Localized sea surface temperature variability in the Yellow and East China Seas

Abstract

The Yellow and East China Seas (YECS) are widely believed to have experienced a robust, large-scale warming during the last decades. However, AVHRR-based optimum interpolation sea surface temperature (SST) after removing annual cycles reveals that long-term SST variabilities are characterized with a temporal pattern consisting of persistent warming by late 1990s, two SST peaks occurred at the late 1990s and mid 2000s, and weak cooling for the last decade. Spatially, significant warming trend downstream of the Changjian River is in contrast to weak cooling on the eastern part of the Yellow Sea. To explain this localized variability, this study adopts cyclostationary empirical orthogonal function analysis that separates principal modes with seasonally cycled loading vectors for the period of 1982&#8722 2014. Among dominant modes, the first and third modes indicate the long-term SST changes for winter and summer respectively, while the second mode captures the interannual variability. The first mode, accounting for 25% of the total variability in the SST anomalies, appears to be associated with the meridional wind anomalies in winter. The principal component for this mode is significantly correlated with the Pacific decadal oscillation index only for the early warming decades by late 1990s and then its correlation sharply drops to an insignificant level, implying that a potential regime shift occurred in dynamical linkals that long-term SST variabilities are characterized with a temporal pattern consisting of persistent warming by late 1990s, two SST peaks occurred at the late 1990s and mid 2000s, and weak cooling for the last decade. Spatially, significant warming trend downstream of the Changjian River is in contrast to weak cooling on the eastern part of the Yellow Sea. To explain this localized variability, this study adopts cyclostationary empirical orthogonal function analysis that separates principal modes with seasonally cycled loading vectors for the period of 1982&#8722 2014. Among dominant modes, the first and third modes indicate the long-term SST changes for winter and summer respectively, while the second mode captures the interannual variability. The first mode, accounting for 25% of the total variability in the SST anomalies, appears to be associated with the meridional wind anomalies in winter. The principal component for this mode is significantly correlated with the Pacific decadal oscillation index only for the early warming decades by late 1990s and then its correlation sharply drops to an insignificant level, implying that a potential regime shift occurred in dynamical linka