UW Partners in Genesis Mission Project to Improve Forecasting of Dangerous Storm Systems

Planette AI, Pacific Northwest National Laboratory (PNNL) and the University of Wyoming have launched DL4MCS, a jointly led Genesis Mission Phase I project supported by the U.S. Department of Energy (DOE) to improve forecasting of large clusters of potentially dangerous thunderstorms.

Called mesoscale convective systems, the clusters can produce intense rainfall, hail, damaging winds and tornadoes, while also delivering a major share of warm-season precipitation across much of the country. Because these storms influence both water availability and extreme weather risk, improving their prediction could help strengthen planning for water resources, energy systems, infrastructure and community resilience.

The Genesis Mission is a historic national initiative led by DOE that is building an integrated science discovery platform by bringing together government, industry, academia and philanthropy to accelerate breakthroughs in energy, scientific discovery and national security through artificial intelligence (AI), supercomputing, quantum systems and advanced scientific instruments.

The goal of the Genesis Mission Phase I awards is to identify promising pathways toward transformative scientific capabilities by designing and demonstrating research workflows that integrate AI with scientific investigation, and testing whether those approaches can improve predictive capabilities, accelerate discovery, enhance experimentation or generate new scientific insights.

DL4MCS -- short for Deep Learning Methods to Enhance Subseasonal Predictions of Mesoscale Convective Systems by Physics-based Forecasting Systems -- addresses that challenge by combining physics-based forecasting with advanced AI methods. The project will develop a hybrid workflow that expands forecast ensembles; calibrates large-scale environmental drivers using ocean and land observations; and downscales coarse forecasts to 6-kilometer resolution to better represent storm initiation, growth and evolution.

The project team will test whether this hybrid approach can substantially improve forecast skill for storm clusters at lead times of seven days to six weeks, a forecasting window that remains especially difficult for today's operational systems. In Phase I, the team will build and validate three major workflow components: AI-based ensemble boosting; observation-informed large-scale forecast calibration; and microphysics-aware downscaling using deep learning.

Under the project, Planette AI is leading development of operationally relevant AI forecasting components; UW is contributing regional modeling and downscaling expertise; and PNNL is contributing strengths in Earth system model development, aerosol-cloud interactions and evaluation of microphysical processes. Together, the partners aim to create a proof-of-concept forecasting pipeline and evaluate its performance against current state-of-the-art operational systems over the last decade of U.S. storm activity.

"The University of Wyoming is excited to contribute its expertise in regional modeling and dynamical downscaling, as well as its responsible integration with AI forecasting, to this effort," says Stefan Rahimi, UW Derecho Professor in the Department of Atmospheric Science. "The ability to translate coarse large-scale forecasts into higher-resolution, decision-relevant guidance is essential for improving real-world preparedness and resilience."

"DL4MCS reflects Planette AI's commitment to delivering more actionable environmental intelligence for high-stakes decisions," says Hansi Singh, founder and CEO of Planette AI. "By combining state-of-the-art AI with proven physical forecasting systems, this project aims to make weeks-ahead storm risk information more useful for the sectors and communities that depend on better foresight."

"Improving prediction of mesoscale convective systems requires advances across scales, from large-scale climate drivers to the cloud microphysics that shape storm behavior," says PNNL atmospheric scientist Susannah Burrows. "This collaboration brings together complementary strengths in Earth system modeling, AI and process-level evaluation to explore a new path toward better subseasonal forecasts."

The DL4MCS team is participating today (Wednesday) in the Genesis Mission Summit in Washington, D.C., together with other selected teams in the initiative's first cohort.