PRACTICAL EVALUATION OF FOUR DIMENSIONAL VARIATIONAL DATA ASSIMILATION USING DUAL NUMBER
-
- NAGANO Takanori
- 大阪大学 大学院工学研究科
-
- IRIE Masayasu
- 大阪大学 大学院工学研究科
-
- OKADA Teruhisa
- 電力中央研究所 環境科学研究所
Bibliographic Information
- Other Title
-
- 二重数を用いた4次元変分法によるデータ同化の実用性評価
- ニジュウスウ オ モチイタ 4ジゲン ヘン ブンホウ ニ ヨル データ ドウカ ノ ジツヨウセイ ヒョウカ
Search this article
Description
<p> This paper presents introduction of auto differentiation with dual numbers to four-dimensional variational data assimilation method (4DVar) to reduce the coding work of tangent linear and adjoint models in Regional Ocean Modeling System (ROMS) and application with a biogeochemical model for Osaka Bay. The practical evaluation of the new 4DVar system was conducted by twin-experiments with vertical profiles of temperature, salinity, chlorophyll, dissolved oxygen at the head of the Bay. The new method which only used nonlinear model improved estimates of the biological state in the Bay as much as the hand-coded original 4DVar system, even though simulation run-time increased due to the calculation in dual parts.</p>
Journal
-
- Journal of Japan Society of Civil Engineers, Ser. B2 (Coastal Engineering)
-
Journal of Japan Society of Civil Engineers, Ser. B2 (Coastal Engineering) 76 (2), I_1003-I_1008, 2020
Japan Society of Civil Engineers
