We attempted to extract chemical parameters characterizing the upper, middle and lower stream of the Tamagawa River in Tokyo, Japan, by Compensation Quantitative Structure-Activity Relationships (CQSAR) method [11, 12] for analysis of incomplete data by perceptron type neural network with three layers.
Dissolved Oxygen (DO) and Total Phosphorus (T-P) were extracted from 12 chemical parameters (Table 2) at 17 points (Table 1) along the main stream of the Tamagawa River (Figure 1) in Tokyo, Japan, for 1997-2001 [10].
DO indicated that water pollution became worse at all points of the Tamagawa River (Figure 4) but T-P shows that water pollution is changed for the better in the middle and the lower stream (Figure 5).
This result suggests that the water pollution is more drastic in the middle stream of the Tamagawa River than the lower stream. The water pollution in the middle stream should be decreased for keeping the Tamagawa River clean.
View full abstract