-
- GC21E-0667: Informed risk assessment for solar radiation management strategies using deep learning and explainability
-
Board 0667‚ Hall EFG (Poster Hall)NOLA CC
Author(s):Generic 'disconnected' Message
Philine Lou Bommer, University of Edinburgh (First Author, Presenting Author)
Carla Roesch, University of Edinburgh
Gabriele Hegerl, University of Edinburgh
As climate change worsens, there is growing interest in solar radiation management (SRM)—methods like altering clouds to cool the planet. While SRM could reduce global or regional temperatures, it carries serious risks and uncertainties, such as unintended climate side effects and governance challenges. Traditional ways of studying SRM rely heavily on complex computer models and expert opinions, which can be slow and hard to scale. To improve this, we propose using deep learning and explainable artificial intelligence (XAI) to create faster and more transparent tools for assessing SRM risks. Our tool would not replace existing models but would support them by quickly analyzing data and helping experts understand how predictions are made. We will test this approach using climate model data and past climate events. By making the AI tool more interpretable, we aim to improve trust, help researchers detect errors, and better understand how SRM actions might play out over time. This approach can lead to quicker, clearer, and more reliable evaluations of SRM, supporting safer and more informed climate decisions.
Scientific DisciplineNeighborhoodType
Enter Note
Go to previous page in this tab
Session
