EOS

body {background-color: #D2D1D5;} Research & Developments is a blog for brief updates that provide context for the flurry of news regarding law and policy changes that impact science and scientists today.

The National Science Foundation (NSF), one of the world’s leading funders of basic research, will “stop awarding all funding actions until further notice,” including awarding new grants and disbursing funds for existing grants, according to Nature. 

Staff at NSF were told of the policy change in a 30 April email. The email did not give a reason for the funding freeze and did not say whether or when the agency would resume awarding funding. 

 
Related

•  Exclusive: NSF Stops Awarding New Grants and Funding Existing Ones
•  NSF Halts Grant Awards While Staff Do Second Review
•  Trump’s Cuts to Science Funding Could Hurt U.S. Economy, Study Shows
• Get Involved: AGU Science Policy Action Center
 

Each year, the agency awards about 12,000 new grants with an average duration of three years, meaning tens of thousands of projects may be affected by the new policy. NSF funds about a quarter of all federally supported research in the United States.

Unless the funding freeze is lifted, it will “destroy people’s labs,” Colin Carlson, an epidemiologist at Yale University, told Nature. The latest development at NSF is a “five-alarm fire for American science,” Carlson wrote in a Bluesky post.

This is going to kneecap science in this country for years. www.nature.com/articles/d41…

— Matt Peeples (@pattmeeples.bsky.social) 2025-05-02T03:52:21.765Z

NSF had already drastically tightened its disbursement of funding—in the past two weeks, the agency terminated more than 1,000 grants, together worth $739 million. Hundreds of those grants were related to diversity, equity, and inclusion or misinformation and disinformation. In April, the agency returned all grant proposals to program officers, asking for extra review of whether each followed directives from the Trump administration. 

In addition to causing turmoil among scientists who are now unsure their work can continue, such dramatic cuts to U.S. research funding may also cause long-term economic harm. One recent study by researchers at American University found that just a 25% reduction in federal funding for scientific research and development would reduce the U.S. Gross Domestic Product (GDP), a measure of economic health, by an amount comparable to the Great Recession and would make the average American poorer.

www.nature.com/articles/d41…I am a scientist and an educator. I cannot believe that everything so many of us have dedicated our adult lives to is being destroyed – burned down – for no reason other than knowledge imperils their vision of the future. This must be reversed. Pls share widely!

— Dr. Mandy Joye (@oceanextremes.bsky.social) 2025-05-02T02:05:48.901Z

“This country’s status as the global leader in science and innovation is seemingly hanging by a thread at this point,” one NSF staff member told Nature.

—Grace van Deelen (@gvd.bsky.social), Staff Writer

These updates are made possible through information from the scientific community. Do you have a story about how changes in law or policy are affecting scientists or research? Send us a tip at eos@agu.org. Text © 2025. AGU. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.

As the world blows past 1.5°C of anthropogenic warming and looks increasingly likely to hit 2.6°C–3.1°C by the end of the century, plenty of controversy still exists, even among geoscientists, about how to slow, stop, or reverse the rapid climate change we are causing. As so many studies have documented, such warming will cause inundation of many coastal cities, trillions of dollars in damage from extreme weather, widespread species extinctions, and unrelenting heat waves. It will also fundamentally threaten financial sectors and economies at all scales.

The scale of mitigation needed to keep warming to below 2°C–3°C goes beyond reducing annual emissions.

One thing is clear: To mitigate these outcomes, humanity’s first priority should be to drastically reduce its annual emissions of roughly 40 gigatons (billion metric tons) of carbon dioxide (CO2), the greenhouse gas most responsible for driving warming. Without this reduction, other measures will be only modestly effective at best.

But unfortunately, at this point, the scale of mitigation needed to keep warming to below 2°C–3°C goes beyond reducing annual emissions. We must also remove and store carbon that has accumulated in the atmosphere.

Reducing Annual Emissions Isn’t Enough

The need for emissions reductions has been articulated accurately, passionately, and compellingly for decades. Yet global emissions continue to set new records, increasing 1% in each of the past 3 years. Meanwhile, even as clean and renewable energy (CRE) growth has recently set its own records, global fossil fuel energy consumption has continued to rise, with oil, gas, and coal still accounting for more than 81% of total energy consumption (only 4% less than 20 years ago).

Even under favorable political conditions, CRE consumption, which as a share of global primary energy consumption is growing at roughly 1% per year, has a long way to go to catch up to the roughly 2% annual growth in global energy consumption. Even once CRE growth catches up, it could take decades to reach something like global energy decarbonization, during which we would emit several times more CO2 than we already have.

Not only has focusing on annual emissions over the past few decades failed to reduce them, but it’s also not our annual emissions today (and into the future) that are causing the 1.55°C of warming we’re witnessing. It’s how much CO2 we have already emitted. Our cumulative emissions of 1.8 trillion tons (1,800 gigatons) of CO2 from energy and industry—heavier than the combined mass of all living things on Earth—taken from geologic reservoirs and dumped into the atmosphere, will stay there (and in the ocean) for thousands of years. Even on that happy day when we finally start reducing emissions, we will be the farthest we have ever been from solving the problem, and in fact, we will still be adding to it.

A Big Opportunity

Scientists and practitioners across many disciplines and sectors can play roles in climate change mitigation. Research in the geosciences is fundamental to understanding carbon reservoirs and fluxes between them, as well as past, present, and possible future effects on climate. But it seems clear by now that more climate science, and even better communication of it, is unlikely to inspire the collective or political action needed to activate significant mitigation. So what else can geoscientists offer?

Some see a role in helping to extract natural resources to fill the staggering projected demand for metals such as copper and rare earth elements and to promote the kind of technology-driven sustainability invoked by the mining industry. Geoscientists also contribute to informing approaches to adaptation and resilience, though neither of those constitute mitigation and, in the long run, they are much more expensive than mitigation. The economic impacts of warming have been estimated to be about 12% of global GDP (gross domestic product) per 1°C of warming, and our current trajectory is projected to reduce global GDP by as much as 40% by 2100, with much greater losses in some regions.

Carbon dioxide removal (CDR) is far less risky than the centuries-long geoengineering experiment of using the atmosphere as a sewer.

The biggest opportunity—and perhaps the biggest responsibility—for geoscientists to contribute to mitigation is through facilitating durable carbon dioxide removal (CDR). Concerns are sometimes raised about CDR as a form of climate intervention, or geoengineering, yet it is far less risky than the centuries-long geoengineering experiment of using the atmosphere as a sewer. Indeed, removing gigatons of CO2 per year is essential to net zero strategies and avoiding disastrous amounts of warming, as unequivocally stated by the Intergovernmental Panel on Climate Change, Energy Transitions Commission, and American Physical Society.

Keys to Carbon Removal

Three principles are generally considered fundamental to CDR. First, CO2 already in the atmosphere must be taken out. This principle distinguishes it from point source carbon capture and storage (CCS), which simply reduces new CO2 emissions from fossil fuel energy and industry sources while competing with clean energy.

The Mammoth direct air capture facility in Iceland, operated by Climeworks, began pulling carbon dioxide from the air in 2024. Credit: ©Climeworks

Many approaches to CDR exist. Direct air capture (DAC), for example, is a rapidly growing method in which CO2 is pulled straight from the atmosphere. Biomass carbon removal and storage (BiCRS) methods capture a fraction of the 480 gigatons of CO2 that plants naturally absorb each year and prevent it from cycling back to the atmosphere by converting biomass to forms that can be isolated and stored.

Other CDR approaches focus on managing ecosystems to stimulate more CO2 removal than would occur naturally, the second of the three principles of CDR. Examples include various strategies for enhanced rock weathering in croplands or forests and for marine CDR, such as using nutrients to promote biomass growth and raising the alkalinity of seawater so it pulls more CO2 from the air.

However CO2 is removed, it must be stored durably, with minimal likelihood to return to the atmosphere for a long time.

Third, and most important, is the fact that however CO2 is removed, it must be stored durably, with minimal likelihood it can return to the atmosphere for a long time. Using captured carbon to create marketable stuff like fertilizer and chemicals may seem economically savvy, but it’s not a durable approach. The entire global industrial demand for CO2 is less than 1% of our annual emissions, and much of this carbon goes right back to the atmosphere or is used for enhanced oil recovery (EOR) to extract more petroleum.

So-called nature- or land-based CDR approaches like afforestation, agricultural practices, and soil management are intuitively appealing alternatives that can remove and store CO2 and, if done right, improve ecosystem health. But these methods are also not very durable. Land plants hold a mass of carbon (~1,650 gigatons in all terrestrial vegetation) almost equivalent to our cumulative emissions, and soils hold 4 times more. However, most of the carbon in plants and soil cycles back to the atmosphere through natural decomposition and disturbances on timescales of years to decades.

Furthermore, anthropogenic warming–driven disturbances to forests and soils, which are becoming bigger and more frequent, may further weaken the durability of nature- and land-based CDR. The 2023 Canadian wildfires alone released almost 3 gigatons of CO2, almost 4 times the annual emissions of global aviation. (These disturbances also threaten to destabilize ancient peat and permafrost, which globally hold a carbon stock equivalent to 5 times our cumulative emissions—yet another reason to pursue CDR.) So although nature- and land-based CDR provides collateral benefits and is inexpensive and ready to deploy, in the context of net zero emissions accounting, it makes sense only as an offset for analogous biogenic (e.g., land use and forestry) emissions, not for the 82% coming mostly from fossil fuel burning.

Apart from the three fundamental principles of CDR, the potential to apply approaches at a large enough scale to make a significant difference is a key consideration. The scalability of DAC on large scales, for example, faces energy and expense concerns. And making a dent in the cumulative emissions load with nature- and land-based approaches like afforestation would require unreasonably huge amounts of land that already has many other competing uses. Meanwhile, the ocean, which already holds about 140,000 gigatons of CO2, offers potential because of its vast size as well as its longer residence times compared with other near-surface reservoirs, notwithstanding questions about its future warming-induced durability.

The Substantial Subsurface

It is becoming increasingly clear that for both capacity and durability, it’s hard to beat subsurface geologic reservoirs.

CDR approaches are diverse and evolving, but it is becoming increasingly clear that for both capacity and durability, it’s hard to beat subsurface geologic reservoirs. The amount of carbon in Earth’s crust is millions of times larger than in all near-surface reservoirs combined, and it stays down there orders of magnitude longer. Estimates suggest that enough subsurface storage capacity exists for at least tens of thousands of gigatons of recaptured CO2, and recent feasibility analyses showed that achieving storage rates of at least 5–6 gigatons of CO2 per year by 2050 is realistic and consistent with current technological trajectories.

Realizing gigaton-scale CDR will be a major challenge—one that requires building  support and further developing the needed methods. A few approaches show the most potential.

Captured CO2 can be compressed and injected as a supercritical fluid (sCO2) into saline aquifers or depleted oil and gas fields deep below fresh groundwater and overlain by impermeable rocks. This approach is likely the main storage route for CO2 captured by DAC, as well as by emissions-sourced CCS, and it is something we already know how to do from decades of practice (albeit mostly for EOR). Under the right conditions, several trapping mechanisms minimize the chances of escape for CO2 stored this way.

At sites like this one, the Icelandic company Carbfix injects carbon dioxide dissolved in water into geologic reservoirs underground, where it reacts with rock to form carbonate minerals. Credit: Siljaye/Wikimedia Commons, CC BY-SA 4.0

Another promising approach is direct mineralization, which involves injecting CO2, either as a supercritical fluid or dissolved in water, into reactive mafic and ultramafic rocks to form carbonate minerals. Use of this method is ramping up to scales of millions of tons per year in some places.

Other, relatively new but promising BiCRS methods that leverage plants’ carbon-capturing power involve subsurface injection (often into depleted oil and gas reservoirs) of biomass-derived carbon in the form of bio-oil, pyrolyzed agricultural or forest waste, or other organic (e.g., municipal or livestock) waste.

Challenges for Geoscientists

Given our still-increasing emissions trajectory and need for scalable carbon storage solutions, it’s hard to imagine that CDR through durable subsurface storage won’t grow in the next few decades, especially if carbon policies and incentives shift from favoring emissions reductions and avoidance to removals. With the fossil fuel industry’s interest in propping up its energy production assets, CDR’s cousin CCS may also proliferate. Either way, it is likely that the subsurface will increasingly be the focus of attention and action.

As this focus grows, we must recognize that the subsurface is an increasingly busy place, where water, energy, and mineral resources—not to mention as much as 90% of all microbial life and 10%–20% of all biomass on the planet—interact. This is where the geosciences come in.

It is time for geoscientists to step up and take on a central role in advancing mitigation solutions.

After a century of the fossil fuel industry directly and indirectly defining much of the discipline’s research and educational emphases, it is time for geoscientists to step up and take on a central role in advancing mitigation solutions, specifically durable carbon storage and responsible subsurface management. There will be no shortage of challenges.

Mining, geothermal, and oil and gas production and disposal activities have already increased subsurface fluid fluxes well beyond pre-Anthropocene rates, and projections of these fluxes in 2050 are many times higher. In the United States alone, in addition to the more than 4 million oil and gas production wells, almost a million underground injection wells dispose of a huge variety of both hazardous and nonhazardous materials and waste.

Scaling subsurface carbon storage to gigatons per year will mean injecting massive quantities of a variety of CO2 and carbon-bearing solutions into a wide range of geologic reservoirs and associated waters, creating not only engineering challenges but also challenges of illuminating the efficacy and hazards of injections under many different conditions. Although we understand relatively well how sCO2 and dissolved CO2 behave in some types of subsurface environments, we know almost nothing about the behaviors of novel carbon storage fluids like bio-oil and slurried or torrefied biowaste.

Hydrogeochemists Ji-Hyun Kim and Rebecca Tyne sample groundwater in the Paradox Basin, Utah, to understand connections among subsurface rocks, fluids, and microbial communities and how they may be affected by anthropogenic activities, including carbon storage. Credit: Jennifer McIntosh

Geoscience’s role in responsible subsurface management will also involve providing new perspectives on basins and igneous provinces to address questions of rock permeability and composition that are important for durable storage, as well as assessing critical risk factors. Risk factors include how fluids migrate and interact with faults and other permeability barriers, the potential for mineral dissolution to mobilize metals and change fluid fluxes, fresh groundwater contamination, and induced seismicity.

Much of this work will necessarily be transdisciplinary, challenging scientists accustomed to traditional and disciplinary emphases to develop shared language and approaches. For example, understanding how carbon storage affects microbial communities (e.g., through species diversity and methanogenesis) and human communities and translating this understanding through public engagement and policies will require geoscientists to collaborate and communicate with biologists, engineers, planners, industry, governments, Indigenous communities, and others.

Rising to the Occasion

Durable carbon storage for CDR may be beneath us literally, but we cannot let it be beneath us figuratively.

Public sentiment toward CDR is improving, although many geoscientists still consider it a distraction from cutting emissions or, worse, a deterrent that will disincentivize emission reductions. But this largely theoretical risk—which, it’s worth pointing out, is also posed by pursuing adaptation and resilience—can be addressed by creating separate targets for CDR and emissions reductions and by other means of deploying CDR strategically. Others may see durable CDR as being complicit with the fossil fuel industry and its tragic delay and distraction tactics or as antithetical to intuitively appealing nature-based approaches.

We need to be clear-eyed about the fact that humanity’s cumulative emissions put us on a path that requires gigatons per year of durable CDR to have any hope of avoiding 2°C–3°C of warming.

But we need to be clear-eyed about the fact that humanity’s cumulative emissions, both to date and in the future (even under optimistic projections), put us on a path that requires gigatons per year of durable CDR to have any hope of avoiding 2°C–3°C of warming. And however it is done, most of that captured carbon needs to be stored in geologic reservoirs.

Developing and responsibly managing subsurface carbon storage pose historic challenges for the geosciences. Rising to meet these challenges will serve society and the planet by helping mitigate disastrous outcomes of climate change. It may also shift long-standing public perceptions of the field as anachronistic and out of touch and create an inspiring mission for new generations of geoscientists.

Author Information

Peter Reiners (reiners@arizona.edu), University of Arizona, Tucson

Citation: Reiners, P. (2025), Real climate solutions are beneath us, Eos, 106, https://doi.org/10.1029/2025EO250168. Published on 2 May 2025. This article does not represent the opinion of AGU, Eos, or any of its affiliates. It is solely the opinion of the author(s). Text © 2025. The authors. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.

Source: Community Science

Communities on small islands are on the front lines of worsening flood risks—not just from severe storms but from persistent tidal flooding events. Scientists estimate that within 15 years, high-tide flood events could triple for two thirds of communities along the East and Gulf Coasts of the United States.

Sea level rise and tidal flooding can vary depending on local land morphology, offshore bathymetry, and wind direction and intensity. To test how a community science partnership might better determine flooding risks, Bertram et al. focused on Little Cumberland Island, Georgia. The community comprises about 40 residences connected by unpaved roads, though no road connects it to the mainland.

In 2021, island residents agreed to allow faculty and students from the College of Coastal Georgia to conduct field research on the island. They asked the researchers to focus on developing a way to predict the frequency and severity of future floods and ultimately provide insight into how to develop more resilient roads.

For the next 2 years, the researchers visited the island every 1–2 months. Each time, an island resident hosted the team for dinner and shared stories about past flooding events and some of their greatest concerns. The scientists shared research updates with the residents, including water pressure recordings in flood-prone areas and comparisons between wind-enhanced high-tide measurements and predicted tidal flooding.

Residents reported that flooding of low-elevation roads has grown more common over time and that this flooding was worse when winds arrived from the northeast. The researchers’ measurements, which supported these observations, allowed the team to determine how wind velocity affects tidal flooding and to predict future flood frequency.

The researchers suggest that grading roadways so they dip downward on the sides, combined with increasing the size of sediment used for the roads from sand to gravel, could be enough to protect the roads until 2030. However, they predict that by 2040, “nuisance flooding” of 30 centimeters or less will double to triple in frequency.

Considering the findings, the researchers suggest that more permanent changes to the roads, such as building a raised wooden bridge, should be implemented within the next decade. They note that though the project was successful at addressing residents’ concerns and incorporating local knowledge, future work could further involve community members in data interpretation and developing recommendations. (Community Science, https://doi.org/10.1029/2023CSJ000058, 2025)

—Sarah Derouin (sarahderouin.com), Science Writer

Citation: Derouin, S. (2025), Flood prediction could boost road resilience off Georgia’s coast, Eos, 106, https://doi.org/10.1029/2025EO250169. Published on 2 May 2025. Text © 2025. AGU. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.

Editors’ Highlights are summaries of recent papers by AGU’s journal editors. Source: Community Science

Two new articles published in Community Science offer evidence and experience-based guidance for doing climate-related research in partnership with Indigenous communities–guidance that applies to community science in general.

This map shows the traditional harvest areas for members of the Organized Village of Kake Tribe and residents of the Community of Kake. Colors represent different Kwaans, or clan groups, who share stewardship with the Keex’Kwaan people of Kake. Credit: Figus et al. [2025], Figure 2

Figus et al. [2025] focus on the experience working in Kake, Alaska using Indigenous evaluation and Ellam-Yua coproduction. Indigenous evaluation is place-based, grounded in Indigenous perspectives, and emphasizes meeting community needs. Ellam-Yua co-production prioritizes processes for equitable collaboration, knowing good practices produce good outcomes, and doesn’t begin with a predetermined scientific goal. Using Indigenous evaluation in the Ellam-Yua co-production allowed for a broader understanding of success, generated a more expansive set of project outcomes, and helped connect climate services with other elements of community wellbeing, including workforce development and healing from trauma. 

Rudolf et al. [2025] provide a way to understand and generalize these findings from Figus et al. [2025] and show how the Indigenous approaches and dispositions can enrich the practice of co-production of knowledge, or CPK. CPK is a process of bringing together diverse perspectives to achieve shared research and practice goals. The paper offers a tool that teams can use to identify individual (perhaps implicit) perspectives on research and how those perspectives interact with other perspectives on research. A second tool helps teams understand different factors that contribute to project success and how they show up in projects.

Together, the research provides guidance for including Indigenous knowledges, practices, and values in community science. As well as showing how including Indigenous knowledges, practices, and values advances the theory and practice of community science overall.

Citations:

Figus, E., Friday, S., O’Connor, J., McDonald, J. J. K. S., James, C., Trainor, S. F., et al. (2025). Sharing our story to build our future: A case study of evaluating a partnership for co-produced research in Southeast Alaska. Community Science, 4, e2023CSJ000073.  https://doi.org/10.1029/2023CSJ000073

Rudolf, M. H. C., Trainor, S. F., O’Connor, J., Figus, E., & Hum, R. (2025). Factors in and perspectives of achieving co-production of knowledge with Arctic Indigenous Peoples. Community Science, 4, e2023CSJ000074.  https://doi.org/10.1029/2023CSJ000074

—Rajul Pandya, Editor, Community Science

Text © 2025. The authors. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.

The Landslide Blog is written by Dave Petley, who is widely recognized as a world leader in the study and management of landslides.

On 28 April 2025, a major debris flow travelled down a channel from a major mountain, Vallunaraju, striking the communities lower down the slope. At least 100 houses were destroyed and two or three people were killed.

I am not in a position to be able to say definitively how this event occurred. Christian Huggel from the University of Zurich has a LinkedIn post that provides some detail. This is a part of what he has posted:-

“Summarizing some information on the recent glacier lake outburst flood (GLOF) that occurred in the early morning of 28 April from one of the glacier lakes at the toe of Vallunaraju (5680 m asl) and badly impacted rural and urban parts of Huaraz in the Andes of Peru. According to videos taken by mountaineers the likely origin of the GLOF (or aluvión) is a rock slope failure into a lake in some 300 m distance of the glacier margin … The analysis suggests that the rock fall triggered an impact wave in the lake with a subsequent debris flow that rushed downvalley along the Casca river, damaged some 100 houses, destroyed about 15 buildings and road infrastructure, and unfortunately also claimed the lives of 2-3 persons. The glacier lake probably formed around the 1970’s as the glaciers of Vallunaraju receded.

“Some lines of evidence suggest that there were rock fall events prior to the 28 April GLOF at this location, including pre-event slope failures likely the day before the disaster.”

This is a cloudy area, so at present I cannot access satellite imagery that shows the slopes affected by the landslide that initiated this event. However, Planet Labs has captured imagery on both 26 April 2025 (before the event) and 30 April 2025 (after the event) that provides some insights into the downstream consequences.

Let’s start with the main channel higher in the slopes of Vallunaraju. This Planet Labs image shows the valley immediately below the steep slope from which this event originated. The marker is at [-9.44993, -77.45431]:-

Planet Labs image before the 28 April 2025 Glacial Outburst Flood (GLOF) / landslide at Vallunaraju in Peru. Image copyright Planet Labs, used with permission. Image dated 26 April 2025.

This is the same area after the landslide / GLOF:-

Planet Labs image of the aftermath of the 28 April 2025 Glacial Outburst Flood (GLOF) / landslide at Vallunaraju in Peru. Image copyright Planet Labs, used with permission. Image dated 30 April 2025.

And here is a slider to compare the two images:-

Clearly, cloud is a major issue in the 30 April 2025 image, and we cannot see the main part of the slope itself, but at the foot of the steep slope extensive scour and erosion is evident, and there is substantial change in the channel below.

This has then led to major impacts in the channel downstream. This is a part of the 30 April 2025 image, with the channel running roughly east to west, with extensive evidence of the aftermath of the debris flow:-

Planet Labs image of the downstream impacts of the 28 April 2025 Glacial Outburst Flood (GLOF) / landslide at Vallunaraju in Peru. Image copyright Planet Labs, used with permission. Image dated 30 April 2025.

Once again, this event highlights the hazards posed by events that occur high in mountain chains, but then travel into populated areas.

Reference

Planet Team 2025. Planet Application Program Interface: In Space for Life on Earth. San Francisco, CA. https://www.planet.com/

Return to The Landslide Blog homepage Text © 2023. The authors. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.

Editors’ Highlights are summaries of recent papers by AGU’s journal editors. Source: Journal of Advances in Modeling Earth Systems 

Understanding and predicting the evolution of the ocean is crucial for improving climate projections and anticipating future environmental changes. A significant source of uncertainty in current oceanic climate models is the accurate representation of mesoscale ocean features, such as eddies and currents.

Silvestri et al. [2025] present a breakthrough in ocean modeling by leveraging GPU-specific programming strategies to accelerate computations drastically. A GPU (Graphics Processing Unit) is a specialized processor designed to perform many calculations simultaneously and process large amounts of data. The model called “Oceananigans” resulting from this study enables routine mesoscale-resolving ocean simulations that were previously impractical due to computational constraints. This work marks a significant step toward next-generation Earth system models, paving the way for higher-fidelity simulations while managing environmental costs in terms of energy consumption.

These findings not only highlight the power of GPUs in climate modeling but also contribute to clarifying the roadmap for porting or redesigning other Earth system components. Moreover, the choice of the Julia programming language opens up numerous opportunities for automatic differentiation, which is fundamental to AI technologies, and for training young researchers in ocean modeling.

Citation: Silvestri, S., Wagner, G. L., Constantinou, N. C., Hill, C. N., Campin, J.-M., Souza, A. N., et al. (2025). A GPU-based ocean dynamical core for routine mesoscale-resolving climate simulations. Journal of Advances in Modeling Earth Systems, 17, e2024MS004465. https://doi.org/10.1029/2024MS004465

—Florian Lemarié, Associate Editor; and Stephen Griffies, Editor-in-Chief, JAMES

Text © 2025. The authors. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.

Coastal waters worldwide are rapidly losing oxygen, causing declines in marine life and affecting communities who rely on the health of coastal waters.

Prominent examples of low-oxygen coastal waters are found in the Baltic Sea, for instance, where oxygen loss in recent decades has led to major ecosystem changes. Potentially toxic cyanobacterial blooms have become frequent and widespread, spawning grounds for cod have been greatly reduced, and fish kills have been observed in coastal waters [Conley et al., 2009]. Similar issues have afflicted the Gulf of Mexico, the Adriatic Sea, the East China Sea, and numerous other areas.

Dead fish are seen in coastal waters near Ostend, Belgium, following a low-oxygen event in 2018. Credit: Grégoire et al. [2023], European Marine Board, CC BY 4.0

The main cause of declining coastal ocean oxygen is well-known: Since the 1950s, phosphorus and nitrogen from agricultural runoff and wastewater have flowed into coastal seas, where they stimulate phytoplankton blooms that, upon their decay, consume oxygen. This process, called eutrophication, is not the only cause of declining oxygen and so-called dead zones in coastal waters: Increasing global temperatures are contributing by reducing both the solubility of oxygen in seawater and vertical mixing of the ocean water column, thereby limiting the aeration of deeper waters [Breitburg et al., 2018].

Indeed, even coastal systems not experiencing eutrophication, such as the Gulf of St. Lawrence in Canada, may be under threat of low oxygen because of changes in ocean circulation linked to climate change [Wallace et al., 2023].

Various means of artificial reoxygenation have been suggested and studied as possible local to regional solutions to coastal oxygen loss.

Long-term reductions in nutrient inputs from land are widely acknowledged as essential to mitigate coastal eutrophication, but such reductions will take time to have an effect. Nutrients have been accumulating in many coastal systems for decades, and switching off inputs will not immediately lead to lower concentrations [Conley et al., 2009]. Moreover, reducing nutrient releases from agricultural lands in many regions is proving challenging. Attempts to curtail the global use of fossil fuels and cut greenhouse gas emissions substantially have also been less successful to date than what is required to affect ocean oxygen [Breitburg et al., 2018].

Amid the challenges of achieving global-scale solutions, various means of artificial reoxygenation have been suggested and studied as possible local to regional solutions to coastal oxygen loss [Stigebrandt et al., 2015]. Yet these approaches come with risks that must be assessed carefully before implementation [Conley et al., 2009].

Such assessments are becoming urgent with the emergence of potential new artificial reoxygenation technologies linked to green hydrogen production. This process of splitting water by electrolysis to generate hydrogen also generates oxygen [Wallace et al., 2023], which could be put to use in coastal waters, particularly where green hydrogen production facilities are located close to the sea.

Oxygen Supply Versus Demand

Coastal seas gain oxygen naturally through air-sea exchange, vertical and lateral mixing of seawater, and photosynthetic production by phytoplankton (Figure 1). They lose oxygen through respiration of organic matter in the water column and the underlying sediment. Surface waters typically remain oxygenated because of rapid air-sea exchange and primary productivity, but in deeper waters, oxygen removal may dominate, especially in systems with limited vertical mixing [Fennel and Testa, 2019].

Fig. 1. The oxygen budget for coastal ocean systems involves several main processes.

The main goal of artificial reoxygenation is to increase the supply of oxygen to deeper waters enough that the water and sediment at the seafloor surface become or remain oxygenated. The oxygen supply needed to achieve this goal depends on the local oxygen demand, which itself depends on the input of organic matter from sinking phytoplankton biomass and the eutrophication history of the system. If organic matter inputs remain high or a lot of organic matter has accumulated on the seafloor, oxygen demand may remain high for a long time. This “legacy” effect can hinder the reoxygenation of a coastal system, as shown in the Baltic Sea [Hermans et al., 2019].

A secondary goal of reoxygenation is to limit recycling of phosphorus from sediments, which, in turn, may reduce the availability of phosphorus as a nutrient for phytoplankton in surface waters. Decreasing how much organic matter is produced and then sinks to the seafloor may lower the oxygen demand for respiration and hence increase oxygen concentrations in deeper waters [Conley et al., 2009].

Inspired by methods used to reoxygenate lakes with some success, two broad approaches have been proposed for artificially reoxygenating coastal systems (Figure 2): bubbling pure oxygen or air into the ocean [Koweek et al., 2020; Wallace et al., 2023] and pumping oxygenated surface water to greater water depths, a process called artificial downwelling [Stigebrandt et al., 2015; Lehtoranta et al., 2022].

Fig. 2. Two key methods for artificially reoxygenating coastal waters include bubble diffusion (left) and downwelling (right). O2, oxygen. Credit: Adapted from Koweek et al. [2020], CC BY 4.0

Bubble diffusers have been used in lakes to oxygenate deep water directly [Koweek et al., 2020] and in shallow coastal systems to destratify and aerate the water column by inducing mixing [Harris et al., 2015]. Artificial downwelling has been tested for local applications in only a few small coastal systems [Stigebrandt et al., 2015; Lehtoranta et al., 2022].

The Imperfections of Artificial Reoxygenation

Studies to date show that artificial reoxygenation can be applied successfully in small estuaries and bays but that its effect lasts only as long as operations are maintained.

Studies to date show that artificial reoxygenation can be applied successfully in small estuaries and bays but that its effect lasts only as long as operations are maintained. This outcome was observed, for example, in two Swedish bays following their reoxygenation through pump-driven downwelling [Stigebrandt et al., 2015; Lehtoranta et al., 2022]. Similarly, when aerators were switched off in a shallow subestuary of the Chesapeake Bay after several decades of aeration, low-oxygen, or anoxic, levels returned within a day [Harris et al., 2015].

The rapid return of anoxia upon discontinuing artificial reoxygenation operations—also known from applications in lakes—illustrates that these approaches alone do not provide permanent solutions to deoxygenation because they do not address its root causes. Moreover, adding oxygen to the water column does not mitigate wider water quality problems. Nuisance algal blooms in many coastal areas will still occur if the availability of nutrients for phytoplankton remains high.

In the Baltic Sea, for example, natural decadal-scale reoxygenation of deeper waters linked to lateral inflow of oxygenated North Sea water does not lead to a removal of phosphorus in the sediment [Hermans et al., 2019]. This lack of an effect results from the highly reducing conditions in the seafloor sediment, which hinder formation of phosphorus-containing minerals. Consequently, reoxygenation of the water column in the Baltic does not necessarily decrease recycling of phosphorus [Hermans et al., 2019], which may continue to fuel cyanobacterial blooms [Conley et al., 2009].

Reoxygenation via artificial downwelling may also be unsuccessful if it causes warming of deeper waters, which is a risk, especially when surface water pumps are used to reoxygenate temperature- and density-stratified coastal waters. Transferring warm surface water to colder, denser depths near the seafloor may weaken stratification and enhance vertical mixing. Although this process may increase the downward transfer of oxygen, it can also boost upward mixing of nutrients, which may enhance biological productivity. This enhancement can ultimately increase the oxygen demand in deeper waters to such an extent that a net decrease in oxygen results [Conley et al., 2009; Lehtoranta et al., 2022].

Warming at depth can also lead to greater metabolic activity and increased respiration of organic matter, further decreasing oxygen concentrations instead of increasing them as intended.

Side Effects on Climate and Habitats

Artificial reoxygenation may have other undesirable effects as well. It can alter the dynamics of greenhouse gases in coastal waters, for example, because increased aerobic respiration increases carbon dioxide production.

Furthermore, bubbling air through shallow coastal waters can enhance upward transport of methane, a potent greenhouse gas, in the water column and its emission to the atmosphere [Lapham et al., 2022]. In eutrophic coastal systems, reoxygenation does not necessarily suppress the release of methane from sediments [Żygadłowska et al., 2024], implying that upon bubbling, methane emissions from coastal waters may be greater than without reoxygenation.

Reoxygenation operations may also alter ocean habitats and have unintended consequences for marine life. Bubbling generates underwater noise, turbulence, and gradients in oxygen pressure that differ from naturally occurring conditions. Artificial downwelling not only changes water column temperatures but also alters vertical salinity distributions, with unknown consequences for marine organisms [Conley et al., 2009; Wallace et al., 2023]. In addition, the return of bottom-dwelling animals with reoxygenation may cause increased sediment mixing that remobilizes sediment contaminants [Conley et al., 2009].

Assessing Artificial Reoxygenation as a Solution

Artificial reoxygenation, when applied, should always be only one of various measures used to improve water quality.

Taken together, the body of evidence from reoxygenation studies to date indicates that long-term improvements in the oxygen levels and quality of coastal waters require reductions in nutrient inputs and greenhouse gas emissions. Hence, artificial reoxygenation, when applied, should always be only one of various measures used to improve water quality.

In heavily managed coastal systems, reoxygenation may be a temporary solution, as illustrated by its successful application in a subestuary of the Chesapeake Bay [Harris et al., 2015]. Elsewhere, such as in the Gulf of St. Lawrence, reoxygenation might be harnessed to maintain the current oxygen state of the system [Wallace et al., 2023]. However, responses to reoxygenation in eutrophic systems with strong legacy effects, where sediments act as a source of nutrients and a sink for oxygen, are very difficult to predict [Conley et al., 2009; Hermans et al., 2019].

The dependence of reoxygenation effects, either from aeration or from pumping, on site-specific biological, chemical, and physical characteristics, which are often poorly known and differ greatly worldwide, also hinders predictions of responses. Yet accurately predicting the effects of artificial reoxygenation before implementing it is critical and consistent with the precautionary principle that in the absence of scientific certainty, we should act to avoid harm.

This principle can be interpreted to suggest that no measures should be taken in some cases and that in other cases, measures should not be postponed because delay could lead to even more harm. Thus, careful case-by-case assessments of the suitability of artificial reoxygenation at given sites are needed—as is careful monitoring when operations are implemented. Modeling studies are valuable for such assessments [e.g., Koweek et al., 2020] but must be paired and validated with field data.

The potential availability of substantial oxygen supplies to support artificial reoxygenation as a result of increasing green hydrogen production further raises the urgency of suitability assessments [Wallace et al., 2023]. If such supplies can be tapped near coastal areas, they may help make artificial aeration operations logistically more viable and sustainable.

Foundations for Responsible Reoxygenation

For areas found to be potentially well suited for artificial reoxygenation interventions, consensus best practices should be followed when initiating pilot studies or larger implementations. As informed by discussions during a recent meeting organized by the United Nations Educational, Scientific and Cultural Organization Intergovernmental Oceanographic Commission’s Global Ocean Oxygen Network, several elements are foundational to these best practices.

Relevant government bodies, such as national and local water management authorities; stakeholders, including representatives of local communities; and scientists should be involved from the outset to safeguard the interests of all parties. Field trials and implementations should consider perceived environmental benefits and risks of the intended intervention, as well as relevant ethical issues, taking into account the intrinsic value of nature.

Monitoring is important for understanding baseline conditions and assessing the effects of reoxygenation on water quality and ecology, including termination effects after an intervention ceases.

Key biological, chemical, and physical parameters of the system where the intervention will occur (as well as of a reference site) should be monitored before, during, and afterward. This monitoring is important for understanding baseline conditions and assessing the effects of reoxygenation on water quality and ecology, including termination effects after an intervention ceases. Continued measurements over years to decades are also critical to determine longer-term effects.

Finally, the results of all field trials, including failures, should be reported completely, transparently, and publicly.

Artificial reoxygenation is unlikely to be a permanent solution to declining ocean oxygen, and it cannot replace essential measures to reduce greenhouse gas emissions and nutrient inputs to ocean waters. But with science-based suitability assessments and ethical, environmentally safe practices, reoxygenation interventions might prove beneficial in some places, allowing temporary mitigation of the detrimental consequences of coastal deoxygenation.

Acknowledgments

This feature article summarizes the discussion of a workshop on marine reoxygenation organized by the Global Ocean Oxygen Network (GO2NE) on 10–11 September 2024. We thank all participants for their contributions: D. Austin, L. Bach, L. Bopp, D. Breitburg, A. Canning, D. Conley, M. Dai, B. DeWitte, H. Enevoldsen, E. Ferrar, A. Galan, V. Garcon, M. Gregoire, B. Gustafsson,, D. Gutierrez, A. Hylén, K. Isensee, R. Lamond, M. Li, K. Limburg, I. Montes, J. Sterling, A. Tan Shau Hwai, J. Testa, D. Wallace, J. Waniek, and M. Yasuhara.

References

Breitburg, D., et al. (2018), Declining oxygen in the global ocean and coastal waters, Science, 359, 1–13, https://doi.org/10.1126/science.aam7240.

Conley, D. J., et al. (2009), Tackling hypoxia in the Baltic Sea: Is engineering a solution?, Environ. Sci. Technol., 43, 3,407–3,411, https://doi.org/10.1021/es8027633.

Fennel, K., and J. M. Testa (2019), Biogeochemical controls on coastal hypoxia, Annu. Rev. Mar. Sci., 11, 105–130, https://doi.org/10.1146/annurev-marine-010318-095138.

Grégoire, M., et al. (2023), Ocean Oxygen: The Role of the Ocean in the Oxygen We Breathe and the Threat of Deoxygenation, edited by A. Rodriguez Perez et al., Future Sci. Brief 10, Eur. Mar. Board, Ostend, Belgium, https://doi.org/10.5281/zenodo.7941157.

Harris, L. A., et al. (2015), Optimizing recovery of eutrophic estuaries: Impact of destratification and re-aeration on nutrient and dissolved oxygen dynamics, Ecol. Eng., 75, 470–483, https://doi.org/10.1016/j.ecoleng.2014.11.028.

Hermans, M., et al. (2019), Impact of natural re-oxygenation on the sediment dynamics of manganese, iron and phosphorus in a euxinic Baltic Sea basin, Geochim. Cosmochim. Acta, 246, 174–196, https://doi.org/10.1016/j.gca.2018.11.033.

Koweek, D. A., et al. (2020), Evaluating hypoxia alleviation through induced downwelling, Sci. Total Environ., 719, 137334, https://doi.org/10.1016/j.scitotenv.2020.137334.

Lapham L. L., et al. (2022), The effects of engineered aeration on atmospheric methane flux from a Chesapeake Bay tidal tributary, Front. Environ. Sci., 10, 866152, https://doi.org/10.3389/fenvs.2022.866152.

Lehtoranta, J., et al. (2022), Different responses to artificial ventilation in two stratified coastal basins, Ecol. Eng., 179, 106611, https://doi.org/10.1016/j.ecoleng.2022.106611.

Stigebrandt, A., et al. (2015), An experiment with forced oxygenation of the deepwater of the anoxic By Fjord, western Sweden, Ambio, 44(1), 42–54, https://doi.org/10.1007/s13280-014-0524-9.

Wallace, D., et al. (2023), Can green hydrogen production be used to mitigate ocean deoxygenation? A scenario from the Gulf of St. Lawrence, Mitigation Adapt. Strategies Global Change, 28, 56, https://doi.org/10.1007/s11027-023-10094-1.

Żygadłowska, O. M., et al. (2024), Eutrophication and deoxygenation drive high methane emissions from a brackish coastal system, Environ. Sci. Technol., 58, 10,582–10,590, https://doi.org/10.1021/acs.est.4c00702.

Author Information

Caroline P. Slomp (caroline.slomp@ru.nl), Radboud University, Nijmegen, Netherlands; also at Utrecht University, Netherlands; and Andreas Oschlies, GEOMAR Helmholtz Centre for Ocean Research, Kiel, Germany

Citation: Slomp, C. P., and A. Oschlies (2025), Could bubbling oxygen revitalize dying coastal seas?, Eos, 106, https://doi.org/10.1029/2025EO250163. Published on 1 May 2025. Text © 2025. The authors. CC BY 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.

Historic wildfires broke out in South Korea in late March 2025, killing 32 people, injuring 45, and displacing about 37,000. In total, the fires burned more than 100,000 hectares (about 247,000 acres), nearly quadruple the area that burned in the country’s previous worst recorded fire season in 2000. (In comparison, the January 2025 Palisades and Eaton Fires in Southern California burned about 91,000 hectares, or 37,000 acres.)

“This study adds to a growing body of science showing how climate change is making weather conditions more favorable to dangerous wildfires.”

A new study by scientists with World Weather Attribution (WWA) suggests that atmospheric warming—caused primarily by fossil fuel burning—made the hot, dry, and windy conditions that drove the South Korean fires about twice as likely and 15% more intense.

About 5,000 buildings burned, including homes, industrial structures, farms, and cultural heritage sites such as the Gounsa Temple in Uiseong, which was originally built in 618 CE.

Credit: World Weather Attribution

“The scale and speed of the fires were unlike anything we’ve ever experienced in South Korea,” said June-Yi Lee, an atmospheric scientist at Pusan National University and the Institute for Basic Science, in a statement. “This study adds to a growing body of science showing how climate change is making weather conditions more favorable to dangerous wildfires.”

Hot, Dry, and Windy

WWA researchers examined the Hot-Dry-Windy Index (HDWI) across the entire country for the month of March. This metric calculates fire risk from temperature, humidity, and wind speed observations.

The combination of high temperatures, low humidity, and high wind speeds that occurred from 22 to 26 March were unusual, even for today’s climate, the researchers found. Such conditions are expected to occur in March only once every 340 years. But this combination of conditions would have been even rarer in a preindustrial climate, occurring only once every 744 years.

The study suggests that the trend in the HDWI was driven primarily by unseasonably high temperatures.

“From March 22–26, the daily maximum average temperature in southeastern Korea averaged around 25°C, which was 10°C higher than the normal March average,” Lee said in a press briefing. Little rain fell in the region this winter, which, combined with high temperatures, led to drier, more flammable fuels. Relative humidity was around 20% at the time of the fires, not unusual for March. Wind speeds on 25 March reached up to 25 meters per second, a short-lived spike that helped the fires spread quickly.

The WWA team also calculated that if the climate warms by another 1.3°C by 2100, the HDWI will continue to increase, with the conditions behind such fires growing another 2 times as likely.

The nature of WWA’s rapid response studies means they are not peer reviewed, but they have published peer-reviewed studies on the methodology they use in all of their analyses. The study marks the World Weather Attribution’s 100th rapid analysis since the organization formed in 2014. It is the sixth to focus on a wildfire.

Credit: World Weather Attribution A Strong Case

ClimaMeter, another project that examines how extreme weather events may have been affected by a changing climate, released a study about South Korea’s wildfires on 25 March. (As with the WWA study, it was not peer reviewed but used peer-reviewed methods.)

ClimaMeter uses a different methodology than WWA does but had similar findings, reporting that the meteorological conditions leading up to the fires were about 2°C hotter, 30% drier, and 10% windier compared with similar past events.

“For the moment, it’s more difficult to prove that climate change did not affect an event.”

Davide Faranda, a physicist with the French National Centre for Scientific Research and coordinator of ClimaMeter, pointed out that their study showed climate change strengthened the meteorological conditions conducive to fires, not necessarily that climate change caused the fires. He was not involved with the WWA study but noted that the two groups’ rapid response studies often arrive at similar or complementary findings.

“For the moment, it’s more difficult to prove that climate change did not affect an event,” he said.

“A decade ago, the influence of climate change on events was less clear. But now it’s undeniable. The wildfires in South Korea are a case in point,” said Friederike Otto, WWA colead and a climatologist at Imperial College London, in a statement.

—Emily Dieckman (@emfurd.bsky.social), Associate Editor

Citation: Dieckman, E. (2025), Climate change heightened conditions of South Korean fires, Eos, 106, https://doi.org/10.1029/2025EO250170. Published on 30 April 2025. Text © 2025. AGU. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.

Source: Community Science

Research shows that communities are best able to mitigate the effects of climate change when they can work alongside scientists on adaptation plans. Hanson et al. recently extended this finding to Indigenous communities in the Colorado Plateau, including members of the Navajo Nation, Hopi Tribe, and Ute Mountain Ute Tribe.

To learn more about the qualities that make climate education most accessible to these groups, the researchers conducted a series of listening circles, interviews, and consultations with Indigenous peoples and Westerners with extensive experience working in Indigenous communities. They collaborated with members of the Nature Conservancy’s Native American Tribes Upholding Restoration and Education, or NATURE, program, which aims to equip Indigenous college students with natural resource management skills.

Several themes emerged. Indigenous students are most likely to engage in climate education when they’re actively recruited for a program, when mentors are willing to learn from students as well as teach them, and when a program emphasizes the value of integrating Traditional Knowledges with Western science, for instance. Small class sizes and ample one-on-one instruction also keep students engaged.

On the basis of these findings, the researchers created a climate module that can be taught as part of a broader college-level environmental science curriculum, for example, as part of a program like NATURE. The module is “menu style,” meaning that instructors and students can choose the activities they find appealing from an array of options. One option is classroom lessons on issues that are relevant to the Colorado Plateau, such as water conservation and cattle management. Another involves field trips, such as a day trip down the Colorado River, during which guides provide insights into how climate change is altering the landscape.

Indigenous students are “uniquely positioned to engage in environmental restoration” because they have deep connections with natural systems, the researchers wrote. This collaboratively designed program could help students achieve this potential, they say. (Community Science, https://doi.org/10.1029/2023CSJ000054, 2025)

—Saima May Sidik (@saimamay.bsky.social), Science Writer

Citation: Sidik, S. M. (2025), Teaming up to tailor climate education for Indigenous communities, Eos, 106, https://doi.org/10.1029/2025EO250166. Published on 30 April 2025. Text © 2025. AGU. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.

Source: AGU Advances

Hundreds of millions of people live in areas that could be affected by volcanic eruptions. Fortunately, clues at the surface, such as earthquakes and ground deformation, can indicate movement within underground magma dikes—sheets of magma that cut across layers of rock. Scientists can use these clues to make potentially lifesaving predictions of eruptions.

But there is room for improvement. Eruption predictions rely on modeling magma dikes, and most models treat magma as a simple Newtonian fluid (like water) whose viscosity stays constant under stress. However, magma’s crystals and bubbles make it more likely to behave as a non-Newtonian fluid whose viscosity decreases under greater stress (known as shear thinning). That’s especially true as it approaches the surface. Ketchup behaves similarly: It pours more easily from a jar when shaken.

Lab experiments by Kavanagh et al. reveal new insights into the potential dynamics of non-Newtonian magma flow in dikes. These findings could ultimately help improve eruption prediction strategies.

To mimic magma dikes, the researchers injected various fluids into a translucent and elastic solid gelatin material representing Earth’s crust. The injected fluids contained suspended tracer particles that could be illuminated by laser light, allowing the researchers to track each fluid’s flow within the forming dike as it traveled up from the injection site to the surface, where it “erupted” from the gelatin. They compared the behaviors of two non-Newtonian shear-thinning fluids, hydroxyethyl cellulose (a thickener often found in cosmetics) and xanthan gum (a thickener often added to foods), to water, a Newtonian fluid.

The experiments showed that the flow patterns of these fluids were very different from the flow patterns of water. However, even though their internal flow patterns differed, all fluids formed dikes with a similar shape and speed as they approached the surface.

These findings suggest that the primary information currently used to predict impending eruptions—such as the shape and speed of magma dikes—does not necessarily correlate with information about magma flow dynamics within the dikes. This result is significant because flow dynamics depend on magma characteristics that can affect how explosive an eruption will be or how quickly or how far the lava will travel.

Further research could help link these findings to real-world geological evidence and explore how they might help to improve eruption forecasting, the researchers say. (AGU Advances, https://doi.org/10.1029/2024AV001495, 2025)

—Sarah Stanley, Science Writer

Citation: Stanley, S. (2025), Matching magma dikes may have different flow patterns, Eos, 106, https://doi.org/10.1029/2025EO250167. Published on 30 April 2025. Text © 2025. AGU. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.

Editors’ Highlights are summaries of recent papers by AGU’s journal editors. Source: Journal of Geophysical Research: Solid Earth

Supercritical fluids—hydrous silicate liquids where water and melt become fully miscible—are believed to play a key role in chemical transport and element redistribution in subduction zones. However, the atomic-scale processes underlying their high mobility are poorly understood.

Chen et al. [2025] use first-principle molecular dynamics simulations to examine the diopside–H2O system over a wide range of water contents, pressures (up to 12 gigapascals), and temperature (3000 kelvin). Their results show that water promotes the breakdown of the silicate network by converting bridging oxygens (BOs) into non-bridging oxygens (NBOs), leading to the formation of smaller, less polymerized silicate clusters with greater diffusivity and structural stability. This depolymerization enhances atomic mobility and reduces viscosity, with strong linear correlations observed between polymerization degree and transport properties. The findings identify water-induced depolymerization as the primary mechanism behind the high mobility of supercritical fluids.

These insights have broad implications for understanding magma transport dynamics and the geochemical signatures—such as uranium-thorium disequilibria—in arc lavas. The study highlights the critical role of water in regulating the structure and dynamics of silicate fluids in subduction-related magmatic and mineralizing processes.

Citation: Chen, B., Song, J., Zhang, Y., Wang, W., Zhao, Y., Wu, Z., & Wu, X. (2025). Water dissolution driving high mobility of diopside-H2O supercritical fluid. Journal of Geophysical Research: Solid Earth, 130, e2024JB030956.  https://doi.org/10.1029/2024JB030956  

—Jun Tsuchiya, Editor, JGR: Solid Earth

Text © 2025. The authors. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.

The Landslide Blog is written by Dave Petley, who is widely recognized as a world leader in the study and management of landslides.

I have frequently highlighted the growing impact of multiple landslide events triggered by extreme rainfall around the world. Whilst there is little doubt that such events are becoming more common, they have occurred through history too. I recently came across a paper (Fiorillo et al. 2019) that documented such an event in 1954. The account is fascinating.

The paper sought to use historic aerial images and topographic data to reconstruct an inventory of the landslide triggered during this event. The location is the area around the villages of Vietri sul Mare and Maiori, which are sited on the beautiful Amalfi Coast in the Campania region of southern Italy. This is the area in the vicinity of [40.67, 14.73] – the Google Earth image below shows the landscape as it is today:-

Google Earth image of the modern setting of the 1954 landslides on the Amalfi Coast in Italy.

The analysis of Fiorillo et al. (2019) indicates that in parts of this area, 500 mm of rainfall fell in the storm that triggered these landslides. They have mapped over 1,500 landslides triggered by the storm – these are shown in the map below:-

Landslides triggered during the 1954 rainfall event on the Amalfi Coast in Italy. Map by Fiorillo et al. (2019).

As the map shows, the density of landslides was extremely high in the hills above Vietri sul Mare and Maiori. The failures were mostly shallow landslides, with many transitioning into channelised debris flows.

In total, it is believed that 316 people lost their lives in this disaster. There is some archive footage of the aftermath in the Youtube video below:-

Reference

Fiorillo, F., Guerriero, L., Capobianco, L., Pagnozzi, M., Revellino, P., Russo, F., and Guadagno, F. M., 2019. Inventory of Vietri-Maiori landslides induced by the storm of October 1954 (southern Italy). Journal of Maps, 15 (2), 530–537. doi: https://doi.org/10.1080/17445647.2019.1626777

Return to The Landslide Blog homepage Text © 2023. The authors. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.

body {background-color: #D2D1D5;} Research & Developments is a blog for brief updates that provide context for the flurry of news regarding law and policy changes that impact science and scientists today.

The EPA plans to cancel 781 grants, almost all focused on environmental justice, according to a court document filed last week.

In Woonasquatucket River Watershed Council v. Department of Agriculture, a coalition of nonprofits is challenging the Trump administration’s freezing of funding from the Biden-era Inflation Reduction Act (IRA) and Bipartisan Infrastructure Law. In the recent court document, Daniel Coogan, an administrator in the Office of Mission Support for the EPA, stated that the agency completed a grant-by-grant review of its awards to ensure that grants aligned with administration priorities. Those that were not aligned were targeted for termination.

All 781 grants targeted for termination fall under programs formed by the IRA, a 2022 law that helped to promote clean energy and bolster environmental projects. Most of the grants are part of EPA programs focused on environmental justice and include projects to help some of the United States’ most environmentally disadvantaged communities be resilient to the effects of climate change and protect residents from pollution. 

According to the court document, 377 grantees have been notified that their funding has been terminated, and the remaining 404 grantees will receive notices within the next two weeks. 

Hundreds of grantees’ projects will be affected by the terminations. In one such project, San Diego nonprofit Casa Familiar expected to receive $12.7 million to help a majority-Latino community obtain low-cost, zero-emission transportation and indoor air monitors and purifiers. The group has been unable to withdraw funds for months and now awaits notification that their grant has been terminated. In another example, the community of Chiloquin, Oregon, now expects that a planned community center and disaster shelter may never be built after the EPA suspended funding for the project.

 
Related

•  EPA is Canceling Nearly 800 Environmental Justice Grants, Court Filing Reveals
•  Democrats Demand EPA Reverse Termination of Grants for Clean Air and Water
•  Court Filing From Daniel Coogan
• Get Involved: AGU Science Policy Action Center
 

The news of the cancellations comes shortly after hundreds of EPA employees working on diversity, equity, and inclusion and environmental justice issues were given notice that they would be fired or reassigned.

According to the Washington Post, experts are concerned that the EPA did not conduct the full grant-by-grant review process required to terminate ongoing grants. “They’re claiming to the court that each one of those was done on an individualized basis, even though they haven’t shown any evidence,” Jillian Blanchard, vice president of climate change and environmental justice at Lawyers for Good Government, told the Washington Post.

—Grace van Deelen (@gvd.bsky.social), Staff Writer

These updates are made possible through information from the scientific community. Do you have a story about how changes in law or policy are affecting scientists or research? Send us a tip at eos@agu.org. Text © 2025. AGU. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.

body {background-color: #D2D1D5;} Research & Developments is a blog for brief updates that provide context for the flurry of news regarding law and policy changes that impact science and scientists today.

In a joint statement yesterday, Mexican and U.S. officials announced that Mexico will immediately transfer some of its water reserves to the United States and also allow a larger share of the Rio Grande River to flow into the United States. This concession from Mexico, which will last through at least October, seems to have averted the threat of additional tariffs and sanctions threatened by President Trump in early April.

Mexico and the United States share several major rivers, including the Rio Grande, the Colorado, and the Tijuana. Control over how much water each country receives from these rivers was set in a 1944 treaty. Under the treaty, Mexico must deliver 1.75 million acre-feet of water to the United States from six tributaries every 5 years, or an average of 350,000 acre-feet every year (An acre-foot is the amount of water needed to cover 1 acre of land to a depth of 1 foot.)

 
Related

•  Mexico to Give U.S. More Water From Their Shared Rivers (New York Times)
•  Welcoming Mexico’s Water Deliveries to the United States and Steps to Meet 1944 Water Treaty Requirements (U.S. State Department)
•  México y los Estados Unidos anuncian acuerdo sobre las asignaciones de agua del río Bravo conforme al Tratado de 1944 (Gobierno de México)
• Get Involved: AGU Science Policy Action Center

The United States and Mexico renegotiated parts of the treaty last year under the Biden Administration, allowing Mexico to meet its treaty obligations with water from other rivers, tributaries, or reserves. Yesterday’s announcement marks a commitment from Mexico to adhere to the amended treaty, rather than striking a new deal.

As climate change has worsened drought conditions in Mexico the country has struggled to meet the obligations of the treaty while supporting its farmers. Mexico’s current water debt to the United States is roughly 1.3 million acre-feet (420 billion gallons). Mexico’s president Claudia Sheinbaum acknowledged this water debt but said that Mexico has been complying with the treaty to “to the extent of water availability.”

In 2020, tensions over these water deliveries boiled over into violence: Mexican farmers rioted and seized control of a dam near the U.S.-Mexico border to halt deliveries. Mexican officials worry that increasing water deliveries during the hottest and driest months of the year will once again spark civil unrest among farmers.

—Kimberly M. S. Cartier (@astrokimcartier.bsky.social), Staff Writer

These updates are made possible through information from the scientific community. Do you have a story about how changes in law or policy are affecting scientists or research? Send us a tip at eos@agu.org. Text © 2025. AGU. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.

Source: AGU Advances

This is an authorized translation of an Eos article. 本文是Eos文章的授权翻译。

人类对地球的改造如此深刻，以至于大气化学家保罗·克鲁岑（Paul Crutzen）和生物学家尤金·斯托默（Eugene Stoermer）在2000年提出，全新世已经结束，“人类世”（Anthropocene）或人类时代已经开始。然而，尽管人类活动引发了如此巨大的变化，国际地质科学联合会（IUGS）去年仍决定不将人类世正式认定为当前的地质时代。现在，参与这一过程的几位科学家发表了一篇评论文章，解释了为什么他们认为应该再给人类世一个被认定为地质时代的机会。

McCarthy等人反驳了针对该提议的两个相关批评：首先，拟议的人类世仅开始于72年前，而每个地质时代通常跨越数百万年；其次，未来不属于地质时间，因此，基于人类将在遥远的未来在地球上留下印记的预期来指定一个时代是不恰当的。

作者认为，人类世的长度无关紧要，因为从功能上讲，地球已经进入了一个前所未有的时期。自20世纪中叶以来，地球的能源消耗量是之前11700年的六倍。由此产生的结果是，全球气温急剧上升，对从海平面到生物多样性再到冰盖等方方面面都产生了广泛的影响。这些变化将是长期的，有些甚至是不可逆转的。作者说，事实上，在如此短的时间内发生如此剧烈的变化，表明地球已经进入了一个新纪元。

一些地层学家认为，划定一个以人类为中心的时代会使地质学变得政治化，但作者认为，忽视数据以维持现状同样具有政治性。同样，有报道称，这个问题在十年内不会被重新讨论，因此我们是否生活在人类世的问题在那之前是确定的，作者对此感到愤慨。“事实并非如此，”他们写道。

—科学撰稿人Saima May Sidik (@saimamay.bsky.social)

This translation was made by Wiley. 本文翻译由Wiley提供。

Read this article on WeChat. 在微信上阅读本文。

Text © 2025. AGU. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.

You can’t teach an old dog new tricks, but can you teach the current generation of Earth scientists about emerging artificial intelligence and machine learning (AI/ML) methods relevant to their research? From our experience helping run a program intended to do just that at the U.S. Department of Energy’s (DOE) Pacific Northwest National Laboratory (PNNL), the answer is yes.

Earth scientists, from those focused on the atmosphere or ocean to those studying the continents or deep subsurface, often work with extremely large—sometimes global—datasets, trying to find patterns among noisy real-world observations. AI/ML is well suited for such tasks.

Relatively few Earth scientists have been trained in artificial intelligence and machine learning (AI/ML) methods, meaning unfulfilled opportunities exist to learn from the growing volumes of Earth science data available.

AI/ML approaches have recently been used, for example, to replace slow, numerical representations of rainfall in a global general circulation model [Gettelman et al., 2021]. Similarly, AI/ML image detection techniques have been used with weather radar datasets to better predict short-term rainfall [Ji and Xu, 2024]. Yet relatively few domain scientists in the field have been trained in these methods, meaning unfulfilled opportunities exist to learn from the growing volumes of Earth science data available.

Several hundred data scientists work at PNNL, and for more than a decade, the lab has developed AI/ML approaches to address critical challenges in scientific discovery, energy resilience, and national security. Recent advancements in computational techniques and methodologies have sparked renewed interest in applying AI/ML across various disciplines. However, connecting the expertise of PNNL’s data scientists to Earth science research at the lab—encompassing atmospheric, hydrological, and environmental sciences—has been a challenge.

Beginning in 2022, researchers at PNNL implemented a two-step approach—a boot camp followed by a hackathon—to prepare their colleagues to incorporate AI/ML into their research effectively. Eighty percent of those who participated in both events are now using ML techniques in their research, and the experience has boosted collaboration between the lab’s data scientists and Earth scientists. The program has also led to innovative new projects, and its initial success suggests it may be a useful model for other organizations.

Boot Camp

Prior to PNNL initiating the program, many of the lab’s Earth scientists expressed interest in learning more about AI/ML and exploring its applicability for addressing a wide variety of science questions.

Atmospheric science in particular offers ideal ground for teaching and applying ML methods because these methods are conducive to tackling many common tasks in the field. For example, they can help fill patchy datasets, such as in time series of satellite imagery [Appel, 2024]; correct biases in gridded data (e.g., overestimations of solar radiation reaching Earth in reanalysis products) [Chakraborty and Lee, 2021]; merge measurements of atmospheric properties into numerical models [Krasnopolsky, 2023]; and iteratively improve models [Irrgang et al., 2021]. Furthermore, the field is ripe with the sort of very large, high-quality datasets that are necessary for applying modern ML methods.

The staff’s interest and the clear relevance of AI/ML for their work motivated development of an initial 10-week boot camp, held in fall 2022, with weekly hybrid (online and in-person) sessions attended by 30–50 people. We enlisted 10 in-house data scientists to design lessons, hands-on tutorials, and activities covering a range of AI/ML methods and tools.

As a result of the boot camp approach, participants gained understanding and appreciation of data curation for AI/ML and the full gamut of AI/ML methods they could use in their research.

The first four sessions introduced participants to the basics of ML, with each session building upon the previous one and focusing on more state-of-the-art approaches. The remaining sessions covered popular deep learning techniques such as convolutional neural networks (CNNs), generative adversarial networks, transformers, and recurrent neural networks. They also covered topics such as how to use the ML libraries Keras and PyTorch, which offer the tools to run these models and other useful resources.

To connect the lessons to the participants’ research interests, each one featured an Earth science–relevant activity, such as using maps of monthly sea surface temperature anomaly data from NOAA satellites with unsupervised learning algorithms to detect the phases of the El Niño–Southern Oscillation (i.e., El Niño and La Niña). The instructors developed and guided participants through virtual notebook environments that included fundamental information (with references) about the topic of the activity and heavily commented model code that could be run interactively. Time was also allotted for participants to better familiarize themselves with the models by running them in parallel on their own research computing environments.

As a result of the boot camp approach, participants gained understanding and appreciation of data curation for AI/ML and the full gamut of AI/ML methods they could use in their research. One remarked that they were impressed by the diversity of applications for ML and said, “I can tell if I continue to work on this skill, it will open a lot of doors and funding opportunities in the future for me.” Another commented, “By the end, I felt my programming skills had improved as well.”

Together with colleagues, one scientist at the lab who took part in the training applied knowledge and code directly from the boot camp material in research exploring stochasticity in aerosol-cloud interactions using field campaign data [Li et al., 2024].

The instructors also reported that participating in the boot camp was worthwhile for several reasons. Each of their lessons and student demonstrations were reviewed by the other instructors, which fostered connections among peers knowledgeable in ML. According to one instructor, teaching their fellow staff also “helped provide context of how valuable my expertise is here at the lab.”

Additional hands-on opportunities were necessary to bridge the gap between learning ML and putting it into practice. So we organized a second learning opportunity—this time a hackathon.

In addition, creating and presenting the weekly lesson plans to an audience with limited knowledge about AI/ML offered opportunities for instructors to improve their teaching skills. Furthermore, the adaptability of the instructional materials to other domain sciences supports the materials’ value, longevity, and easy reuse in future trainings and research.

One year after the boot camp, participant responses to a questionnaire indicated that though many had gained literacy in ML, most had not taken the next step to start incorporating ML methods into their research. The results also showed that additional hands-on opportunities were necessary to bridge the gap between learning ML and putting it into practice. So we organized a second learning opportunity—this time a hackathon—focused on pairing ML experts and data scientists with domain scientists who share common research interests.

The Hackathon

Twenty-five domain and data scientists, many of whom had participated in the boot camp, took part in the 6-week hackathon, which began in January 2024. The domain scientists involved work in various areas of Earth science and as part of DOE projects such as the Atmospheric Radiation Measurement user facility and the PNNL-led Addressing Challenges in Energy: Floating Wind in a Changing Climate (a DOE Energy Earthshot research center), as well as NASA’s Aerosol Cloud Meteorology Interactions over the western Atlantic Experiment project.

In preparing for the course, we discovered that these scientists often had trouble formulating research questions suited to ML methods and selecting which ML method to use. Prehackathon brainstorming sessions proved critical to success. During the first prehackathon meeting, the organizing committee gathered participants virtually to group the domain scientists by their topics of interest—vegetation-atmosphere interactions, clouds and precipitation, aerosols and aerosol-cloud interactions, hydrology, and wind energy—and to brainstorm potential research questions to address.

Each of the five groups then pitched project ideas to the participating ML experts and data scientists, who selected which team to join. With the teams assembled, each further workshopped a research question within their topic focus area—as well as which ML methods to use—that they could address within the duration of the hackathon. For example, one team chose to use a CNN model to identify open- versus closed-cell atmospheric convection in radar data, which helps explain distributions of clouds and rainfall.

During the hackathon, all the teams met weekly to discuss progress and exchange ideas for continuing work. This assessment method allowed the domain scientists to engage further with experts in the PNNL ML community, who provided feedback and answers to follow-up questions, such as how to prepare data for use in the ML models. Data preparation proved to be the most time-consuming step for the domain scientists because of the challenges of correctly formatting time series and gridded atmospheric datasets (e.g., temperature, relative humidity, and pressure) before they were fed into the models.

At the end of the 6 weeks, four of the five project groups had successfully processed their data and run them through their models to achieve results related to their initial questions. The fifth group, upon reflection, agreed that selecting an overly broad research question hindered progress on their project. Their experience underscored the importance of clearly defining a focused research question—and an appropriate ML approach—with cross-disciplinary consultation among scientists.

Soon after the hackathon concluded, a representative from each team presented their project during a seminar. A postseminar Q&A about the projects with staff who had not participated in the hackathon was positive and engaging, indicating a base level understanding of AI/ML methods within the division that was not present before the boot camp.

Fostering an AI-Literate Workforce

With growing datasets of Earth observations and ongoing computing advancements, AI/ML is an increasingly useful tool to aid in skillfully assessing conditions and processes in the Earth system.

Jingjing Tian presents results from the hackathon at the HydroML Symposium in May 2024. Her project involved training a convolutional neural network (CNN) model to detect open versus closed convection using weather radar data. Credit: Andrea Starr/Pacific Northwest National Laboratory

At PNNL, more than 20% of the research workforce is advancing AI and its applications in science. The initial goal of the recent training activities was to further grow ML expertise and implementation specifically within the lab’s Atmospheric, Climate, and Earth Sciences (ACES) division. The lessons and successes of these activities suggest that other organizations similarly seeking to expand their use of AI/ML may benefit from the model of PNNL’s approach.

The different approaches of the boot camp and the hackathon allowed instructors to meet participants at their preferred comfort level and cater to different learning styles.

The boot camp created a long-term, structured environment for a large number of staff to better understand the increasingly complex ML landscape, whereas the follow-up hackathon allowed a smaller group of eager staff to be coached in a faster-paced environment to produce deliverables. The different approaches of the boot camp and the hackathon allowed instructors to meet participants at their preferred comfort level and cater to different learning styles.

The results demonstrate that although learning new skills in AI/ML takes time, the effort is worthwhile and a collaborative, cross-disciplinary environment accelerates such learning. Staff self-reported that work done during the boot camp and hackathon had resulted in three conference presentations, including at the HydroML 2024 Symposium, and two publications (another is still in preparation).

Furthermore, PNNL reported an uptick in proposals from its Earth scientists for various internal funding opportunities focused on leveraging AI/ML methods. More proposals means more competition for funding, which should drive innovation and ultimately lead to stronger projects moving forward.

Another lesson from our experience was that sourcing instructors from within PNNL (i.e., ML experts who are already colleagues of Earth scientists in the ACES division) facilitated future collaborations between data and domain scientists and new research opportunities that wouldn’t have been possible previously. One of the participating AI/ML experts noted to us that “after the hackathon, many lab scientists reached out to me for help in implementing ML/AI algorithms into their work,” leading to multiple collaborations.

Hackathon participant Sha Feng’s comments offer additional, anecdotal evidence of the success of PNNL’s program: “Participating in the hackathon has been a transformative experience,” Feng said. “By bridging the gap between atmospheric science and data science, we have created a foundation for future projects that leverage the strengths of both fields.”

We plan to continue to bridge such gaps at PNNL—and we support other organizations doing the same—to advance applications of AI/ML to address crucial questions about our planet, from the atmosphere to the ocean to the solid Earth.

Acknowledgments

We acknowledge the instructors who took part in the boot camp and hackathon: Peishi Jiang, Tirthankar “TC” Chakraborty, Andrew Geiss, Sing-Chun “Sally” Wang, Robert Hetland, Rachel Hu and Danielle Robinson from Amazon Web Services, Erol Cromwell, Maruti Mudunuru, Robin Cosbey, Samuel Dixon, and Melissa Swift. We also acknowledge the work of colleagues who contributed to this article and supported these efforts: Sing-Chun “Sally” Wang, Court Corley, Larry Berg, Timothy Scheibe, Ian Kraucunas, and Rita Steyn.

References

Appel, M. (2024), Efficient data-driven gap filling of satellite image time series using deep neural networks with partial convolutions, Artif. Intell. Earth Syst., 3, e220055, https://doi.org/10.1175/AIES-D-22-0055.1.

Chakraborty, T. C., and X. Lee (2021), Using supervised learning to develop BaRAD, a 40-year monthly bias-adjusted global gridded radiation dataset, Sci. Data, 8(1), 238, https://doi.org/10.1038/s41597-021-01016-4.

Gettelman, A., et al. (2021), Machine learning the warm rain process, J. Adv. Model. Earth Syst., 13(2), e2020MS002268, https://doi.org/10.1029/2020MS002268.

Irrgang, C., et al. (2021), Towards neural Earth system modelling by integrating artificial intelligence in Earth system science, Nat. Mach. Intell., 3, 667–674, https://doi.org/10.1038/s42256-021-00374-3.

Ji, C., and Y. Xu (2024), trajPredRNN+: A new approach for precipitation nowcasting with weather radar echo images based on deep learning, Heliyon, 10(18), e36134, https://doi.org/10.1016/j.heliyon.2024.e36134.

Krasnopolsky, V. (2023), Review: Using machine learning for data assimilation, model physics, and post-processing model outputs, Off. Note 513, 32 pp., Natl. Cent. for Environ. Predict., College Park, Md., https://doi.org/10.25923/71tx-4809.

Li, X.-Y., et al. (2024), On the prediction of aerosol-cloud interactions within a data-driven framework, Geophys. Res. Lett., 51, e2024GL110757, https://doi.org/10.1029/2024GL110757.

Author Information

Lexie Goldberger, Peishi Jiang, Tirthankar “TC” Chakraborty, Andrew Geiss, and Xingyuan Chen (xingyuan.chen@pnnl.gov), Pacific Northwest National Laboratory, Richland, Wash.

Citation: Goldberger, L., P. Jiang, T. Chakraborty, A. Geiss, and X. Chen (2025), A two-step approach to training Earth scientists in AI, Eos, 106, https://doi.org/10.1029/2025EO250160. Published on 29 April 2025. Text © 2025. The authors. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.

Editors’ Highlights are summaries of recent papers by AGU’s journal editors. Source: Journal of Advances in Modeling Earth Systems

Climate models are essential tools for understanding and predicting our planet, but accurately setting their many internal parameters is complex and has been a labor-intensive manual task in the past.

Elsaesser et al. [2025] showcase a method using machine learning to automatically tune, or “calibrate,” the NASA GISS climate model against real-world observations. The authors develop a neural network surrogate of GISS ModelE to efficiently explore different parameter settings, creating a collection of well-performing model versions known as a calibrated physics ensemble. A key success was significantly improving the model’s simulation of challenging features such as shallow cumulus clouds and Amazon rainfall—longstanding modeling challenges—without negatively impacting, for example, radiation fields.

This work represents an important advance, moving automated calibration techniques from theoretical research into practical application for large-scale climate modeling. It brings us an essential step closer to more trustworthy climate predictions. 

Citation: Elsaesser, G. S., van Lier-Walqui, M., Yang, Q., Kelley, M., Ackerman, A. S., Fridlind, A. M., et al. (2025). Using machine learning to generate a GISS ModelE calibrated physics ensemble (CPE). Journal of Advances in Modeling Earth Systems, 17, e2024MS004713. https://doi.org/10.1029/2024MS004713

—Tapio Schneider, Editor, JAMES 

Text © 2025. The authors. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.

Where do meteorites come from? A new analysis of 75 fall events suggests that meteorites with different geologies travel from different places in the asteroid belt, which separates Mars and Jupiter. Researchers traced some types of meteorites to particular asteroid families, creating a geologic map of meteorite origins. Most meteorites were generated by just a few recent collisions between asteroids.

“Understanding the asteroid belt is really looking into the past, into the formation of the solar system, and into all the dynamics that happened at that time,” said Peter Jenniskens, coauthor on the new analysis and a meteorite astronomer at the SETI Institute in Mountain View, Calif. Those early interactions and collisions matter because much of the water on Earth and a lot of the organics likely came from primitive asteroids, he added.

Tracking Falls

Spacecraft have returned small volumes of material from the Moon, comets, and asteroids, but meteorites remain the primary way that scientists get their hands on space rocks.

“By reconstructing where specific meteorite types formed, we gain a clearer picture of the compositional and thermal gradients that existed when the solar system was young,” said Michaël Marsset, an astronomer at the European Southern Observatory in Santiago, Chile. “This has major implications for understanding how habitable environments emerge, not just here but potentially in other planetary systems as well.” Marsset studies small solar system objects and Earth impactors and was not involved in the new study.

But matching a meteorite to the asteroid it came from is a tall task.

“Asteroids in space look quite a bit different than the meteorites that we have in our laboratories.”

“Asteroids in space look quite a bit different than the meteorites that we have in our laboratories, because the asteroids in space are covered by regolith and debris and they are exposed to solar radiation and solar wind,” Jenniskens said. A meteorite might come from an asteroid’s interior, which could look entirely different from its surface. That makes it challenging to use astronomical observations alone to match meteorites to their asteroid parents.

When someone witnesses a meteorite falling to Earth, scientists can try to backtrack its orbit to a point of origin. Combining this information with the meteorite’s geochemistry, mineralogical structure, and age, they can then figure out which asteroid or asteroid family—a group of asteroids that originate from the same collision event—sent it hurtling toward Earth.

The trouble is that meteorites fall more or less at random, Jenniskens explained. It has taken a while to document enough falls to spot patterns, he said. Just 6 years ago, there were fewer than 40 meteorite falls with well-measured trajectories.

“The number of falls has doubled since that time,” Jenniskens said.

Meteorite researchers have set up more than 2 dozen global camera networks that have detected many of these recent falls—roughly 14 falls per year. Also, the rising popularity of dash cameras and doorbell cameras has contributed to the surge of recent detections.

In the new analysis, about 36 of the 75 falls were recorded by residential security cameras, Jenniskens said. People report fireball sightings and submit videos for analysis. “We really depend on the citizen science.”

Meteorite Ancestry

Jenniskens and his colleague Hadrien Devillepoix of Curtin University in Perth, Australia, reviewed the trajectories, geochemistry, mineralogy, and size of 75 meteorites. They also looked at the meteorites’ ages, calculated on the basis of how long a rock’s surface has been exposed to cosmic rays.

Though a few asteroids are suspected sources of certain meteorite types, a meteorite’s age was often the key factor in figuring out which asteroid family produced the meteorite. The positions and movements of asteroids within a family evolve in a predictable way over time, and if this so-called dynamical age matched a meteorite’s cosmic ray age, that family was more likely to be the meteorite’s source.

NASA’s Dawn spacecraft orbited asteroid 4 Vesta and mapped its surface geology and chemistry. Debris from impacts that made some of these craters makes it way to Earth as HED meteorites. Credit: NASA/JPL-Caltech/UCAL/MPS/DLR/IDA, Public Domain

Most of the meteorites originated from a handful of asteroid families, and different classes of meteorites could be traced to different parts of the asteroid belt.

Jenniskens and Devillepoix confirmed that very low iron LL-type meteorites, such as the Chelyabinsk meteorite, originated from the extensive Flora family in the inner asteroid belt. They tracked H-type chondrites to debris clusters in the Koronis, Massalia, and Nele families. They also traced low-iron L chondrites to the Hertha asteroid family, rather than to the previously determined Massalia family.

“Hertha is covered by dark rocks that were shock blackened, indicative of an unusually violent collision,” Jenniskens said. “The L chondrites experienced a very violent origin 468 million years ago when these meteorites showered Earth in such numbers that they can be found in the geologic record.”

“It turns out that, yes, our HED meteorites seem to come from Vesta, not from its family.”

Marsset has also worked to trace meteorites to their asteroid origins, though his team used astronomical observations of asteroids and numeral modeling, rather than meteorite data. “Even with these different approaches, we’re mostly converging on similar conclusions,” Marsset said. “Where we disagree, well, that’s part of the fun! For example, I’d gladly bet a pint with Dr. Jenniskens and Dr. Devillepoix that L chondrites come from the Massalia family, not Hertha,” he joked.

The team also looked at howardite, eucrite, and diogenite (HED) meteorites, achondrites that have long been tied to the Vesta asteroid family. According to the new analysis, the volume of HED material that made its way to Earth must have come from a collision so large that it only could have happened on Vesta itself. (Vesta is the second-largest object in the asteroid belt.) What’s more, the cosmic ray exposure ages of HED meteorites closely match the ages of particular impact craters on Vesta’s surface that were mapped by NASA’s Dawn spacecraft.

“It turns out that, yes, our HED meteorites seem to come from Vesta, not from its family,” Jenniskens said.

Decoding Solar System History

“What’s remarkable about this work is the broader picture it starts to paint,” Marsset said. “We are finally able to map specific classes of meteorites that fall on Earth to distinct regions in the asteroid belt and to specific asteroid families.… That’s a major step toward understanding the compositional structure of the asteroid belt and, ultimately, how our solar system formed and evolved.”

But it’s just as important to understand where meteorites aren’t coming from, he pointed out.

“While one might expect the meteorite flux to represent a broad sampling of material from across the entire asteroid belt, we now know that it is actually dominated by a few recent fragmentation events,” Marsset said. “This insight helps us better understand the natural sampling bias in the meteorites we collect on Earth, and it also highlights which asteroid populations are underrepresented. That, in turn, can guide the targets of future space missions aimed at filling in those missing pieces.”

—Kimberly M. S. Cartier (@astrokimcartier.bsky.social), Staff Writer

Citation: Cartier, K. M. S. (2025), A geologic map of the asteroid belt, Eos, 106, https://doi.org/10.1029/2025EO250165. Published on 28 April 2025. Text © 2025. The authors. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.

Source: Global Biogeochemical Cycles

Glaciers provide a unique opportunity for researchers to measure levels of atmospheric carbon deposition. Unlike other terrestrial ecosystems, these slow-moving rivers of ice do not have other large reservoirs of soil or vegetation that might obscure how much carbon they receive from the atmosphere.

In most terrestrial ecosystems, dissolved organic matter comes from plants and soil and can contain both organic carbon and black carbon (the sooty black product from wildfires and burning fossil fuels). In glaciers, organic matter is predominately derived from in situ microbial production and atmospheric deposition. Both can contribute to downstream food webs and broader biogeochemical cycling.

Understanding how glaciers get their carbon, including how much comes from atmospheric deposition, can help scientists understand how human activity affects the glacier carbon cycle and ecosystems.

Holt et al. investigated dissolved organic matter in the meltwater from 10 glaciers across Alaska, Switzerland, Kyrgyzstan, and Ecuador. By examining dissolved organic carbon and black carbon isotopes, as well as molecular-level composition, researchers found that anthropogenic pollutants significantly influenced the composition of dissolved organic matter in glaciers and that this influence varied by region.

The researchers collected samples from each glacier outflow stream and determined the age of the dissolved organic carbon in the samples. These ages offered an isotopic signature of their sources. For instance, younger samples might originate from wildfire material and microbial activity on the glacier surface, whereas older material more likely originated from ancient carbon sources, namely, fossil fuels.

Each region displayed different amounts of dissolved organic carbon linked to anthropogenic atmospheric pollution, ranging from 12% to 91%, with a median of 50%. Carbon from fossil fuels was more prevalent in the dissolved organic matter of the Alaskan glacier. In Ecuador, there was a higher relative contribution of carbon from biomass burning, such as wildfires, and in situ microbial activity. The exact source, age, and makeup of dissolved organic carbon and dissolved black carbon varied between different glaciers outflows. But overall, the researchers say, fossil fuels are affecting the carbon content in glacier outflow globally, with implications for the ecosystems that depend on them. (Global Biogeochemical Cycles, https://doi.org/10.1029/2024GB008359, 2025)

—Rebecca Owen (@beccapox.bsky.social), Science Writer

Citation: Owen, R. (2025), Glaciers offer clues into the path of fossil fuel pollution, Eos, 106, https://doi.org/10.1029/2025EO250161. Published on 28 April 2025. Text © 2025. AGU. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.

Editors’ Highlights are summaries of recent papers by AGU’s journal editors. Source: Earth’s Future 

Coral atoll islands are particularly vulnerable to sea level rise and climate change. In rural islands where active coastal sediments processes are occurring, wave-driven sediment deposition can raise islands’ crest on their oceanward side.

Roelvink et al. [2025] show that coastal morphodynamic models are now able to provide quantitative insight into these phenomena. Specifically, they show that in the natural islands of Fiyoaree (Maldives), the sediment accumulation on the island crests can mitigate the projected increase of overwash during extreme wave events by a factor of three. Their modeling framework also confirms the benefits of adaptation measures aiming at protecting corals, particularly in reducing incoming wave energy. As climate is warming due to anthropogenic greenhouse gas emissions, increasing sea surface temperatures are causing widespread bleaching and mortality of corals, raising the urgent question of limits to coral adaptation, even at 2 degrees Celsius of global warming.

Hence, the study opens the way for future research exploring these limits in a quantitative manner, while also reminding us about the urgency of mitigating climate change to avoid irreversible losses and damages.

Citation: Roelvink, F. E., Masselink, G., Stokes, C., & McCall, R. T. (2025). Climate adaptation for a natural atoll island in the Maldives – predicting the long-term morphological response of coral islands to sea level rise and the effect of hazard mitigation strategies. Earth’s Future, 13, e2024EF005576. https://doi.org/10.1029/2024EF005576

—Gonéri Le Cozannet, Associate Editor, Earth’s Future

Text © 2025. The authors. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.