High-impact weather prediction through data assimilation
  • Takemasa Miyoshi, PRACE, XSEDE, RIKEN, Compute Canada
  • June 2016

Slide contents

  • “Big Data Assimilation”
  • Who am I?
  • http://data-assimilation.riken.jp/
  • Global 870-m simulation (Miyamoto et al. 2013)
  • Computers keep advancing…
  • A simulated study using the T30/L7 SPEEDY AGCM
  • Advantage of large ensemble
  • A real-world study using the NICAM
  • With subsets of 10240 samples
  • Sources of Big Data
  • Global Observing System
  • Observation data (6-h period)
  • Observation data (6-h period)
  • Next-generation geostationary satellite
  • L1 products (Improvement of temporal resolution)
  • Big Data Assimilation
  • Data Assimilation (DA)
  • Data Assimilation (DA)
  • DA workflow
  • Data size in NWP
  • Workflow with current data size
  • Workflow with extreme-scale
  • Toward next 20 years of DA
  • Near-real-time SCALE-LETKF
  • Severe disaster case in Sep. 2015
  • [mm/h]
  • [mm/h]
  • Himawari-8: A new generation satellite
  • Typhoon Soudelor (2015)
  • LETKF
  • Band 14
  • Precipitation patterns
  • 100x
  • =(Gp:) Wow
  • Data Assimilation(Gp:) +
  • (Gp:) +
  • =(Gp:) +
  • Revolutionary super-rapid 30-sec. cycle
  • File I/O Middleware (FARB)
  • Evaluation of FARB
  • A case selected for the first offline study
  • 30-sec. and 5-min. evolutions
  • First results
  • Advantage of 100-m simulation
  • 9/11/2014 morning, biked to office
  • 9/11/2014 morning, biked to office
  • 9/11/14 morning torrential rain
  • Summary and a perspective
  • Revolutionary super-rapid 30-sec. cycle