Feed aggregator

Rainwater connects the biogeochemical cycles of land and the atmosphere. It's effectively a chemical conveyor belt that delivers compounds to the ground during a process called wet deposition.

Arctic sea ice moves and spreads in ways that wind alone cannot explain. A new UC Riverside-led study finds that collisions between individual pieces of ice provide a surprisingly simple explanation for these movements, potentially offering scientists a better way to predict where sea ice will travel as the Arctic warms.

Source: Journal of Geophysical Research: Biogeosciences

Rainwater connects the biogeochemical cycles of the land and the atmosphere. It’s effectively a chemical conveyor belt that delivers compounds to the ground during a process called wet deposition.

For decades, research on airborne chemicals and wet deposition focused on inorganic compounds in rainwater, particularly nitrate and sulfate, which are key drivers of acid rain. Lacking the means to detect volatile organic compounds (VOCs) in rainwater, scientists long assumed that rainwater-based cycling of these compounds, which can be found in some petroleum fuels, paint thinners, and pharmaceuticals, was negligible.

But recent improvements in analytical techniques revealed that VOCs are present in rainwater in higher concentrations than scientists thought. In a new review, Yañez-Serrano et al. estimate that of the roughly 460 teragrams of carbon delivered via wet deposition every year, about 100 teragrams come from wet deposition of VOCs. And climate change and human activities are exacerbating VOC emissions.

The authors argue that dissolved atmospheric volatile organic compounds have been “largely invisible” to the biogeosciences and are a missing link in carbon cycle modeling. VOCs are highly reactive and biologically labile, so even small concentrations can affect water chemistry, plants, and microorganisms at local to regional scales. Particularly in carbon-limited ecosystems, even a small amount of VOC deposition can trigger a rapid biological response.

The authors present key knowledge gaps and suggest next steps for the research community. VOCs pose unique analytical challenges because they are highly reactive and chemically diverse. But both detection and modeling will need to further improve to better assess the role of VOCs in the biogeochemical cycle and in fields like air pollution monitoring, the authors say.

As scientists collect more VOC monitoring data, modeling will need to incorporate it. Most atmospheric carbon budgets currently omit or underestimate water-soluble VOCs and their subsequent oxidized products, leaving this highly reactive form of carbon out of the picture. Modelers will need to explore VOC mechanics and transport and incorporate them into Earth system models.

Finally, observations of and experiments on the effects of VOC wet deposition on terrestrial ecosystems and biogeochemical cycles will be needed to understand the potential damage or disruptions VOCs could be causing and how those may change as climate change continues.

As climate change and human activities exacerbate VOC emissions and change rainwater patterns, understanding VOCs as an overlooked part of Earth’s biogeochemistry has, perhaps, never been more crucial. (Journal of Geophysical Research: Biogeosciences, https://doi.org/10.1029/2026JG009767, 2026)

—Rebecca Dzombak, Science Writer

Citation: Dzombak, R. (2026), Volatile organic compounds have been underestimated in the carbon cycle, Eos, 107, https://doi.org/10.1029/2026EO260296. Published on 16 September 2026. Text © 2026. 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.

In late April 2026, a weakened polar vortex drove the outbreak of a cold Arctic air mass that threatened European crops with frost just as the growing season was starting. Temperatures in some places dropped 10°C–15°C relative to the seasonal average, posing particular hazards for vineyards and orchards.

Severe cold air outbreaks (CAOs) like this aren’t uncommon: Another event, in February 2021, sent an Arctic blast that dropped temperatures precipitously through the central United States, causing prolonged, widespread, and costly power outages in Texas especially.

Outbreaks of cold, dry air from the Arctic, as well as influxes of warm and moist air masses to the region, influence and are being influenced by rapid changes occurring in the Arctic climate system.

Increasingly, outbreaks of cold, dry air from the Arctic, as well as influxes of warm and moist air masses to the region, influence and are being influenced by rapid changes occurring in the Arctic climate system, including enhanced warming and dramatic sea ice loss. These air mass transport events, which link the Arctic atmosphere with midlatitudes, are also often accompanied by extreme weather.

Air masses transform significantly as they travel over open ocean and sea ice. For example, their temperature and moisture contents, as well as cloud and precipitation patterns, evolve. These transforming characteristics determine both local and large-scale effects of weather events along the air mass transport paths. Seasonal variability, especially related to the influx of solar radiation (absent during polar night), can have dramatic influences on these air mass transformations. However, the processes driving these influences are not well understood.

Specifically, representations of the development of clouds and the atmospheric boundary layer (the layer closest to Earth’s surface) during air mass transport events to and from the Arctic, particularly during the polar night of winter, are a significant weakness of numerical weather and climate models.

Overcoming these deficiencies, which would help us understand and forecast the potential impacts of these events, can be achieved through coordinated observations intended to contrast seasonally varying influences. Such measurements are exceedingly rare, especially in winter, but forthcoming efforts can capitalize on newly developed techniques for observing air masses firsthand and for modeling their movement and transformation.

An Arctic-Midlatitude Link

The Arctic is a crucial regulator of the global climate system and is linked to processes that occur far beyond its boundaries. Atmospheric interactions between the Arctic and midlatitudes take place primarily through meridional (north to south or south to north) air mass transports, including CAOs and warm air intrusions (WAIs; Figure 1).

Fig. 1. Typical patterns of (a) cold air outbreaks (CAO) and (b) warm air intrusions (WAI) over the Fram Strait between Greenland (top left corner) and Svalbard, Norway (island in the top right), are shown. The color scale in (a) indicates the marine CAO (MCAO) index, in which higher values denote larger temperature differences between the surface and the lower atmospheric layer typical for stronger CAOs; in (b), the color scale indicates vertically integrated water vapor (IWV) contents in the atmosphere. Higher IWVs are typical of strong WAIs. Arrows indicate wind directions and velocities. The data come from the ERA5 (the fifth-generation global climate and weather reanalysis dataset produced by the European Centre for Medium-Range Weather Forecasts) reanalysis by Hersbach et al. [2020].

These dynamic exchanges influence near-surface air temperatures, sea ice extent and thickness, cloud formation, and the balance between incoming solar and outgoing terrestrial radiative energy. They also control the movement of energy (heat), moisture, and pollutants into and out of the Arctic, stimulating feedbacks that operate on local, regional, and global scales [Pithan et al., 2018]. For example, WAIs inject humidity (i.e., water vapor, a major greenhouse gas) into the Arctic, significantly increasing near-surface air temperatures, especially in winter. This humidity also promotes cloud development that further heats the Arctic surface and thus can limit sea ice growth.

Collectively, these exchanges and feedbacks determine how the Arctic shapes large-scale atmospheric circulation and weather patterns—both within and outside the Arctic—that often contribute to extreme, societally impactful events like severe winter storms, cold snaps, heat waves, and significant melting events [Zhang et al., 2025]. Given the wide-ranging implications of meridional air mass transports, observing, understanding, and modeling their fundamental drivers and behaviors, many of which vary seasonally and spatially, are essential steps for scientists.

Seasonal Contrasts Shape Dynamic Processes

The most striking seasonal difference in the Arctic arises from the extreme contrast between polar night and polar day.

The most striking seasonal difference in the Arctic arises from the extreme contrast between polar night and polar day that is driven by the profound effects of solar radiation “switching off and on.” This annual back-and-forth fundamentally alters the region’s radiative energy budget: During the polar night of winter, the surface predominantly loses energy, whereas in summer it gains energy. The radiative energy budget, in turn, modulates energy exchanges at the surface and the vertical distribution of heating and cooling throughout the atmosphere.

Radiative changes strongly influence the properties and structure of the atmospheric boundary layer through effects on vertical stratification and mixing processes. For instance, temperature inversions, which inhibit vertical mixing, tend to be stronger and more frequent in winter, whereas summer conditions can promote more vertical mixing. The radiative budget also affects turbulent and conductive heat fluxes at the surface, near-surface temperature variations, sea ice melting, and other surface processes and properties.

Seasonality affects north–south temperature gradients, further shaping atmospheric transport dynamics and influencing the frequency, properties, and intensity of key events like CAOs and WAIs. Cloud characteristics, including their coverage, thermodynamic phase (liquid versus ice), and precipitation type and frequency, also differ markedly between seasons. For example, the transition into the summer melt season supports a shift from mostly icy to more liquid-containing clouds [e.g., Lac et al., 2026], a change with significant influences on atmospheric boundary layer structure, radiative balance, and more.

Little is known about how seasonal variability affects air mass transformations.

Atmospheric composition also evolves over the year. Alongside variations in mixing and transport pathways, aerosol sources and properties shift because of changes in sunlight-driven photochemical activity, precipitation-driven removal of aerosol particles, surface conditions (e.g., melt pond formation), and biological production of particles and aerosol precursor gases [Schmale et al., 2021]. These aerosol transitions, in turn, affect cloud properties, precipitation efficiency, and possibly cloud lifetimes.

At the surface too, conditions are far from static. The frozen winter surface environment transitions into one with increasing meltwater, open ocean, and moisture availability, before eventually transitioning back after summer passes. These changes influence surface albedo, snow depth, surface roughness, and ice thickness, all of which modulate the exchange of heat, moisture, momentum, and matter between the atmosphere and surface.

Indeed, many drivers of, and constraints on, air mass transports vary seasonally. However, little is known about how seasonal variability affects air mass transformations, including rates of spatiotemporal changes in temperature and moisture that affect cloud formation and longevity, precipitation patterns, radiative properties, and surface interactions. In short, these transformation rates ultimately control the spatial extent of air masses and their impacts on people, yet they remain poorly understood.

Establishing Foundations for Accurate Models

The reliability and credibility of models for weather forecasting and for understanding long-term climate trends and feedbacks depend on strong foundations of observational ground truthing. These observations help models accurately represent fundamental physical processes such as Arctic amplification, meridional transport, and Arctic climate dynamics. However, building these foundations for the Arctic remains a major challenge.

A central issue is the particular scarcity of ground truth data collected over sea ice and during winter [Jung et al., 2016]. The lack of data makes it difficult to evaluate and improve models because key processes are insufficiently constrained.

Sea ice near Svalbard is seen through the clouds in spring 2026. Credit: Marcus Klingebiel

Characterizing the Arctic accurately across all seasons is essential because the effects of seasonal changes in the radiative energy budget extend beyond any single season. For example, although sea ice melting is limited during winter, the amount of ice growth during the dark season strongly influences how the ice evolves later in the year. The importance of interseasonal linkages like this one suggests that weaknesses in model representations of any single season can propagate through simulations over time, significantly affecting the longer-term trajectory of the Arctic that they portray.

With the limitations of available satellite data, in situ and remote sensing observations from ground-based and aerial platforms are indispensable.

Many current modeling tools, including global and regional climate models, reanalyses, and numerical weather prediction systems, exhibit substantial biases, particularly in representing clouds, humidity, and lower atmospheric structure. For example, models do not represent the height of the atmospheric boundary layer over sea ice well, and they often overemphasize cloud glaciation. These deficiencies are most pronounced in winter and in processes tied to net surface energy exchange, such as cloud phase partitioning, which are also among the processes least constrained by satellite observations.

With the limitations of available satellite data, in situ and remote sensing observations from ground-based and aerial platforms are indispensable. They provide high-resolution measurements that not only help calibrate and validate satellite data and support reanalyses but also enable a deeper understanding of the underlying mechanisms at work. Such observations are also crucial for improving model parameterizations and, ultimately, for building more reliable modeling tools.

Coordinated Campaigns Can Close Critical Gaps

Closing critical observational and modeling gaps and advancing predictive representations of Arctic weather and climate dynamics, including meridional air mass transport events, require coordinated research efforts. Field campaigns involving a variety of data-collecting platforms are needed to quantify seasonally and spatially varying processes and feed model development. Targeting the drastically underobserved winter season, for example, would help to isolate effects of solar radiation and other varying conditions.

Moreover, work is needed to implement and improve upon quasi-Lagrangian observational strategies [Wendisch et al., 2025]. These techniques focus on following a moving air mass and sampling it repeatedly as it travels and evolves, combining aircraft, ship-based, and ground-based measurements together with modeling to help constrain air mass transformation processes and rates [Karalis et al., 2025]. Initial quasi-Lagrangian aircraft-based measurements showed promise for estimating air mass temperature and moisture change rates, though they did not always agree well enough with model simulation results [Wendisch et al., 2025]. Further, cloud and precipitation change rates have not yet been derived from these observations.

A central objective of future campaigns should be directly observing and quantifying seasonal differences in the heating or cooling and the drying or moistening of air masses to guide model improvements. Upcoming initiatives offer valuable opportunities to expand the observational footprint and address this objective.

This fall, the Tara Polar station is scheduled to embark on an 18-month expedition—the first of 10 intended missions—to track atmospheric (and oceanic) conditions while drifting with Arctic sea ice. And in coming years, the Contrasting Polar Night & Day (CONIDA) airborne campaigns will follow CAOs and WAIs in the European Arctic, first during winter (CONIDA-Night, November–December 2028) and then during summer (CONIDA-Day, June–July 2029), to study transformations of clouds, precipitation, and radiative processes.

The multifaceted strategies of these efforts—more of which are needed—should provide the comprehensive, multiscale observational constraints we need to robustly characterize and model the processes governing and responding to Arctic seasonal variability, particularly meridional air mass transformations and their broader climate connections. These improvements will, in turn, allow us to sharpen forecasts of the effects of warm air arriving in the Arctic and of the cold air blasts that occasionally send midlatitude temperatures plummeting.

Acknowledgments

We thank Felix Pithan for his valuable discussion regarding this article and Marcus Klingebiel for providing Figure 1. This work was funded by Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) project ID 268020496–TRR 172 and project ID 316646266 SPP 1294.

References

Hersbach, H., et al. (2020), The ERA5 global reanalysis, Q. J. R. Meteorol. Soc., 146, 1,999–2,049, https://doi.org/10.1002/qj.3803.

Jung, T., et al. (2016), Advancing polar prediction capabilities on daily to seasonal time scales, Bull. Am. Meteorol. Soc., 97(9), 1,631–1,647, https://doi.org/10.1175/BAMS-D-14-00246.1.

Karalis, M., et al. (2025), Lagrangian single-column modeling of Arctic air mass transformation during HALO–(AC)³, Atmos. Chem. Phys., 25, 13,177–13,198, https://doi.org/10.5194/acp-25-13177-2025.

Lac, J., et al. (2026), Understanding the spring cloud onset over the Arctic sea ice, Atmos. Chem. Phys., 26, 4,189–4,213, https://doi.org/10.5194/acp-26-4189-2026.

Pithan, F., et al. (2018), Role of air-mass transformations in exchange between the Arctic and mid-latitudes, Nat. Geosci., 11, 8050812, https://doi.org/10.1038/s41561-018-0234-1.

Schmale, J., P. Zieger, and A. M. L. Ekman (2021), Aerosols in current and future Arctic climate, Nat. Clim. Change, 11, 95–105, https://doi.org/10.1038/s41558-020-00969-5.

Wendisch, M., et al. (2025), Observed and modeled Arctic airmass transformations during warm air intrusions and cold air outbreaks, Atmos. Chem. Phys., 25, 15,047–15,076, https://doi.org/10.5194/acp-25-15047-2025.

Zhang, X., et al. (2025), Weather and climate extremes in a changing Arctic, Nat. Rev. Earth Environ., 6, 691–711, https://doi.org/10.1038/s43017-025-00724-4.

Author Information

Manfred Wendisch (m.wendisch@uni-leipzig.de), Leipziger Institut für Meteorologie, Universität Leipzig, Germany; Matthew D. Shupe, Cooperative Institute for Research in Environmental Sciences and the National Snow and Ice Data Center, University of Colorado Boulder; also at Physical Sciences Laboratory, NOAA, Boulder, Colo.; Susanne Crewell, Institut für Geophysik und Meteorologie, Universität zu Köln, Cologne, Germany; Felix Ament, Meteorologie, Universität Hamburg, Germany; and Gunilla Svensson, Department of Meteorology and Bolin Centre for Climate Research, Stockholm University, Sweden

Citation: Wendisch, M., M. D. Shupe, S. Crewell, F. Ament, and G. Svensson (2026), The effects of polar night and day on Arctic air masses, Eos, 107, https://doi.org/10.1029/2026EO260293. Published on 16 September 2026. Text © 2026. 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.

At just an inch (2.5 centimeters) long, the sand hoppers skipping across California's Isla Vista Beach at dawn don't exactly look like a force of nature. But a new study reveals that these shrimp-like creatures might be doing more to shape the landscape in which they live than any other animal on Earth.

Across its basins and thousands of meters down, the Mediterranean Sea is getting warmer and saltier. Now, a new study finds it is also doing so at an increasingly rapid pace. The speedup is largest in the central and eastern Mediterranean and strongest of all in the Adriatic, where it reaches all the way to the sea floor.

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

Changes in how often earthquakes occur can reveal the evolution of stress on faults, providing a window into processes that are otherwise difficult to observe. Jiang et al. [2026] introduce T-rate, a Bayesian method that evaluates many possible stress histories to reconstruct stress changes from earthquake records and quantify the associated uncertainty. The method represents each episode of stress loading as a flexible rise, steady phase, and decline. Tests with simulated earthquake sequences show that T-rate can recover complex stress histories under a range of conditions.

Applied to the 2008 Reno–Mogul swarm, the preferred model identifies four loading phases, including a sharp increase beginning about three days before the magnitude 5.1 earthquake. The timing of this late increase remains relatively stable across the tested data-processing choices. Combined with independent GPS measurements of ground deformation and observations of rapid earthquake migration, the result points to slow fault motion without detectable earthquakes as a plausible source of the final loading phase. By turning widely available earthquake records into uncertainty-aware estimates of stress evolution, the openly available T-rate tool complements ground-deformation measurements, especially where they are sparse, and supports studies of earthquake swarms and other transient fault processes.

Citation: Jiang, Y., Trugman, D. T., & González, P. J. (2026). Bayesian inference of complex stress evolution in rate-and-state governed faults constrained by seismicity rate observations. Journal of Geophysical Research: Solid Earth, 131, e2026JB033922. https://doi.org/10.1029/2026JB033922

—Bogdan Enescu, Associate Editor, JGR: Solid Earth

Text © 2026. 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.

"Sediment and fossil records are a major pillar of our understanding of how the climate works—but climate scientists have gotten something wrong about Australia's ice age," says Dr. Alex F. Wall, lead author of a new research paper published in Quaternary Science Reviews.

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

Researchers have calculated that Greenland and Antarctica have lost 11 trillion metric tons of ice since the 1970s, which has led to a global rise in sea level of more than 3 centimeters.

“Just over three centimetres of sea-level rise may sound small, but that puts another six to nine million people at risk of coastal flooding and erosion,” said Andrew Shepherd, head of the School of Geography and Natural Sciences at Northumbria University and founder of the Ice Sheet Mass Balance Intercomparison Exercise (IMBIE), in a press release. 

 Related

•  Read the paper: “Mass balance of the Greenland and Antarctic ice sheets from the 1970s to 2023”
•  Ice Sheet Mass Balance Intercomparison Exercise
•  Antarctic Ice Sheet Has Lost a Connecticut-Sized Amount of Ice Over the Past 30 Years
 

IMBIE was founded to bring together data from various satellite missions in order to build a clearer picture of how the Greenland and Antarctic ice sheets have changed over the last few decades. Prior to today’s report, published today in Scientific Data, IMBIE’s most recent assessment calculated a rise of about 21 millimeters to global mean sea level between 1992 and 2020.

In the new assessment, researchers drew on data from 27 satellite missions spanning more than 50 years. The vast majority—about 84%—of the calculated ice loss came from glaciers and ice sheets discharging directly into the ocean, with only 16% of the loss caused by melting at the surface.

This “tells us the long-term trend is being driven by the dynamic response of the ice sheets to a warming ocean,” said Inès Otosaka lead researcher and professor of geography and natural sciences Northumbria University, in a press release.

Though global ice loss slowed between 2020 and 2023, overall ice loss has risen steeply since the 1980s: In Greenland, annual ice loss went from 60 billion metric tons in the 1980s to 264 billion metric tons in the 2010s. In Antarctica, annual loss went from about 48 billion metric tons in the 1980s to 202 billion metric tons in the 2010s.

Through most of the twentieth century, sea level rose by about 1.4 millimeters per year. Between 2006 and 2015, the rate was more than twice as high, at 3.6 millimeters per year.

—Emily Gardner (@emfurd.bsky.social), Deputy Editor

These updates are made possible through information from the scientific community. Do you have a story about science or scientists? Send us a tip at eos@agu.org. Text © 2026. 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.

The longest satellite record ever assembled shows that Greenland and Antarctica have lost 11.3 trillion metric tons of ice since the 1970s, raising global sea level by more than 3 centimeters (1.2 inches)—and that most of the loss has come from glaciers flowing faster into the ocean, not from melting at the surface.

A new paper (Wang et al. 2026) suggests that the bedrock failure that initiated the Nepal landslide disaster may have displayed measurable precursory deformation.

Inevitably, one of the key discussion points in the aftermath of the 26 August 2026 Nepal landslide disaster is whether the failure could have been anticipated. Opinions differ, but a key starting point is the identification of precursory signals in the days, weeks or months leading up to the failure.

The focus is inevitably on INSAR, but the terrain in the source area of the landslide (on the north face of Langtang Lirung) is unforgiving to this technique. An alternative lies in pixel offset tracking from optical imagery, which can extract movement in more challenging terrain.

In the journal Landslides, a paper published with impressive rapidity (Wang et al. 2026 – the article is paywalled but this link should provide access) uses the pixel offset tracking technique on Sentinel-2 satellite data to examine deformation of the glaciers in the area of the initial failure. They have compared movement of the main glacier, which was not involved in the landslide, and of the hanging glacier under which the landslide occurred. In the Google Earth image below, this is the portion of the glacier under the location marker (the hanging glacier) and main glacial area upslope:-

Google Earth image from 2017 showing the site of the 26 August 2026 landslide in Nepal.

Wang et al. (2026) have measured movement of these two parts of the glacier from 2018 to now. Until the early part of 2026, the two glaciers moved at a linear rate, although the main glacier moved at a much higher rate than the hanging glacier. But, from late 2025 or early 2026, the hanging glacier started to move faster, and from that point onwards showed a dramatically accelerating trend to failure.

Presumably, this is the development of failure in the underlying bedrock being transmitted through to the ice. By the end, the movement rate had increased from about 7.4 m per year to about 55 m per year.

Wang et al. (2026) have a slightly strange interpretation of the mechanical processes that led to this behaviour that I must admit I don’t really understand. To me, this is likely to have been a progressive failure of the bedrock, which is known to show this style of deformation. I would be very interested in seeing a plot of 1/velocity against time for the period of acceleration, which will give an indication as to whether there was a brittle failure mechanism in play.

But nonetheless, this is a fascinating result that might suggest that the initiating bedrock failure that started the disastrous cascade in Nepal demonstrated precursory signals. Please note though that this does not mean that the event itself could have been reliably predicted nor that a warning system can be easily developed.

Reference

Wang, T., Sassa, S., Yang, W. et al. 2026. Precursor of a powerful large glacier landslide causing tsunami floods in the 26 August 2026 Nepal disaster. Landslides. https://doi.org/10.1007/s10346-026-02842-6.

Return to The Landslide Blog homepage Text © 2026. 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.

SumamryThe introduction of full waveform inversion has been the starting point of four decades of investigation in seismic imaging, which aim to reduce the ill-posedness of this nonlinear and quantitative imaging method. In this review, we propose a comprehensive overview of these investigations. Beyond simply cataloging proposed approaches, this work aims to classify them, identify similarities and differences between methods, establish connections with common concepts and practices in full waveform inversion, and emphasize the few core ideas underlying this long-term research effort. We introduce a general mathematical framework that encompasses all proposed methods, which we use throughout this study to support their presentation, analysis, and comparison. The concluding discussion should promote a better understanding of the current state of research on this topic. In particular, we examine which methods have been applied to field data and which have not, highlighting the main difficulties and the requirements such techniques must fulfill to transition to such applications.

SumamryElectrical resistivity models derived from magnetotelluric (MT) measurements across the Khangai Dome, an intracontinental plateau in central Mongolia, have identified several intriguing features, including a locally thinned lithosphere and extended fluid-rich domains in the lower crust. However, due to the limited spatial coverage of these studies, whether those features extend beyond central Mongolia could not be assessed. This is an important question with major scientific implications for understanding the geodynamic evolution of the region. The surrounding areas also host significant mineral deposits and geothermal resources, whose origin and interpretation depend on the characterization of deep structures to be fully understood. In this study, we report on MT data from 378 new locations to the west, east, and north of the Khangai Dome, acquired between 2020 and 2023. A new 3-D electrical resistivity model of the region was derived using a combination of the new and previously acquired datasets. This new model covers an area of approximately 900 × 1250 km2, substantially expanding the spatial coverage compared to previous models. The key methodological novelty of this study is the implementation of a recently developed, scalable, and open-source 3-D inverse solver based on the integral equation approach and non-local parametrization. The 3-D model fits the observed data remarkably well, and the recovered features show good agreement with those from previously published models obtained with different solvers and subsets of the data. The model reveals an upper crustal dichotomy, with very high resistivity in the northern region and low resistivity in the southern region. The boundary between these regions follows the Main Mongolian Lineament in the central and eastern regions and the Ikh-Mongol arc system in the western region. In the lower crust, the resistivity is generally lower, and several long, laterally extended, very low-resistivity anomalies are observed. These anomalies extend westward from central Mongolia but are bounded to the east by a high-resistivity anomaly whose location coincides with the Mogod fault zone. The aforementioned features highlight the control of major tectonic boundaries on the electrical resistivity distribution. An upper-mantle low-resistivity anomaly below the Khangai Dome, previously interpreted as a locally thinned lithosphere with an upwelling asthenosphere, is imaged not only beneath central Mongolia but also, to some extent, in the western region. The model also provides new information on the structure of major features across central-western Mongolia, such as the extensive Bulnay fault. Taken together, these new constraints on lithospheric properties and architecture advance our understanding of the mechanisms that shaped the region and its subsequent evolution.

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

Yellowstone is among the United States’ most popular national parks, with more than 4.5 million visitors every year. So it’s probably no surprise that park staff does some picking up after visitors. What may be surprising is that the staff has already recovered a record-setting 20,000+ pieces of trash, 5,600 rocks and sticks, and nearly 300 hats from the park’s hydrothermal areas so far this year, according to a USGS press release.

It’s no small feat to dislodge all of this trash from the park’s iconic hydrothermal features, and members of the park’s Geology Program are behind the effort. Much of it is done with shovels or short-handled grabber tools, but some efforts are more complicated, requiring fishing rods, strainers, slotted spoons, and grabber poles up to 30 feet long.

 Related

•  Caps, trash, and other funky hydrothermal finds: 2026 edition
•  Hot Spring Cleaning!
•  Hydrothermal hats and visitor safety: Walking in the boots of the Yellowstone National Park Geology Team
 

Most of this litter was left unintentionally; regardless, the USGS’s accounting of its cleanup serves as a good reminder for visitors to never throw anything into a hot spring or geyser, to hold on (or remove) their hats, and to stash their trash safely until they can get to a trash can.

Below is a list of some of the most notable finds of 2026.

The 280 hats recovered from Yellowstone’s hydrothermal areas this summer included several hats that appear to be from the park’s gift shops. Credit: Credit: National Park Service photo by Margery Price, September 2026 The Yellowstone Geology team collected more than $18.00 in coins from the park’s hydrothermal areas during summer 2026. Credit: National Park Service photo by Margery Price, September 2026

1. 36 pairs of sunglasses. All the better to see those geysers with, we suppose.
2. More than 280 hats. Hats galore! Some had place names, like San Diego, Bears Ears National Monument, Miami, or Yellowstone itself. Sun hats and tie dye hats, patriotic hats and pawprint hats. A baseball cap with the words “COWBOY HAT” written on it. These accessories were likely blown off by wind, and park visitors did the right thing by leaving them behind and not venturing into the geothermal areas to rescue them.
3. 54 camera lens caps. At least it wasn’t the cameras themselves? There did appear to be one children’s camera among the objects, though.
4. A single jingle bell.
5. Six Invisalign retainers! Oof, this one might have been more painful to lose than the 5 AirPods also found in the hydrothermal areas.
6. $18 in coins, including Canadian dollars, euros, pesos, and yen. There were a few dollar bills as well. Most of these appear to have been thrown into hot springs or geysers on purpose, which is not allowed and will not make your wishes come true. If you want to throw your coins somewhere, throw them in a real wishing well, or to us here at Eos.
7. Lots of Yellowstone paraphernalia. In addition to the hats, staff found six Yellowstone pressed pennies, 7 Polaroid photos (several of which were of the hydrothermal feature in which the photos were found), six Junior Ranger badges, and at least one Yellowstone pin. There’s something tragically poetic about this.
8. A single Uno card (yellow 4).
9. A tube of gouache paint (also yellow).
10. Three fortune cookie fortunes: “You are the center of every group attention,” “Trust in the journey, for it is leading you to blessings,” and one that isn’t quite readable.
11. 3 golf tees (there is not a golf course within the park).

—Emily Gardner (@emfurd.bsky.social), Deputy Editor

These updates are made possible through information from the scientific community. Do you have a story about science or scientists? Send us a tip at eos@agu.org. Text © 2026. 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.

The same amount of rain falling on the same landscape at different times can send very different amounts of water into its rivers, according to a new study in Nature Water. The research offers a global picture of how catchments behave—not where rivers are or how much water they carry, but how consistently landscapes turn rainfall into river flow.

From a mountainside overlooking Utah's Great Salt Lake, a time traveler turning back the clock 20,000 years would watch a dramatic transformation of the landscape. The shallow, salty lake below would swell into an immense body of fresh water, at one point reaching nearly 1,000 feet (305 meters) deep and covering more than 10 times the lake's modern area. Mountain ranges would become islands, and the new shoreline would move far beyond its current boundary.

Irrigating U.S. crops prevents higher greenhouse gas emissions by avoiding land conversion, new research by Colorado State University shows. Agricultural irrigation allows the United States to grow more food on less land. If irrigation were suddenly shut off, the land-use change required to make up for lost yields would produce significantly more greenhouse gas emissions than irrigation emits—363 times more.

The most productive scientific programs are not those in which researchers know the answers to their questions in advance. Instead, they are those organized around questions that are genuinely open, consequential, and resolvable by investigation.

The discourse surrounding underground hydrogen storage (UHS) and natural geological hydrogen—increasingly discussed in recent years as a potentially important part of a lower-carbon energy system—offers such questions: How much of the hydrogen injected underground can be recovered later? Will Earth’s own hydrogen ever accumulate in commercially useful quantities?

The barriers should be the focus of research. They define what needs to be known, why it is not yet known, and what the consequences of knowing it would be.

Potential applications of UHS and natural hydrogen are rooted in real physical possibility. Yet deep uncertainties also exist that no honest assessment can resolve with currently available data. Addressing these uncertainties requires investigation through a rigorous scientific program justified on its own terms, regardless of preexisting ideas of whether commercial applications will eventually succeed.

The barriers facing applications of UHS and natural hydrogen are not evidence that researching them is futile, as some observers have suggested (e.g., in recent work questioning self-replenishing production and commentary urging scientific rigor over hype).

In fact, the opposite is true. The barriers should be the focus of research. They define what needs to be known, why it is not yet known, and what the consequences of knowing it would be.

Fixed Constraints for Subsurface Hydrogen

Though they may seem unrelated, UHS and natural hydrogen exploration share identical scientific foundations in the subsurface behavior of molecular hydrogen (H2). H2 is the smallest and lightest molecule in nature. It leaks through rock and cement far faster than methane (natural gas) and rises buoyantly along fractures and faults. Furthermore, many underground microbial communities consume H2 wherever water and suitable chemical partners are present.

Whether one is considering injecting and storing hydrogen underground or whether Earth has already done that on its own, the underlying science is the same. Research efforts looking into those questions are not parallel; they are studying the same geological process from opposite ends.

Hydrogen contains roughly 3.5 times less energy per unit volume than methane at any given pressure and temperature, a consequence of their respective molecular weights and bond chemistries. This physical fact sets an inescapable constraint for applications of either UHS or natural hydrogen exploration—no engineering advance can close that gap.

The Bulqizë mine in Albania hosts one of the few documented cases of intense natural hydrogen degassing from an underground mine, illustrating the tantalizing promise of natural hydrogen as a resource. Credit: Frédéric-Victor Donzé

It might seem reasonable to expect that this disadvantage would narrow at the high pressures found in deep reservoirs, where gases compress and pack more efficiently. Under reservoir conditions, however, methane actually compresses more efficiently than hydrogen, meaning the energy gap is, if anything, slightly greater at depth.

To be economically viable, hydrogen-generating or storage systems must compensate through high generation volumes, efficient trapping, and minimal microbial losses.

This gap is not a verdict against hydrogen as an alternative to methane. But it is a calibration that every economic model and every research design must incorporate. To be economically viable, hydrogen-generating or storage systems must compensate through high generation volumes, efficient trapping, and minimal microbial losses. The understanding of all these factors, apart from the volumetric comparison to methane, is far from complete—and is where vital science lives.

Underground Storage: Open Questions, Not Dead Ends

Whether geological formations can serve as reliable, large-scale hydrogen traps is a genuine open question. Some challenges involve engineering problems likely to be solved: For example, hydrogen embrittlement of steel, which is used throughout hydrogen infrastructure, from injection wells to pipelines and storage tanks, is a real, but tractable, problem.

The deeper unknowns are geoscientific and fall into three classes.

The first is geological specificity: The formations best suited to hydrogen storage—salt deposits and depleted gas fields with intact caprocks, for example—are not necessarily distributed where they are most needed operationally (Figure 1). Whereas some facilities could have a direct demand for hydrogen, others that generate surplus energy from renewable sources could use this energy to hydrolyze water, and the resulting hydrogen could be stored for use when energy demand exceeds supply. Quantifying the mismatch between storage locations and operational needs has barely begun.

Possible storage options for underground hydrogen include salt deposits, saline aquifers, and depleted hydrocarbon reservoirs with intact caprocks. Credit: Miocic et al., 2023, https://doi.org/10.1144/SP528-2022-88, CC BY 4.0

Second is the biological sink: Subsurface microbial communities can consume hydrogen at significant rates over operational timescales. Pilot experiments injecting hydrogen underground have documented substantial losses due to microbial activity—field pilots report losses from a few percent up to more than 50%. But how much consumption occurs varies across geological settings, and whether it can be predicted or suppressed remains unknown. Current estimates of full-system, round-trip efficiency (the percentage of energy retrieved from a storage system compared to the total energy initially used to charge it) span from 25% to 40%, reflecting genuine uncertainty.

The third unknown is caprock integrity: Hydrogen’s much higher diffusivity means that seals adequate for methane—shales, evaporites, and tight carbonates, for example—may not work for H2. However, experimentally based knowledge about hydrogen diffusion through these rocks under reservoir conditions is thin.

These unknowns aren’t dead ends; they outline valuable research directions.

Natural Hydrogen: Generation Is Certain, Accumulation Is Not

Earth’s crust generates hydrogen continuously. Serpentinization of iron-rich rocks, decomposition of water by natural radioactivity (radiolysis), and fluid circulation along active faults all contribute to H2 production within the lithosphere.

How much of this hydrogen reaches the atmosphere remains poorly constrained, though. Documented point sources account for far less than 1 megaton per year globally. Diffuse degassing across the ~2.5 million square kilometers of serpentinized terrains worldwide, potentially the dominant pathway, is essentially unmeasured, with estimates ranging from roughly 0.02 to a few megatons per year.

Whether any of this geologic hydrogen accumulates underground in commercially useful concentrations is also unclear. The one producing hydrogen well, located in Mali, yields the energy equivalent of roughly 30 kilowatts of continuous power: scientifically significant, but not a template for commercial production.

The geological conditions required for large, long-lived hydrogen accumulations are demanding.

Indeed, the geological conditions required for large, long-lived accumulations are demanding.

In many environments, dissolved hydrogen in groundwater never reaches the saturation needed to exsolve into a separate gas phase at all. Where it does, microbial communities that consume hydrogen may be active. Furthermore, gas accumulation requires a porous reservoir sealed by a caprock that can withstand hydrogen’s powerful tendency to escape. No such accumulation has been confirmed anywhere in the world; the Mali borehole, for example, taps a shallow (~120 meter) aquifer that continuously degasses hydrogen, not a sealed pressurized reservoir.

Also, unlike thermogenic methane, abiotic hydrogen generation can be continuous, making the sources of surface seeps fundamentally ambiguous. The same signal measured at the surface could indicate a large column of trapped hydrogen or a flow-through system with no accumulation at all. Distinguishing between these scenarios requires subsurface data—gathered from wells, seismic surveys, and pressure and geochemical measurements—that do not exist yet for most prospective terrains.

This catalog of uncertainties could be read as an argument against research. If commercial-scale UHS appears unlikely within a realistic timeline and if commercial production of natural hydrogen faces structural barriers, why treat the underlying science as indispensable?

The question is fair. We suggest, however, that scientific research into UHS and natural hydrogen production is justified on grounds that do not depend on confident commercial outlooks.

The Need for Research Now

The knowledge gaps described above cannot be closed within the compressed timelines that commercial deployment demands or by market forecasting.

The Cedars ophiolite north of San Francisco is a site of active, low-temperature serpentinization where hyperalkaline springs have been studied for decades as a natural analogue for subsurface hydrogen generation. Credit: Frédéric-Victor Donzé

They require systematic, hypothesis-driven scientific investigation. This research will require subsurface data, pilot injections, reactive transport experiments, and mechanistic characterization of microbial consumption and caprock/unconventional reservoir behavior. Ultimately, it will reveal the specific geological, biological, and thermodynamic conditions under which UHS and exploration for natural hydrogen could succeed.

A useful precedent comes from the development of enhanced geothermal systems (EGS): For more than 4 decades, systematic characterization of fracture networks, rock mechanics, and fluid-rock interactions in hot, dry rock proceeded despite deep uncertainty about commercial viability. That patient, systematic science—not optimistic projections—is what eventually let a handful of projects demonstrate technical feasibility only in the past few years.

The lesson is not that EGS were destined to succeed, but that sustained research was what allowed the question to be answered at all.

A research program that maps the subsurface, showing where storage is feasible and where natural accumulations can persist, will provide actionable knowledge about specific formations, specific depths, and specific trapping configurations. The economic value of such knowledge is asymmetric: Credible answers about viability in specific places and conditions could prevent costly commitments in the wrong places and, conversely, could direct exploration to potentially productive formations.

Billions of euros, dollars, yuan, and dinars invested in infrastructure will be guided, or misguided, by the quality of the science available when decisions are made. Indeed, major funding programs in Europe, North America, and China are already channeling substantial resources toward developing UHS and natural hydrogen projects. Those commitments will not wait for full scientific resolution of uncertainties.

Uncertain science, pursued systematically and transparently, is a far better foundation for public decisionmaking than perhaps overly optimistic commercial narratives that mask the unknowns.

In that context, the responsibility of the research community is not to endorse or reject individual programs, but to ensure that the knowledge base on which they rest is as rigorous as possible. Uncertain science, pursued systematically and transparently, is a far better foundation for public decisionmaking than perhaps overly optimistic commercial narratives that mask the unknowns.

Beyond these pragmatic considerations, understanding the subsurface behavior of molecular hydrogen—among the least characterized topics in applied geochemistry—has intrinsic scientific value. Hydrogen’s generation by serpentinization and radiolysis, its migration along faults, its interaction with minerals and fluids, its microbial consumption, and its occasional trapping are fundamental concerns of planetary chemistry. These concerns are directly relevant to illuminating the formation and early differentiation of Earth and other rocky planets, how life on Earth originated, and the functioning of today’s deep biosphere.

The tools required for this science, such as trace gas sensors, reactive transport models, and equations of state for high-pressure gas mixtures, already serve applications in geothermal monitoring, carbon storage, and astrobiology. The current mobilization of funding around hydrogen creates an unprecedented window of opportunity to accelerate investigations into these questions. But that window will not stay open indefinitely.

The Choice Ahead for Hydrogen Scientists

The deepest argument for pursuing research into UHS and natural hydrogen accumulation within a unified framework is not administrative, but scientific. Both are governed by the same set of physical constraints on generation, trapping, sealing, and microbial loss.

The uncertainties of each track are precisely what make dialogue between them productive. A subsurface environment inhospitable to developing storage today is also one where natural accumulations are unlikely to have survived. Conversely, identifying where natural hydrogen has persisted provides an empirical blueprint for where engineered storage might succeed. What limits one therefore informs the other.

Scientists studying underground hydrogen face a choice. They can court favor with hydrogen technology advocates and funders—a path that can bring faster funding and greater visibility—by framing their work as steps toward commercial deployment while risking future damage to their credibility if promises outpace evidence. Or they can frame their work as the rigorous investigation of a geochemical system whose commercial potential is uncertain but whose scientific significance is not.

The second framing is harder to sell. It is also more honest, more durable, and, ultimately, more powerful because it produces knowledge that retains its value regardless of which way commercial verdicts fall.

The open questions cataloged should not be obstacles to research. They are the research. The science required to find out whether UHS and natural hydrogen exploration are likely to succeed or fail has independent value that holds across every possible outcome.

Author Information

Laurent Truche (laurent.truche@univ-grenoble-alpes.fr) and Frédéric-Victor Donzé, Institut des Sciences de la Terre, Université Grenoble Alpes, Grenoble, France; and Chris Ballentine, Department of Earth Sciences, University of Oxford, U.K.

Citation: Truche, L., F.-V. Donzé, and C. Ballentine (2026), Underground hydrogen: Two high-reward bets, one indispensable science, Eos, 107, https://doi.org/10.1029/2026EO260292. Published on 15 September 2026. This article does not represent the opinion of AGU, Eos, or any of its affiliates. It is solely the opinion of the author(s). Text © 2026. 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: AGU Advances

Gaining a better understanding of how extreme rainfall is evolving in the tropics under global warming is essential, as this region plays a central role in the Earth’s climate system and hosts some of the world’s most intense precipitation. Yet, despite its global significance, the tropics remain among the most poorly monitored regions, making improved observations and analyses critical for advancing climate science and informing adaptation.

Observed hourly extreme rainfall in tropical Singapore (1980–2024). Temporal trends in the annual number of events and accumulated amount of extreme rainfall (rainfall above a certain intensity threshold). The shaded area denotes the interquartile range (25th–75th percentile) of the 100-meter gridded rainfall reanalysis. “n.s” in the upper right indicates a non-significant trend. Credit: Zhuang et al. [2026], Figure 1a

By combining a unique dataset of weather radar observations and 122 rain gauges covering a 730 km² area in Singapore, Zhuang et al. [2026] reconstruct gridded rainfall fields at 100-meter spatial resolution over a 45-year period. They show that tropical extreme rainfall responds to warming in a fundamentally different way than commonly assumed—extremes intensify through greater spatial extent and longer persistence, while peak rainfall intensities remain largely unchanged. These findings provide new insight into how rainfall extremes evolve under global warming and offer essential information for improving climate risk assessment and adaptation strategies.

Citation: Zhuang, Q., Peleg, N., Prein, A. F., Babovic, V., & Fatichi, S. (2026). Equatorial extreme convective storms are expanding and becoming more persistent. AGU Advances, 7, e2026AV002370. https://doi.org/10.1029/2026AV002370

—Alberto Montanari, Editor-in-Chief, AGU Advances

Text © 2026. 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.

In early July 2026, Stanford postdoctoral researcher Alex Honeyman and graduate student Mark Leone attached a box the size of a briefcase to a fire truck in Colorado that was about to be dispatched to a wildfire 300 miles (480 kilometers) away.