Editors’ Vox is a blog from AGU’s Publications Department.
The supply of magma from the Earth’s mantle is a primary source of heat to volcanic arc crust, where the heat is then dissipated through various processes. Much of this magmatic heat is dissipated as heated water, or aqueous fluid.
A new article in Reviews of Geophysics compares 11 different volcanic-arc segments where heat discharge via aqueous fluid has been well-inventoried to better understand the factors that influence this p
The supply of magma from the Earth’s mantle is a primary source of heat to volcanic arc crust, where the heat is then dissipated through various processes. Much of this magmatic heat is dissipated as heated water, or aqueous fluid.
A new article in Reviews of Geophysics compares 11 different volcanic-arc segments where heat discharge via aqueous fluid has been well-inventoried to better understand the factors that influence this process. Here, we asked the authors to give an overview of heat discharge from volcanic arcs, how scientists measure it, and what questions remain.
Why is it important to study how heat is dissipated from volcanic arcs?
The heat from these magmas matters for geothermal energy, patterns of groundwater flow, and the patterns of volcanic activity at the surface.
Volcanic arcs are the chains of volcanoes on top of subduction zones. They can produce some of Earth’s most explosive and hazardous eruptions. But much of the magma beneath the surface never erupts. Nevertheless, the heat from these magmas—and the simple fact of their existence and abundance—still matters for geothermal energy, patterns of groundwater flow, and the patterns of volcanic activity at the surface.
What are the main modes in which heat is discharged from volcanic arcs?
Heat at volcanic arcs can be carried by magmas, transmitted through the crust conductively, and carried by water seeping slowly through the crust. At the base of the crust, magmas are probably most important, with conduction coming in second. But as magmas move upwards through the crust, some of them solidify and impart their heat to their surroundings where it is transferred by conduction. Within a few kilometers of the surface, fluids seeping through the crust begin to take up all that heat, and so if we can quantify the heat carried by those fluids, we can retrace it to its origins in magmas.
How do scientists measure these different forms of heat loss?
Scientists measure the heat carried by erupting magmas using satellites, or by adding up the erupted masses and making an estimate of how much energy was released by cooling from eruption temperatures to solid igneous rocks at Earth’s surface. Conductive heat flow is measured by drilling holes in the Earth’s crust to see how quickly it gets hotter with depth.
Measuring the heat carried by aqueous fluids in the crust is in some ways the trickiest. One approach is to find all the springs where hot or even slightly warm water is trickling out and measure the temperature and discharge to estimate how much extra heat that water is carrying.
What are the challenges and uncertainties in measuring hydrothermal heat discharge?
One challenge is that many springs are only slightly warmer than you’d expect. There is good data for many hot springs, but there are data tracking these ‘slightly warm’ springs for only a subset of arcs. Another challenge is that warm underground fluids can flow laterally, so you have to try to account for that. This is not an uncertainty in hydrothermal discharge, but one additional big uncertainty for our study, where we were trying to quantify the proportion of magmas that freeze underground versus erupting, is in the estimates of how much magma has actually erupted through time.
What are some of the factors that influence hydrothermal heat loss?
A major goal of our paper is to try to quantify these hidden magmas.
A main factor that influences hydrothermal heat loss is the magmas that solidify underground. This link is the key motivation for our study. A major goal of our paper is to try to quantify these hidden magmas—how much magma needs to intrude the crust beneath the surface to supply the hydrothermal heat fluxes that we observe? The composition of magmas influences how much heat they can release. The depth at which magmas are emplaced also matters, because magmas that intrude the shallow crust eventually cool to lower temperatures than magmas emplaced in the lower crust and therefore release more heat.
What are the remaining questions or knowledge gaps where additional research efforts are needed?
A big outstanding challenge is combining estimates from hydrothermal data of how much magma is coming into the crust – like ours – with other approaches to estimating the same thing. The magmatic systems beneath volcanoes span the crust. At the base of the crust, you have magma supply, sort of like the water main feeding your plumbing system. Despite how fundamental magma supply is, we know remarkably little about it. It’s exciting to think about how the rates of magma supply could vary through time and space and why. Applying a range of techniques—including geophysical imaging, hydrothermal budgets, gas and igneous geochemistry, and petrology—to understand these questions could really be a game changer.
Editor’s Note: It is the policy of AGU Publications to invite the authors of articles published in Reviews of Geophysics to write a summary for Eos Editors’ Vox.
Citation: Black, B. A., S. E. Ingebritsen, and K. Sawayama (2026), Hydrothermal heat flow as a window into subsurface arc magmas, Eos, 107, https://doi.org/10.1029/2026EO265017. Published on 28 April 2026.
This article does not represent the opinion of AGU, Eos, or any of its affiliates. It is solely the opinion of the author(s).
Every chip fabricated in a semiconductor plant needs ultrapure water. Most nuclear reactors need water as a coolant and neutron moderator. Every artificial intelligence (AI) data center drinks between 1 million and 5 million gallons of water a day, with thirst often peaking during drought.
Water runs through every technology priority the United States has named, yet the word does not appear once in “Launching the Genesis Mission,” an executive order (EO) released in November 2025. As describ
Every chip fabricated in a semiconductor plant needs ultrapure water. Most nuclear reactors need water as a coolant and neutron moderator. Every artificial intelligence (AI) data center drinks between 1 million and 5 million gallons of water a day, with thirst often peaking during drought.
Water runs through every technology priority the United States has named, yet the word does not appear once in “Launching the Genesis Mission,” an executive order (EO) released in November 2025. As described in the EO, the Genesis Mission is a “dedicated, coordinated national effort to unleash a new age of AI-accelerated innovation and discovery that can solve the most challenging problems of this century.”
Led by the Department of Energy (DOE), the initiative aims to build an integrated AI framework that would harness federal scientific datasets to accelerate breakthroughs in advanced manufacturing, biotechnology, critical materials, nuclear fission and fusion energy, quantum information science, and semiconductor development. The scope of the mission is comparable to that of the Manhattan Project.
Since the announcement, the DOE has listed “Predicting U.S. Water for Energy” among its 26 Genesis Mission Science and Technology Challenges. The project is also soliciting proposals in three water-related focus areas.
This framework provides a foothold for hydrology in the Genesis Mission, but it is scoped narrowly around water as a supply variable for energy production.
In reality, water is a crosscutting constraint that will help determine whether the mission’s priorities translate into deployable outcomes. The hydrology community now has a seat at the table, and if it moves first and positions water security as one of the “most challenging problems of this century,” the Genesis Mission can become the sandbox in which AI reshapes how the country measures, models, and manages water.
Making this happen will require that the DOE and the Office of Science and Technology Policy charter a hydrology workstream inside the Genesis Mission, with interagency delivery involving the U.S. Geological Survey (USGS), NOAA, the Bureau of Reclamation, the EPA, and partners at state, regional, and community levels. Here is what we think that workstream should look like:
A water-centric Genesis Mission architecture supports seven hydrological components that both feed into and receive decisions from the Genesis AI platform. Each component maps to a section of this article. Credit: Amobichukwu C. Amanambu. Click image for larger version.
While the existing challenges reflect some of these components, others will require coordinated effort from the hydrology community to bring into the Genesis Mission’s scope.
Build the Water Corpus Genesis Will Need
The Genesis Mission EO instructs the DOE to create an American Science and Security Platform to provide the public, scientists, agencies, and policymakers access to crucial scientific datasets.
The good news is that accessible water data systems already exist across several federal agencies and academic research centers. The USGS National Water Information System tracks real-time and historical water quality and use across the country. NASA’s Earth Science Data Systems Program provides open access to Earth science observations. NOAA’s National Water Center, the first federal facility dedicated to national water resource forecasting, operates the National Water Model, which continuously forecasts flows on 2.7 million stream reaches across the continental United States. The Catchment Attributes and Meteorology for Large-Sample Studies (CAMELS) dataset, currently hosted by the National Center for Atmospheric Research, provides data tailored for hydrological research on hundreds of river basins, and the Caravan framework pulls together multiple large-sample meteorological and hydrological datasets at a global scale.
What is missing is a unified, AI-ready repository that brings federal, state, and community data together.
What is missing is a unified, AI-ready repository that brings federal, state, and community data together. Building one is hard. Water data are fragmented, inconsistent, and often entirely absent. Consistent, reliable data for groundwater, withdrawals, reservoir operations, and water quality are especially difficult to obtain.
Local resistance to sharing data is real. In Texas, for example, landowners hold private property rights over groundwater and have opposed metering and reporting requirements imposed by groundwater conservation districts. In California, agricultural well owners fought metering mandates for years before the Sustainable Groundwater Management Act compelled local agencies to begin tracking withdrawals. Tribal nations face a different concern: Water data collected on Indigenous lands has been misrepresented in federal datasets that were modeled without accounting for Indian country, leading many nations to restrict access to their data as an exercise of sovereignty.
Practical steps toward building a unified AI-ready repository include tiered access and licensing for different stakeholders, clear provenance tracking for all data reported, financial and educational incentives for stakeholders for reporting, and targeted gap filling. Where measurements are missing, AI can fuse remote sensing with gauged records and operational logs—but only if the results carry honest uncertainty estimates tied to real decisions.
Get the corpus right, and it will outlive any single program name. It becomes infrastructure the country can lean on.
Develop Shared Hydrologic Foundation Models
The Genesis Mission EO directs the DOE to provide “domain-specific foundation models across the range of scientific domains covered.”
Hydrology has a head start. Long short-term memory (LSTM) networks are a key type of neural network designed to last thousands of time steps. Hydrology LSTMs trained on CAMELS data have already matched traditional conceptual models for daily streamflow discharge prediction. Open-source Neural Hydrology tools serve as baselines for regional runoff prediction. These predictions may serve as precursors to the foundation models the Genesis Mission envisions and building blocks from which they could be developed.
The process of scaling up these tools is not straightforward, however. A hydrologic investigation of snowmelt-driven streams in Colorado will not require the same spatiotemporal data as tile-drained fields in Iowa, for example. A hydrology-specific foundation model must take nuanced requirements into consideration and provide a clear path for managing and exploiting a variety of datasets.
Google’s Flood Hub shows what is possible: Its AI-enabled flood forecasts now cover more than 80 countries. However, Flood Hub’s core model code and trained weights remain proprietary, meaning researchers can use the forecasts but cannot rebuild or adapt the underlying models. Genesis, if well positioned, can fill that accessibility gap by producing foundation models for water that are reusable, reliable, and openly governed.
Build a National Water Digital Twin
The EO prescribes an integrated AI platform combining foundation models with simulation tools to stimulate AI-enabled innovations.
That architecture is exactly what a digital twin requires. Europe’s Destination Earth initiative is already building digital twins for weather extremes and nonstationary conditions on the Large Unified Modern Infrastructure (LUMI) supercomputer. The United Nations–led AI for Good initiative has prioritized water applications, warning that fragmented national efforts risk duplicating work.
If the United States aims for global strategic leadership in AI-accelerated science, water infrastructure cannot be an afterthought.
A water digital twin earns its keep when it makes the consequences of choices visible, in terms of flows, levels, temperatures, and risks to people and ecosystems.
Rather than starting from scratch, a water-centric Genesis Mission would unite existing federal models—the National Water Model, reservoir simulators, and groundwater codes—in a single digital twin. AI can become the thread that stitches them together, correcting biases and providing numerical solvers to enforce mass and energy balance.
What should this twin actually do? Help a dam operator decide whether to release water ahead of a storm. Tell planners where a new data center can draw cooling water without drying up a stream. Flag which coastal defenses will fail first under rising seas.
A water digital twin earns its keep when it makes the consequences of choices visible, in terms of flows, levels, temperatures, and risks to people and ecosystems.
Turn Basins into AI Test Beds
The Genesis Mission promotes AI-directed experimentation and directs the DOE to keep a record of robotic laboratories and production facilities in which such experimentation could be conducted. Hydrological field sites belong in that inventory. The National Ecological Observatory Network already operates 81 sites with standardized measurements of meteorology, surface water, groundwater, and biodiversity. The Critical Zone Collaborative Network instruments catchments to track water-soil-vegetation interactions over decades.
Formalizing these networks as AI test beds would link field observations back into the water digital twin so that experiments and models continually sharpen each other. Imagine mobile sensors steered by AI agents during a storm or aquifer recharge experiments designed by algorithms and verified in real time. That feedback loop is what separates a useful model from a decorative one.
Expand Water Challenges on the Genesis Mission List
The Exchange and What’s at Stake
Allowing water security to flow through the diverse components of the Genesis Mission would benefit both the policies championed by the mission itself and the hydrology community.
The Genesis Mission gets real-world, noisy test beds where AI proves value beyond benchmarks, a domain to stress test climate and infrastructure investments, and scientists trained in both AI and the stubborn realities of rivers, aquifers, and pipes.
Hydrology gets resources for shared data infrastructure, foundation models and instrumented basins no single lab can support, a seat when rules for AI and national scientific infrastructure are negotiated, and a chance to reset practices around openness, collaboration, and equity.
Earlier this year, the DOE released 26 Genesis Mission Science and Technology Challenges, and “Predicting U.S. Water for Energy” was among them. The accompanying funding call (DE-FOA-0003612) solicits proposals on cloud microphysics, coupled surface water–groundwater modeling, and seasonal to multiyear prediction, all framed around energy needs and flood resilience.
These inclusions are a significant win for a hydrology component to Genesis, but several urgent challenges sit outside their scope. Can AI close the gap between a flood forecast issued 12 hours out and the 48 hours emergency managers actually need? Can it map compound extremes, in which drought, heat, and infrastructure failure collide in the same week? Can it redesign monitoring networks so that coverage follows risk rather than where gauges happened to be installed a century ago? Integrating energy and water systems is equally urgent: Floods have caused 80% of major U.S. grid outages since 2000, while drought-driven water stress curtails cooling at thermoelectric plants and reduces hydropower output, exposing how deeply energy infrastructure depends on hydrologic extremes.
The water footprint of new AI infrastructure deserves a place on that list. A separate executive order (14318, “Accelerating Federal Permitting of Data Center Infrastructure”) is already fast-tracking expansion of data center construction, and a single hyperscale facility can consume 1 million to 5 million gallons of water daily. Emerging research shows how withdrawals at that scale can push streams below ecological thresholds during low flows.
Make Hydrology the Conscience of AI Governance
The EO directs the DOE to set data access rules and clarify policies for ownership, licensing, trade secret protections, and commercialization of products and tools associated with it.
Three principles should anchor such policies for AI use in water security.
First, Indigenous and community data rights must be embedded in every major AI water security effort, in line with the collective benefit, authority to control, responsibility, and ethics (CARE) principles for Indigenous data governance.
Second, AI’s own water footprint, through electricity generation and cooling, must be treated as a design constraint. Transparent reporting, stress-based siting, and efficiency targets will prevent hydrology in Genesis from being self-defeating.
Third, the DOE should define what failure looks like. Missing a flood crest portends loss of lives and livelihoods and breaches of treaties. Accountability standards must be measurable, and they must ask not just how accurate the forecast was on average, but who bore the cost when it was wrong.
A single executive order will not solve the country’s water security problems, and a single challenge topic will not either.
But the Genesis Mission has provided a seat at a table that did not exist 6 months ago. Whether the hydrology community treats it as a ceiling or a foundation depends on what happens next. Europe’s Destination Earth and the United Nations’ AI for Good water initiatives are already moving.
American hydrology now has a seat at the table. We should take it.
Kratzert, F., et al. (2019), Toward improved predictions in ungauged basins: Exploiting the power of machine learning, Water Resour. Res., 55, 11,344–11,354, https://doi.org/10.1029/2019WR026065.
Xiao, T., et al. (2025), Environmental impact and net-zero pathways for sustainable artificial intelligence servers in the USA, Nat. Sustainability, 8, 1,541–1,553, https://doi.org/10.1038/s41893-025-01681-y.
Zhang, L., et al. (2025), Foundation models as assistive tools in hydrometeorology: Opportunities, challenges, and perspectives, Water Resour. Res., 61, e2024WR039553, https://doi.org/10.1029/2024WR039553.
Author Information
Amobichukwu C. Amanambu (acamanambu@ua.edu), Department of Geography and the Environment, The University of Alabama, Tuscaloosa; and Jonathan Frame (jmframe@ua.edu), Department of Geological Sciences, The University of Alabama, Tuscaloosa
Citation: Amanambu, A. C., and J. Frame (2026), The Genesis Mission needs hydrology: Here’s how to incorporate it, Eos, 107, https://doi.org/10.1029/2026EO260131. Published on 28 April 2026.
This article does not represent the opinion of AGU, Eos, or any of its affiliates. It is solely the opinion of the author(s).
Login
