New article in Science: Still difficult to forecast volcanic eruptions

In a new article in the journal Science, Magnús Tumi Guðmundsson, professor of geophysics at the Institute of Earth Sciences at the University of Iceland, and Christopher J. Bean, professor at the Dublin Institute for Advanced Studies, say that despite significant advances in recent decades in monitoring volcanoes, scientists are far from being able to forecast exactly when eruptions will start, their duration or magnitude.

The article belongs to the class of perspectives articles in Science, titled "Forecasting volcanic eruptions across scales", written at the request of the editor of Science, one of the most prestigious science journals worldwide. In the article, Magnús Tumi and Bean discuss one of earth sciences largest challenges: if, and then how, it’s possible to forecast volcanic eruptions, not only whether an eruption is imminent, but also their magnitude, evolution and impact.

The authors argue that volcanoes are among the most complex natural phenomena on Earth, and even though it is often possible to see that an eruption is likely, predicting its size can be very difficult. They point out that eruptions range from small effusive eruptions to massive, catastrophic eruptions that impact the climate and communities worldwide. Luckily, those kinds of eruptions are rare but can have a significant impact when they do occur.

Magnús Tumi is well-known in Iceland and has been among the country’s leading geophysicists and volcanologists for decades. He has repeatedly been called upon by the government and media to assess the status and development of volcanic eruptions in Iceland.

Iceland plays a key role

When it comes to volcanic activity and eruption forecasts, Iceland is prominent in the article. Iceland is effectively a living laboratory in geophysics, seismology and volcanology where volcanic activity, plate-boundaries and glaciers meet, making them impossible to ignore. Icelanders literally live on the boundary between two tectonic plates, and thus, volcanic eruptions have shaped both the land and the Icelandic nation for centuries.

The authors point out that volcanoes can give very different warnings. Some show clear signs of impending eruption for weeks or months, such as earthquakes, ground deformation and changes in geothermal heat. The warning signs for the eruption in Eyjafjallajökull in 2010, e.g., were relatively clear when compared to signs from many other volcanoes — and are often mentioned as an example of how closely monitored volcanoes can give scientists warnings about impending eruptions before they start.

The article also notes that other volcanoes can erupt at a very short notice. Hekla is specifically mentioned as an example of volcanoes where short-term precursors have sometimes lasted less than an hour before eruption. This makes eruption forecasting particularly difficult and underscores the importance of continuous monitoring.

The article also discusses recent events on the Reykjanes Peninsula in 2023, when a large magmatic intrusion (a dyke) formed in a few hours without an eruption. Later, comparable magmatic intrusions led to eruptions. This shows that large-scale major unrest related to magma movement in the Earth’s crust does not always automatically lead to an eruption.

Great advances in monitoring

Magnús Tumi and Bean argue that advances in technology have transformed how scientists can monitor volcanoes. They can now use seismographs, GPS stations, satellite and radar measurements, and new fibre-optic sensors that can detect tiny movements in the Earth’s crust in real time. Those kinds of measurements can provide important information about magma accumulation beneath volcanoes and support risk assessment.

Despite all these advances, according to the two scientists, it is still difficult to estimate how much magma lies beneath volcanoes and how it will behave. For example, even though clear warning signs were detected before eruptions in Holuhraun in central Iceland in 2014-2015 and in Hawaii in 2018, it was not possible to foresee the caldera collapses that occurred in Bárðarbunga in 2014-2015 and in Kilauea 2018. Consequently, the eruptions were an order of magnitude larger than originally predicted. The uncertainties these events reflect are one of the main limitations to current eruption forecasts, according to Magnús Tumi.

Large datasets and advances in analytical procedures could improve forecasting

Both Magnús Tumi and Chris Bean think that further progress in forecasting volcanic eruptions will be based on combining various methods. It would be important to combine the experiences from earlier eruptions, real-time monitoring data, physics-based models, and new methods of data analysis, e.g., artificial intelligence. Thus, it may be possible to detect behavioural patterns in volcanoes that are difficult to detect with conventional methods alone.

They point out that uncertainties will always accompany eruptions because of their complexity.

Important for future communities

At the end of the article, the scientists emphasise that a better understanding of volcanoes isn’t important only from a scientific perspective, but it is key for community safety. The article further reflects on how scientists have, for years, not only argued for the importance of robust monitoring and research to mitigate the effects of natural hazards but have also demonstrated that such work delivers clear results through innumerable examples.

The authors say that significant advances in volcano monitoring have occurred concurrently with rapid technological changes. Modern measurement methods using seismographs, GPS equipment, satellites, and other real-time systems have greatly improved scientists' ability to monitor changes in the Earth’s crust and assess the course of events.

Better volcanic eruption forecasts can thus aid authorities and the public in responding sooner to natural hazards and in mitigating risk to people and financial losses from seismic and volcanic activity.

The article by Magnús Tumi Guðmundsson and Christopher J. Bean was published in Science on May 7 2026.