April 28, 2025
On April 28, 2025, our CTO Paraskevi Vourlioti presented how we run hashtag WRFDA data assimilation workflows on high-performance computing systems. The session demonstrated how advanced numerical weather prediction techniques benefit from HPC in practice.
Key highlights from the presentation:
• Why data assimilation needs HPC: Incorporating real-world observations into WRF-DA (3D-Var & 4D-Var) is computationally intensive—HPC clusters dramatically accelerate these calculations.
• Improving forecast accuracy: Leveraging HPC lets WRF ingest larger volumes of satellite, radar, and in-situ data for higher-resolution simulations, producing more accurate initial conditions in less time.
• Optimizing WRF on supercomputers: Practical tips on tackling I/O bottlenecks, scaling across hundreds of cores, and optimizing memory usage for efficient model runs.
• Future outlook – HPC & AI: Insights on combining AI (e.g., neural nets for parameter tuning or hybrid modeling) with HPC-driven assimilation to push weather and climate science forward.
Thank you to the EuroCC@Greece team for hosting and the HPC weather modeling community for the engaging discussion.
🔗 Download the slides (PDF) link
