Beneath the western edge of Naples, the ground has been rising. Since 2005, the floor of the Campi Flegrei caldera — a restless volcanic basin ringed by neighborhoods that are home to hundreds of thousands of people — has swelled upward, breaking into earthquake swarms that rattle Pozzuoli and its surroundings. In the spring of 2024 the unrest reached a new pitch: on 20 May 2024, at 18:10 UTC, a magnitude 4.4 earthquake struck near the Solfatara crater, the strongest instrumentally recorded event the caldera had ever produced [source: Communications Earth & Environment, 2025]. It was later matched and then exceeded — another magnitude 4.4 on 13 May 2025, and a magnitude 4.6 offshore on 30 June 2025 [source: Civil Protection, 2026].
Coverage of these events tends to arrive wrapped in a single word: supervolcano. The word does real work — it makes you pay attention — but it also quietly promises something the science cannot deliver: that we know what happens next. The honest story of volcano monitoring is more interesting than the headline, and it turns on separating three things the word "supervolcano" tends to fuse together: what instruments actually measure, what those measurements mean for the chance of an eruption, and what has been verified versus merely feared.
The ground is moving — and we can measure it precisely
Start with what is not in doubt. The unrest at Campi Flegrei is real, and it is measured with remarkable precision. The phenomenon has a name — bradyseism, "slow movement" — and it describes the way the caldera floor slowly inflates and deflates as pressure changes beneath it. The current inflation phase began in 2005 and has not stopped [source: Civil Protection, 2026].
The numbers are concrete. At Rione Terra in Pozzuoli, the point of maximum ground deformation, the cumulative uplift reached roughly 163.5 centimeters by the end of March 2026, with about 25.5 of those centimeters accumulating since January 2025 alone [source: Civil Protection, 2026]. A peer-reviewed reconstruction put the total at around 130 centimeters by July 2024 — already almost 40 centimeters higher than the peak reached during the previous major crisis of 1982–84 [source: Communications Earth & Environment, 2025]. The rate of rise is not steady; it has at times exceeded 30 millimeters per month before easing back toward 20 millimeters per month or less [source: Civil Protection, 2026].
These figures come from the same toolkit that watches every well-instrumented volcano on Earth. The U.S. Geological Survey groups the measurements into a few families: ground deformation, tracked by GPS stations and satellite radar (InSAR) that can detect a caldera swelling by centimeters; seismicity, the earthquakes and tremor that, as the USGS puts it, almost always precede an eruption; and volcanic gases, whose changing chemistry can betray magma or hydrothermal fluids on the move [source: USGS, 2025]. Read against a known background, a sustained departure from normal is the signal observatories are built to catch.
So the first layer is settled. We can see the ground move, count the earthquakes, and sample the gas, and we can do it to a precision that would astonish a geologist of a century ago. The hard part is not measurement. The hard part is meaning.
Why measuring is not predicting
Here is the distinction that most coverage collapses. Measuring unrest is not the same as predicting an eruption, and the people who do this work are careful to say so. The USGS describes its own product as a forecast, not a prediction — a probabilistic statement about what is likely and where the hazard would fall, not a deterministic claim about the day, size, and place an eruption will occur. Those forecasts, the agency says plainly, are "becoming much more reliable, but they remain imperfect" [source: USGS, 2025].
The gap is not a failure of instruments; it is the nature of the system. A rising, shaking caldera is a body under stress, and stress can resolve in more than one way. It can build toward an eruption. It can also plateau, or relax, and subside without one. Campi Flegrei is itself the proof: during the 1982–84 crisis the ground rose by nearly two meters and the caldera did not erupt [source: Civil Protection, 2026]. The current episode has been climbing for two decades, through thousands of earthquakes, and has not erupted either. Unrest is something like a necessary ingredient of an eruption without being a sufficient cause of one. It correlates with eruptions across the geologic record; it does not, by itself, cause the next one.
This is why the honest forecasters lean on probabilities and comparisons rather than dates. The clearest success story, Mount Pinatubo in the Philippines in 1991, did not come from a formula. It came from watching seismicity, deformation, and gas all escalate together, fast, at a densely monitored volcano, and reading that convergence correctly in time to evacuate and save thousands of lives [source: USGS, 2025]. The lesson volcanologists actually took from Pinatubo was not that eruptions can be predicted like eclipses. It was that a well-monitored volcano in a rapid run-up can sometimes be forecast well enough to act — which is a more modest and more useful claim.
What is actually driving Campi Flegrei
The correlation-versus-causation problem gets sharper when you ask why the ground at Campi Flegrei is rising. The intuitive answer — magma is pushing up toward the surface — turns out to be the one the evidence does not clearly support.
The 2024 magnitude 4.4 earthquake became a natural test of the question, and the researchers who analyzed it reached a specific conclusion. Writing in Communications Earth & Environment, they found that pressurized fluids — hot water and gas in the shallow hydrothermal system — are "largely preferred" as the driver of the present uplift, and that those fluids "play a crucial role in the seismogenic process, by increasing local stress and reducing rock strength" [source: Communications Earth & Environment, 2025]. In plainer terms: the swelling and the earthquakes can be explained largely by gas and water pressurizing the rock a couple of kilometers down, not by fresh magma climbing toward an exit.
Italy's Civil Protection Department frames it compatibly. Officials point to deep-seated magma as the ultimate root cause — the heat engine that keeps the whole system charged — while stating there is "no proof of rising magma" toward the surface [source: Civil Protection, 2026]. The two accounts fit together rather than clashing: deep magma supplies the heat and gas, pressurized fluids in the shallow system produce the uplift and the quakes, and no one is observing a shallow magma pathway opening toward an eruption. That is why the caldera has sat at a yellow alert — "attention," the second of four levels — with scientists judging that further damaging earthquakes are very likely even as an eruption is not judged imminent [source: Communications Earth & Environment, 2025]. The most probable near-term hazard at Campi Flegrei is a strong earthquake, not lava.
"Supervolcano" versus the actual odds
Then there is the word itself. Campi Flegrei earns the "supervolcano" label from deep in its past: roughly 39,000 years ago it produced the Campanian Ignimbrite, one of the largest European eruptions in the last 200,000 years, rated VEI 7 [source: Global Volcanism Program, 2025]. That is a genuinely enormous event, and it is the reason the caldera is famous. But it also invites a specific confusion, because the formal definition of a "super-eruption" sits a full step higher still.
The USGS reserves the term for a magnitude VEI 8 eruption — one that expels more than 1,000 cubic kilometers of material [source: USGS, 2025]. The Campanian Ignimbrite, for all its scale, was VEI 7, below that bar. The point is not to split hairs about classification; it is that "supervolcano," in popular usage, smuggles in the worst imaginable case and attaches it to a caldera whose realistic near-term behavior is nothing of the kind.
How unlikely is the worst case? The USGS has put a number on the closest analogue, Yellowstone, and the number is instructive precisely because of how the agency handles it. The yearly probability of another caldera-forming eruption there is about 1 in 730,000 [source: USGS, 2025]. And rather than present that as a precise forecast, the USGS undercuts it in the same breath: the figure comes from simply averaging the intervals between three past eruptions, which "is hardly enough to make a critical judgment," because "catastrophic geologic events are neither regular nor predictable" [source: USGS, 2025]. The most likely future eruption at Yellowstone, the agency adds, is not a supereruption at all but a lava flow, and there is no evidence a catastrophic one is imminent [source: USGS, 2025].
Campi Flegrei carries its own reality check in the historical record. Its only eruption in recorded history was Monte Nuovo, over roughly a single week from 29 September to 6 October 1538 — a small event that built a tuff cone about 123 meters high, after years of the same bradyseismic uplift being measured today [source: INGV, 2025]. If the caldera were to erupt in the near term, that localized, modest scale is the realistic reference point, not the caldera-forming catastrophe the word "supervolcano" conjures. The worst case belongs to the geological record and to modeled scenarios; it is not a forecast of what 2026 holds.
Iceland: monitoring at its best, and still only hours of warning
For a sense of how good volcano forecasting can get — and how bounded it still is — look north to Iceland. On the Reykjanes Peninsula, along the Sundhnúkur crater row near the town of Grindavík, a run of nine eruptions unfolded between 18 December 2023 and early August 2025, when the ninth ended around 4–5 August [source: Icelandic Meteorological Office, 2025].
This is nearly an ideal case for monitoring. The magma is basaltic and relatively shallow, the plumbing is comparatively simple, and the Icelandic Meteorological Office watches it continuously with satellite radar and GNSS. Between eruptions, the instruments record magma reaccumulating beneath the Svartsengi area and the ground steadily rising again; each new eruption is heralded by an intense, short seismic swarm as magma cracks its way toward the surface [source: Icelandic Meteorological Office, 2025]. Forecasters can say, with real confidence, that another eruption is likely and roughly where it will break out.
And yet even here the warning of the exact moment often comes in hours, not weeks. If Iceland — shallow, basaltic, simple, exhaustively instrumented — yields only short notice of the precise event, then Campi Flegrei, with its deep and fluid-dominated system and its ambiguous signals, is harder still. Better monitoring narrows the uncertainty; it does not abolish it.
What to watch
So can we predict a supervolcano? The honest answer is that we can measure one with extraordinary precision, forecast its hazards in probabilities, and act on those forecasts when a run-up is fast and clear — and that this is not the same as naming the day. At Campi Flegrei the verified picture is sober rather than apocalyptic: real, measured uplift; the strongest earthquakes in the instrumental record; a system that the evidence ties to pressurized fluids more than to rising magma; a yellow alert whose most likely near-term hazard is a damaging quake, not an eruption; and, if an eruption did come, a Monte Nuovo far more plausible than a Campanian Ignimbrite.
Three things are worth watching from here. First, the signals themselves — whether deformation and seismicity keep escalating or ease, and above all whether gas chemistry ever shows fresh magma moving toward the surface rather than fluids venting from below. Second, the alert level, which is the distilled judgment of the scientists closest to the data; a move from yellow toward orange would mean the experts, not the headlines, had reweighted the odds. Third, the language: the difference between a report that says the ground rose by a measured amount and one that says an eruption is coming is the difference between what we know and what we fear. Volcano monitoring, at its best, is the discipline of holding the first tightly and the second lightly.