R-CCS Cafe 第207回 第1部

We have been developing the regional ensemble data assimilation system SCALE-LETKF for improving numerical weather prediction (NWP) methods. For example, we developed methods to assimilate the latest-generation geostationary satellite Himawari-8 data and tested in an intense typhoon case. The results showed that the assimilation of the Himawari-8 radiance observations greatly improves the representation of the typhoon structure and the forecast skill of the typhoon intensity. Another application was summer-time convective rainfalls observed by the latest phased-array weather radar every 30 seconds. To fully exploit this new radar observation, we developed the workflow that can perform the assimilation of radar data and extended forecasts in real time. The system successfully ran in the summer of 2020 in real time and provided weather forecasts to the public. Although many achievements were obtained with the K computer and other High-Performance Computing (HPC) systems so far, Fugaku would provide new opportunities for further advancing NWP and data assimilation. For example, Fugaku would allow us to increase the ensemble size so that we can resolve the probability density function of forecast errors more accurately. Further, it would be possible to increase the NWP model resolution. Besides, we would have a chance to use a sophisticated cost-demanding algorithm for modeling clouds. The algorithm can predict lightning flashes as well, and we can test the assimilation of lightning flash observations on Fugaku.