The shaking ground of Campi Flegrei

Solfatara. The earthquake swarms are focussed between here and Pozzuoli harbour in the distance

I loved Naples. It is a lively Mediterranean city where there is always something going on. The people are amazing. I was told (a long time ago) by someone whose wife was from there that they were visiting the city, and his wife went somewhere and had told him in no uncertain terms to stay with the car. Which he did, but not at the car -stretching legs, walking a bit, always within sight of the car. When he came back, sure enough, the car had been robbed – unseen by the watching wanderer. You have to admire them. When I was there I was told which particular street to avoid if I valued my life, and had no problems – perhaps because I did not bring a car! The city is sandwiched between two volcanoes: Vesuvius, an excellent tourist introduction to what a volcano should look like, and Campi Flegrei, looking nothing like that. As Hendrik pointed out to us many years ago, the danger is in the hidden one, in typical Neapolitan fashion looking for an opportunity to erupt unseen while in full sight, like that unknown Neapolitan robbing the car.

The most recent eruption in Campi Flegrei started off-shore before migrating to land. It produced a nice new mount on the coast, did a lot of damage to the nearby area, wiped out an old harbour and caused half the population of Naples to flee towards it to have a good look – while it was still erupting. This all happened in 1538, and is described in a recent post. This was the first eruption (as far as we know) in 3000 years, was relatively small and was preceded by months (or years or even decades) of earthquake swarms. There had been plenty of warning. The first recorded earthquake swarm was in 1470-1472. After that, there was inflation, just a few meters (!) at first but eventually reaching 14 meters. There were a few more earthquake swarms, but they accelerated about 18 months before the eruption and became damaging in the months before the eruption.

After 1538, very little happened. There were major earthquake swarms in 1582 and 1594, but that did not lead to eruptions. Nothing significant was reported for centuries afterwards.

Of course, there were other problems. The area suffers from uplift and deflation -bradyseism- of up to 12 meters – the old roman town is now that far below sea level. One of the Roman towns in the area was at Pozzuoli, in the heart of Campi Flegrei. The unstable leveling was possibly one reason why it didn’t age well. A new city was build further down the coast, away from the unstable ground. This was called ‘New City’ of NeaPolis, which became known as Naples. (Actually it was ‘new’ compared to a hill side settlement much closer to modern Naples, but it is a nice story.)

The instability came back to Pozzuoli in 1950, after centuries of rest, when there was 70cm of uplift over two years. And in 1969 locals again began to notice uplift on the shoreline, affecting bridges and wharfs. Measurements confirmed that between 1968 and 1970, there had been almost a meter of uplift. There were minor earthquake swarms in 1970/71, too small to be felt. Things quickly calmed down again. (There was a suggestion that some of the earthquakes that were detected later in the 1970’s were actually from fishing with explosives!) But in 1982, uplift resumed and in May 1983, earthquake activity restarted in earnest with a M3.4 event. Over the next two years, there were multiple swarms with events reaching M4.0. Uplift over this period reached 3 meters. People were evacuated. But again, things calmed down and by 1990, measurements showed some deflation. Obviously there was no problem. Never mind that this was first time since the 1538 eruption that there had been such uplift and earthquake swarms – Naples deals with the present, not the past, and whatever happened a decade before can not be important.

But it did not last. Subsidence came to an end in 2005, and over the 20 years since, activity has increased – slowly enough for people not to readily notice, but the INGV (Osservatorio Vesuviano, with a notable naming focus on only one of the volcanoes under its remit) was on the case. They have installed many monitoring stations and produce regular bulletins. Over the past few years, casual observers have also been able to notice the changes. Solfatara is more active. Earthquakes are becoming more frequent. The sea is falling.

Let’s first look at the region. The map below shows the Naples area with Vesuvius, Pompei (of course) and other famous locations. The red oval shows the (very) rough outline of the overlapping calderas of Campi Flegrei.

The next map zooms in on this caldera. It shows the location of the volcanic features (old eruptions) of this region, all of which happened after the last major explosion a very long time ago. The 1538 eruption, Monte Nuovo, is number 9 on the map.

The must-have map of volcanic features of Campi Flegrei. Click on image for full resolution. Monte Nuovo is number 9. The current seismicity is centred around number 25. Solfatara is number 23, just above it.

Earthquake swarms

The earthquake activity after 2005 was initially minor, and shallow at 2 km depth or less. From 2012 it became more intense and since 2020 it has really taken off. The plot below comes from the monthly bulletins of the INGV and includes events up to April 2024 (so not this month’s swarms). It shows the rapid increase in number, and also the large increase in energy since 2023. The depth of the earthquakes remained initially around 2 km but more recently has also occurred at a depth of 4km. Much of this bradyseism, by the way, will be due to the sloshing around of underground heated water/steam. The hydrothermal system extends to 2-3 km below the surface and sees the effect of extra heat.

The swarm in October 2023 saw a M4.2 event, possibly (?) the largest Campi Flegrei earthquake for centuries. May 2024 saw this surpassed with an M4.4. (It was advertised as the ‘largest since 40 years’ but the largest one 40 years ago was ‘only’ an 4.0)

Earthquakes are mainly from three locations. One is an apparent fault in the middle of the bay, slightly curved. The second is the Starza fault along the coast, stretching from Pozzuoli eastwards. More recently, the earthquakes have been along only part of this fault, at the eastern end near Solfatara. Note that earthquakes in swarms can be hard to localize precisely.

High quality volcanic-tectonic earthquakes for the period 2005–2019. The size of the epicentres are shown as a function of the magnitude according to the legend on the bottom right.

What is the significance of the depth of the earthquakes? The events at 2 km occur at a calcium-rich cap layer. This is a thin rock cap later that separates the sedimentary layer below from the volcanic deposits (pyroclastics) above. Gasses and fluids rising from below get stuck against this cap: this hydrothermal stress bends the cap elastically and this causes the earthquakes. 2 km is about the depth where water can become steam, so it is likely that this earthquake zone is due to water vapour collecting below (and breaking through) this cap.

The sedimentary layer goes down to about 3km. Below this is the bedrock, affected by the heat from below, with another cap-rock separating the two. Since 2020, more of the earthquakes have been taking place around this deeper cap or even the bedrock below it. This may be mainly caused by CO2 degassing of magma. Magma intrusions have tended to stay within the bedrock and solidify there. The liquid magma is deeper down, at 7 to 8 km. (There is also a much deeper magma reservoir.) If magma is rising in significant quantities, it is likely that we will see earthquakes to this larger depth of 7-8 km, which we do not do at present. It is however possible that some magma did rise to 3 to 4 km in recent episodes: it has been proposed that the pressure increase was caused by the emplacement of new sills in 1950, 1970 and 1980. However, there was no proof that magma has collected at this depth – but see below.


The current activity is comparable to the events of 1983 and 1984, but with a much longer build-up to the activity. Since 2016 there has been uplift of around 1 meter at Pozzuoli (1.2 meters since 2005). The figure below shows uplift since 2021, accounting for half the total since 2016. It is focussed right at the coast, close to Solfatara. (It is interesting that the strong increase in activity in 2020 happened when inflation had recovered the highest surface level reached in 1984, although this may be just a coincidence.)

During the current inflation period, there has also been a slow change in the gas emissions at Solfatara. The fraction of CO2 over H2O has been steadily increasing, and there is also a bit more CO. This indicates that the gas emissions are coming from a reservoir that is getting a bit hotter, from 215C before 2005 to around 245C at present. There was a spike to 265C in 2020 – one wonders whether the reservoir caught covid. The temperature at which the gas reaches the surface has increased only a little, though. But the site is now considered too dangerous for tourists and it has been closed since 2019.


Seismic mapping has recently shown (a paper to be published in July – see references) that underneath Solfatara, the depth at which water becomes steam has grown shallower since 1984. This can be explained by an increasing temperature of the water, and provides independent evidence for this change.

Location of the reservoirs “A’, B’ and ‘C’. Locations are quite approximate, read off from

The same seismic mapping has also shown that there are two somewhat shallow reservoirs within the caldera. One (‘A’ on the map) is located underneath the bay at some 2.5 km depth, and the other (‘B’) under land at some 3.5 km depth near Solfatara. After 1984, the reservoir underneath the bay appears to have cooled, while the one near Solfatara has grown hotter: there has been a change in the pathway used for the heating. Neither reservoir appears to contain significant amounts of magma, though. Think hot water.

As always in volcanic regions, areas without earthquake are possible carriers of magma. The recent mapping has shown such a zone at a depth of 5 km not far from the shore line, which has risen to 4.5 km depth since 2019: this is zone ‘C’ on the map. Interestingly, this zone did not exist in 1984. A possible interpretation is that perhaps 15 years ago a bit of magma migrated up and formed this new sill. This may be what has been driving the sharp increase in seismic activity since 2019.

Current events

In the last few days, there has been a strong, though fairly brief, earthquake swarm. The swarm followed a minor swarm four days earlier and two swarms in April. The current series started just before 6pm (UTC) on May 20 and culminated in the evening with a strongest reported event at M4.4 – the precise value may be subject to change. More than 150 quakes were measured. After midnight the swarm calmed down, although some events were still continuing and around 4pm (UTC) on the 21st, they picked up again a bit with a minor burst. From a report in a local news site:

The seismic swarm that has been underway in the Campi Flegrei since yesterday ‘has slowed down its pace, but it is not over yet’, volcanologist Mauro Di Vito, director of the Vesuvian Observatory of the National Institute of Geophysics and Volcanology, told ANSA.

There are no signals that allow us to say whether and for how long the swarm will continue nor whether this seismic swarm will be followed by others, but ‘given the deformation of the ground that is affecting the area it is clear that we also expect other events’, he said.

‘In the meantime, we are working to collect new elements: ‘we are making measurements of all possible parameters’ and we are implementing the monitoring network in areas that have been less neglected so far.’

All phenomena linked to the condition of the magma are also monitored: ‘gas analyzes are highlighting an increase in temperatures and pressurisation of the upper hydrothermal system, with values ​​of the gas emitted equal to 4,500 tonnes of CO2 per day emitted by the sulphate system in locality Pisciarelli’. Similar measures have been extended in the gulf.

In the Vesuvian Observatory, meanwhile, researchers and technicians have been working for 24 hours now: ‘the staffing pace is becoming intense, there is great pressure and we would need more staff’. Many INGV researchers spontaneously went to help the workers in the Monitoring Room, where earthquakes continued to be recorded during the night with a frequency of 10 or 20 seconds.

The director of the Vesuvian Observatory Mauro Di Vito in the monitoring room of the INGV in Naples. Photo by Luigi Avantaggiato. Source:

Locations of earthquakes in the May 20 swarm, between Pozzuoli and Solfatara. Source


What will happen next? That is impossible to predict. INGV is reporting that no change is being seen in the inflation rate. That already suggests this is just part of the longer term activity. There is no sign of imminent changes.

Models of the number of earthquakes suggest that we are in a regime where strain is increasing. That may resolve itself in a rupture, which can be gentle (an opening forms in the crust which lets gasses escape, as may have happened in 1984) or instantaneous, i.e. where a fault gives way in a larger earthquake. The fault is not particularly long and therefore such an event would not be that strong, perhaps M5, but a shallow earthquake of such size in a densely populated area could still be damaging.

There is no indication of an impending eruption. But of course, this caldera has history. If we get a series of earthquake swarms which are propagating in depth and location, then a dike may well be forming. Otherwise, if a fracture grows in the crust, then the hydrothermal system may find its way more easily to the surface. As the activity is close to Solfatara, the likely escape path is there. In that case, phreatic explosions may occur. It seems best that Solfatara remains closed to tourists.

Should Pozzuoli be evacuated again? That is not for us to decide! But it is a difficult one. It can only be done once. If an evacuation is ordered and nothing happens, many people will not leave next time. To evacuate half a million people may take three days. In 1538, combined with modern instrumentation, this would have given plenty of time. So it may be best not to fire off the evacuation order until INGV can see the white in the eyes of a forming dike. In that case the biggest risk is phreatic explosions which can occur without warning.

But at the moment, it is time mainly to call for calm (also for the earthquakes) and get on with life. It is what Naples does best.

Albert, May 2024

Further reading,175%20cm%20in%201982–84.

470 thoughts on “The shaking ground of Campi Flegrei

  1. On Husafell live webcam I’ve noticed that there is a pulsating of degassing:
    From time to time you get a big white plume of gas that hides the eruption like a cumulus cloud. Then the gas cloud disappears and lets us see the eruption clearly. Althoug the effusive lava eruption looks very steady, it’s possible that we get an eruption with different degassing phases and related eruption phenomena. Fagradalsfall 2021 showed how this can look like, f.e. the lava geysir, but also pulsating effusive activity. The lava/magma now looks very gasrich. So it’s possible that we get a more gas-driven eruption style different to the eruption during April.

    • At the same time there were two earthquake swarms today close to Reykjavik in the direction towards hengill: “Hellisheiðarvirkjun”

      The first swarm began around 9:30, the second after 13:30. Does this area belong to Hengill or Brennisteinsfjöll?

      • If there they get fires like Grindavik, it would be bad for the capital and would likely interrupt the road towards the east.

      • That’s on the Hengill system. Specifically on the southwest edge of its central heights. Dikes don’t erupt there, rather to the north and south. The earthquakes look to be centered about a mile west of the starting point of the southern rift system. That mostly flows south to the sea, but some portion could potentially slip north of Brennisteinsfjöll in the direction of Reykjavik. Realistically Brennisteinfjoll is more of a risk for that, has bad habit of flowing into southern suburbs of Reykjavik. Hengill though would sever the southern road into Reykjavik. Its northern rift zone is also probably a bigger risk for a maar type eruption then Krysuvik, which has not had eruptions right around the actual lake since the ice age.

  2. Google Earth is good stuff as one can explore any volcano you want instantly and learn features and landforms on them. I have noticed that they are huge general diffrences between alkaline and thoelitic fluid fissure vents and cones in morphology. Alkaline volcanism almost always produce cinder cones, while thoelitic vents produce spatter cones, even huge thoelitic vents haves a spatter like quality unless the fountains gets really tall. This is clearly visible in Hawaii where the thoelitic main shield volcanoes generaly does not form alot of cinder cones, unless the fountains gets very tall and focused like at Puu Oo and Iki, while the alkaline postshield stuff produce cinder cones at nearly every fissure.. obivious its the diffrence in gas content between alkaline and subalkaline mafic magmas. Hawaiian shield lavas are very gas rich as well and specialy so for sulfur.. but alkaline magmas haves higher co2 and higher water content and haves a high sulfur content too. Magmatic water is defentivly a real rocket gun in magmas. Alkaline fissure cones tends to be of cinder / scoria type and these are incredibley common on alkaline areas and Islands where they can dot entire landscapes

    La Palma was a spectacular example so gas rich it have gas vents near the top of the cone at near supersonic speed a real blowtorch, while very fluid lavas issued at same time from vents in its base. Even the very fluid lavas in Congo produce paricutin like cones due to high gas content making taller fountains and therefore more and cooler fragmentation than at Iceland and Hawaii, La Palma 2021, Fogo 2014 basanites looked similar fluid the near vent flows but had lava fountains that looked strombolian/ almost subplinian in look.. high gas content. Strongly alkaline open lava lakes are also more churned up than subalkaline ones, many mildly alkaline subduction lavas also have alot of gas due to recycled slab volatiles ( Marums lava lakes ). And indeed monogenetic cinder cones are very common too among subduction magmas partly due to the volatiles but many almost all subduction magmas are also more viscous than oceanic basalt. Cinder and scoria cones are also the norm in low sillica alkaline continetal fields and those tends to be very gas rich too.

    Truely high alkaline eruptions are also rather very rare ( well Nyiragongo and Nyiramuragira are more or less non stop erupting ) but these are open conduits and constantly degassing so not typical of what it may look like. Nyiramuragira before 2015 when it was more closed as a system had many fissure eruptions and they looking at photos does not look like thoelitic fissure vents at all despite being just as fluid.. the lava fountains are taller and more fragmentation and more cinder cones. La Palma pretty much confirmed that behaviour and morphology, among alkaline vents and even more alkaline like Melilitites and Kimberlites stuff just blows up into maars

    Its hard for me to learn stuff and think abstract likey due to past lead poisoning from making home made lead oxide batteries without good ventilation or any protection, but Im trying to write a tourist New VC post on a volcanic subject .. and was possible to pass high school years ago

  3. HVO put out some video of the eruption. It looks like fresh magma not stuff stored at shallow depth for a few months, I guess there was just an aseismic dike.

    It also looks like the fissure system is still growing southwest, and might have stopped and restarted a few times over by now too. No maps of the actual flow field yet but it doesnt seem to have been something of a flash lava flood like 1974.

    Its pretty much doing what I proposed a few months ago. Intrusion late January, refilling to pressure point and then eventually failing in the same place leading to an eruption this time. The rifting events in this area before now had always been isolated primary intrusions that were never followed up because activity moved to the ERZ, with 1974 being a rare case of the primary dike erupting first time. But not this time, the rift was reused. Which makes it now quite likely for more eruptions to break out here in the next few months to years.

    • HVO are suggesting the eruption is being fed by magma from the previous dike. It’s also very clear this is only a minor eruption so far, maybe comparable to the early phase of the Geldingadalir eruption.

      • Yes I read that too just after posting. I also just read it could be over now but its never so easy to tell at this point.

        Its also then pretty likely we get follow up eruptions in this area until an eruption breaks out somewhere lower down. Or that this eruption keeps opening fissures further southwest, the last dike reached as far southwest as the north side of the Kamakaia Hills area so if the dike us basically being pressurized again anywhere between these areas could see lava.

          • Yes that was my thought too. I guess the seismic SWRZ can in fact do slow eruptions. Maybe only the 1823 flow was a Halemaumau drainout and the other SWRZ eruptions in the 1800s were all ‘true’ rift eruptions after all.

            I guess that all depends on how this new rift acts in the coming years. A lot of the 1960s eruptions were small but got a lot bigger and eventually spawned Mauna Ulu, and tangentially Pu’u O’o too. I dont think something that big is going to exist in this spot in 2035 but I would be surprised if this is it with the current info.

      • That was my feeling. Of course the original source is the same, so it will be the same composition. I don’t expect this to become a major eruption. But it is interesting that this has not happened for almost 50 years

    • The eruption was over after 12 hours. A bit like Mauna Loa’s 1975 short eruption.

Leave a Reply

Your email address will not be published. Required fields are marked *