PREDICTION OF THE CATCH OF JAPANESE SARDINE LARVAE IN SAGAMI BAY USING A NEURAL-NETWORK

We attempted to forecast the catch of post-larval stages of Japanese s ardine with total length 19-35 mm exploited each year by troll and bea ch seine fisheries in Sagami Bay, Japan, during March and April. In th e forecasting system, the feed forward (layered) type of neural networ k was utilized. The system for forecasting the catch in Sagami Bay dur ing March and April was developed on the basis of (a) predicted hydrog raphic conditions (occurrence rates of the Kuroshio path types and dis tance between the axis and Cape Iroh-zaki in March and April) as predi cted in the previous paper, (b) hydrographic data from November (previ ous year) to February (current year) and (c) Japanese sardine catch da ta in various landing regions in the previous year. The predicted valu es of catches agreed well with the observed catches. Upon investigatio n of the weights and threshold values in the trained neural network, t he distance between the Kuroshio axis and Cape Iroh-zaki was found to significantly affect the predictions. We also examined how the data le ngth for learning of the neural net affects the prediction. It appears that the neural network is a practical tool for predicting the catch.