A neural network-based cluster technique called self-organizing map (SOM) determines 12 distinguishable patterns derived from the sea surface temperature (SST) anomaly fields in the South Indian Ocean for 1901–2010 in austral autumn (March–May). Two out of twelve patterns (patterns 1 and 12) are characterized by a dipole pattern with opposite SST anomalies in the Southwest Indian Ocean and to its northeastern side, which is similar to the so-called Indian Ocean Subtropical Dipole (IOSD), and they are named as traditional IOSD (T-IOSD) in the present work. Patterns 2 and 11 show a similar dipole pattern of a northeast-southwest orientation, which can be significantly distinguished from T-IOSD due to their distinct location of action centers, amplitudes and evolution processes. These two patterns were called new IOSD (N-IOSD). For both positive and negative positive phase of T-IOSD, the anomalous dipole pattern is first observed near Madagascar in previous spring and it migrates eastward and forms the typical pattern of T-IOSD in early summer. Differently, for positive phase of N-IOSD, significant SST anomalies appear to the west of Australia continent at first in late spring, and a well-organized dipole pattern could not be observed until late summer, approximately one or two months later than the formation time of T-IOSD. The major asymmetry of N-IOSD characterizes one month later of the formation time for the negative phase of N-IOSD, compared with its positive phase. The possible mechanisms responsible for different evolution process of T-IOSD and N-IOSD events were examined. The results show that latent heat flux, not surprisingly, plays an essential role in the formation of both T-IOSD and N-IOSD. However, the contribution of meridional advection by anomalous oceanic current to the N-IOSD event is much more important than that to the T-IOSD.
Read full abstract