Predicting U.S. Water for Energy

Sunlight shines across the surface of blue ocean

Challenge

Water availability is essential for expanding production and utilization of energy, as well as the nation’s health and security. However, there are fundamental scientific gaps in our understanding of terrestrial and atmospheric systems that limit our ability to predict water resources, especially on the time scale of weeks to years.

AI Solution

AI capable of multi-scale temporal reasoning could tackle three interrelated grand challenges: cloud physics, surface and subsurface water flows, and the broader hydrologic cycle. AI could improve, accelerate, and couple exascale-class modeling systems through advances in model initialization, and develop surrogates trained on DOE’s atmospheric and terrestrial observations and laboratory data, at a fraction of the computational cost of existing models. AI-based model diagnostics for enhanced analysis could refine a model-observational system aligned with decision-making needs.

Justification

DOE is the only agency with AI expertise, advanced computing, and integrated modeling capabilities (e.g., the DOE Energy Exascale Earth System Model, or E3SM), and infrastructure for field research necessary to meet the challenge of providing accurate information on surface and ground-water availability on the time scales of weeks to years.

National Impact

Solutions to these long-standing science challenges will radically improve America’s ability to anticipate water supply in the context of changing water availability, demands, energy technologies, and ambitions for energy expansion.

Aligned Actions from the National Laboratories

Real-Time, AI-Based Water Intelligence for U.S. Energy Systems

National Challenge: Predicting U.S. Water for Energy
Leading Lab: Oak Ridge National Laboratory

Water availability increasingly constrains U.S. energy production and the resilience of critical infrastructure, yet existing modeling systems are too slow and too coarse to support real-time decision-making.

To address this national challenge, Oak Ridge National Laboratory is leading the development of ORBIT, a unified AI foundation platform that transforms global weather forecasts into sub-kilometer, high-resolution environmental intelligence. Trained on DOE exascale supercomputers and deployable anywhere, ORBIT reduces simulation times from days to seconds while delivering up to 99% prediction accuracy.

This breakthrough enables real-time water forecasting for hydropower operations, flood response, and grid reliability, democratizing access to advanced forecasting capabilities across the energy sector.