This is an online zoom workshop on Wednesday, 23 Oct 2024 followed by an exercise session focusing on Jupyter notebooks on 24 Oct 2024. In this workshop we are aiming to help existing climate and weather researchers to get hands-on experience of various SOTA (State-of-the-Art) deep learning models in Weather and Climate Studies. The aim of this workshop is to get an overview of the recent development of the field and let you try some of our examples and then start training your own climate/weather deep learning models for your own research. 

Office Hour

Starting from 29 Oct 2024, as discussed in the workshop, we offer a fortnightly drop-in session on Tuesdays 1:00-2:00pm AEDT. If you want to discuss any of the models and datasets at NCI, or more broadly how to use them in your research, please join the session on the following date. No need to register in advance.

Scheduled Zoom session for 2024 [https://anu.zoom.us/j/87443993007?pwd=hCAp6aUsZ9zW9OPN6d5qUSSuwbKk6X.1]

  • 29 October, 2024
  • 12 November, 2024
  • 26 November, 2024 
  • 10 December, 2024

Scheduled Zoom session for 2025 [https://anu.zoom.us/j/88928263598?pwd=mM17Lya7ayuXX6D97piPuNRoNNk8Ch.1]

  • 7 January, 2025
  • 21 January, 2025
  • 4 February, 2025
  • 18 February, 2025
  • 4 March, 2025
  • 18 March, 2025
  • 1 April, 2025
  • Further dates TBC

 Please double check details on this page before joining the zoom session. Further date and time will be updated on this page.


In order to participate in the exercise session on the second day please check the prerequisites well in advance of the workshop. 

Note: If you are not an existing NCI user, then you will need to create an account as per the instructions below.

Prerequisites for Day 2

  1. a NCI account ( create an account ) with valid password 
  2. Membership of the following projects:
    1. vp91 (join) [the training project which we will use for this workshop's compute and storage]
    2. dk92 (join) [for environment modules and some examples]
    3. wb00 (join) [for NCI-WeatherBench and ClimateNet datasets]
    4. rt52 (join) [for ERA5 datasets]
    5. ob53 (join) [for BARRA2 datasets]


Day 1

Click to Register 

1:05-1:10

Dr. Ben Evans

Opening

1:10 - 1:30

Dr. Yue Sun

A gentle Introduction to the Interactive Workflow of Developing Deep Learning Models 

Presentation Slides 

1:30 - 2:15

Dr. Maruf Ahmed

Dr. Edison Guo

Describing Deep Learning Models for Climate and Weather Studies

  • Forecasting Models: (DLWP-CS, ClimaX) on NCI-WeatherBench, (Pangu-Weather, GraphCast) on ERA5 
  • NCI-FourCastNeXt: train modified FourCastNet at NCI 
  • Downscaling Models: CorrDiff on BARRA2, Latent Diffusion Model

Presentation Slides

2:15 - 2:30

Tea Break

2:30 - 3:15

Dr. Yue Sun

Dr. Rui Yang


Describing Deep Learning Models for Climate and Weather Studies

  • Physical Law Incorporation: SFNO on ERA5, ClimODE on NCI-WeatherBench, NeuralGCM, Prithvi.WxC
  • Segmentation Models: CGNet on ClimateNet

Presentation Slides

3:15 - 3:30

Dr. Rui Yang

Overview: Facilitating AI/ML Processes with NCI

  • Datasets: ERA5, WeatherBench, BARRA2 and more
  • Environments: Modules compatible with (ARE VDI/JupyterLab, PBS Jobs) x Examples
  • Models and Training/Inference examples that are tested on Gadi and ready to use

Presentation Slides

3:30 - 4:00

Dr. Ben Evans

Discussions

  • Feedback on the models. Suggestions to further improvements.
  • What more support do you need?
  • What models/algorithms are you interested in / working on?


Day 2

Click to Register 

10:00 - 10:30

Ms. ZhuoChen Wu

ARE jobs setup

10:30 - 12:30

Dr. Edison Guo

Dr. Maruf Ahmed

Dr. Rui Yang

Dr. Yue Sun


Self-Directed Examples using ARE JupyterLab

[Recommend to focus on a notebook showing in the first four bullet points during the training]


Exercise Example Direct Links

InstructorsExamplesNotebooksARE Job Instruction
Maruf Ahmed

DLWP-CS on NCI-WeatherBench

https://github.com/nci/NCI-DLWP-CS.gitsearch "Queue" on the page NCI-DLWP-CS Notebooks

Inference Notebooks:

  • Pangu-Weather
  • GraphCast
  • FourCastNet
  • FourCastNet v2
https://github.com/nci/NCI-AI-Models.gitseearch "Queue" on the page ECMWF's inference AI-Model notebooks 
Edison GuoCorrDiff on BARRA2 inference/g/data/dk92/notebooks/examples-aiml/corrdiff/inference/CorrDiff_inference.ipynbsearch "Queue" on the page NCI-CorrDiff
Rui YangCGNet on ClimateNet/g/data/dk92/apps/climatenet/24.02/notebooks/ClimateNet/train_eva_pred.ipynbsearch "Queue" on the page Evaluation of CGNet Model Using the ClimateNet Dataset
Yue Sun

SFNO on derived ERA5

  • /g/data/dk92/notebooks/examples-aiml/sfno/shallow_water_model.ipynb
  • /g/data/dk92/notebooks/examples-aiml/sfno/res_invar.shallow_water_model.ipynb
  • /g/data/dk92/notebooks/examples-aiml/sfno/weather_era5.ipynb
search "Queue" on the page Spherical Fourier Neural Operators

ClimODE on NCI-WeatherBench

/g/data/dk92/notebooks/examples-aiml/climODE/climODE_global.ipynbsearch "Queue" on the page ClimODE

NeuralGCM

  • /g/data/dk92/notebooks/examples-aiml/neuralgcm/neuralgcm_inference_deterministic.ipynb
  • /g/data/dk92/notebooks/examples-aiml/neuralgcm/neuralgcm_finetuning.ipynb
search "Queue" on the page NeuralGCM

Prithvi.WxC Model Family

  • /g/data/dk92/notebooks/examples-aiml/prithvi-WxC/inference.ipynb
  • /g/data/dk92/notebooks/examples-aiml/prithvi-WxC/autoreg.prediction.ipynb
  • /g/data/dk92/notebooks/examples-aiml/prithvi-WxC/downscaling.ipynb
  • /g/data/dk92/notebooks/examples-aiml/prithvi-WxC/parametrization.ipynb
  • /g/data/dk92/notebooks/examples-aiml/prithvi-WxC/finetuning_param.ipynb
search "Queue" on the page Prithvi.WxC Model Family
  • No labels