INDENTIFYING PATTERNS OF SATTELITE IMAGERY USING AN ARTIFICIAL NEURAL NETWORK

Abstract

An artificial neural network analysis based on the self-organizing map (SOM) was used to examine patterns of satellite imagery. This study used 3 × 4 SOM array to extract patterns of satellite-observed chlorophyll-a (chl-a) along the southern coast of the Lesser Sunda Islands from 1998 to 2006. The analyses indicated two characteristic spatial patterns, namely the northwest and the southeast monsoon patterns. The northwest monsoon pattern was characterized by a low chl-a concentration. In contrast, the southeast monsoon pattern was indicated by a high chl-a distributed along the southern coast of the Lesser Sunda Islands. Furthermore, this study demonstrated that the seasonal variations of those two patterns were related to the variations of winds and sea surface temperature (SST). The winds were predominantly southeasterly (northwesterly) during southeast (northwest) monsoon, drived offshore (onshore) Ekman transport and produced upwelling (downwelling) along the southern coasts of the Lesser Sunda Islands. Consequently, upwelling reduce dSST and helped replenish the surface water nutrients, thus supporting high chl-a concentration. Finally, this study demonstrated that the SOM method was very useful for the identifications of patterns in various satellite imageries.