The objectives of this research were to determine spatial patterns of fish well and this spatial patterns were closely associated with physical habitat and chemical water quality. Such relations of physical habitat and chemical water quality to high trophic-level biota (i.e., fish) are very complex in analyzing the cause-effective relations due to large ecological dataset, thus an approach of self-organizing map (SOM) has been proposed in elucidating complex ecological relations of aquatic ecosystems. Communities from up-stream to down-stream of Geum River using an approach of Self-Organizing Maps (SOMs) and analyze their relations to water chemistry. For the study, fish and water samplings were conducted in 130 different streams and rivers during 2008 - 2009. The analysis of data using the SOMs model showed that fish community had longitudinal gradients of up-stream, mid-stream and down-stream. The clustering of the trained SOMs units reflected the stream morphology, land-use pattern and water quality, resulting in influenced the ecological trophic compositions and tolerance of top-level fish in the aquatic ecosystem