Seasonal and interannual patterns of sea surface temperature in Banda Sea as revealed by self-organizing map

Seasonal and interannual variations of sea surface temperature (SST) in the Banda Sea are studied for the period of January 1985 through December 2007. A neural network pattern recognition approach based on self-organizing map (SOM) has been applied to monthly SST from the Advanced Very High Resolution Radiometer (AVHRR) Oceans Pathfinder. The principal conclusions of this paper are outlined as follows. There are three different patterns associated with the variations in the monsoonal winds: the southeast and northwest monsoon patterns, and the monsoon-break patterns. The southeast monsoon pattern is characterized by low SST due to the prevailing southeasterly winds that drive Ekman upwelling. The northwest monsoon pattern, on the other hand, is one of high SST distributed uniformly in space. The monsoon-break pattern is a transitional pattern between the northwest and southeast monsoon patterns, which is characterized by moderate SST patterns. On interannual time-scale, the SST variations are significantly influenced by the El Niٌo-Southern Oscillation (ENSO) and Indian Ocean Dipole (IOD) phenomena. Low SST is observed during El Niٌo and/or positive IOD events, while high SST appears during La Niٌa event. Low SST in the Banda Sea during positive IOD event is induced by upwelling Kelvin waves generated in the equatorial Indian Ocean which propagate along the southern coast of Sumatra and Java before entering the Banda Sea through the Lombok and Ombai Straits as well as through the Timor Passage. On the other hand, during El Niٌo (La Niٌa) events, upwelling (downwelling) Rossby waves associated with off-equatorial divergence (convergence) in response to the equatorial westerly (easterly) winds in the Pacific, partly scattered into the Indonesian archipelago which in turn induce cool (warm) SST in the Banda Sea.