Imme Ebert-Uphoff
Colorado State University-Fort Collins, CO·Earth & Environmental Science
Climate models and weather reanalyses, which are reconstructions of past weather from observations, describe the atmosphere at large scales. Clouds and storms, however, form at much finer scales that these datasets cannot show directly.
- Fields
Earth & Environmental Science
Computer Science
Mathematics & Statistics
- Topics
interpretable AI
explainable AI
machine learning
- Grants
- 1 known grant, through Aug 2027
What they work on
- Build simple, interpretable AI models that map large-scale environmental variables from reanalysis products to satellite-derived cloud properties, and use them to estimate cloud variability before the satellite era.
- Combine mathematical and AI methods to analyze present and future climate model simulations, linking broad environmental conditions to the three-dimensional structure of clouds and precipitation in a warmer climate.
- Develop interpretable AI methods that build in mathematical feature engineering from invariant and equivariant frameworks such as topological data analysis and harmonic analysis.
More about the research
Linking the two is key to predicting hazardous weather today and to understanding how storms may change in a warmer climate. This project brings together mathematicians, data scientists and atmospheric scientists. They will use artificial intelligence (AI) to connect broad environmental conditions with detailed cloud features. The team focuses on interpretable AI, models whose reasoning humans can follow, so that the models can be used to discover new knowledge rather than only to make predictions.
Why it matters
Linking large-scale weather conditions to cloud and storm structure could improve prediction of hazardous weather in the present climate. It could also clarify how storms may change in a warmer future, information that public safety officials and policy makers need. More interpretable AI methods could extend AI-based discovery to many other geoscience problems.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Guiling Wang
University of Connecticut, CT·Earth & Environmental Science
Develop machine learning and deep learning models that predict global vegetation properties from selected climate variables, trained on satellite remote sensing, reanalysis data and CMIP6 Earth System Model output.
Zhanqing Li
University of Maryland, MD·Earth & Environmental Science
Determine how cloud-surface coupling, the link between clouds and the lowest atmospheric layer, changes the influence of aerosols on clouds over urban, forested and coastal land surfaces.
Steven Greybush
Pennsylvania State University, PA·Earth & Environmental Science
Design and optimize a deep learning nowcasting system for convection initiation that combines forecasting, diagnosis and generation of storm scenarios, trained on high-resolution satellite, radar and surface weather data.
The school
Colorado State University-Fort Collins: 23 labs on PI Pages, $466M a year on research, 276 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.