CAIG: Toward a Deeper Understanding of Cloud Processes and Future Storm Modes using AI

This project, a collaboration of mathematicians, data scientists, and atmospheric scientists, seeks to gain a deeper understanding of the formation of clouds and storms in both the present and future climate of our planet. To do this, the project will develop novel algorithms leveraging recent advances in Artificial Intelligence (AI) and advanced mathematics that are contributing to the creation of Interpretable AI, in which AI models and their outputs are more human-understandable.

The anticipated benefits for public safety are twofold: (1) to improve the prediction of hazardous weather conditions in the present time, and (2) to better understand hazardous weather conditions in the future climate, thus providing much needed information for policy decisions. This work will assist in developing a globally competitive STEM workforce by training several scientists in the development of mathematical and AI approaches and their use for geoscience applications.

Additionally, this project is expected to lead to new connections between mathematical methods and geoscience applications which will bring new opportunities to both fields.

This project has several goals and associated activities. The first goal is to gain a better understanding of the connections between large-scale environmental characteristics from reanalysis products and smaller-scale satellite-derived estimates of cloud properties. The proposed approach aims to use relatively simple and interpretable AI models to create a mapping between these two datasets.

This mapping will be used to study current cloud variability overlapping with the satellite record and estimate historical cloud variability before the satellite era. The second goal is to discover how large-scale environmental variables, which cannot show detailed weather processes, relate to finer-scale elements like the three dimensional structure of clouds and precipitation, especially in a future warmer climate scenario.

The proposed approach is to combine mathematical and AI methods to analyze climate model simulations for current and future climates to connect broad environmental conditions with specific cloud features. This activity will help prepare for the future climate, benefiting society by enhancing understanding of weather patterns and their impacts. The third goal is to develop Interpretable AI methods by integrating mathematical components, including mathematical feature engineering using equi-/invariant mathematical frameworks, such as topological data analysis and harmonic analysis.

These research activities are expected to lead to new connections between mathematical methods and geoscience applications, bringing new opportunities to both fields. Furthermore, expanding the use of mathematical methods to make AI models more interpretable creates new opportunities to use AI methods for knowledge discovery in a wide range of geoscience applications.

This award by the Division of Research, Innovation, Synergies, and Education within the Directorate for Geosciences is jointly supported by the National Discovery Cloud for Climate initiative of the Office of Advanced Cyberinfrastructure within the Directorate for Computer and Information Science and Engineering and by the Division of Mathematical Sciences within the Directorate for Mathematical and Physical Sciences.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.