Will a supervolcano erupt before 2050?

Compared to Yellowstone, Campi Flegrei exhibits greater recent seismic activity and ongoing ground deformation. Yellowstone maintains a NORMAL alert status and experienced a pause in ground uplift in January 2026, while Campi Flegrei has shown continuous uplift and recent seismic events ^[1][2][3][4]. Despite these differences, neither volcano currently displays strong earthquake swarms or rapid uplift exceeding background levels that are typically identified as eruption precursors ^[6].

Yellowstone shows minimal activity, with no signs of imminent eruption. Its current status indicates very low volcanic activity, with no evidence of impending eruption. In April 2026, Yellowstone recorded 97 earthquakes, with the largest event reaching magnitude M2.5 ^[1]. Ground uplift at Yellowstone paused in January 2026, and there are no signs of magma accumulation beneath the caldera. The annual probability of a Yellowstone supereruption is exceptionally low, estimated at 1 in 730,000 ^[1][2][5].

Campi Flegrei's unrest is slowing, driven by hydrothermal processes. Its ongoing unrest, while notable, is decelerating and primarily influenced by hydrothermal systems. In one week in April 2026, Campi Flegrei registered 21 earthquakes, with a maximum magnitude of M1.9 ^[3]. The ground uplift rate was recorded at 10 mm per month from February to April 2026, contributing to an approximate total uplift of 26.5 cm since 2025 ^[3][4]. Although unrest at Campi Flegrei has been continuous since 2005, current activity is driven mainly by hydrothermal dynamics rather than deeper magmatic migration ^[7][8].

Significant alert level elevation requires dramatic deformation and seismicity increases. For the U.S. Geological Survey (USGS) or similar agencies to significantly elevate a supervolcano's alert level, precursors would involve dramatic increases in deformation and major increases in seismicity ^[2]. The USGS does not rely on set thresholds for changing alert levels, instead opting for case-by-case decisions based on all available monitoring data ^[2].

Yellowstone and Long Valley show distinct precursor patterns. At Yellowstone, precursors to an eruption would include 'dramatic increases in the rates of deformation and obvious shallowing of the deformation source,' coupled with 'major increases in seismicity' ^[1]. Yellowstone's monitoring employs continuous GPS stations to measure caldera rim uplift or subsidence, and volcano notices report changes in deformation alongside located earthquakes ^[5][6]. An example from January 2026 noted subtle uplift and background seismicity, comprising 100 located earthquakes with the largest being M2.6 ^[5]. For Long Valley Caldera, unrest is characterized by recurring earthquake swarms and inflation of the resurgent dome, with geodetic deformation tracked using GPS and tiltmeters ^[3][4].

USGS alert levels define escalating eruption potential and hazards. USGS alert levels escalate from ADVISORY, indicating elevated unrest, to WATCH for heightened unrest with increased eruption potential ^[7][8][9]. The highest level, WARNING, is issued for a hazardous eruption that is imminent, underway, or suspected ^[7][8][9].

The U.S. Geological Survey (USGS) estimates a very low annual Yellowstone supereruption probability. Specifically, the USGS calculates the yearly probability of another caldera-forming Yellowstone supereruption at approximately 1/730,000, or 0.00014% ^[1]. This figure, however, is a crude average derived from limited historical data, and the timing of catastrophic supereruptions is inherently unpredictable ^[1]. The USGS also emphasizes that Yellowstone is not "overdue," as volcanic activity does not follow predictable schedules, and supereruptions are not evenly spaced in time, limiting the utility of probabilistic framing ^[3].

Prediction markets indicate higher supervolcano eruption probabilities by 2050. These markets suggest significantly higher probabilities for a supervolcano eruption occurring before 2050 ^[2]. For instance, one market shows a 'Yes' price of 31.5%, while another indicates a 14% 'Yes' for resolution before January 1, 2050 ^[2]. A direct numerical comparison between these market-implied probabilities and the USGS annual figure is conceptually sensitive. This is primarily because the USGS figure specifically pertains to Yellowstone caldera-forming events, whereas the market's scope likely encompasses any global supervolcano ^[1][2]. Furthermore, the USGS reiterates that eruptions are neither regular nor predictable, which complicates any direct comparison between these differing assessments ^[3][2].

Since 2020, satellite monitoring, particularly InSAR, significantly advanced in tracking pre-eruptive activity. Advancements in InSAR include improved revisit frequency and spatial coverage, now enabling sub-weekly observations and spatial resolutions of less than one meter, with global coverage provided by satellites such as COSMO-Skymed, TerraSAR-X, and ICEYE ^{[4][5][1][2][6][7][8]}. This technology can detect millimeter-level ground deformations, offering crucial information about magma movements, magma chamber pressurization, and localized deformation within craters that can signal shallow magma ^[3][2]. Algorithms such as LiCSAlert actively monitor volcanoes for unrest by detecting changes in deformation rates or styles, thereby improving forecasting, detection, and tracking of eruptive activity ^{[9][3][1][4][10][11]}. Additionally, multi-parameter analysis combines various satellite data on thermal anomalies, deformation, gas emissions, and ash plumes for deeper insights into complex volcanic processes ^[1].

Simultaneously, seismic sensor networks also show substantial progress in monitoring volcanoes. Modern broadband seismometers, strategically deployed, detect a wide array of seismic frequencies, including volcanic tremors, volcano-tectonic earthquakes, and long-period events ^[3]. AI and machine learning techniques are increasingly applied to seismic data analysis, enabling volcanologists to pinpoint pre-eruptive patterns and anomalies with greater accuracy and earlier warning ^{[3][12][13][14][15]}. Seismic tomography techniques now create three-dimensional images of a volcano's internal structure, mapping the location and movement of magma ^[3]. Improvements in real-time data processing, facilitated by parallel, cloud, and edge computing, allow for quicker analysis and dissemination of warnings ^[12]. Furthermore, earthquake sensors have improved the detection of infrasound from distances of tens to hundreds of miles ^[16], and the use of existing optical fiber networks for seismic monitoring is being explored, with sensors on these cables potentially relaying earthquake information faster than satellite communication ^[17][18]. Integrated monitoring networks now combine high-precision sensors, satellite technologies, and geochemical methods, utilizing AI and satellite imaging for enhanced early warning capabilities ^[9][3].

Several active volcanic systems globally possess the geological potential for a Volcanic Explosivity Index (VEI) 8 eruption. Such an event involves the ejection of over 1,000 cubic kilometers of volcanic material and could significantly impact global climate ^[2][6][7]. Beyond Yellowstone and Campi Flegrei, systems such as Toba, Taupō, the Altiplano-Puna Volcanic Complex, and Island Park Caldera are recognized for this potential. While Campi Flegrei has experienced massive eruptions, it has not produced a confirmed VEI 8 event ^[2][6]. Global risk databases, including the NOAA NCEI "Significant Volcanic Eruptions Database" and the Smithsonian Institution's Global Volcanism Program, aid in tracking and assessing these volcanic hazards ^{[8][9][10][11]}.

Past VEI 8 eruptions occurred globally at several sites. Toba in Indonesia produced the Younger Toba Tuff eruption approximately 75,000 years ago, registering as a VEI 8 and potentially a VEI 9 ^[1][2]. New Zealand's Taupō also experienced a VEI 8 event with its Oruanui eruption around 26,000 to 27,000 years ago ^{[3][4][5][2][6][7]}. In South America, the Altiplano-Puna Volcanic Complex, particularly calderas like Vilama (8.4-8.5 million years ago) and Cerro Guacha, has a history of VEI 8 eruptions ^{[4][5][12][5]}. The Island Park Caldera in the USA, associated with the Yellowstone hotspot, generated a VEI 8 eruption 2.1 million years ago, known as the Huckleberry Ridge Tuff ^[5][6]. Other notable systems with VEI 8 potential include Long Valley Caldera and Valles Caldera in the United States, both recognized as supervolcano sites ^{[13][14][12][14][4]}. Japan's Aira Caldera, formed by a massive eruption approximately 22,000 years ago, is an active complex ^[3]. Additionally, the Calabozos system in the Southern Andes Volcanic Zone and Okataina in New Zealand are considered large enough to potentially host a VEI 8 eruption and possess sizable magma chambers ^[15].

The Kalshi contract “KXERUPTSUPER-0” concerning whether a supervolcano will erupt before Jan 1, 2050, currently indicates an implied “Yes” probability of around 14% and a “No” probability of around 86% ^[1][2]. This market, which opened on Jul 18, 2025, will close at 2050-01-01T04:59:00Z and resolves to “No” if no eruption occurs by Dec 31, 2049, 11:59 pm ET ^[2]. Scientists generally agree that volcanic timing is inherently uncertain, and precursors for a catastrophic eruption, such as earthquake swarms or rapid ground deformation, would likely be detectable for weeks or even months to years beforehand ^[3].

Regarding Yellowstone, it is not considered “overdue” for an eruption based on its past events at approximately 2.08, 1.3, and 0.631 million years ago, averaging roughly 725,000 years between them ^[4]. However, the U.S. Geological Survey cautions that such eruption intervals are not meaningfully predictive and also highlights concerns about melt fraction and magma availability ^[4]. The most recent supereruption at Yellowstone, the Lava Creek supereruption, has a preferred 40Ar/39Ar isochron date of 631.3 ± 4.3 ka ^[5].

In a different system context, research on Campi Flegrei unrest offers a decade-scale pathway. One scenario-based forecast indicates that, with a maximum uplift/accumulation rate of 8×10^6 m3/y, the reservoir could reach eruption-compatible conditions within the next two decades, potentially extending into the 2040s. This, however, does not guarantee an eruption ^[6][7].