| layout | title |
|---|---|
default |
Machine Learning for Earth System Modeling |
The response of clouds to future warming is a leading source of uncertainty in long-term climate projections. A fundamental blocker is the sheer computational complexity associated with explicitly resolving certain subgrid processes, like clouds and storms. Because of this computational complexity, operational climate models used for experiments and projections rely on "parameterizations" designed to crudely approximate the effects of such processes. As a consequence, key uncertainties remain with respect to changes to precipitation as well as the frequency and magnitude of extreme events.
The climate science community needs YOUR help to design Machine Learning (ML) models that can emulate subgrid effects of atmospheric storms, clouds, turbulence, rainfall, and radiation inside E3SM-MMF, a multi-scale climate model actively supported by the Department of Energy (DoE). With better models, we can bring greater clarity to the hazards associated with our warming future and empower policymakers with the knowledge necessary to mitigate them.