ONE WEEK WAVE PREDICTION METHOD BY NEURAL NETWORK USING GLOBAL WAVE FORECAST DATA
-
- Tracey H. A. TOM
- (株)ハイドロ総合技術研究所
-
- KIM Sooyoul
- 鳥取大学 工学研究科
-
- TAKEDA Masahide
- 東亜建設工業(株)技術研究開発センター
-
- KURAHARA Yoshinosuke
- 東亜建設工業(株)技術研究開発センター
-
- HARA Chisato
- 東亜建設工業(株)技術研究開発センター
-
- NISHIYAMA Yamato
- 東亜建設工業(株)技術研究開発センター
-
- KAWASAKI Koji
- (株)ハイドロ総合技術研究所 研究開発センター
-
- MASE Hajime
- 京都大学 東亜建設工業(株)
Bibliographic Information
- Other Title
-
- 全球波浪予報値のニューラルネットワーク変換による高精度1週間波浪予測の試み
- ゼン キュウ ハロウ ヨホウチ ノ ニューラルネットワーク ヘンカン ニ ヨル コウセイド 1シュウカン ハロウ ヨソク ノ ココロミ
Search this article
Abstract
<p> Currently there are three kinds of global wave forecasts for more than one week in the world; however, when targeting a location nearshore on the coast of Japan, the accuracy of their predictions, from the spatial resolution point of view can only be estimated. This study proposed to improve the one week wave prediction accuracy using an artificial neural network with global wave forecasts data, similar to our Group Method of Data Handling (GMDH) model of one week ahead wave prediction. At the target NOWPHAS location of Port Hitachinaka, the predictions of significant wave heights agreed well with the observed ones when using two global wave data by both the NOAA wave model and ECMWF wave model; for significant wave periods, the data by the Japan Meteorological Agency wave model and ECMWF wave model gave the best performance. It was found that it is not necessary to use all three kinds of global wave prediction values to improve prediction values. The results of this study using an artificial neural network are similar to those using GMDH method.</p>
Journal
-
- Journal of Japan Society of Civil Engineers, Ser. B2 (Coastal Engineering)
-
Journal of Japan Society of Civil Engineers, Ser. B2 (Coastal Engineering) 75 (2), I_133-I_138, 2019
Japan Society of Civil Engineers