Japan Geoscience Union Meeting 2018

Presentation information

[EE] Poster

A (Atmospheric and Hydrospheric Sciences) » A-AS Atmospheric Sciences, Meteorology & Atmospheric Environment

[A-AS01] High performance computing for next generation weather, climate, and environmental sciences

Sun. May 20, 2018 10:45 AM - 12:15 PM Poster Hall (International Exhibition Hall7, Makuhari Messe)

convener:Hiromu Seko(Meteorological Research Institute), Chihiro Kodama(Japan Agency for Marine-Earth Science and Technology), Masayuki Takigawa(独立行政法人海洋研究開発機構, 共同), Takemasa Miyoshi(RIKEN Advanced Institute for Computational Science)

[AAS01-P06] The computational aspect of the SCLAE-LETKF data assimilation system for rapid-update-cycle, high-resolution radar data assimilation

★Invited Papers

*Guo-Yuan Lien1, Seiya Nishizawa1, Ryuji Yoshida1, Hisashi Yashiro1, Tatiana Martsinkevich1, Takumi Honda1, Shigenori Otsuka1, Takemasa Miyoshi1, Hirofumi Tomita1, Yutaka Ishikawa1 (1.RIKEN Advanced Institute for Computational Science)

Keywords:SCALE-LETKF, Radar data assimilation

We have developed the SCALE-LETKF system, utilizing the Scalable Computing for Advanced Library and Environment-Regional Model (SCALE-RM) model and the Local Ensemble Transform Kalman Filter (LETKF). The initial purpose is to use this system for the rapid-update-cycle, high-resolution assimilation of the Phased Array Weather Radar (PAWR) data. An ultimate goal of real-time PAWR assimilation with 100 ensemble members, at 100-m model resolution, and with a 30-second update cycle, using the full capacity of the K computer, has been set for this development. This requires very careful design in every part of the code, including computation and I/O, to achieve high parallelization efficiency to meet the goal. Memory space needs to be thriftily used in single processes to allow processing the big observational data. Besides, separate execution of many small programs, which is typical in ensemble data assimilation systems, needs to be avoid; instead, only two MPI programs, the model and the data assimilation programs, are executed for the entire data assimilation cycles. In the past three some years, we have made remarkable progress of the code development towards this goal, although the actual real-time operation has not been achieved yet. Meanwhile, we believe that the SCALE-LEKTF system has become a useful tool for broad mesoscale data assimilation studies. This presentation will summarize our achievement so far with the SCALE-LETKF system in the computational aspect.