This post will be a bit different from our normal description of a volcanic system or field. It is a (however poor) attempt to at least set up a framework for starting to think about exceptionally violent volcanic eruptions.
This question fell out of online discussion of the Hunga Tonga Hunga Ha’apai eruption: What if exceptionally violent volcanic eruptions were far more common than anyone thought? There are a lot of bunny trails that fall out of this particular question. How do you measure violence? What defines exceptional violence? What are past examples? Most importantly, how many of these have we missed?
Hat/tip to long-time supporter and collaborator of this site, Bruce Stout for raising the question.
There are a handful of candidates for an initial list. In order, they are Hunga Tonga Hunga Ha’apai (2022), Krakatau (1883), Hatepe (Taupo, 180 or 233 AD), Santorini (1600 BC), Macauley (5700 BC). There are most certainly more, but this is a good of a place to start as any. If the readers want to suggest additions to the initial list, I will look into the eruptions and try to see what happened.
While this is a great question, there are a lot of things that make it difficult to answer, most of them associated with the sorts of things that define an exceptionally violent eruption (lots of material ejected in a very short period of time). While we are pretty good at defining this today mostly because of better instrumentation, the farther back we look, the less actual real time observations and data we have, so we have to infer what happened based on indicators like volumes erupted, coverage of that material, with the most poorly defined variable being how quickly the event took place (time compressed) all helping define how violent it was.
That last point is important, as today we only have two historic datapoints for large, exceptionally violent volcanic eruptions. Both of these disturbed the atmosphere worldwide – Hunga Tonga Hunga Ha’apai (2022) and Krakatau (1883) – producing atmospheric shock (acoustic gravity) waves that circled the globe multiple times. None of this new instrumentation existed for earlier eruptions. Both of these also produced tsunamis, as did Santorini and likely Macauley.
We do know a few things, though. All known and suspected violent eruptions have large quantities of water involved. All of them are grey eruptions, explosive rather than effusive. All of them are connected to a subduction zone. Problem with this list is that it includes lot of volcanoes.
We can also figure out volumes of erupted stuff and look its makeup. More violent eruptions tend to eject more airfall. Slightly less violent eruptions tend to eject more ignimbrites. Eruption times in exceptionally violent large eruptions tend to be a few hours. Less violent, though larger eruptions tend to be several hours to a few days in length.
An Example of a Violent Eruption
At some level, a violent eruption needs to eject the maximum amount of volcanic material in the shortest possible time. Visible shock waves are always a good sign. Generally, this correlates with a high eruption plume for larger eruptions. But small eruptions can be violent too. An Aug 29, 2014, explosion from the Rabaul Caldera’s active cone Tavurvur is a great example.
Note several things. The plume is gray. It got to altitude very quickly. The explosion was quick enough and powerful enough to create a shock wave that was both visible and audible. There was a lot of large debris blown out of the conduit along with the plume.
An explosion like this requires presence of some amount of water. This is generally dissolved in the magma and flashes to steam when the conduit is depressurized. The quicker the flash, the bigger the boom. While Tavurvur is located inside the Rabaul Caldera and on the shoreline with available seawater, I have not come across anything that would characterize seawater penetration into the foot of the volcano or the conduit. On the other hand, I don’t know of anything that would keep seawater out of the foot of the volcano either. And the more water you get into the foot of the volcano, the more water you end up in close proximity to the magma conduit.
Hunga Tonga Hunga Ha’apai (2022)
The most recent of these extremely violent eruptions took place Jan 15, 2022. It was notable for a couple things. First, it was by far, the smallest of the five eruptions we are going to discuss, at a VEI 6. Total ejecta is also the smallest, with an initial estimate just under 7 km3. But it was the first of these to be viewed with the full array of satellite and modern sensors. These sensors recorded a plume height of 58 km. Tsunamis were recorded worldwide, including the Mediterranean Sea. The explosion was heard as far away as Alaska, nearly 9,400 km N. Barometers recorded multiple passages of the atmospheric shock wave as it circled the globe. The main eruption deepened the caldera by 700 m. Big chunks are missing from the caldera rim walls, particularly in the southern part of the caldera. It is being refilled by material as the walls collapse inward.
The volcano is a submarine volcano that rises 2,000 m from the sea floor, within 150 m of the surface. It is topped with a 4 km caldera. The only parts of the caldera rim above sea level are the neighboring islands of Hunga Tonga and Hunga Ha’apai. These merged for a while following a 2015 series of eruptions. They were separated in 2022 before the main eruption by a violent eruption.
The main volcano had its last major eruption somewhere around 1108. Historic eruptions took place in 1912, 1937, 1988, 2009, and 2014 – 2015. The islands have been around for a long time, with pyroclastics dated 1040 – 1180 AD. The closest inhabited island is Tongatapu, Tonga’s main island 65 km S.
Initial activity leading up to the 2022 eruption started 20 Dec 2021 with a large plume. Explosions put a plume high enough to trigger an aviation ash alert from the Darwin VAAC. Explosions were heard up to 170 km away. Eruptions continued, with satellite images on Dec 25 showing a larger island. Activity decreased and was dormant until Jan 14 ejecting a 20 km plume. Most of the island collapsed into the caldera during this eruption, though it was just a precursor to the main event the following day. No resulting tsunami was generated by this collapse.
The main eruption on Jan 15 put a plume 58 km into the atmosphere. There were numerous reports of booms around the Pacific rim. The eruption triggered an 18 m tsunami on the western side of Tongatapu, 20 m on Nomukeiki, both of which are 65 km S and N of the eruption respectively. Tsunamis of 10 m were reported on islands greater than 85 km from the eruption. It was one of the most widespread and destructive tsunamis known from a volcano.
The eruption cut internet connection between Tonga and the rest of the world for days. Some of the broken submarine cable may not be repaired for months. As of this writing, less than 10 died from this eruption, four on Tonga, two in Peru and two in California. The eruption destroyed most of the aerial part of the merged islands. While the volcano is as yet still too dangerous to approach for a detailed survey, there is speculation that the cable may have been cut or buried by a debris avalanche off the side of the volcano.
Magma for the 2009 and 2014 – 2015 eruptions was evolved, having risen and spent a while in a magma chamber 5-8 km below the volcano. Magma in the 2022 eruption was much more juvenile, having risen quickly risen through the chamber. Given the size of the eruption, there was relatively little ash emitted. Depth of the eruption may have made a difference. Not too deep to suppress the blast, but deep enough that the eruptive column saw a lot of water.
Most submarine eruptions don’t produce high plumes. This one has so much upward momentum while rising that it couldn’t stop and overshot its buoyancy altitude. The top immediately subsided, and was forced back up again by new material on its way up, not unlike a bobber in the water, creating visible gravity waves radiating outward. These ripples propagated well into the mesosphere and ionosphere. The plume produced over 200,000 lighting strikes in the first hour of the eruption. Fortunately, there was not a lot of sulfur dioxide in the plume.
Another oddity from this eruption was that the tsunami generated was arrived much earlier and was much larger than expected. The blast and M 5.8 earthquake took place within 5 minutes of one another, with the blast coming first. The earthquake did not occur on a fault. The guess is that it was caused by some sort of collapse in or outside of the volcano due to the blast. Earthquakes of this size do not generally create tsunamis, though a warning was issued. It ended up being non-predictive, as the waves were both early and larger than expected.
The tsunami arrived earlier than predicted across the Pacific basin, 2.5 hours early in Japan, 3 hours early in Sydney. The tsunamis and air pressure wave from the blast are related, with the tsunami generated by the interaction of acoustic gravity waves (explosion) with water gravity waves (tsunami). A small tsunami was generated in Mayaguez, Puerto Rico. Barometric pressure changes show the acoustic wave arrived just before the tsunami, meaning the acoustic wave was a local source generating a tsunami in the Caribbean Sea.
Long period acoustic waves (over an hour in period) travel around 1,225 km/hr, faster than the speed of tsunamis through deep water, 800 km/hr. The 1883 eruption of Krakatau is the only comparable example of a distant tsunami generated by a volcanic explosion. Barographs of that explosion are similar to those from Tonga, with very long periods, over an hour for one full cycle. Acoustic gravity waves matching wavelengths and phase velocities of water gravity waves (tsunamis) amplify the height of the tsunami. This is why tsunamis generated by this explosion were much higher than those predicted by an earthquake generated tsunami.
Large quakes have continued around the volcano since the Jan 15 eruption. The island is not locally monitored. Quakes are thought to be due to fresh magma rising to refill the chamber depleted by the eruption.
Krakatau (1883)
Activity at Krakatau has been going on for at least 60 ka, the formation of the first of at least two calderas at its location in the Sunda Strait between Java and Sumatra. The original volcano destroyed a shield perhaps 2,000 m high, 10 – 12 km in diameter some 60 ka, creating a caldera some 7 km in diameter. The islands of Sertung and Panjang and the basement of the main island of Rakata are remnants of the original caldera rim. Following the collapse, a basaltic cone some 800 m tall grew on the caldera rim in the vicinity of the current remnant of Rakata island. This was followed by two andesitic cones, Danan and Perbuatan (Perboewatan) in the caldera. These three overlapping cones merged, creating the main island destroyed in the 1883 eruption.
There is a proposed eruption, which was probably not caldera forming dated either 416 AD or 535 AD, possibly both, though the 535 AD date is based on sketchy local reports. The VOGRIPA database lists 416 AD as the only other substantial eruption from Krakatau, a VEI 4.3 that ejected 0.1 km3 of rhyolite. The 1883 eruption was a VEI 6.7 that ejected just over 48 km3 of material, creating a 7 km diameter caldera. Column height was in the neighborhood of 50km, near the theorical limit of height for volcanic eruption columns.
Eruptive activity in 1883 began May 20, with an explosive eruption from the Perbuatan vent, 120 m above sea level. The eruption produced a minor pumice fall (gray ash) 30 cm thick near the vent. During the next three months, minor explosive eruptions continued intermittently, producing minor amounts of ashfall. By July, both Perbuatan and Danan were both active. By Aug, new ash was 50 cm thick on the beach, all three vents were active, and steam was issuing from at least 11 other vents. Violence of the eruption at this point produced what was described as unusually high tides in the Strait.
The main eruption began on Aug 26, with an increase in intensity of explosive eruptions. This produced the first Plinian column 25 – 26 km high, dropping pumice fall deposits with lithic fragments.
The catastrophic phase of the eruption took place Aug 26 – 27. It first deposited a Plinian pumice fall unit 5 – 20 cm thick. Fall and surge deposits are interbedded above this layer, with total thickness 5 – 9 m. There are small pyroclastic flow units bedded with these layers thought to be momentary instabilities in the eruption column. There were four large explosions on Aug 27. The 1000 blast was the most powerful, heard 4,800 km away. It was the first to be recorded by barometers as the wave it created circled the world at least seven times.
Column collapse and pyroclastic flow generation probably began with the first of several large explosions. These covered the remaining islands with a 40 – 60 m layer of multiple pyroclastic flow units. The sequence of events for the final phase of the eruption appears to be a sub-Plinian eruption from 0500 – 1000. Flank collapse of two thirds of the pre-existing volcanic system, triggering the largest explosion and major tsunami (36 – 40 m). Burning ashes reported at distance are interpreted as the blast from this collapse. Ignimbrites were emplaced 1000 to late in the evening. Caldera collapse took place somewhere around 1000. Note that the new caldera is located to the W of the former main island, making it impossible for the bulk of the island to disappear into the caldera, though some of it certainly did. Flank collapse / debris avalanche is the only explanation for its disappearance.
A widespread, voluminous high energy pyroclastic flow likely was emplaced at the same time as the largest explosion, around 1000 on Aug 27, likely ejected by the collapse of some portion of the main portion of Krakatau Island into the caldera. Pyroclastic flows into the water separated into a dense, high-energy component that traveled below the surface of the water on the ocean floor. This was one of several causes of multiple tsunamis. The more dilute component traveled over the surface of the water up to 80 km from the source. The 1000 explosion was followed by at least two hours of mud rain, either due to phreatomagmatic explosions when the main pyroclastic flow entered the sea or the ash plume ejected with the pyroclastic flow.
Pyroclastic surges reached the southern coast of Sumatra around 30 minutes after the 1000 explosion. There is no description or geologic evidence of the surge on the coast of Java, though at least one ship near Java reported hot hurricane winds 80 km from the vent. The flow was still hot enough to char wood in Sumatra, but not shells and coral.
Tsunamis (unusual waves) were not reported on Java before the major tsunami of Aug 27. They were reported in western Sumatra during the evening of Aug 26. Explosions during that evening generated 1 – 2 m high tsunamis as far as 150 km from the volcano. Some of the tsunami deposits inland were reworked pumice rafts. Tsunamis generated by the eruption killed 36,000 on either side of the Sunda Strait.
Entry of pyroclastic flows into the ocean does create tsunamis. But this does not explain the massive size of the 1000 tsunami, which has been blamed on caldera collapse. Instead, it appears that not only did the caldera collapse, but there was a hot sector collapse, directed blast, of the main island, emplacing a debris avalanche deposit toward the N, not unlike what we have seen from Mount Saint Helens or Bezymianny. This destroyed the Perbuatan and Danan cones and removed much of Rakata. Sonar surveys of the sea floor to the N of the caldera found hummocky blocks identical to those of other flank collapse / debris avalanche failures worldwide. Additionally, pyroclastic flows tend leave flat surfaces, though both were involved in this eruption. Camus, et al in 1992 presented data including sonar images making a good case for a debris avalanche.
The magma system beneath Krakatau before the 1883 eruption is thought to have been a relatively shallow magma chamber at 4-5 km, with at least 12.5 km3 DRE of eruptible magma. The chamber was zoned with cool rhyodacite, slightly hotter dacite to andesite. Hotter basalts are injected into the chamber from time to time, eventually destabilizing the upper layers. Erupted magmas were mixed and ejected banded pumices.
As a coda to the 1883 eruption, in 1927, activity from the Krakatau caldera breached the ocean surface, building the island of Anak Krakatau (Child of Krakatau) on the caldera rim. Over the years, increasingly violent Strombolian eruptions built a cone over 400 m high with an island area approaching 2 km2. During a particularly violent eruption Dec 22, 2018, a second vent formed at the foot of the cone somewhere in the vicinity of the shoreline or a bit offshore. Activity progressed to the point where the island suffered a flank collapse, with about two thirds of it sliding down the caldera wall into the bottom of the caldera. This created a 5 m tsunami that inundated neighboring islands. The island resumed eruptions following the flank collapse, which continue today.
Krakatau is and continues to be an exceptionally dangerous system prone to flank collapses.
Part 2 of 2 will discuss the three remaining eruptions comparisons and conclusions.
Additional information – Hunga Tonga Hunga Ha’apai
Smithsonian GVP – Hunga Tonga Hunga Ha’apai
Hunga Tonga Hunga Ha’apai – Wiki
Immense crater hold created Tonga volcano, J Amos, BBC News, May 25, 2022
Dramatic changes at Hunga Tonga – Hunga Ha’apai, NASA Earth Observatory, Apr 10, 2022
NASA mission finds Tonga eruption effects reached space, NASA, May 10, 2022
Why the Tongan eruption will go down in the history of volcanology, A Witze, Feb 9, 2022, Nature
Hunga Tonga volcano spewed ash 36 miles high, M Cappucci, Washington Post, Mar 5, 2022
Why the volcanic eruption in Tonga was so violent, and what to expect next, The Conversation
Additional Information – Krakatau
Krakatau revisited: the course of events and interpretation of the 1883 eruption, S Self, Oct 1992
The collapse of Anak Krakatau volcano: a scenario envisaged, V Cigala, EGU Blogs, Feb 2019