Knowing and understanding weather is hugely important, from an industrial angle (e.g., agriculture, traffic management, power plant scheduling) as well as a personal angle (e.g., planning outdoor activities, commute, insurance coverage). Weather modelling is founded on physics models, but these models are often too limited in comparison with the true complexity of atmospheric processes, both in terms of equations missing some effects and finite resolution in solving them. Machine learning is starting to fill these gaps, but must provide principled uncertainty quantification to be useful and trusted.
Aims
In this workshop, we will discuss how probabilistic methods can improve the quality of forecasts by giving us predictive error bars and new ways of incorporating data as well as physical knowledge into weather models. We discuss various challenges in weather and climate science, including modelling, forecasting, data assimilation, and downscaling, all of which pose different technical challenges that need addressing. This workshop brings together domain experts and machine learning researchers in order to discuss problem settings, existing approaches, and ideas where and how probabilistic, data-driven approaches could fill some gaps. We encourage contributions of ideas, open problems, concepts, preliminary results, and other unpublished work to be discussed at the workshop. We will implement a light-touch reviewing process, but we expect to accept nearly all submissions. The workshop will provide a space for formulating open problems and brainstorming ideas.
Registration
Please register to attend the workshop using this link. Note that details on abstract submission are given below.
Abstract Submission
We invite submissions of short abstracts (1 or 2 pages) for consideration as posters or short presentations during the workshop. Submissions that explore probabilistic methods in any area of weather and climate science are welcome. We reference a few areas of interest below:
- Weather forecasting
- Climate model emulation
- Data assimilation
- Extreme events
- Constraining uncertainty in climate
- Climate model ensembling
- Hydrology
- Clouds and aerosols
- Air quality
- Downscaling
We particularly welcome submissions that focus on climate and weather related issues that affect lower and middle income countries.
This workshop will not publish proceedings, and previously published work is welcome.
The link for abstract submission is here. If you are unable to submit your abstract with the link, please email the abstract to climateworkshop23@gmail.com.
If your abstract is accepted, we expect you to present your work in person at the event.
Submission deadline: 6 March 2023
Acceptance notification: 8 March 2023
Sponsor
Partners
