Machine Learning for Climate and Weather Working Group Announce

Hi,

This month’s meeting of the Machine-Learning Working Group will take place this Friday. Meeting will be at the usual time, 2pm (AEDT), and we will have a talk by Belinda Trotta (BoM), ML for post-processing at BoM.

Zoom details: Zoom link
Meeting ID: 831 8730 7759
Password: 123456

Agenda

Facilitator: Ryan Holmes
Co-chairs: Tennessee Leeuwenburg, Sanaa Hobeichi, Vassili Kitsios, Ryan Holmes (outgoing), Micael Oliveira (ACCESS-NRI liaison), Taimoor Sohail (incoming), and Yue Sun (NCI liaison)

  • Acknowledgement of country
  • Updates from ACCESS-NRI
    • Release of the ACCESS-NRI’s work plan for FY26-27, first draft. Seeking feedback from the community. Please see the announcement for details.
  • Updates from NCI
    • New models available
    • New dk92 environment anemoi/26.04
  • Updates from the Chairs
  • Updates from the Community & new community member introductions
  • Presentation by Belinda Trotta (BoM) (Details provided below).
  • Discussion (ongoing, dependent on time available in meeting):
    • Plans for using project nm47 (100TB gdata storage, 875kSU/quarter) on NCI.
    • Proposed projects:
      • Coarsening of the fine-scale atmosphere for cross-scale learning. Outputs: Coarse-grained fields and tendencies including fine-scale transports of prognostic variables. nm47 resources: 50TB - two months

Details of this month’s presentation:

Presenter: Belinda Trotta | Research Scientist at the Bureau of Meteorology

Title: ML for post-processing at BoM

Abstract: This talk will describe how the Forecast Improvement team uses machine learning for post-processing weather forecasts. I will focus particularly on Rainforests, our ML model for post-processing rainfall, and describe the current production version, as well as some enhancements we are developing (neural network version, extension to AIFS input). I’ll also discuss our work on applying statistical calibration methods to ML weather models, and some very recent work training a simple version of Nvidia’s ModAFNO model for temporal interpolation.

Biography: Dr Belinda Trotta is a research scientist in the Forecast Improvement team at BoM. Her research focuses on forecast calibration, with an emphasis on machine learning techniques. She led development of Rainforests, the Bureau’s new ML rainfall calibration methodology, which won the award for Best Scientific Publication of 2024. Belinda holds a PhD in pure mathematics from La Trobe University.