[M-GI24] Data assimilation: A fundamental approach in geosciences

[MGI24-P07] Intercomparison and ensemble project of coastal ocean prediction models in Japan

*広瀬 直毅¹、渡邉 修一²、木戸 晶一郎³、大石 俊⁴、広瀬 成章⁵、坂本 天⁶、印 貞治² (1.九州大学応用力学研究所、2.海洋科学振興財団、3.海洋研究開発機構、4.理化学研究所、5.気象庁気象研究所、6.オーシャンアイズ)

キーワード：海洋モデル、データ同化、沿岸海洋力学、マルチモデルアンサンブル

Ocean data assimilation and prediction models have been consistently advanced and become more and more popular these days. But the variety and discrepancy of systems might lead the user oscillations. Thus we decided to start intercomparison and ensemble project of most coastal ocean prediction models in Japan including DREAMS, MOVE, JCOPE, LORA, and SEAoME. First, we select a few small zones to closely compare the differences among these ocean DA models. The intercomparison is more important than the comparison to observation data at the initial phase. One purpose of this intercomparison is to define metrics to effectively measure the coastal ocean dynamics. Second, in-situ and remotely-sensed measurement data are used to find the problems of each ocean model. We probably need to repeat the forward and inverse estimation processes to improve the forecast performance of individual models. Third, the ensemble combinations are finally conducted and the multi-model ensemble prediction will be provided with the support of Japan Marine Science Foundation. The small experimental zones will be extended to all coastal waters along the Japanese archipelago after 2028.