Rhyolite has more silica, this makes it more viscous, more explosive and in turn more dangerous. Basalt is the opposite, fluid, well-behaved, safe. This could be a phrase out of any geology textbook, I can almost feel some readers getting ahead of me and thinking what I am obliged to say. But there are exceptions!
Exceptions don’t make the rule unimportant, because it is a really important rule. As you turn up the silica, andesite, dacite, rhyolite it becomes increasingly violent until you get to the pinnacle of volcanic explosivity, Taupo, Toba. Yet for a volcanologist, or anyone who wants to dive deep into volcanic knowledge, being aware and understand all the possible situations is key. This is why this series is about the exceptions.
I bring to light an eruption mechanism that has so far been in the shadows. I don’t know of anyone who has described it before separately, which it deserves to be, or explained the way it works, this is possibly because they are rather unusual events, I have only been able to confirm 3 cases since 1900. It is possibly the pinnacle of basaltic explosivity, nothing compared to what a silicic caldera can do, but still quite dangerous, deadly eruptions.
For this kind of eruption it is first important to understand well how caldera collapses happen. Volcanic calderas form in mighty events where a magma reservoir empties and the roof comes falling down over it. This is not exactly slow or fast, it is best described as episodic, the roof falls in series of steps known as collapse-events, each collapse-event generates a peculiar earthquake with very long period waves (VLP) and a focal mechanism called CLVD. The focal mechanism is created by the ring shape of the caldera fault that generates the earthquake. With each time, the roof drops on the top of the reservoir and repressurises it so that it will result in a surge on the ongoing eruption/intrusion.
Kilauea demonstrated dramatically how this works in 2018 with a total of 62 collapse events, each a M 5.3 earthquake. But the step-like nature of a caldera formation was well know before, Bardarbunga in 2014, Piton de la Fournaise in 2007, every single caldera collapse observed has taken place this same way, even in silicic volcanoes. With Katmai in 1912 it formed in 4 earthquakes of M 6.5-7.0.
There are 2 main ways a caldera can collapse. The first is by explosive eruption which drains the reservoir upwards, this is reserved for silicic volcanoes with high viscosity magmas, dacite, rhyolite and trachyte, once the collapse is ongoing more primitive stuff can get flushed out. The other is when a reservoir drains laterally to feed a dyke intrusion or a lateral eruption, this is far more common in basaltic volcanoes but silicic systems can do it as well.
With that part explained we are off to Galapagos to make a shield volcano blow up.
Fernandina rises above the waves of the Pacific Ocean as a classic Galapagos shield, with that overturned soup bowl shape topped by a big caldera that is so popular in Martian volcanoes. This is the fiery heart of the archipelago, the most frequent eruptor and closest volcano to the hotspot. As a good shield volcano its eruptions are effusive, gentle, from fissures, and any other day Fernandina would have just done that. But the 11th of June of 1968, it decided to throw an enormous 0.6-1.9 km3 explosive event, a VEI 4-5, and also resulted in a caldera 1.5 km3 in volume. How does a volcano go from effusive eruptions to this? There is more than meets the surface, lets see.
Looking at the volumes is a good way to start, the caldera is 1.5 km3, it collapsed like a trapdoor which is no surprise since Galapagos is all about trapdoor calderas. There is an additional excavated crater at the eruption vent with a volume of 0.3 km3. The explosion volume is estimated in 0.6-1.9 km3, however we need to consider that only 5-10 % of the eruption was juvenile, this means a very small amount was fresh magma, the rest was old rock, called lithics (gabbro, fragments of lava flows…), at most there was only 0.046 km3 DRE of magma used in the eruption, then where did the caldera come from, and all the lithics?
The lithics seem to come from the crater that was excavated in the caldera wall because when 0.3 km3 of rock are grounded to ash it falls within the range of tephra volume. Almost the entire eruption material came from blasting away the caldera wall.
There is just about one explanation I can see for the 1.5 km3 caldera, a submarine flank eruption. Fernandina is just the tip of the iceberg, it is an enormous mountain that rises 5 kilometers above the ocean floor, its submarine flanks are made of countless craters and fissures. There is enough information to picture how this likely went, first there was a subaerial effusive eruption sighted by a ship 20 days before the explosion, that is May 21, at this point a dyke had reached the surface and formed a fissure. This same dike probably fed the submarine eruption as it kept propagating down the flank, evidence is that the first week of June has intense seismic unrest and earthquakes up to M 4.6. After June 7 three days of seismic quiet follow, sounds like the start of the eruption, the dyke releases pressure and stops producing earthquakes. The volume of lava erupted may well be greater than that of Kilauea 2018 or Holuhraun. Judging from the caldera it formed we are talking of one of the largest effusive eruptions of the post-Laki era.
Now back to the explosion, what caused it? My initial theory (before I started researching for the series) was that some of the most powerful eruptions of basaltic volcanoes were due to a laterally-triggered caldera collapse which causes water to make its way into the magma reservoir and blow up, relatively simplistic, and as it turned out relatively wrong.
First, there was a lake inside the caldera before 1968, everything below the lake must be filled with groundwater, the lake was also there after the events, so yes the water survived the caldera collapse and and a VEI4-5 explosive eruption, water is more resilient than it seems is it? There are few eyewitnesses of the events, Fernandina is remote and uninhabited, the ash was also distributed over the ocean away from the islands. We do know when it started, on the morning of June 11 an enormous pillar of steam rose 22 km into the atmosphere and started to spread out, at this point there was still no ash but this is a huge amount of steam, it starts to become obvious that the eruption was a gigantic steam blast.
The climactic phase starts in the afternoon, a loud powerful boom is heard all over the islands as far as 220 km. A strong earthquake is also felt at that same time, an author estimates it to have been close to M 6 based on intensity reports, however it is missing from the seismic records that should have been more than sensitive enough. I don’t know what to make out of this phantom earthquake. Pyroclastic surges sweep over the west flank of the volcano amidst ceaseless lightning and red flashes, ashfall fell on a ship 350 km away.
While it is not known when the eruption ended the strongest phase appears to be over by morning June 12, now problem is, problem with my initial theory, that the first collapse event doesn’t happen until later that day, an M 5.1, from here on the caldera forms in periodic large earthquakes every 6 hours for 2 days followed by more days of decreasing magnitude earthquakes as the collapse gradually slows down. The theory is off somewhere, if collapse was the eruption trigger then the blast would have waited for the first M 5.1.
By this point I had an improved model on how the eruption worked but it wouldn’t be until I inspected my last case of study that I found the strong enough evidence my search was aimed at. The enlightenment came from a japanese volcano.
Miyakejima is an island 150 km to the south of Tokyo, it is best known for gas masks. If you look it up on google a lot of black and white photos of people wearing gas masks will show up, I don’t know why, because actually those images are not from Miyakejima. But it is true that residents were required to carry gas masks at all times when allowed to return to the island in 2005, after a years-long evacuation, they would have to put it on if the volcanic gas concentration rose high enough. The reason for this? After Miyakejima underwent a collapse in 2000 it was followed by more than 2 years of high SO2 emissions. At its peak in late 2000 Miyakejima was releasing 54 kt of sulphur dioxide per day, without a reference that is nothing more than a number, but it happens that this number is greater the estimated degassing of all other non-erupting volcanoes of Earth together, it is higher than the rate of degassing of Kilauea back during its 2018 eruption, and by 2003 the amount of SO2 emitted was similar to that of the 1991 eruption of Pinatubo!
Big question is how Miyakejima managed to produce such massive gas emissions while not erupting. It was suggested that the caldera collapse had been responsible, however Kilauea dropped to less than 1 kt/day after its collapse ended so it can’t be that simple. So far this was a mystery, but no more.
In 2000 a 1.2-2.0 km3 dyke and 0.6 km3 caldera collapse hits this corner of the world. You may have noticed the difference between the 2 volumes, the dyke intruded westward for 30 km below the ocean towards nearby Kozushima island, it has been suggested that the volcano of Kozushima started supplying the dyke as well. It is also possible that deep levels of storage at Miyakejima provided more magma.
The rifting event, as it might be called given the scale of the intrusion, has initiated caldera formation at Miyakejima by July 8, a small explosion and caldera collapse take place together. Conditions for the first month of caldera deepening were dry, DRY, that is important. A later lake that formed around September-October at the bottom shows the water table was roughly 450 meters below the ground, enough it seems to inhibit phreatomagmatic eruptions.
There were just a few explosions, probably driven by magmatic gasses, like at Kilauea in 2018, SO2 separates from the magma due to the dropping pressure, when the roof comes falling, it pushes the gas out in a smallish explosion.
Collapse turns for much wetter conditions on August 10, a stronger explosion to 9 km produces mudflows on the flank of the island, at noon a tall white steam plume rises. Inside the caldera a little crater next to the wall ejects small explosions and base surges from a mud pool and mud flows down towards the center of the caldera. At this point the water table/hydrothermal system has surfaced but only partially in one corner, the rest of the crater floor is dry.
August 18 is the climax, the last collapse-event and largest explosion. The ash plume rises up to 16 km, the timing of the events is highly interesting. Explosion started at 17:05 and peaked at about 18:15 when the last collapse-event of the sequence struck, however the explosion was well underway an hour before the roof failed again proving a problem to the initial theory. Some have argued the explosion was magmatic because it contained a higher proportion of juveniles compared to other phases, but it also contains accretionary lapilli, this kind of volcanic product only shows up in steam-rich plumes.
One last explosion takes place on August 29 and sends a pyroclastic surge into a residential area, luckily it was cold and diluted so there was no one’s death to grieve. It is really surprising that the residents stayed there in the island during the whole eruption only to go on an evacuation more than 4 years long. It also shows the eruption was relatively small, a low-end VEI 3, it was more dangerous than a pure dry collapse like that of Kilauea in 2018, but much smaller than a pure wet collapse like that of Fernandina 1968.
Unraveling the mechanism
But if Miyakejima stands out for something that is the massive SO2 emissions that followed the event, this is something that didn’t happen after Kilauea, Bardarbunga or Piton de la Fournaise had their caldera collapses, and they were well monitored. It needs an explanation and a good one because it is highly abnormal, almost unbelievable. There is one fundamental difference with Miyakejima that I can see, the hydrothermal system was at the surface.
After a caldera collapses the magma reservoir is at a very low pressure until it reinflates, at this low pressure the SO2 can escape the magma easily so it creates the condition for strong degassing, but that is not enough, the hydrothermal system of Miyakejima must have played a role in transporting the available SO2 to the surface. In contrast Kilauea had extremely low emissions, why? As SO2 rises it encounters a layer of water and it is lost by reacting into sulfuric acid, this a reason why HVO thinks degassing in currently so low, dissolved into the lake…
The hydrothermal system of Miyakejima managed to transport gas to the surface where those of Kilauea, Piton de la Fournaise and Bardarbunga failed. And here is my theory, by being at the surface the hydrothermal system can flash into steam, this creates a suction force on the underlying system and a rapid plume of gas, vapor and water forms that transports SO2 rapidly to the surface without being lost. The thick glacier of Bardarbunga or the few upper hundred meters of rock at Kilauea confined the water and did not allow it to flash into steam, and by the time Kilauea had formed a lake it had reinflated enough. But there is more, I think there is yet a 3rd element, a really important one, lets go back to Fernandina.
Before 1968 there was a lake at Fernandina, a normal looking lake with no important hydrothermal phenomena going on. But that June 11 of 1968 the people of a fishboat would be looking at massive 22 km high column of steam, this is no longer the cold hydrothermal system of earlier years, something had changed. What? Well by June 11 the volcano would have been through 3-4 days of flank eruption and the pressure of the magma chamber would be VERY LOW, well that’s it! The low pressure of a magma chamber, the deflation before a collapse event allows a better contact between water and the magma reservoir providing the strong heat flux capable of powering a VEI 4-5 steam blast or a violent plume of fluids that transports SO2 to the surface.
The exact physics mechanism is speculative but the way I think it goes is that water simply runs into the magma reservoir in the low pressure before a caldera collapse. The pressure from magma and volcanic gasses keeps the water away, but as magma drains it reaches a point where it loses the “protection”, a possible way I picture it happening is a cavity at the roof of the reservoir filling with supercritical water and volcanic gases then convecting upwards and heating up the shallow hydrothermal system, water becomes superheated. Fernandina was a ticking bomb on June 11.
The big moment arrives. In the same way as Miyakejima it starts from the surface where water can flash into steam, a big enough steam explosion may suddenly reduce the pressure in the hydrothermal system at depth, superheated water here exists at temperatures of 300 °C, in liquid state, but if pressure would drop… The hydrothermal system literally blows up, flashes into steam blasting rock into ash and blocks, projected into the atmosphere. This is what must have happened at Fernandina the afternoon of June 11, it must have also been involved at a smaller scale in the eruptions of Miyajima of August 10, 18 and 29. Smaller probably due to the lesser amount of water available. Kilauea or Bardarbunga couldn’t because the hydrothermal system was not close enough to the surface for it to flash into steam.
The implications of this are actually amazing, it is a exception to the rule volcanoes need pressure to erupt, it is an important rule, a very useful one, almost always true but here it is the opposite. Extreme deflation and depressurization of a volcanic system can lead to eruptions as well.
New questions arise, how do we predict these events? Which volcanoes are susceptible of producing them? Which are the main hazards? These and others aspects we will look into in upcoming posts.
How calderas form:
On the 1968 eruption of Fernandina, by Tom Simkin and Keith A. Howard.
Chronology and images of the 2000 Miyakejima events: