Beneath Maine’s salt marshes, underground chemistry is anything but static. As water migrates through soils and rocks, dissolved minerals, reactive chemicals, and microbial metabolism continually reshape the subsurface. The result is a dynamic system where microbes can accelerate mineral dissolution or formation, steer fluid pathways, and determine whether pollutants break down—or persist and spread.
For years, large-scale models used to plan energy projects and assess contamination have relied on simplifying assumptions about microbial communities. In many cases, microbes are treated as fixed “ingredients,” even though real organisms rapidly adapt to temperature, nutrients, salinity, and local chemistry. Bridging molecular biology with physical modeling has remained a persistent technical bottleneck—largely because genomic data is high-dimensional and difficult to scale.
Now researchers at the University of Maine, led by Jiaze Wang and Amanda Albright Olsen, are attempting to close that gap through the U.S. Department of Energy’s Genesis Mission. The project applies artificial intelligence to extract which biological processes matter most for specific underground reactions. Rather than feeding raw genomes directly into geochemical simulations, the team aims to translate genomic signals into actionable model parameters.
The core technical challenge is coupling. Subsurface simulators represent transport and reactions at geological scales, while microbiology captures mechanisms at molecular resolution. The team’s strategy uses AI to identify dominant microbe-driven processes—then embeds those relationships into existing subsurface frameworks to improve predictions of microbe–mineral–groundwater interactions under changing environmental conditions.
Because microbial community effects can shift with salinity and oxygen availability, coastal systems provide a fast-moving laboratory. In the first phase, the researchers will test their approach using real-world coastal wetland data spanning locations from Lake Erie to the Chesapeake Bay, where rapid environmental turnover improves observational constraints.
Salt marshes also share practical relevance for managers and planners. They are exposed to saltwater intrusion and groundwater contamination, yet remain ecosystems that communities value and already monitor. That combination makes them an ideal proving ground for methods intended to support decision-making beyond the coast.
The team’s expected deliverables include a library of microbial genetic information formatted for subsurface physical models, an upgraded version of a widely used simulation tool, and a faster AI emulator that can reproduce large-scale predictions efficiently.
Ultimately, the approach could generalize to subsurface problems worldwide, including siting for energy infrastructure, forecasting contamination fate, and improving responsible extraction of critical minerals. The same coupling logic—how biology and geochemistry interact over space and time—may also inform risk assessments in polar and other rapidly changing environments.
This Genesis Mission effort highlights a broader shift: making microbial behavior legible to large-scale prediction systems, so underground models can evolve from approximate snapshots to mechanistic, data-driven forecasts.
Subject of Research: Microbial processes in subsurface chemical reactions; coupling molecular biology with physical models
Article Title: (Not provided)
News Publication Date: (Not provided)
Web References: https://www.energy.gov/undersecretaryforscience/genesis-mission/genesis-mission
References: (Not provided)
Image Credits: Photo courtesy of the University of Maine.
Keywords: microbial genomics, artificial intelligence, subsurface modeling, geochemistry, groundwater contamination, salt marshes, porous media, environmental remediation, AI emulators, Genesis Mission
