海洋理工学会誌
Online ISSN : 2188-3262
Print ISSN : 1341-2752
ISSN-L : 1341-2752
原著
ニュートラルネット解析を利用した東北海域の海況パターンの予測
為石 日出生小松 輝久青木 一郎杉本 隆成
著者情報
ジャーナル フリー

1996 年 2 巻 2 号 p. 97-104

詳細
抄録

Importance of forecasting of fisheries condition utiling oceanographic features has been now well established and this is going to play a very important role in sustenable management of fisheries resources in feature. In this study, it was attempted to forecast the oceanographic feature pattern from the oceanographic information using neural network. As the factors of input unit layer, hydrographicprincipal factors (warm-core ring off Sanriku, warm-core ring off Kushiro.nearshore intrusion of Kuroshio, offshore intrusion of Kuroshio, first branch of Oyashio, second branch of Oyashio) were utilized in this study off Tohoku. Hidden layerwas fixedfor only one layer.and it had fourhidden nervecells.The abovefactors from 1981 to 1993ware made use of in the system of neural network for forecasting oceanographic features in 1994 and 1995. And then.this system was valued by comparison of estimated pattern of oceanographic features with real one. It was made certain that the estimated pattern of oceanographic features agreeable with the real one within 60 nautical miles, and this neural network system would be useful for application of forecast of oceanographic feature pattern.

著者関連情報
© 2018 海洋理工学会
前の記事 次の記事
feedback
Top