Feed aggregator

How are cold air masses advancing in the United States connected to fertilizers carried by "flying rivers" from Africa that nourish the soils of the Brazilian Amazon? An article published in Geophysical Research Letters reveals an atmospheric connection between these distant regions.

We replaced the stove with plywood, turning the kitchen of the dive boat into an impromptu research lab. Plugging in wires and connecting tubing, we assembled a scientific instrument within the cramped cabin. Then we cast off into Halifax Harbor, Canada, surveying the turquoise waters for signs of an unusual test: could we use the ocean itself to remove carbon dioxide from the air?

Imagine being tasked with counting every blade of grass in a field, noting every single species as you go. This is not far from the challenge many scientists face when analyzing microscopic samples packed with thousands of tiny particles.

What's the link between the global economy and the climate? Consumption drives extraction and carbon emissions. But there is more. The inequalities of the global economy don't just shape what goes into the atmosphere. They affect our understanding of the climate and our perspectives when it comes to possible solutions.

SummaryMicroseismic source location is essential for seismic monitoring and subsurface resource exploitation. Both traveltime inversion and waveform stacking methods suffer from limited accuracy when processing low signal-to-noise ratio (SNR) data under complex velocity models. Existing deep learning approaches mainly employ purely data-driven strategies without physical constraints, exhibiting limited capability to suppress large and unexpected location errors. We propose a physics-constrained deep learning method for microseismic source location that integrates the physical principles of cross-correlation stacking (CCS) imaging into network training. The method incorporates a joint loss function combining source imaging quality loss and traveltime consistency loss, with a Pareto dynamic weighting strategy to balance different loss components. Synthetic experiments on the Marmousi velocity model demonstrate that the joint-constrained method reduces the mean absolute error (MAE) from 34.09 m to 27.91 m compared to the purely data-driven approach. The maximum error decreases from 280.18 m to 130.38 m, a 53.5% reduction, demonstrating effective suppression of large location errors. The trained network achieves single-event imaging prediction in 0.04 s, providing a 75-fold speedup over the 3 s required by conventional CCS. The proposed method shows great potential in near-real-time microseismic monitoring with dense arrays.

When Cyclone Gabrielle tore through New Zealand's Tairāwhiti region in 2023, it left behind more than silt and floodwaters.

Publication date: Available online 5 May 2026

Source: Advances in Space Research

Author(s): Anatolii Alpatov, Erik Lapkhanov

Publication date: Available online 5 May 2026

Source: Advances in Space Research

Author(s): Danial Abdollahi, M.Mahdi Alizadeh

Publication date: Available online 5 May 2026

Source: Advances in Space Research

Author(s): Chengzhi Xie, Xiao Chen, Tianle Wei, Yuxin Ding, Yuhuan Cui, Shuang Hao

Publication date: Available online 4 May 2026

Source: Advances in Space Research

Author(s): Yufei Guo, Zixuan Zheng, Yingjie Wang, Yanli Feng, Chen Li, Junhua Zhang

Publication date: Available online 4 May 2026

Source: Advances in Space Research

Author(s): Andrea Aprovitola, Luigi Iuspa, Giuseppe Pezzella, Antonio Viviani

A huge 11,000-carat ruby has been discovered in Myanmar, state media reported Friday, one of the largest ever found in the country renowned for its precious gemstones.

Antarctica is being ravaged by a triple-whammy of climate chaos that has melted sea ice to record lows, a new study has revealed. For decades, the frozen wilderness at the bottom of the world defied global warming trends, with ice levels actually growing—until 2015 when it suddenly reversed.

Thirty years ago, the blockbuster movie Twister featured a group of academics putting themselves at risk by chasing tornadoes in the name of science. Although the Hollywood story entailed a surfeit of sensationalism, special effects, and unrealistic stereotypes, the movie got a few things right. Specifically, the scientists were trying to study tornadoes using a large number of spatially distributed, home-built, low-cost (and potentially sacrificial) sensors.

Today, we commonly refer to the coordinated use of tens to hundreds of similar sensors that are spread out as “large-N” sensing. Such sensor distributions have led to important advances in seismology and infrasound science, where they have improved our understanding of seismic ground motion and helped shed light on volcanic eruption dynamics [e.g., Rosenblatt et al., 2022; Anderson et al., 2023].

The benefits of large-N networks and arrays include robust spatial sampling and signal extraction from noise. They are also advantageous for detecting small signals, sensing natural hazards in remote environments, and offering critical redundancies for sensors at risk from lava or debris flows, wildfire, weather, or even malicious mammals.

Since 2013, our research group in the Department of Geosciences at Boise State University (BSU) has worked to study infrasound from geophysical phenomena by capitalizing on the benefits of low-cost, large-N sensing technology [e.g., Slad and Merchant, 2021]. More than a decade on, this effort has yielded scientific successes from a variety of environments, and it is continuing to evolve.

Large-N Sensing for Infrasound

Many violent natural processes, including landslides, volcanic eruptions, earthquakes, avalanches, and meteors, produce infrasound.

Many violent natural processes, including landslides, volcanic eruptions, earthquakes, avalanches, and meteors, produce infrasound, defined as low-frequency sound below the threshold of human hearing (less than 20 Hertz). Such events may create audible sound as well, but the subaudible band is often much more energetic in terms of sound intensity, and it has long wavelengths that can propagate long distances with little attenuation. These characteristics make infrasound especially valuable for remote sensing of natural phenomena.

Our group at BSU grew more interested in developing our own inexpensive infrasound sensing solutions after costing out technology for commercial data logging systems, the compact electronic devices that record and store sensor data. These systems can be far more expensive than infrasound transducers—the sensors that actually detect sound—themselves.

The cost element became particularly relevant after we lost instrumentation deployed at the summit of Chile’s Villarrica volcano when it erupted a 2-kilometer-tall lava fountain on 3 March 2015 [Johnson et al., 2018]. In an instant, our hardware, including seismic and infrasonic sensors and their commercial multichannel data loggers, was entombed beneath falling lava. This financial loss incentivized our work to develop low-cost loggers that would match the technical specifications and fidelity of commercial systems.

The result was the customized Gem infrasound logger, which we created using the widely available and very economical Arduino open-source electronic prototyping platform and its low–power consumption microcontroller. The Gem is an all-in-one infrasound sensor and data logger with a high dynamic range (millipascals to 100 pascals), a 100-hertz sample rate appropriate for infrasound, and a built-in GPS for precise timing and synchronization [Anderson et al., 2018].

Although we initially conceived of the Gem as an alternative to commercial loggers to be deployed as single stations or in small arrays, we quickly realized its potential for use in high-density distributed sensing arrays that enabled new detection capabilities. In particular, its small package size (it has about the dimensions and weight of a paperback novel) and its ease of deployment—simply insert alkaline batteries, place it on the ground, and turn it on—have opened opportunities for rapid, large-N deployments in difficult-to-access environments.

Early Successes for the Gem Volcán Villarrica, near Pucon, Chile, is seen in 2025 (left). The volcano regularly releases gas from a small lava lake recessed deep within the summit crater (right). Credit: Jeffrey B. Johnson

The Gem’s inaugural field mission came in January 2020 during a return to Villarrica, where activity had returned to normal following its 2015 paroxysmal eruption [Rosenblatt et al., 2022]. Typical activity in the volcano’s normal state includes open-vent degassing from a small lava lake recessed deep within the summit crater, which produces its famously powerful volcano infrasound [e.g., Johnson et al., 2012].

To capture Villarrica’s infrasound in detail, a four-person team from BSU climbed the 3,000-meter-tall glaciated volcano and quickly installed 16 sensors around the crater rim, as well as another 16 sensors along an 8-kilometer linear transect from the summit down the northern slope (Figure 1). This unique sensor distribution permitted us to capture the infrasound wavefield and how it interacts with topography in unprecedented detail.

Fig. 1. (a) Oblique and (b) plan views of Villarica’s summit region were created from structure-from-motion surveys in 2020. Red triangles and circles indicate locations of Gem sensing packages. (c) Also in 2020, Jake Anderson adjusts a cable suspended across the volcano’s crater that held a Gem sensor (circled). (d) In 2025, Jerry Mock unloads Gem systems at Villarica’s summit during another data collection campaign there. Click image for larger version. Credit: Jeffrey B. Johnson

Deploying such an array configuration using much heavier, larger, and power-intensive conventional instruments would have taken far more time and resources, as well as a bigger group. With the Gems, however, the installation was feasible for our small team, each member of which could easily carry eight instruments and the batteries needed to power them.

To monitor volcanoes with infrasound, it is necessary to understand the influence of atmospheric effects.

Once in place, these sensors collected continuous data during the 2-week study that were used to quantify the diffraction of sound coming out of the volcanic crater [Rosenblatt et al., 2022] and to measure the sound’s attenuation as it propagated away. Such studies are important for investigating time-varying atmospheric parameters such as changing temperatures and winds, which can affect infrasound transmission, diminishing its amplitude or even—in extreme cases—completely silencing it in an acoustic shadow zone [Johnson et al., 2012]. To monitor volcanoes with infrasound, it is necessary to understand the influence of atmospheric effects.

Months later, another opportunity arose to demonstrate the Gems’ capability for large-N infrasound sensing. During the early days of the COVID-19 pandemic, on 31 March 2020, a magnitude 6.5 earthquake occurred near Stanley, Idaho. The earthquake, the largest in the state since 1983, kicked off an energetic aftershock sequence, with more than 700 magnitude 3 or greater earthquakes occurring in 6 months. Most of these events produced significant local infrasound radiation, or “airquakes,” caused by ground-atmosphere coupling [e.g., Johnson et al., 2020].

Pandemic-related precautions inhibited a large team from venturing as a group into the field. However, a lone BSU researcher (coauthor Jacob Anderson), trudging through forest terrain and deep snow on skis, was able to deploy and activate 22 Gems in less than 4 hours in early April, thanks in part to the sensors’ compact size and ease of deployment.

This array captured hundreds of local infrasonic aftershocks within about 25 kilometers of their epicenters. It also recorded a far larger event 700 kilometers away, the 15 May magnitude 6.5 Monte Cristo earthquake in Nevada. The array detected the epicentral infrasound from the distant earthquake source, as well as infrasound from numerous secondary sources, including mountain ranges throughout the western United States that reradiated the ground motion as infrasound (Figure 2) [Anderson et al., 2023].

Fig. 2. This map shows source region(s) of infrasound associated with the May 2020 Monte Cristo earthquake in Nevada that was detected by an array of Gem infrasound sensors deployed at the PARK site near Stanley, Idaho. Click image for larger version. Credit: Adapted from Anderson et al. [2023], CC BY 4.0

Detecting all these distinct signals was possible because of the enhanced array processing capabilities provided by the large number of sensors. Anderson et al. [2023] showed that when the data were processed from 3-sensor subsets of the 20+-sensor array—instead of from the whole array—it was possible to detect only the most intense earthquake infrasound arrivals. In other words, the larger array had much greater fidelity and sensing capabilities than smaller distributions of sensors.

During its 2-month deployment, the Stanley array also detected sounds from other distant nonearthquake sources, including waterfalls 195 kilometers away and thunder more than 900 kilometers away [Scamfer and Anderson, 2023]. Such enhanced detections, facilitated by large-N sensing, demonstrate an improved capacity to monitor a range of Earth phenomena continuously over a wide range of distances.

Putting Sensors in Harm’s Way

Since those proof-of-concept deployments, Gems have been used to monitor snow avalanches, lahars, river flow discharge, stratospheric sounds (while mounted aboard a solar balloon), and numerous volcanoes during field experiments [e.g., Tatum et al., 2023; Bosa et al., 2024; Rosenblatt et al., 2022; Brissaud et al., 2021]. Given their ease of use, small size, and low replacement cost, they’ve also been tested in hazardous environments where the risk to more expensive hardware could be considered unreasonable.

The motivation to put sensors in harm’s way is to gain insight into geophysical phenomena by recording subtle signals close to the source that may not be detectable from farther away.

The motivation to put sensors in harm’s way is to gain insight into geophysical phenomena by recording subtle signals close to the source that may not be detectable from farther away. For example, at Villarrica, Rosenblatt et al. [2022] suspended a Gem on a cable 100 meters above a lava lake to collect infrasound data from a unique, bird’s-eye perspective over the crater (Figure 1c). (Stringing the cable across the crater proved far more challenging than deploying the sensor itself, which slid down the cable until finding its resting place at the bottom of the cable’s arc.)

In another case, we landed a pair of Gems on the ground near a frequently exploding crater at Fuego volcano in Guatemala using a drone (see video below). We later retrieved one of the sensors from high on the volcano’s flanks. Another was lost because high winds initially posed too great a risk to fly the drone back for it. Then the following day after the wind subsided, we could not locate the stranded Gem, which was probably a casualty of a nighttime explosion.

Drone footage and infrasound recordings were collected during an explosion of Fuego volcano on 4 February 2024. Pa = pascals. Credit: video: Jerry C. Mock; animation and infrasound: Jeffrey B. Johnson

Our group at BSU also has nascent interest in using Gems to study fire in natural environments. Wildfires produce infrasound from a spatially extensive source region corresponding to actively burning areas. Because of the source complexity and the fact that fire infrasound is low amplitude and tremor-like [Johnson et al., 2025], enhancing signal-to-noise ratios in recorded infrasound is critical. This enhancement is enabled by using large-N monitoring networks, making infrasound wildfire surveillance a promising area of investigation.

Low-cost, rapid infrasound deployments could one day be used as an effective operational tool.

Toward this objective, our group installed 76 sensors ahead of a prescribed burn in Reynolds Creek, Idaho, in October 2023 to begin developing infrasound as a tool for monitoring and mapping wildfire. We have also deployed Gems for infrasound studies of naturally occurring wildfires, such as the Emigrant wildfire in Oregon in August and September 2025 (Figure 3). During that active wildfire response, a team safely and quickly installed tens of sensors within a matter of hours in an area facing dynamic hazards from the rapidly expanding fire, which eventually covered 33,000 acres (about 13,354 hectares). Luckily, no instruments were lost, and the data have shown the potential to track a wildfire as it advances.

Preliminary results suggest that low-cost, rapid infrasound deployments could one day be used as an effective operational tool. For example, in firefighting responses, infrasound might complement intermittent aerial observations, from aircraft or drones, because it provides a continuous record of fire activity. Infrasound surveillance might also be able to “hear” combustion sources within a burn area that is obscured to optical sensing because of clouds or nightfall.

Fig. 3. (a) The spread and severity of the 2025 Emigrant Fire in Oregon, as calculated from prefire (21 August) and postfire (18 October) Sentinel-2 satellite images, are shown. Inset maps show the distribution of 37 Gem sensors rapidly deployed in three arrays. (b) Smoke from the fire rises from the landscape on 31 August during deployment of the sensors. (c) Following the fire, one sensor that had been melted by the fire was recovered with its data card still intact (red circle). dNBR = differenced normalized burn ratio. Click image for larger version. Credit: (a) and (b): Madeline A. Hunt; (c): Jacob F. Anderson The Evolution of Low-Cost Sensors

Five years ago, the single-sensor Gem was a cutting-edge infrasound logging solution. While it remains a powerful and economical tool for large-N arrays and for sensing in hostile environments, it is evolving.

Boise State University researchers (left to right) Madeline Hunt, Owen Walsh, Jerry Mock, and Jacob Anderson prepare to deploy Gem sensors in Idaho’s Sawtooth Mountains in January 2024. Credit: Jeffrey B. Johnson

We have now developed the Gem into an even more versatile version called the Aspen, which can log four independent sensors at a sample rate of 200 hertz, double that of the Gem. The Aspen retains the small size, low weight, low power consumption, and low cost of the Gem, but with the capability to record higher-resolution 24-bit, time-synchronized data from a triaxial seismic sensor and an infrasound transducer.

Recording synchronous seismoinfrasonic data on the same logging platform offers the advantage of sensing both ground shaking and infrasonic oscillations. The ability to measure waves propagating in the ground and in the air simultaneously could facilitate work in the growing field of environmental seismology, which focuses on geophysical sources at Earth’s surface like debris flows and volcanoes.

Although we have focused on seismoacoustic geophysical measurements in our work, the concept of gathering data with low-cost instrumentation in harm’s way or from coordinated arrays of numerous sensors holds promise across Earth and environmental sciences. Such approaches could be used, for example, with tiltmeters (which measure slope changes), gravity meters, or near-infrared thermometers (e.g., optical pyrometers), all of which would offer additional data streams complementing seismoacoustic observations in geophysical studies of volcanoes.

With the diversity of emerging uses, it’s clear that large-N sensing—infeasible or cost prohibitive in many cases until recently—could transform how we measure many facets of Earth, helping to reveal the inner workings of volatile volcanoes, twisting tornadoes, and more.

Acknowledgments

More information about low-cost infrasound sensing solutions can be found at https://sites.google.com/boisestate.edu/infravolc/home. Development of the Gem infrasound logging platform was supported by a grant from the National Science Foundation (EAR-2122188).

References

Anderson, J. F., et al. (2018), The Gem infrasound logger and custom‐built instrumentation, Seismol. Res. Lett., 89(1), 153–164, https://doi.org/10.1785/0220170067.

Anderson, J. F., et al. (2023), Remotely imaging seismic ground shaking via large-N infrasound beamforming, Commun. Earth Environ., 4(1), 399, https://doi.org/10.1038/s43247-023-01058-z.

Bosa, A. R., et al. (2024), Dynamics of rain-triggered lahars and destructive power inferred from seismo-acoustic arrays and time-lapse camera correlation at Volcán de Fuego, Guatemala, Nat. Hazards, 121, 3,431–3,472, https://doi.org/10.1007/s11069-024-06926-1.

Brissaud, Q., et al. (2021), The first detection of an earthquake from a balloon using its acoustic signature, Geophys. Res. Lett., 48, e2021GL093013, https://doi.org/10.1029/2021GL093013.

Johnson, J. B., et al. (2012), Probing local wind and temperature structure using infrasound from Volcan Villarrica (Chile), J. Geophys. Res., 117, D17107, https://doi.org/10.1029/2012JD017694.

Johnson, J. B., et al. (2018), Forecasting the eruption of an open-vent volcano using resonant infrasound tones, Geophys. Res. Lett., 45, 2,213–2,220, https://doi.org/10.1002/2017GL076506.

Johnson, J. B., et al. (2020), Mapping the sources of proximal earthquake infrasound, Geophys. Res. Lett., 47, e2020GL091421 , https://doi.org/10.1029/2020GL091421.

Johnson, J. B., J. F. Anderson, and K. Yedinak (2025), Infrasound produced by a small pile fire, Appl. Acoust., 231, 110559, https://doi.org/10.1016/j.apacoust.2025.110559.

Rosenblatt, B. B., et al. (2022), Controls on the frequency content of near-source infrasound at open-vent volcanoes: A case study from Volcán Villarrica, Chile, Bull. Volcanol., 84(12), 103, https://doi.org/10.1007/s00445-022-01607-y.

Scamfer, L. T., and J. F. Anderson (2023), Exploring background noise with a large‐N infrasound array: Waterfalls, thunderstorms, and earthquakes, Geophys. Res. Lett., 50, e2023GL104635, https://doi.org/10.1029/2023GL104635.

Slad, G., and B. Merchant (2021), Evaluation of Low Cost Infrasound Sensor Packages, Sandia Rep. SAND2021-13632, Sandia Natl. Lab., Albuquerque, N.M., https://doi.org/10.2172/1829264.

Tatum, T., J. F. Anderson, and T. J. Ronan (2023), Whitewater sound dependence on discharge and wave configuration at an adjustable wave feature, Water Resour. Res., 59, e2023WR034554, https://doi.org/10.1029/2023WR034554.

Author Information

Jeffrey B. Johnson (jeffreybjohnson@boisestate.edu), Jacob F. Anderson, Madeline A. Hunt, Owen A. Walsh, and Jerry C. Mock, Department of Geosciences, Boise State University, Idaho

Citation: Johnson, J. B., J. F. Anderson, M. A. Hunt, O. A. Walsh, and J. C. Mock (2026), Sensing the sounds from Earth’s hazardous environments, Eos, 107, https://doi.org/10.1029/2026EO260142. Published on 8 May 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.

Emissions from urban areas account for about a tenth of the global methane budget, according to a new analysis of satellite data published in the Proceedings of the National Academy of Sciences of the United States of America. And those emissions grew by about 10% from 2020 to 2023, despite cities’ pledges to slash them.

Methane is a potent greenhouse gas, and it’s shorter lived in the atmosphere than carbon dioxide. That means cutting methane emissions would have great benefits for the climate over the short term. Oil and gas operations and agriculture are major sources of methane, but so are cities and their infrastructure.

“Cities have started attempting to reduce their methane emissions, and we hope to be able to monitor this,” said Erica Whiting, a graduate student in climate and space science at the University of Michigan. Most efforts to account for urban methane emissions—from wastewater treatment plants, landfills, leaky natural gas infrastructure, and other sources—have relied on ground-based measurements and on inventories that estimate emissions on the basis of activities, said Whiting. Most of these studies have looked at a handful of cities, typically in North America and Europe.

In contrast, Whiting said her team’s study is one of the first to use satellite data to monitor urban methane emissions over time. Satellite monitoring offers long-term, often global, measurements and can provide a clearer picture of how mitigation efforts are developing.

Falling Short

A growing number of cities are aiming to reduce carbon emissions, and the new data show many of them are not on track.

A growing number of cities are aiming to reduce carbon emissions, and the new data show many of them are not on track. Whiting’s study included 92 cities around the world, including 51 members of a coalition called C40, which was founded in 2005. This 96-country coalition is working toward the goal of cutting greenhouse gas emissions by half by 2030, including a 34% decrease in methane emissions. These numbers are aligned with the goal of limiting global warming to 1.5°C over preindustrial levels.

Whiting’s team analyzed methane data from the satellite-based TROPOMI (Tropospheric Monitoring Instrument) from 2019 to 2023. TROPOMI launched in 2017, making it possible to continuously monitor methane and other gas concentrations around the world. TROPOMI data showed that from 2019 to 2020, urban methane levels fell. But from 2020 to 2023, emissions grew 10% in C40 cities and 12% in non-C40 cities. The study focuses not just on urban centers but also on their outlying areas, where known methane sources such as landfills and wastewater treatment plants are often located.

The Tropospheric Monitoring Instrument (TROPOMI) aboard the Sentinel-5P satellite measures the potent greenhouse gas methane. In snapshots over urban areas, higher methane concentrations are depicted in warmer colors. Credit: Erica Whiting

The current study can’t point to what accounts for these trends, said Whiting. However, she said, urban populations grew during the study period, which could be a contributor to the cities’ growing emissions.

“In most regions of the world, there is no evidence that methane emissions from cities are decreasing at all.”

Rob Jackson, an Earth system scientist at Stanford University and chair of the Global Carbon Project, noted that it’s hard to know how to interpret the increase in emissions because the study period includes the era of the COVID-19 pandemic lockdowns, which caused major changes in people’s behavior and associated drops in anthropogenic emissions in 2020. (However, counterintuitively, the early 2020s actually saw a spike in overall methane emissions, which some scientists attribute to wetlands and changes in atmospheric chemistry.) Nevertheless, he said the data show that the world is not on track to decrease urban methane emissions. “In most regions of the world, there is no evidence that methane emissions from cities are decreasing at all,” he said.

“This work clearly shows that major cities worldwide are not reducing methane emissions at a rate consistent with the Global Methane Pledge,” Jackson said. This international agreement, made in 2021, has reduction goals that align with those of the C40 coalition: decrease global methane emissions by at least 30% relative to 2020 levels by 2030. The European Commission and 159 countries are participating in the pledge.

Whiting hopes better data will help. City and regional governments can use data from satellites to support and monitor ongoing efforts to lower methane emissions. “We’re excited to have this approach to monitor changes, and it should be useful for urban planning,” she said.

Zachary Tofias, director of food and waste at C40 Cities, noted via email that the organization was not involved with the design of the study. He pointed to several recent large-scale composting and other waste management facilities recently commissioned by member cities that should help bring down methane emissions going forward. The increasing availability of satellite and aerial monitoring data, he said, “provides an amazing additional tool for cities and facility managers to understand and address methane leaks from waste-disposal sites.”

—Katherine Bourzac (@bourzac.bsky.social), Science Writer

Citation: Bourzac, K. (2026), Urban methane emissions are rising, despite cities’ pledges, Eos, 107, https://doi.org/10.1029/2026EO260143. Published on 8 May 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.

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

Permafrost beneath Arctic roads is warming and becoming less stable, creating growing risks for northern infrastructure. Yet predicting how frozen ground will evolve remains difficult because subsurface conditions vary sharply over short distances, observations are sparse, and conventional process-based models are not easy to update as new field data arrive. In a new study, Gou et al. [2026] address that challenge at an embankment road in Utqiaġvik, Alaska, using fiber-optic temperature measurements collected along a 100-meter transect to track how shallow ground conditions change through time. Rather than treating monitoring and modeling as separate tasks, the authors link them in a framework designed to evolve with the physical system itself.

What stands out here is not simply the use of machine learning, but the way the authors build a physics-informed digital twin for permafrost under infrastructure. Their framework embeds a neural network within a heat-transfer solver, so the governing physics remain central while the model can still update uncertain soil properties as new observations arrive. This study moves beyond black-box prediction toward an interpretable, updateable system that can reconstruct subsurface temperature fields, infer thermodynamic properties such as unfrozen water content and thermal conductivity, and then test those inferences against independent DAS data, borehole temperatures, and laboratory measurements. This makes the work more than a site-specific modeling exercise; it offers a credible pathway toward near-real-time permafrost forecasting and infrastructure monitoring in a rapidly warming Arctic.

Framework of the proposed digital twin model. The neural network (NN) takes soil temperature at each lateral position as input and outputs six unknown parameters that vary laterally with distance. These parameters are embedded in the heat‐transfer equation through constitutive relationships, and the resulting system is solved using a finite difference method (FDM). The difference between predicted and observed temperatures is computed and defined as “loss,” and the loss gradients are backpropagated to update the NN parameters. Credit: Gou et al. [2026], Figure 2

Citation: Gou, L., Xiao, M., Zhu, T., Martin, E. R., Wang, Z., Rocha dos Santos, G., et al. (2026). Physics-informed digital twin for predicting permafrost thermodynamic characteristics under an embankment road in Utqiaġvik, Alaska. Journal of Geophysical Research: Earth Surface, 131, e2025JF008787. https://doi.org/10.1029/2025JF008787

—Xiang Huang, Associate Editor, JGR: Earth Surface

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.

Author(s): A. Frazzitta

Plasma discharge devices have recently emerged as compact and versatile tools for particle beam manipulation. Building upon the active plasma lens (APL) and its curved extension, the active plasma bending, this work introduces the concept of the plasma discharge undulator (PDU). In a PDU, a high-cur…

[Phys. Rev. E 113, 055204] Published Fri May 08, 2026

Author(s): Jingchun Li, Zhaoyang Lu, Jianqiang Xu, Wei Chen, Jiaqi Dong, Jingting Luo, and Yong Liu

The fast ion (FI) on turbulent transport is one of the key topics of magnetic confinement fusion. This work focus on the impact of FI pressure gradients on turbulence in a high-β plasma scenario using gyrokinetic simulations. Linear analyses reveal that FIs strongly stabilize ion temperature gradien…

[Phys. Rev. E 113, L053201] Published Fri May 08, 2026

In April 2026 I recorded 36 fatal landslides causing 90 fatalities, the lowest monthly total for 2026 to date.

This is my regular update for the number of fatal global landslides, focusing on March 2026. As usual, this data has been collected in line with the methodology described in Froude and Petley (2018) and in Petley (2012). References are listed below – please cite these articles if you use this analysis. Data presented in these updates should be treated as being provisional at this stage as I will reanalyse them prior to formal publication, and other events will emerge.

The headline figures are as follows:

March 2026: 36 fatal landslides causing 90 fatalities;

This is an interesting result, unusually showing that fatal landslides in April were substantially lower than for any of the preceding months in 2026. This is the updated annual chart by month:-

The number of global fatal landslides in 2026 by month to the end of April.

Loyal readers will know that I like to present the running total using pentads (five day blocks). This is the cumulative total pentad graph to the end of Pentad 24 (which captures all of the events to the end of April):-

The cumulative total number of global fatal landslides in 2026 by pentad to the end of April.

Thus, whilst April 2026 was unexceptional compared with the previous months of this year, the number of fatal landslides was still above the long term mean. Overall, 2026 continues to run extremely hot, exceeding even the record-breaking year of 2024.

We now start to enter the crucial period of much higher global fatal landslide occurrence. Whilst in the long term dataset this acceleration typically occurs in June (or even July), in recent years it has happened in May, as the 2024 line shows. I will watch with great interest to see what happens this month.

As I always stress, the occurrence of fatal landslides prior to the South and East Asia rainy seasons is not a predictor of what will happen in that period. Interestingly, the WMO is forecasting a below average summer monsoon in South Asia.

References

Froude, M. and Petley, D.N. 2018.  Global fatal landslide occurrence from 2004 to 2016.  Natural Hazards and Earth System Sciences 18, 2161-2181.

Petley, D.N. 2012. Global patterns of loss of life from landslides. Geology 40 (10), 927-930.

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.

SummaryWe present a new method, ODFTEX, for calculating evolving crystal preferred orientation (CPO) in deforming aggregates of olivine plus orthopyroxene undergoing dynamic recrystallization. The model is based on a continuum description of texture in terms of the orientation distribution function (ODF), which satisfies an evolution equation that we solve numerically. The model thus delivers the ODF directly, rather than a collection of grain orientations like most alternative models. Recrystallization is represented by a source term in the evolution equation, defined in such a way that crystals poorly oriented for slip recrystallize most rapidly. The model has only a single free parameter, the recrystallization rate, which we calibrate against a laboratory experiment on an olivine aggregate deformed in simple shear. We illustrate the predictive power of ODFTEX by using it to calculate evolving CPO along pathlines in a two-dimensional convective flow and a three-dimensional subduction zone flow. ODFTEX is computationally about 6-7 times faster than the D-Rex model of Kaminski et al.(2004).