Development of data assimilation methods and observation systems for a wind digital twin in urban areas

Onodera, Naoyuki   ; Shimokawabe, Takashi*; Idomura, Yasuhiro   ; Kawamura, Takuma ; Asahi, Yuichi   ; Hasegawa, Yuta   ; Ina, Takuya ; Shimomura, Kazuya ; Inagaki, Atsushi*; Hirano, Kohin*; Shimose, Kenichi*

The project goal is to realize real-time wind prediction in urban areas by assimilating observed data into real-time wind simulations on GPU supercomputers. In FY2022, the first year of the project, we developed a dynamic optimization method for model variables by applying a particle filter (PF) based data assimilation method to reproduce wind conditions in the atmospheric boundary layer with high accuracy. The numerical simulations for the field experiment in Oklahoma City showed improvements of about 10 % for the standard deviation error of the all-day velocity compared to the results without the application of PF. In addition, a multi-scale analysis based on boundary conditions given by a geographic information system (GIS) and a cloud-resolving numerical model (CReSS) was realized for the Tokyo metropolitan area.