Climate & Sustainability Thrust: Five Years of Advancing AI for Climate Science and Sustainability
Climate and sustainability represent a very broad topic that cuts across multiple thrusts within the Institute, which aligns with the IBM’s priority areas.
In this thrust, we are particularly interested in innovative solutions that enable timely and accurate analysis and prediction of the changing climate and its impacts, supported by high-performance compute and innovations in cloud computing.
In the past five years, the researchers in this thrust have significantly advanced state-of-the-art in the following areas, including (1) developing next-generation computing systems to support distributed discovery and modeling of multi-modal geospatial data and power grid data; (2) advancing operation and governance of multi-cloud and hybrid environments that remove the barriers of using secure and distributed climate-relevant and power data; (3) identifying critical applications that drive innovations in infrastructure and AI; (4) improving foundation models for accelerating AI for Climate and Power.
Major Accomplishments
Building the Next Generation of Geospatial Foundation Models
Highlight:
- ClimateBench-M: A Multi-modal Climate Data Benchmark, https://github.com/iDEA-iSAIL-Lab-UIUC/ClimateBench-M
- GFM-Bench: a new evaluation benchmarking toolkit for geospatial foundation models, https://huggingface.co/GFM-Bench
- MC-Search (ICLR’26 Oral): the first benchmark for agentic MM-RAG with long, step-wise annotated reasoning chains spanning five representative reasoning structures, https://mc-search-project.github.io/
Advancing AI for Real-World Climate Applications
Highlight applications such as:
- Weather forecasting using multi-modal foundation models
- Climate impacts on agriculture and finance
- Extreme weather events prediction
- Electric power grids
- CO2 conversion and utilization
Building a Research Community
Highlight:
- 14 UIUC faculty collaborators from CS, iSchool, ECE, Climate, Meteorology & Atmospheric Sciences, Prairie Research Institute, Integrative Biology
- 3 major geospatial and multi-modal learning benchmarks
- 5 software packages
- 35+ Publications in ICML, ICLR, NeurIPS, KDD, WSDM, etc.
- Multiple keynote and invited talks every year
- Fellow of the Royal Society of Canada (RSC) in the Category of Academy of Science, ACM Fellow, IEEE Fellow, ACM Distinguished Member, NSF CAREER Award, University Scholars
- Promotion to Full Professors
- WSDM 2024 Test of Time Award
- Best Poster Award, Rising Star in EECS Award
- Best paper awards at ICLR 2026 workshop, ICCV 2025 workshop
- Bests of LoG 2025