Volcanoes are monitored through a layered system of instruments and techniques, each tuned to detect a different physical signal: ground shaking, surface swelling, gas emissions, sound waves too low for human ears, heat radiation visible from orbit, and even the passage of subatomic particles through solid rock. No single measurement tells the whole story. Instead, observatory scientists combine data streams from seismometers, GPS receivers, gas analyzers, satellites, and other tools to build a picture of what is happening beneath and around a volcano at any given moment. The challenge is that every volcano behaves differently, and the signals that precede an eruption at one site may look nothing like the signals at another.
Seismic Monitoring
Seismometers are the backbone of volcano surveillance. Networks of sensitive ground-motion sensors, often installed on the flanks and summit of a volcano, continuously record vibrations caused by processes deep underground. These vibrations are not all alike, and their character tells scientists different things about what is going on inside the volcano.
Volcano-tectonic earthquakes, the most familiar type, result from rock fracturing under stress, often triggered by magma forcing its way through the crust. Long-period events, by contrast, are linked to the movement of fluids (magma or superheated water and gas) through cracks and conduits. A third category, hybrid earthquakes, shows features of both: they begin with a sharp fracture onset but trail off with the resonant signature of fluid interaction.
1SpringerOpen (Bulletin of Volcanology). Capturing expert uncertainty: ICC-informed soft labelling for volcano-seismicityA sudden increase in any of these earthquake types, or a shift from one type to another, can signal that magma is on the move. Swarms of small earthquakes migrating upward through the crust are among the most reliable precursors volcanologists look for. Continuous tremor, a sustained low-frequency vibration rather than discrete quakes, often indicates that magma or gas is actively flowing through the plumbing system. Seismic networks at well-monitored volcanoes can detect changes in activity within minutes, making them the first line of defense in eruption forecasting.
Ground Deformation
When magma accumulates underground, it pushes the overlying rock outward and upward. This swelling, sometimes just millimeters per day, can be tracked with remarkable precision using several complementary instruments. GPS receivers (more accurately called GNSS stations), strainmeters, and tiltmeters are the most widely used tools for continuous deformation monitoring at active volcanoes. Each captures a slightly different aspect of surface movement, and using them together is often necessary to detect the small variations tied to volcanic activity.
2PubMed Central. Thirty years of ground deformation monitoring at Stromboli volcanoTiltmeters, which measure tiny changes in the angle of the ground surface, can be especially sensitive. At Kīlauea volcano in Hawaii, tiltmeters have tracked inflation and deflation events tied to the pressurization of the shallow magma reservoir. An increase in radial tilt corresponds to inflation of the summit and a rise in the level of the lava lake, while deflation signals the opposite. Tiltmeters at Kīlauea offer better sensitivity and finer time resolution during these events than GPS alone.
3PubMed Central. Relative seismic velocity variations correlate with deformation at Kīlauea volcanoFrom space, a radar technique called InSAR (Interferometric Synthetic Aperture Radar) can map ground deformation over wide areas without anyone needing to set foot on the volcano. Satellites send radar pulses toward the surface and measure how the reflected signal changes between passes. By comparing images taken weeks or months apart, scientists can produce detailed maps of how the ground has risen, sunk, or shifted. During the 2021 eruption of Fagradalsfjall volcano in Iceland, researchers used a combination of persistent and distributed scatterer techniques to produce dense deformation maps of the area surrounding the eruption site.
4PubMed Central. Time-series InSAR measurement using ICOPS and estimation of along-track surface deformation using MAI during the 2021 eruption of Fagradalsfjall Volcano, IcelandInSAR is particularly valuable for remote or poorly instrumented volcanoes. A satellite can observe dozens or hundreds of volcanoes on a regular schedule, flagging deformation even at sites with no ground-based sensors. The limitation is that clouds, dense vegetation, and rapid surface changes (like fresh lava flows) can interfere with radar measurements, and the time gap between satellite passes means that fast-evolving events may be missed.
Volcanic Gas Emissions
Magma contains dissolved gases, mainly water vapor, carbon dioxide, and sulfur dioxide. As magma rises toward the surface and pressure drops, these gases escape. Tracking what comes out of a volcano’s vents and fumaroles, and in what proportions, gives scientists a window into the depth and state of the magma below.
One of the key instruments is the MultiGAS, which can measure concentrations of multiple gas species simultaneously. In a recent study at Soufrière Hills Volcano on the island of Montserrat, helicopter-borne MultiGAS measurements revealed distinct spikes in sulfur dioxide and hydrogen chloride within a larger carbon dioxide-rich plume. That kind of compositional detail helps scientists understand what part of the magmatic system is degassing and whether the plume chemistry is changing over time.
5PubMed Central. Quantification of low-temperature gas emissions reveals CO2 flux underestimates at Soufrière Hills volcano, MontserratDrones have become increasingly important for gas sampling, especially at volcanoes where it is too dangerous or impractical for people to approach the plume. Researchers have successfully flown multirotor drones carrying lightweight gas sensors and miniaturized spectrometers at volcanoes in Italy, Costa Rica, and Nicaragua, at flight heights ranging from about 450 to 3,300 meters. At Turrialba volcano in Costa Rica, a small spectrometer mounted on a drone measured sulfur dioxide on traverses beneath the plume, producing total flux estimates of roughly 1,600 to 1,800 tonnes per day across two passes.
6Atmospheric Measurement Techniques. Implementation of electrochemical, optical and denuder-based sensors and sampling techniques on UAV for volcanic gas measurements: examples from Masaya, Turrialba and Stromboli volcanoesLong-term geochemical monitoring of hot springs and fumaroles near a volcano can also reveal precursory changes. At Meakandake volcano in Japan, decades of sampling showed that chloride and sulfate concentrations in a hot spring at the volcano’s western foot began rising roughly two years before phreatic eruptions occurred. The increase pointed to a growing supply of hot, chemically enriched water from depth, effectively serving as an early warning signal of volcanic unrest.
7ScienceDirect (Elsevier) / Journal of Volcanology and Geothermal Research. Chemical and thermal changes in hot spring waters and fumarolic gases related to volcanic unrest at Meakandake volcano, Japan: Results of long-term geochemical monitoring from 1986 to 2022Infrasound and Acoustic Detection
Eruptions are loud, but much of their sound lies below the range of human hearing. Infrasound, acoustic waves with frequencies below 20 Hz, has evolved over the past two decades from an academic curiosity into a routine monitoring tool used by volcano observatories worldwide to detect, locate, and characterize volcanic activity.
8Bulletin of Volcanology. Volcano infrasound: progress and future directionsInfrasound is useful because it connects directly to what is happening at the vent. The pressure waves produced by an eruption can be linked to the rate at which gas escapes from the volcanic conduit, giving scientists a real-time proxy for eruption intensity.
9Geophysical Research Letters. Volcanic eruptions observed with infrasound And because low-frequency sound waves lose very little energy as they travel through the atmosphere, and because atmospheric waveguides can channel them over huge distances, explosive eruptions can be detected thousands of kilometers from the source.
10Journal of Volcanology and Geothermal Research. An overview of volcano infrasound: From hawaiian to plinian, local to globalThis long-range detection capability makes infrasound especially important for remote volcanoes that lack ground-based instrument networks. The Comprehensive Nuclear-Test-Ban Treaty Organization operates a global network of infrasound stations originally built to detect nuclear tests, but these same stations regularly pick up volcanic eruptions. Combined with local infrasound arrays installed on or near specific volcanoes, the technology provides both close-up detail and global coverage.
Thermal Monitoring from Space
Active volcanic surfaces radiate heat, and satellites equipped with thermal infrared sensors can detect that radiation from orbit. Instruments on satellites like MODIS and VIIRS, originally designed to spot wildfires, are also sensitive to the thermal emissions of volcanic surfaces, from incandescent lava flows to hot fumarole fields.
11Remote Sensing of Environment. An initial comparison of the thermal anomaly detection products of MODIS and VIIRS in their observation of Indonesian volcanic activityThermal satellite data is most useful for tracking the onset and progression of eruptions in near-real time, particularly at volcanoes that are not directly observed by people on the ground. A sudden thermal anomaly appearing where none existed can be the first satellite-based sign of new eruptive activity. Because these satellites scan the entire globe multiple times per day, they provide a systematic check on volcanic activity worldwide, even at volcanoes that no one is watching from the ground.
Tracking Volcanic Ash for Aviation
One of the most consequential applications of volcano monitoring is keeping aircraft safe from volcanic ash, which can destroy jet engines, sandblast windscreens, and clog sensors. Since the mid-1990s, Volcanic Ash Advisory Centers around the world have issued forecasts of ash cloud location and movement for the aviation industry. These forecasts rely heavily on satellite data because of its excellent spatial and temporal coverage, which makes it critical for validating ash dispersion models.
12Journal of Geophysical Research: Atmospheres. Using Simulated Radiances to Understand the Limitations of Satellite‐Retrieved Volcanic Ash Data and the Implications for Volcanic Ash Cloud ForecastingAsh advisory centers combine satellite imagery with atmospheric transport models that simulate where wind will carry the ash cloud over the coming hours and days. Pilots and air traffic controllers rely on these products to reroute flights. The system works well when eruptions are large and visible, but thin, diffuse ash clouds can be difficult to distinguish from ordinary weather clouds on satellite images, which remains an area of active research.
Monitoring Volcanoes Underwater
Roughly three-quarters of Earth’s volcanic activity takes place on the ocean floor, far from view. Monitoring submarine volcanoes poses obvious challenges: you cannot install a tiltmeter on a vent at the bottom of the sea the way you would on a mountainside. The primary remote tool for tracking undersea eruptions is hydroacoustic monitoring, picking up the sounds that travel through the ocean.
The International Monitoring System hydrophone arrays, part of the global nuclear test detection network, have turned out to be powerful tools for this purpose. At Wake Island in the northwestern Pacific, the HA11 hydroacoustic station has detected a large number of signals originating from Ioto (formerly Iwo Jima), with analysis suggesting that roughly 75% of those signals were associated with undersea eruptions.
13PubMed Central. Interpretation of detections of volcanic activity at Ioto Island obtained from in situ seismometers and remote hydrophones of the International Monitoring System The same station has been used for systematic monitoring of submarine volcanic activity across the broader Izu-Bonin-Mariana arc.
14Seismological Research Letters. Monitoring Submarine Volcanic Activities in the Izu–Bonin–Mariana Arc Using the International Monitoring System Hydrophone Array SignalsThis kind of repurposing, using infrastructure built for one objective to monitor volcanoes, highlights how creative volcano monitoring often has to be. Dedicated submarine volcano observatories do exist in some locations, with seafloor-mounted seismometers and pressure sensors connected by cable to shore-based stations. But these are expensive and limited to a handful of sites. For the vast majority of submarine volcanoes, remote hydroacoustic detection is the only practical option.
Seeing Through Rock with Muons
One of the more exotic monitoring techniques involves cosmic-ray muons, high-energy particles that rain down from space and can pass through hundreds of meters of solid rock. Denser material absorbs more muons, so by placing a detector at the base of a volcano and measuring which muons make it through, scientists can build a density map of the volcano’s interior, somewhat like a medical CT scan but at a vastly larger scale.
At Mt. Iwodake in Japan, a portable muon telescope was used to image the shape and density distribution of the volcanic conduit beneath the crater floor.
15Geophysical Research Letters. Cosmic‐ray muon imaging of magma in a conduit: Degassing process of Satsuma‐Iwojima Volcano, Japan At Sakurajima volcano, also in Japan, ongoing muon imaging revealed a direct physical mechanism linking eruption frequency to ground deformation. When eruption frequency was low, a dense plug formed in the conduit, trapping pressurized gas beneath it and causing the volcano to inflate. When eruptions were frequent, the plug was absent, gas escaped freely, and the edifice deflated.
16Geophysical Research Letters. Muon Imaging of Volcanic Conduit Explains Link Between Eruption Frequency and Ground DeformationThat finding has broader implications: it offers a physical explanation for deformation patterns observed at other volcanoes, where scientists see inflation but are uncertain whether it signals an impending eruption or simply a temporary blockage in the plumbing. Muon imaging is still limited by its slow data-collection rate and the need for suitable detector placement geometry, but it provides information about the interior of a volcano that no other technique can.
Hydrothermal Systems and Electrical Surveys
Water heated by magma circulates through volcanic rock, creating hydrothermal systems that can extend far beyond the visible crater. Mapping these systems tells scientists about the internal structure of a volcano and where heat and fluids are concentrated. One approach uses a combination of electric self-potential surveys, which measure natural voltage differences on the volcano surface caused by flowing groundwater, and magnetotelluric surveys, which probe subsurface electrical resistivity. Researchers applied both techniques at five large stratovolcanoes in Japan and found a consistent pattern: good conductors in the subsurface corresponded to hydrothermal zones, with a low-permeability clay layer forming a cap above the hot zone.
17Journal of Geophysical Research: Solid Earth. Groundwater flow and hydrothermal systems within volcanic edifices: Delineation by electric self‐potential and magnetotelluricsUnderstanding the hydrothermal system matters for hazard assessment because steam-driven (phreatic) eruptions, which can occur with little warning, originate from the sudden flashing of pressurized water to steam. Knowing where hot fluid accumulates and how it is sealed helps scientists estimate where and how violently such an eruption might occur.
Machine Learning and Eruption Forecasting
All the instruments described above generate enormous amounts of data, and one of the frontiers in volcanology is using machine learning to sift through it. Neural networks trained on seismic features can classify a volcano’s state of alert and estimate the probability of an eruption in the near term, providing a tool for probabilistic short-term forecasting that can be transferred across different volcanic systems. One recent approach reported the ability to forecast eruptions with lead times ranging from at least hours to several days.
18Frontiers in Earth Science. Universal machine learning approach to volcanic eruption forecasting using seismic featuresThe appeal of machine learning is its potential universality. Each volcano has its own personality, and experienced volcanologists at a given observatory develop intuition about their specific volcano over years or decades. An algorithm that can learn from multiple volcanoes and transfer those patterns to a new site could be transformative for the many volcanoes that lack dedicated observatories. The risk, of course, is over-reliance on algorithms trained on limited eruption histories. False alarms erode trust, and missed eruptions cost lives. These tools are best understood as aids to human judgment, not replacements for it.
The Global Monitoring Gap
Despite all these techniques, many of the world’s active volcanoes are barely monitored at all. Ground-based monitoring is inadequate or lacking at many active volcanic sites, and satellite observations are increasingly seen as the way to fill those gaps and provide operational awareness of unrest and eruptions.
19Bulletin of Volcanology. Steps toward a satellite-based global volcano monitoring and early warning system: from pilot to demonstrator to GVEWERSThere are roughly 1,500 potentially active volcanoes on land, and hundreds more underwater. A fraction of these have dense, modern instrument networks. Many have a single seismometer, or none at all. The reasons are predictable: cost, logistics, remoteness, and the political reality that volcanoes in wealthy countries tend to be better monitored than those in developing nations, even though the latter often pose a greater risk to densely settled populations nearby. Efforts to build a satellite-based global volcano early warning system aim to at least ensure that no eruption goes entirely undetected, even if the fine-grained detail of a well-instrumented observatory is out of reach.
Physical Sampling and What Rocks Reveal
Not everything about a volcano can be measured remotely or in real time. Sometimes the most useful information comes from physically collecting rocks, ash, and lava and analyzing them in a laboratory. The chemistry and texture of erupted material record the conditions that existed underground before the eruption: the temperature, pressure, water content, and how long the magma spent sitting in a reservoir before being expelled.
At Aira Caldera in Japan, researchers analyzed crystals in erupted material and found that eruptible magma had short residence times in the shallow reservoir, on the order of hundreds to a couple of thousand years, with no progressive increase across successive eruptions.
20PubMed Central. Submillennial timescales of eruptible rhyolite magma: Converging evidence from Aira Caldera, Japan That kind of finding informs long-term hazard assessments by constraining how quickly a dangerous body of magma can assemble beneath a caldera. It is a different kind of monitoring, not real-time but retrospective, and it provides context that no instrument network can replicate on its own.
When Communities Are the Sensors
In many volcanic regions, local and indigenous knowledge systems operate alongside modern technology as a parallel form of monitoring. In Indonesia, home to more active volcanoes than any other country, communities near volcanoes use cultural practices like word-of-mouth networks, mosque announcements, and the kentongan (a bamboo slit drum) to disseminate warnings. People also interpret environmental cues, changes in animal behavior, water levels, or the smell of sulfur, as indicators of volcanic unrest.
21International Journal of Disaster Risk Reduction. The inclusivity of volcanic hazard early warning systems: Experiences of persons with disabilities in IndonesiaThese practices are not relics of a pre-scientific era. They fill real gaps. Technology-based warning systems can fail during power outages, may not reach people with disabilities, and often assume a level of literacy or smartphone access that not everyone has. Cultural practices function as trusted channels precisely because they are embedded in daily life and do not require any infrastructure beyond the community itself. The most robust early warning systems are those that integrate both formal scientific monitoring and the locally adapted practices that have kept people alive near volcanoes for generations.