Machine Learning for Atmospheric Sciences: Values and Controversies
LOCATION
Virtual - Hosted on Zoom
Machine learning (ML) is becoming a trending, and yet controversial, tool for scientific research. There have been rapidly growing numbers of studies utilizing ML in weather and climate related studies, where ML presents unprecedented potentials beyond traditional tools and models. Nevertheless, debates on how we can understand, use, and interpret this “black box” remain concerning to many.
In this workshop, we attempted to showcase and discuss the values and limitations of ML in atmospheric sciences. We also aimed to foster discussions on how we can maximize the value of ML bounded by its limitation and caveats. More importantly, we’d like to hear from attendees on how you see the future development and potential of ML in the broad atmospheric science community. We had the pleasure to have invited senior scientists and young careers to demonstrate both the overall development and case-by-case application of ML in atmospheric science.
A minute's silence was held at the start of the event to commemorate Her Majesty Queen Elizabeth II, our former Patron.
Agenda
14th Sep 2022 13:00 - 17:00
Session Chair: Alcide Zhao
Time | Title | Speaker |
---|---|---|
13:00 | Welcome and Introduction | |
13:05 | General Introduction into Machine Learning the context of tackling climate change | Dr K Tokarska |
13:30 | Overview of Machine Learning in Atmospheric Sciences | Dr I Colfescu |
13:55 | Challenges and Limitations of Machine Learning for Atmospheric Sciences | Dr P D Duben |
14:20 | Comfort Break | - - - |
14:40 | Weather Forecast and Climate Prediction | M Schultz |
15:00 | Systematically Generating Hierarchies of Machine-Learning Models, from Equation Discovery to Deep Neural Networks | T Beucler |
15:20 | Thaw Slumps in Siberia | Dr Yang |
15:40 | Machine learning parameterizations: opportunities and headwinds | Dr N Brenowitz |
16:00 | Generative Adversarial Networks for Improving Precipitation Fields from Earth System Models | P Hess |
16:20 | General Discussions | - - - |
Registration
REGISTRATION IS NOW CLOSED
Resources
Machine learning (ML) is becoming a trending, and yet controversial, tool for scientific research. There have been rapidly growing numbers of studies utilizing ML in weather and climate related studies, where ML presents unprecedented potentials beyond traditional tools and models. Nevertheless, debates on how we can understand, use, and interpret this “black box” remain concerning to many.
In this workshop, we attempted to showcase and discuss the values and limitations of ML in atmospheric sciences. We also aimed to foster discussions on how we can maximize the value of ML bounded by its limitation and caveats. More importantly, we’d like to hear from attendees on how you see the future development and potential of ML in the broad atmospheric science community. We had the pleasure to have invited senior scientists and young careers to demonstrate both the overall development and case-by-case application of ML in atmospheric science.
A minute's silence was held at the start of the event to commemorate Her Majesty Queen Elizabeth II, our former Patron.
Agenda
14th Sep 2022 13:00 - 17:00
Session Chair: Alcide Zhao
Time | Title | Speaker |
---|---|---|
13:00 | Welcome and Introduction | |
13:05 | General Introduction into Machine Learning the context of tackling climate change | Dr K Tokarska |
13:30 | Overview of Machine Learning in Atmospheric Sciences | Dr I Colfescu |
13:55 | Challenges and Limitations of Machine Learning for Atmospheric Sciences | Dr P D Duben |
14:20 | Comfort Break | - - - |
14:40 | Weather Forecast and Climate Prediction | M Schultz |
15:00 | Systematically Generating Hierarchies of Machine-Learning Models, from Equation Discovery to Deep Neural Networks | T Beucler |
15:20 | Thaw Slumps in Siberia | Dr Yang |
15:40 | Machine learning parameterizations: opportunities and headwinds | Dr N Brenowitz |
16:00 | Generative Adversarial Networks for Improving Precipitation Fields from Earth System Models | P Hess |
16:20 | General Discussions | - - - |