FPWG Workshop Program.pdf (85.9 KB)
Join here for online or Griffin Room
Our team at NCI have been working on Climate/Weather models for the last ten months or so. We currently released several examples together with ready to use environment, input data, scripts and jupyter notebooks. You can find ClimaX, FourCastNet and FourCastNeXt mentioned in Cat’s talk here Machine Learning Models for Climate & Weather - Specialised Environments - Opus - NCI Confluence. There are also other released examples and more to come, such as PanguWeather and GraphCast.
For more general AI/ML examples, see AI/Machine Learning - Specialised Environments - Opus - NCI Confluence. All the examples have their own scripts/jupyter notebooks ready in /g/data/dk92/notebooks/, for example, SFNO in /g/data/dk92/notebooks/examples-aiml/ and the general neural operators in /g/data/dk92/notebooks/examples-julia/.
The corresponding example page on opus.nci.org.au contains the step-by-step instruction of how to run those examples on Gadi.
3 Likes
This is a useful review paper on Machine Learning for numerical weather and climate modelling written by @CatherinedeBurgh-Day and @Tennessee
1 Like
Note: slight change to the program, we will be shifting Steve’s talk to first thing after lunch
Hi all,
Thank you for your part in making our first Forecasting and Prediction WG meeting on ML for modelling and prediction such a success!
Here are the presentations from the day that we have received permission to share:
Getting Started
-
Getting started with Machine Learning - Sanaa Hobeichi (269.2 KB download)
-
Key developments in ML for weather and climate modelling in the last ~18 months - Catherine de Burgh-Day (11.6 MB download)
Applications of ML
-
Siteboost: Gradient Boosted Decision Trees for Site Optimised Forecasts - Tennessee Leeuwenburg (1.8 MB download)
-
Random Forests for identifying predictors of flash droughts. - Pallavi Goswami (10.0 MB download)
-
Dynamic Bayesian networks for evaluation of Granger causal relationships in climate models - Terry O’Kane (5.1 MB download)
-
Harnessing the Potential of Remote Sensing and Machine Learning for Improving Bushfire Fuels Data - Abolfazl Abdollahi (11.6 MB download)
-
A framework for embedding ML physics into climate models - Steve Sherwood (5.1 MB download)
-
A deep learning model for forecasting global monthly mean sea surface temperature anomalies using CNN (Unet-LSTM) and transformers (TUnet) - John Taylor (11.9 MB download)
-
Cutting the cost of downscaling using Machine Learning - Sanaa Hobeichi (5.1 MB download)
-
ML Big and small - what hardware and software do you need? - Tennessee Leeuwenburg (7.4 MB download)
2 Likes