This paper provides an artificial neural network model for assessing the environmental change induced by human impacts with long term developments in Tokyo Bay. Highly-interconnected network of non-linear neurons are shown to be very effective in estimating the seasonal change of water quality and habitats in the shallow water zone.
The neural network constructed herein consists of a number of simple and interconnected processing elements and evaluates the interactions of biotic, chemical and physical processes in the mud flats. The back-propagation training algorithm is introduced to minimize the mean square error between the actual output of a multilayer feed-forward processing and the desired output. As a verification of the procedure using the neural network proposed here, the final outputs from the network are compared with the observational data. A successfully trained neural networks are set up and its usefulness is examined.