探讨了采用主成分分析对渤海冰情进行定性、定量预测方法。结果表明:在主成分得分散点图上不同冰情年份的落区有一定的规律,根据年份的落点位置可以对冰情等级做出定性预估;采用主成分回归(PCR)方法建立的定量预测模型,具有较好的历史拟合。对2005—2014年的独立样本试报试验,相关系数为0.76,通过了α=0.01的显著性检验。
Based on the sea-ice grade data in the Bohai Sea from 1960 to 2013 and the NCEP/NCAR reanalysis data at 500 h Pa and 1000 h Pa, the atmospheric circulation factors associating closely to sea-ice grade were selected by using the statistical methods of moving average, the correlation analysis and the principal component analysis(PCA), and the qualitative and quantitative prediction method for sea-ice conditions in the Bohai Sea was investigated. The results show that there is some rule of falling areas in different sea-ice grades at the PCA scatter chart, the qualitative prediction of sea-ice grades in the Bohai Sea can be estimated according to the falling areas of the year. The quantitative forecast model is established by the principal component regression method(PCR) and has a good historical fitting rate. The correlation coefficient is 0.76 with significant at the confidence level of 0.01 for the independent samples test from 2005 to 2014.
出处
《海洋预报》
2015年第6期74-79,共6页
Marine Forecasts
基金
河北省科技厅渤海海冰发生规律及预测技术研究(14273501D)
关键词
环流异常
环流因子
主成分分析
海冰预测
circulation anomalies
circulation factors
PCA
sea ice prediction
