[H-DS11] Tsunami and tsunami forecast

[HDS11-03] Tsunami Data Assimilation of High-Frequency Radar-Derived Surface Currents: Accounting for Measurement Error Distribution

*Muhammad Irham Sahana¹, Ryotaro Fuji^1,2, Tomoyuki Takahashi⁶, Hirofumi Hinata^1,3,4,5 (1.Graduate School of Science and Engineering, Faculty of Engineering, Ehime University, 2.Hydrologic Cycle and Marine Department, Kokusai Kogyo Co., Ltd., 3.Institute for the Promotion of Science and Technology, Center for Marine Environmental Studies, Ehime University, 4.Institute for Collaborative Relations, Center for Disaster Management Informatics Research, Ehime University, 5.South Ehime Fisheries Research Center, Ehime University, 6.Graduate School of Societal Safety Sciences, Faculty of Societal Safety Sciences, Kansai Univeristy)

Keywords:Tsunami Data Assimilation, High-frequency Radar, Measurement Error, Stable Assimilation

Assimilating tsunami-induced current fields using high-frequency (HF) radar has been increasingly discussed. To measure the surface current velocities, utilizing two or more radars located at different sites is required. However, the performance of surface current velocity measurement depends on the azimuthal difference between crossing radar beams at the measurement points. Thus, the appropriate estimation of the errors is needed to mitigate the adverse effects of HF radar measurement errors. In this study, we estimated the beam angle-dependent measurement errors of the east-west and north-south velocity components through a principal component analysis (PCA) and incorporated them into a tsunami data assimilation, based on the optimal interpolation method for HF radar-derived surface current velocities. Assuming a time-independent and uniform standard deviation of Gaussian noise (STD= 5 cm/s) in fifteen independent assimilation experiments, we could reasonably assimilate tsunami-induced velocities and estimate the tsunami wave height properly. A significant improvement of up to 29% (on average) in assimilation performance was observed at coastal stations, particularly for a scenario with a 1-m initial maximum sea surface height, and a uniform water depth of 500 m (resulting in a 9 cm/s tsunami-induced velocity at radar coverage). Considering the beam angle dependency on the measurement errors distribution effectively reduced the impact of error-induced tsunamis, leading to stable and reasonable assimilations, with lower standard deviations across multiple assimilation performances.