Volcano Forecasting: Can We Predict Eruptions Like Weather?

The Quanta Podcast examines whether volcano forecasting can resemble weather forecasting. Host Hannah Waters talks with volcanologist Robin George Andrews about why predicting eruptions is hard, what magma is doing underground, and how new data collection and AI might uncover the physics of volcanic activity.

• Understanding volcano unpredictability: magma lies underground and eruptions vary widely
• Forecasting is probabilistic, not certain, relying on signals and odds
• Advances in remote sensing, seismic data, and geochemistry could reveal underlying physics
• Optimistic path toward archetype based forecasts that improve safety for nearby populations

Overview

The podcast features a conversation between host Hannah Waters and Robin George Andrews, a volcanologist and science writer, about the feasibility of forecasting volcanic eruptions in the way weather is forecast. The discussion surfaces the core idea: while weather forecasting benefits from constant access to atmospheric data, volcanoes operate in a largely unseen interior where magma can reside for long periods, making precise predictions difficult and often probabilistic rather than certain.

Why Forecasting Volcanoes Is Difficult

The guest explains that magma spends most of its life underground, and scientists cannot directly sample magma deep beneath the surface. Volcanoes are idiosyncratic; while they obey the same fundamental physics, each system can respond differently to similar triggers. There is no single red alert that tells you an eruption is imminent; some volcanoes erupt after centuries of quietude, others remain noisy for long periods before a major event. The fundamental problem centers on pressure, heat, and gas, and how these factors interact with magma viscosity and depth to determine eruptive style.

What We Know About the Subsurface

The discussion covers magma as the planet’s pressure release, with magma chambers that may be complex, possibly nested, and not a single cavern. Observations indicate that magma reservoirs could be more like a sponge with porous networks and crystal matrices than a simple pocket. Depths range from a mile to tens of miles below the surface. Geologists have glimpsed magma-related processes through indirect means, such as geothermal testing and rare drill events, but direct access remains limited and difficult to interpret.

How Eruptions Are Triggered

The big picture is that eruptions are driven by pressure differentials and the ability (or inability) of magma to trap gases. If magma becomes sufficiently hot, gas builds up, or structural failure releases the confining pressure, magma can move toward the surface. The eruption style depends on magma viscosity and gas content: runny magma tends to produce lava flows, whereas sticky magma can produce explosive eruptions. Because of this complexity and the diversity of volcanoes, attribution of a specific eruption to a single cause is often not possible in advance.

Forecasting Methods and Data

Forecasting involves monitoring a volcano's “normal symphony” of signals: earthquakes from rock fracturing, ground deformation measured by satellites and GPS, and gas emissions as magma rises and degasses. For volcanoes that erupt frequently, researchers build probabilistic forecasts based on observed patterns. Remote sensing has improved monitoring for less accessible volcanoes, enabling deformation measurements from space even when on the ground sensors are sparse or compromised by lava flows.

The episode highlights a research effort led by the Universities of Bristol and East Anglia that uses dense sensor networks and machine learning to identify causative links in volcanic activity rather than simple correlations. The aim is to move beyond post hoc explanations toward physics grounded in robust data, bridging observations with physical models and simulations.

Challenges, Opportunities, and the Road Ahead

Despite advances, there is no pan volcanic forecasting solution. A major challenge is the need to extract reliable physics from noisy, sparse, and diverse data. Scientists hope to develop archetypes of volcano behavior (for example, Hawaiian style eruptions versus Mount St. Helens style explosive events) that can be generalized across contexts. The potential is significant: better forecasts could inform evacuation planning and save lives, but it requires sustained investment, collaboration across nations, and careful management of public expectations.

The conversation ends with a nod to a broader cultural question about time and science, recommending Jo Marchant’s talk/book In Search of Now for a perspective on the present moment, science, and perception.

Takeaways and Looking Forward

The episode paints an optimistic but cautious path toward volcano forecasting: a combination of dense instrumenting, AI driven pattern recognition, and physics based modeling could eventually yield reliable probabilistic forecasts and reduce risk for people living near volcanoes.

Guest Recommendation

Robin encourages listening to Jo Marchant’s talk on the philosophy of the present moment, a topic that resonates with the perception of time in science and the human experience of forecasting and risk.