[A-AS02] From weather predictability to controllability

[AAS02-01] Analysis of ensemble forecasts for a convective squall line performed with a 1000-member real-time ensemble Kalman filter system with 30-second update

*James David Taylor¹, Arata Amemiya¹, Shigenori Otsuka¹, Yasumitsu Maejima¹, Takumi Honda¹, Takemasa Miyoshi¹ (1.RIKEN Research center for computational science)

Keywords:data assimilation, fugaku, LETKF

Accurate simulation of fast-evolving, fine structured convective systems that can generate extremely intense, localized rainfall remains one of the difficult challenges in numerical weather prediction (NWP). To meet this challenge, a convective-scale NWP modelling system was developed capable of performing real-time extended 30-minute precipitation forecast using observations from a multi-parameter phased array weather radar (MP-PAWR). The system, known as the SCALE-LETKF, combines the Scalable Computing for Advanced Library and Environment (SCALE) model with the local ensemble transform Kalman filter (LETKF) data assimilation system, and was developed and designed on the Fugaku supercomputer to take advantage of its exceptional computing power. In the summer of 2021, a demonstration of the real-time SCALE-LETKF capabilities was performed, using an ensemble size of 1000 members cycled every 30 seconds with observations from a MP-PAWR. During the demonstration, 30-minute extended precipitation forecasts were initialised with 10 members after each 30-second update. This study examines ensemble forecasts for a convective squall line and compares them to nowcasts generated from a simple advection model to demonstrate the benefits of this unique NWP system and to investigate how improvements in the model may be achieved