We investigated seasonal and annual variation of demersal fish collected by otter trawl in the eastern Yellow Sea from 2004 to 2006. A total 145 species under 63 families were collected during the survey and were consisted of cold and warm water fishes. Number of individuals of small sized fish was relatively abundant in the summer and autumn than the winter and spring, whereas biomass was high in the autumn with growing migratory fish.
To analyze spatio-temporal dynamics of the fish data, we implemented a self-organizing map (SOM), an unsupervised artificial neural network. Based on SOM, the samples were classified into two main clusters. The samples from the offshore areas with deeper, high temperature and salinity including Yellow Sea Bottom Cold Water (YSBCW) was grouped, while the other group was characterized the samples in the coastal areas with shallow and high seasonal variation of temperature and salinity. Furthermore, the main clusters were divided to two sub-clusters of south and north areas.