Figuring out the eruption history of Afar volcanoes is taking longer than I expected, so in the meantime, I will have to post about other topics. And lately, one that has been present in my volcano discussions, here and elsewhere, has been about caldera volcanoes. It’s nothing new that calderas generate all sorts of admiration among us, and quite understandably. Be it islands that disappear into the sea, 40-m tall tsunami run-ups, tens of kilometers of land swept and buried by burning clouds of pyroclastic material, coignimbrite clouds shooting into the mesosphere, global atmospheric pressure waves and meteo-tsunamis, or volcanic winters, the sheer scale of these eruptions is mind-blowing.
Some elements of caldera volcanism are surprisingly poorly known, even among researchers who publish articles, particularly where it comes to the plumbing structures beneath calderas. I’ve learned a lot about the plumbing structures found in old eroded volcanoes, and how they can inform the activity of the present ones. I also know much about basaltic caldera systems whose behaviour can be used to better understand their silicic counterpart. The reason is that basaltic calderas are easier to monitor (less obliterating) and have faster life-cycles, so observations of caldera collapse and explosive activity triggered by them are more abundant. However, it’s quite an extensive topic, so this time I will only be able to skim the surface, literally, since I will be looking at surface features only (calderas themselves, ignimbrites, ring faults, and uplift structures).
I will also focus on the Tibesti Mountains in the Sahara Desert, which have great examples of caldera systems that are not covered in vegetation and are in various stages of erosion: from pyroclastic shields that have their ignimbrite shields intact, to eroded calderas which have exposed cone sheets, plugs, and ring dikes, and lastly in the north part of the massif, the nearby Uweinat mountains, and areas close to the Red Sea, systems that are eroded down to their plutons.
The caldera
The main obvious feature of a caldera volcano is, of course, the caldera itself. It’s a crater in the ground that forms when the magma chamber feeding a large eruption (or dike intrusion) collapses inward as it’s emptied. There used to be a notion that first the eruption happened, and then, at the end of it, the caldera would suddenly collapse once the chamber had fully emptied, a notion that has probably greatly harmed the understanding of calderas, since, for a while, impeded making a connection between ring dikes and ignimbrites. Now we know, cause we have seen it with basaltic calderas, that they collapse gradually. The caldera collapses that have been well observed in recent decades have been related to flank effusive eruptions or large dike intrusions, like Mijakejima in 2000, Piton de la Fournaise in 2007, Bárðarbunga in 2014, or Kilauea in 2018. In these cases, the caldera collapse was gradual, associated to the lateral draining of the magma chamber, and happened through slip in a ring fault complex. The caldera collapse of Hunga Tonga in 2022 was not well observed due to the limited monitoring and the catastrophic nature of the event, though seismically it was similar.
Calderas are generally elliptical in shape, from perfect circles to ovals, although they can have more complex shapes at times. Calderas formed in explosive eruptions tend to be deeper than their Hawaiian-style counterparts (that arguably can be only about one meter deep if, for example, Sierra Negra’s ring fault’s singular collapse event in 2018 counts as a caldera), though the width is usually far greater than the depth. For example, Hunga Tonga’s 2022 caldera is about 700 m deep below the previous caldera floor and about 4 km wide, while Tambora’s 1815 caldera is 1000-1200 m deep and 6-7 km wide; Krakatau is shallower at a few hundred meters but has seen some filling since the formation. Older calderas are usually much shallower and almost never over 1 km in depth (maybe 2 km exceptionally). In part because many older calderas have likely seen an amount of reinflation and lava filling, but also, there’s probably a rough limit around 1 km depth. The largest calderas, like Yellowstone or Toba, can be about 100 km across. Since large calderas are always larger horizontally than they are vertically, this should imply caldera-forming magma bodies to be shaped like a pancake. That said, the collapsing block that is bound by a ring fault, known as a structural caldera, tends to be slightly smaller than the full caldera diameter, because the steep sides collapse inward and can also be enlarged by post-eruptive erosion.
Caldera of Tambora volcano, formed in the infamous VEI-7 eruption of 1815
Caldera-forming ignimbrites
Other than the collapse caldera itself, the most obvious landform that is left is the ignimbrite. Originally, the term ignimbrite applied to pyroclastic flow deposits that are welded, but now it’s used for all pyroclastic flow deposits and is often used in the context of PDCs associated with caldera-forming eruptions, like Krakatau in 1883. An ignimbrite rock is made of pyroclastic lapilli and blocks of fresh magma or old rock, found inside a matrix of finer particles. Ignimbrites are emplaced at enormous temperatures that can cause pyroclasts carried by the cloud to fuse into a compact rock. The weight of the ignimbrite also contributes to flattening pumice fragments inside the deposits into an elongated shape known as fiamme, which is very characteristic of welded ignimbrites. In extreme cases, it can weld into a lava-like layer, with columnar jointing that can even flow after being deposited.
Image of an ignimbrite rock, the dark bits inside are pyroclastic fragments flattened by their still hot temperature and weight. By Beatrice Murch, in Wikimedia, https://es.wikipedia.org/wiki/Ignimbrita#/media/Archivo:Museo_de_La_Plata_-_Ignimbrita.jpg
Caldera-forming ignimbrites are the most powerful eruptions on planet Earth. They consist of pyroclastic flows, a towering wall of dark clouds, carrying ash, pumice, scoria, or spatter, moving at speeds of probably over 100 km/h, demolishing and incinerating everything in its path, leaving behind an incandescent glow as if the ground itself were on fire (some Tambora narrations seemingly allude to this). If it were a small pyroclastic flow, it wouldn’t be so impressive, but a typical silicic caldera-forming ignimbrite will cover a radius of about 40 km all around the caldera edge, a distance that is far greater than normal. Even the pyroclastic density currents that destroyed Saint-Pierre in 1902 “only” covered a width of about 8 km at a distance of about 6 km away from the volcano, and it was a surge that was more dilute, in terms of the amount of pyroclastic material it carried, than caldera-forming ignimbrites. A typical silicic caldera ignimbrite will often achieve a circumference of 300 km near its outermost edge. This means that a distance similar to that between Washington and New York, or between London and Paris, is being swept over, at one given moment, by a “tsunami” of burning pyroclastic material, often hot enough to weld itself into a hard rock layer. To give another example, over 4 million people live within the area that was devastated by the ignimbrite of the Campi Flegrei volcano 40,000 years ago.
Mountains will normally deflect the worst of an ignimbrite, but it can still make it over. In the Tibesti Mountains (Chad), in the Sahara Desert, calderas often produce ignimbrites, and due to the arid environment, they are easy to see. Here I’ve seen ignimbrites of the Yirrigue caldera that overtopped mountain ridges that were 500 m tall at distances of 30 km away from the caldera rim! Such is the magnitude of these eruptions.
Ignimbrite cover of two major Tibesti (Chad) calderas, Yirrigue and Voon, which are 12-15 km wide. Yirrigue ignimbrites spanned a radius of 40 km around the caldera edge, in places overtopping 500 m tall mountain ridges. Voon is older and less well preserved, but likely affected a similar area.
The three young (<3 Ma) pyroclastic shields of the Tibesti. Emi Koussi terminated with major ignimbrite activity around 2.4 Ma. Tarso Voon hasn’t been dated yet, but basaltic cones that date the end of its silicic activity are less eroded than those of Emi Koussi, so it likely formed around 1 Ma (million years ago). Yirrigue has a dated ignimbrite of around 400 ka (thousands of years ago) and a likely much younger Trou au Natron ignimbrite, plus central conduit lava effusion reaching into the Holocene, so it is likely still being constructed.
The ignimbrite deposits can come in many different colours, white, gray, pink, or nearly black. They tend to get darker with stronger welding and more mafic compositions. Caldera-forming ignimbrites occur with many different compositions, including rhyolite, dacite, andesite, basalt, trachyte, trachyandesite, phonolite, and even foidite, and almost always they come from crystal-poor magma. Silicic ones are not even necessarily the most common; for example, Hunga Tonga in 2022 was a basaltic-andesite eruption, while Krakatau was dacite, and Tambora was trachyandesite (crystal-poor and thus relatively fluid, probably not much more viscous than Etna). The largest ignimbrites are those of high-silica rhyolite. The *most intense known event, and probably volcanic eruption altogether, is the 1-million-year-old Kidnappers ignimbrite in the Taupo Volcanic Zone, which engulfed a radius of up to 200 kilometers around the caldera, which is a bit over half of the North Island of New Zealand, plus likely a substantial extent of ocean.
*Volcanoguy brought up even larger ignimbrite sheets erupted from calderas in Nevada, US, around 23-29 Ma ago. These ignimbrites, like the Pahranagat Tuff, the Nine Hill Tuff, or the Tuff of Campbell Creek, travelled up to 300 km from their parent calderas (though probably aided by sloping ground, and not the relatively flat surface the Kidnappers Ignimbrite affected), and it’s thought that dilute surges may have reached even farther.
Image of the Grey’s Landing ignimbrite, which was the result of a super-eruption of the Yellowstone hotspot 8.7 million years ago. This layer of rock was emplaced by a pyroclastic density current! From USGS
The Tibesti Volcanic Province has many nice examples of ignimbrite deposits, easily distinguishable in the desert environment. Here is a gray ignimbrite from Trou au Natron caldera, probably around ~100,000 years old or less, that filled valleys excavated into older (~400,000 years old) bright-colored Yirrigue caldera ignimbrites. Both appear welded and have a polygonal joint surface due to thermal contraction.
Dark gray ignimbrite (of around 2.3 million years ago) in the upper flanks of Emi Koussi shield volcano, in the Tibesti Mountains. The surface has polygonal joints from thermal contraction, and it seems to have a hard and smooth surface, as very hot, near-vent tephra welded together. The final ignimbrite sheets of Emi Koussi were dark-colored (trachyte and trachyandesite in composition) and probably rich in scoria or spatter, while the earlier ignimbrites of the volcano are white, which I don’t think have been sampled, but I think might be rhyolitic and rich in pumice. White pumiceous pyroclastic flow deposits (less welded) from post-caldera plinian eruptions of phonolite composition took place around 1.3 million years ago and cover the uppermost ignimbrite in places. The phonolite pumice flows formed during collapses of the smaller, nested Era Kohor crater.
This location (19°44’5.15″N, 18°17’4.67″E), on the west foot of Emi Koussi, shows an ignimbrite that draped the slope of a small valley and has rheomorphic flow features. The pyroclastic material was so hot that it continued to flow, in a manner resembling lava (rheomorphic flow), after it was deposited as rock, 25 km away from the caldera rim. The ignimbrite flowed down the side of the valley, making wrinkles or foliation patterns in the surface of the layer that resemble ropy lava.
Some calderas will only collapse once, but many go through multiple cycles of collapse. The explosive layers will stack on top of each other and build a vast circular edifice with shallow slopes and a surface that tends to erode quickly, known as a pyroclastic shield. There are many more pyroclastic shields than is generally thought, but it’s in the Tibesti where the most famous such volcano is found (Emi Koussi). In general, all or almost all of the central volcanoes of the Tibesti are pyroclastic shields. Most of them have undergone extensive erosion, which has removed most of their ignimbrite covers; however, the three youngest volcanoes, formed in the last 3 million years, preserve massive pyroclastic aprons. Emi Koussi itself is a magnificent edifice approximately 2 km tall above its base, with an 11 km wide caldera on top, overlooking the Sahara Desert, and a total volcano diameter of 60 km. The volcano is made of ignimbrites formed in caldera events, which are mixed with silicic lava flows supplied by radial intrusions from the central reservoir. I have seen around 11 well-distinguishable welded ignimbrites on the W side of the shield, one of which has been dated to around 2.4 Ma. The early ignimbrites have generally bright colours and are strongly welded. The final ~4 ignimbrites are darker; they are incredibly welded near the rim of the caldera, and almost seem lava-like, but, in the lower flanks, they are darker and more weakly welded, less resistant to erosion than the first ignimbrites. I think the final ones carried very dense material, maybe scoria or even molten spatter that welded well near the caldera but fell out before reaching the lower flanks, which instead concentrated lighter tephra.
A series of about eleven ignimbrite layers are stacked on top of each other on the west foot of Emi Koussi volcano (Tibesti mountains), at a distance of 25 km from the caldera. These came from repeated caldera collapses.
The next major volcano to be constructed was Tarso Voon (post-caldera basaltic cones are younger than post-caldera cones of Emi Koussi), which is actually even more of a pyroclastic shield. It’s 1 km tall and 70 km in diameter. It had an initial effusive stage, but then entered a phase of almost exclusively caldera-forming eruptions. The north flank, in particular, is a massive stack of ignimbrite layers with no sign of lava flows. They are harder to count than with Emi Koussi, but there seem to be around 14 caldera-forming ignimbrites stacked over the north flank, getting generally darker and less welded as time went on, eventually producing ~3 layers that were practically black in color and very weak to erosion, alternating with white pumice when near the caldera rim, which were overall the hardest to separate and most striking to see in Google Earth. There was also a substantial hiatus between the stronger early ignimbrites and the later weaker events, because the first ignimbrite that marks the change in eruption style locally fills valleys excavated into the first series of more strongly welded layers.
The Voon pyroclastic shield in the Tibesti is formed from about 14 ignimbrite layers, here marked in different colours over the north flank of the volcano. Earlier ignimbrites were more strongly welded, while late ignimbrites are less welded and more easily eroded. The last eruptions formed very dark layers, almost black, that may have been trachyandesite or even trachybasalt pyroclastic flows.
The youngest central volcano in the Tibesti is the Yirrigue pyroclastic shield. Due to having little erosion, its eruption history is harder to know, but the wall of the youngest caldera collapse, Trou au Natron, exposes a sequence of ignimbrite deposits, some of them emplaced upslope against the sides of the Yirrigue caldera. Though it’s hard to tell for sure, there are probably around 6 Yirrigue ignimbrites exposed, and a last ignimbrite that probably came from the collapse of Trou au Natron itself. One Yirrigue ignimbrite has been dated at 430,000 years old. There was a long lull between the last Yirrigue ignimbrites and the one that likely formed Trou au Natron, which filled valleys excavated into the older pyroclastic material. The nested Tarso Tousside stratovolcano resembles a laccolith of about 7 by 5 km in size and about 500 m in thickness, draped in trachyandesite and trachyte lavas, which I think represents a future caldera-forming reservoir built on the edge of the Yirrigue caldera.
Trou au Natron in the foreground is the youngest caldera collapse in the Tibesti, probably no more than 100,000 years old. Its wall exposed about 6 ignimbrite layers of the larger Yirrigue caldera behind (marked with colours, together with the one that formed Trou au Natron). In the upper right corner towers the small Ehi Timi stratovolcano, probably younger than Yirrigue but older than the Trou au Natron ignimbrite. A deep circular crater rimmed with dark material on the right is the Doon Kinimi vent that sourced a (likely VEI 5) explosive eruption with 20 km long plinian ignimbrite deposits, which is the youngest major explosive eruption of the Tibesti. Tarso Tousside volcano, made of effusive activity, with dark-colored trachyandesite lavas, rises in the center of the Yirrigue caldera, having erupted multiple times in the Holocene (judging from the state of the lava flows) and likely hiding a laccolith uplift.
Pyroclastic shields also happen underwater and are actually the only true submarine shields, since basaltic volcanoes are actually quite steep underwater. Pyroclastic shields are found around calderas and reach diameters of 150 km or more, since ignimbrites can travel longer distances as turbidity currents underwater. The submarine currents create dunes that advance away from the volcano in a concentric array, creating a vast pattern of ripples around them. Some pyroclastic shields around Tonga submarine calderas are as voluminous as 4000 km3, and I think their parent calderas may have collapsed as many as 50 times or so. In fact, it’s oceanic volcanic arcs where pyroclastic shields must be most common, given they are rare in continental arcs but abundant in more oceanic-influenced arcs, like Central America, Kamchatka, New Britain, Izu-Bonin, Tonga-Kermadec, or the South Sandwich Islands, that together add up the vast majority of such volcanoes in the world.
Large submarine caldera immediately west of Tongatapu island, surrounded by a concentric array of ripples produced by turbidity pyroclastic currents during (likely multiple) caldera collapses. There are much bigger pyroclastic shields to the north, but these have poor bathymetry. The floor of the caldera has been pushed upwards into a piston/trapdoor, which I’ve observed has been a regular source of vertical-T CLVD earthquakes over the past decades, meaning earthquakes related to piston uplift of a ring fault structure. There have been no fewer than 12 CLVD earthquakes since 1979, which must be roughly around 20 meters of uplift. Capture from NOAA Bathymetric Data Viewer (https://www.ncei.noaa.gov/maps/bathymetry/)
Resurgence structures and earthquakes
There used to be an old notion (not so old, really) that a volcano would go through a life cycle. Start with mafic lavas that build into a lofty stratovolcano or shield, then gradually evolve and terminate in caldera collapse, after which the magma chamber is destroyed and the volcano has to restart again. However, as far as I’ve seen, reality is quite different; some volcanoes never collapse, while others can collapse many times. It’s not just the calderas in the Tibesti; here in Europe, we have some calderas that, in the last ~400,000 years, have sourced multiple explosive caldera-forming eruptions. Santorini, for example, has at least four such events described, and Pantelleria five. Many recent caldera-forming eruptions, like Krakatau in 1883, Tambora in 1815, or Hunga Tonga in 2022, are from volcanoes that have seen earlier caldera events. In fact, Hunga Tonga is thought to have seen caldera formation only 900 years ago, and have a total of ~3 prehistoric ignimbrite layers.
There is, therefore, the question of how magma reservoirs can quickly build themselves up and generate repeated caldera events. And I think there’s plenty of evidence that this happens through the rapid resurgence of the caldera floor. Many large calderas and other volcanoes show imposing resurgent structures that reflect endogenous growth. Since its supereruption 74,000 years ago Toba caldera in Indonesia has seen its floor rise like a trapdoor, encompassing Lake Toba, an area of about 1300 km2, to at least 1 km tall near the tallest part of the structure, with Samosir Island as the tip of this resurgent block.
Samosir Island in Lake Toba is the tip of a rising trapdoor block. Here seen in Google Earth with exaggerated relief.
Since magma intrusions occupy a certain volume of space, it’s logical to expect a magma reservoir will displace a similar volume away in order to intrude, and the easiest way (the one that poses the least resistance) is upwards. Resurgence is a very common process in calderas and is likely the way that the magma reservoir is formed again after collapse, it can occur very rapidly in some cases and allow the caldera to go for another collapse without building new plumbing, simply reinflating the existing structure. There are two main forms of resurgent structures. One is that of piston/trapdoor uplift, where the volcano uses the existing ring fault system of a caldera to uplift the block overlying the reservoir. The other is a dome uplift where the reservoir simply displaces the rock upwards into a dome without employing a ring fault. Probably the strength of the overlying rock plays an important role in which of the two mechanisms is used; also, the piston uplift might be chosen by volcanoes that have already collapsed and therefore have a ring fault to use, while new magma reservoirs are likely to do dome uplift, which is something often seen around plutonic intrusions. Piston or dome resurgence would be the surface equivalents of pistolith and laccolith intrusions, respectively. Dome resurgence can also be recognized by an axial graben that forms along the top of the uplift due to the stretching as it’s bent into a dome.
A mix of piston and trapdoor uplift in Suswa caldera (Kenya), seen with exaggerated relief, and formed after non-explosive collapse during the Holocene (probably during a large dike intrusion).
Caldera of Pastos Grandes (Bolivia), showing dome uplift of the floor, with the characteristic axial graben that is associated with this form of uplift.
Piston uplift affecting the entire floor of Tat Ali caldera in Afar, Ethiopia, which has locally pushed the floor above the caldera rim. The relief is exaggerated.
With the ring faults of caldera volcanoes comes another unique phenomenon, which is that of ring fault CLVD earthquakes. Ring faults have a unique shape, a cylinder instead of a plane, so they result in a unique signature when picked up by seismic stations. In particular, there is a focal mechanism or “beach ball” diagram that can result from the slip in a volcanic ring fault, which is a CLVD (compensated linear vector dipole) earthquake. There are two polarities. Vertical-T CLVD earthquakes that are associated with the rise of a caldera, and vertical-P CLVD earthquakes associated with the fall of a caldera.
For example, Sierra Negra volcano in the Galapagos Islands, which possesses a trapdoor/piston structure over its central caldera, produced in 2018 events of both polarities that were observed with a GPS located near the crest of the trapdoor (T. Shreve and F. Delgado, 2023). After years of inflation, the caldera of Sierra Negra had bent upwards into a slight dome and on June 26, 2018, the strain was suddenly released; the ground around the piston fell while its edge jumped upwards. The rock had returned to a relaxed shape, but with a now taller central block. This event produced a vertical-T CLVD earthquake with a magnitude of 5.4 that uplifted by about 1.5 m the trapdoor that had already experienced some uplift over the preceding years. This earthquake also initiated the intrusion that led to an eruption hours later, which would eventually open a vent near the coast of the island and turn into a voluminous lava flow. The CLVD activation of the volcano had already been seen in the preceding 2005 eruption, as well as the 1996 Gjalp eruption of Bárðarbunga, and I’m not too sure, but from memory I think the 1994 eruption of Rabaul (Papua New Guinea) too. And it’s likely that vertical-T CLVD earthquakes are a common trigger for eruptions in ring-fault calderas; they might locally induce strains that help magma-filled cracks to develop and initiate intrusion, something to watch out for in Campi Flegrei, that is probably capable of these. But, let’s go back to the 2018 eruption of Sierra Negra. Once the eruption was underway, the caldera was deflating and now strain was building again in but in a different sense, by dropping into a bowl. The center of the caldera had subsided more than 5 meters by the time the ring fault broke again; now a M 4.9 vertical-P earthquake dropped the trapdoor structure by about half a meter.
The other monster looming over Tongatapu island and the Tongan capital is a submarine caldera sometimes designated “Volcano 1”, that lies immediately west of the island. Frequent CLVD earthquakes that must be related to caldera floor uplift are shown with their beach-ball diagrams. Since 2009, these powerful ~M 5.5 ring fault earthquakes have been happening at intervals of 3-4 years, and I think inflation rates of the caldera floor must be very roughly around 1 m per year. The last earthquake came less than a year ago. Note that the quake locations are a bit off, which is due to the sparse station coverage of the global seismic network.
CLVD earthquakes often have M 5-6 so and their focal mechanisms (from anywhere in the world) are known. This allows us to detect calderas that have gone under the radar and also reveal the rise and fall of caldera structures worldwide. And oh boy, is this a great thing.
As frequent lurker Thomas well knows, the king of ring fault earthquakes is the Icelandic volcano Bárðarbunga. In the years 1973-1996, Bárðarbunga produced a series of 20 vertical-T CLVD earthquakes related to the rise of the caldera, the last of which triggered the Gjalp eruption. Then, in 2014-2015, another series of caldera earthquakes came, but with the reverse polarity, as the Holuhraun fissure eruption ran its course. Since 2018, we have been back to vertical-T earthquakes; there have been 15 earthquakes larger than M 4.9 coming from the caldera at increasingly short intervals. On this day last year, Sept 3, a M 5.1 event took place, and starting with that earthquake, the caldera has been rocked by 7 CLVD events of M 5.1-5.3 in what must be a truly enormous uplift rate of the caldera floor. Easily as many meters of uplift as earthquakes have hit the caldera, if not more.
Sequences of vertical-P CLVD earthquakes have occurred in association with caldera collapses, mostly related to lateral draining; Vailulu’u (submarine caldera in Samoa) in 1995, Miyakejima (Izu-Bonin, Japan) in 2000, Piton de la Fournaise (Reunion) in 2007, Bárðarbunga (Iceland) in 2014-2015, Kilauea (Hawaii, US) in 2018, Hunga Tonga in 2022, a submarine caldera near Sofugan in the Izu-Bonin back-arc in 2023, and maybe other submarine calderas. Isolated CLVD events have also happened in Sierra Negra, Ambrym (Vanuatu) in 2018, and Fentale (Ethiopia) this year, associated with lateral draining.
Going over to the other ring, fault polarity, the only regular subaerial sources of vertical-T CLVD events are Sierra Negra and Bárðarbunga, subglacial in Bárðarbunga’s case; other such volcanoes are underwater or largely so, leaving Sierra Negra as the only actively rising (and falling) piston-caldera on the planet that can be closely observed. The DESMOS/Unzen caldera, which has a mapped trapdoor structure, near the Manus Basin spreading ridge (offshore Papua New Guinea), is likely the source of four vertical-T earthquakes that happened in that area during the 90s. Other sources are located in island arcs. Izu-Bonin, south of Japan, has Sumisujima and Kita-Ioto as regular CLVD producers. Kita-Ioto is located immediately north of Ioto; both “Iotos” must have massive rates of inflation, but Ioto is dome-type inflation, of at times as much as 1 meter per year, and doesn’t use a ring fault, while Kita-Ioto uses a ring fault to push an intracaldera block. In the South Sandwich Islands near Antarctica, both Southern Thule and Zavodovski have sourced these earthquakes. And in the Tonga-Kermadec arc, there’s another pair of caldera volcanoes with vertical-T ring fault earthquakes, Curtis Island, and an unnamed caldera west of Tongatapu Island. Hunga Tonga was also likely a regular ring-fault shaker before its 2022 collapse; it sourced a clear M 5.3 vertical-T CLVD earthquake in 1994, and a less clear M 5.2 in 2001, plus a M5.1 earthquake on 31 May 1988 for which no focal mechanism is available, but I think was a shift in the ring fault that triggered a fissure eruption the next day.
The Izu-Bonin arc south of Japan has a remarkable occurrence of CLVD earthquakes related to rising and collapsing calderas, as well as additional resurgent structures in other volcanoes.
The caldera west of Tongatapu Island is sometimes known as “volcano 1”, though I’d propose to call it Havea Hikuleʻo (Tongan goddess of making volcanic islands) or “Hiku”. It is a small caldera of dacite composition, 6 by 4 km in size, and about 300 m deep underwater. It’s also the second volcano to the south of Hunga Tonga and a potential threat to the area in the form of volcanic tsunamis. I’ve found the volcano to be a remarkable conjunction of caldera features, which I’ve already shown in a pair of images higher up, a pyroclastic shield surrounds a resurgent caldera with a block inside that has both a bit of piston and trapdoor characteristics, and which overtops the caldera rim elevation (so is probably not too far from caldera collapse) and might be the fastest rising piston-caldera structure in the planet other than Bárðarbunga, producing powerful vertical-T CLVD earthquakes at intervals of 3-4 years since 2009. Inflation rates are probably comparable to Ioto at its best, probably around a meter per year or more over the ~4 km long piston uplift, and in fact, at this moment, it’s Ioto and Havea Hikule’o/Volcano 1 that are my preferred candidates for future caldera collapse. Maybe around 0.1-0.2 km3 of magma has been added to the reservoir in about 15 years, in the form of endogenous growth.
As can be seen, caldera resurgence is a common, though often overlooked, volcanic phenomenon. It often seems as if it were something unusual and a thing of the past, that is just a side-effect of collapse, when in reality it’s happening in many volcanoes around the world right now, and is likely a fundamental process of a multi-collapse caldera. A cycle of rise and fall through which the volcanic edifice has its main form of volume growth.
With this said, I’ve more or less covered the surface features of a caldera system and how geomorphology can help to understand the way these systems work. To understand the serious threat they pose to us, or simply how pretty mountains came to be. I would continue talking about cone sheets, ring dikes, or plutons, but this article is already way too long, so that will have to wait for some other time. There’s still a lot we don’t know about calderas, but hopefully this post can help. This is it for now, see you in the comment section!
Bibliography
Roche, O., Henry, C. D., Azzaoui, N., & Guillin, A. (2022). Long-runout pyroclastic density currents: Analysis and implications. Geology, 50(10), 1172–1176. https://doi.org/10.1130/g50215.1
Deniel, C., Vincent, P. M., Beauvilain, A., & Gourgaud, A. (2015). The Cenozoic volcanic province of Tibesti (Sahara of Chad): major units, chronology, and structural features. Bulletin of Volcanology, 77(9). https://doi.org/10.1007/s00445-015-0955-6
Gourgaud, A., & Vincent, P. (2003). Petrology of two continental alkaline intraplate series at Emi Koussi volcano, Tibesti, Chad. Journal of Volcanology and Geothermal Research, 129(4), 261–290. https://doi.org/10.1016/s0377-0273(03)00277-4
Jordan, N. J., Rotolo, S. G., Williams, R., Speranza, F., McIntosh, W. C., Branney, M. J., & Scaillet, S. (2017). Explosive eruptive history of Pantelleria, Italy: Repeated caldera collapse and ignimbrite emplacement at a peralkaline volcano. Journal of Volcanology and Geothermal Research, 349, 47–73. https://doi.org/10.1016/j.jvolgeores.2017.09.013
Polo-Sánchez, A., Flaherty, T., Hervé, G., Druitt, T., Fabbro, G. N., Nomikou, P., & Balcone-Boissard, H. (2023). Tracking timescales of magma reservoir recharge through caldera cycles at Santorini (Greece). Emphasis on an explosive eruption of Kameni Volcano. Frontiers in Earth Science, 11. https://doi.org/10.3389/feart.2023.1128083
Brenna, M., Cronin, S. J., Smith, I. E., Pontesilli, A., Tost, M., Barker, S., Tonga’onevai, S., Kula, T., & Vaiomounga, R. (2022). Post-caldera volcanism reveals shallow priming of an intra-ocean arc andesitic caldera: Hunga volcano, Tonga, SW Pacific. Lithos, 412–413, 106614. https://doi.org/10.1016/j.lithos.2022.106614
Shuler, A., Nettles, M., & Ekström, G. (2012). Global observation of vertical‐CLVD earthquakes at active volcanoes. Journal of Geophysical Research Solid Earth, 118(1), 138–164. https://doi.org/10.1029/2012jb009721
Shreve, T., & Delgado, F. (2023). Trapdoor fault activation: A step toward caldera collapse at Sierra Negra, Galápagos, Ecuador. Journal of Geophysical Research Solid Earth, 128(5). https://doi.org/10.1029/2023jb026437
Gregg, P. M., Zhan, Y., Amelung, F., Geist, D., Mothes, P., Koric, S., & Yunjun, Z. (2022). Forecasting mechanical failure and the 26 June 2018 eruption of Sierra Negra Volcano, Galápagos, Ecuador. Science Advances, 8(22). https://doi.org/10.1126/sciadv.abm4261
Fantastic overview of caldera dynamics, Hector. Timely even, considering my own machinations…
On the Ioto situation, the current eruption seems like all the others since 2024, small and not very long lasting. More interested in the current increase in uplift. I was considering writing up a post on the possibility of it doing one of these major Vertical CLVD quakes and uplift event. I think, looking at the current situation, it’s hard to reconcile what is exactly going on. Horizontal deformation rates have been stable the last 5 years and yet there was a slowdown in uplift in conjunction with the west portion being displaced eastward and the east portion being displaced westward. I was pondering the possibility of it preparing one of those quakes.
What do you think of these oddities?
Thanks Tallis!
“it’s hard to reconcile what is exactly going on.”
I’ve been having problems in finding any information at all. It does look like things are growing more unstable and that the already massive trachyte dome uplift keeps getting more magma in at rates as high as they can get in a subduction zone caldera, so I don’t think it will end well. The question is how long it will take and how the events will proceed.
https://geodesy.unr.edu/NGLStationPages/stations/J605.sta
https://www.data.jma.go.jp/vois/data/report/activity_info/329.html
These past few eruptions are likely consisting of degassed magma from the shallowest portion of the magma, erupting through the current cone-sheet intrusion.
On it’s own, this is of no concern, similar things have happened at other systems with no catastrophic eruption. It would take a significant amount of time to expel the useless magma at the current pace.
What worries me is the accelerating deformation and it’s odd patterns that hasn’t had an easy explanation. Ioto is in the same position as CCN, as it is already loaded but still struggling finding an adequate path for a significant eruption
I am far, far less knowledgeable than others here, so I will defer to anyone who wants to correct me if this is nonsense, but I wonder if Ioto’s long-term hyper-uplift without going big implies a certain level of stability to the system? As in, the surrounding rock is pliable enough to stretch along with the intrusion with only occasional minor breaks like we’re seeing now.
This thought makes less sense now that I’m looking at it written down, but I can’t really wrap my head around how it’s able to do this with such an extreme supply without bursting for centuries or millennia on end.
This just says biggest is >=VEI-4 without constraining it:
https://vogripa.org/searchVOGRIPA.cfc?method=detail&id=1297
“I wonder if Ioto’s long-term hyper-uplift without going big implies a certain level of stability to the system?”
There’s definitely some stability, but that may be coming to an end, given that earlier eruptions were phreatic, while 2023 was magmatic. The central dome uplift is also about 200-300 m above the submarine caldera rim, so that could be a lot of magma built up over the years (I doubt it can take much more, though, of course, with caldera systems, a little more can be a really long time).
My rough back-of-the-envelope math is that Iwo Jima has probably experienced in the ballpark of 20 km^3 of inflation since its last big bang. That is quite a bit of DRE, though more Krakatau-scale than Tambora-scale. The stuff that erupted was a bit flat, but hard to imagine that the bulk of that 20 km^3 is anywhere near properly degassed or chilled.
I think I remember you saying in your “Volcanic Poker” article that you thought CCN was capable of a larger eruption than Ioto. Is that still the case for you?
It’s a bit more complicated now with more context. There is little information about the size of Ioto’s inner system and the size of It’s intrusion that it’s impossible to get a proper comparison on the 2. I’m still of the opinion that CCN has the capacity for a larger eruption while Ioto is more likely 5o pull it off due to very different circumstances.
CCN probably has over 4,000 km3 of melt within it’s system and it’s hard for.me to imagine Ioto being as big as that. Ioto has a far shallower chamber and NO shortage of ways of going boom though.
My thing is, I have some doubts now about CCN being able to pull off a low-end VEI 6, let alone anything higher, based on reading through the comments on your recent CCN article. My takeaway from everyone else there was that a VEI-6< eruption was theoretically possible, but given some of the other factors and questions to consider about the arc CCN is in, as well as the volcano itself (which admittedly are more than likely impossible to answer definitively), a Pinatubo scale event would be the biggest CCN could produce. Not a Hatepe sized event, to use the example I think you used to measure CCN's capabilities.
Great article, Hector!
I find it curious that both P-CLVD and T-CLVD earthquakes in the examples you cited were all in the M5 range (give or take). Same for Bard’s P and T – CLVD ring fault quakes (also trap door) that seem to consistently source as M5-ish events…both during and after the Holuhraun eruption. Is there something special about the stiction involved with volcanoes that form caldera’s that limits how much strain/stress a fault can hold before failing?
“Is there something special about the stiction involved with volcanoes that form caldera’s that limits how much strain/stress a fault can hold before failing?”
There’s a bit of a range that these earthquakes can span as a whole. The limited range in magnitude tends to be for an individual volcano. The caldera west of Tonagatapu almost always does M 5.5, Bardarbunga and Hunga Tonga prefer closer to M 5. Curtis Island, I think, is close to M 6. Each volcano has a unique fault structure, and the size of the earthquakes stays similar because of that, but varies from one to another.
At a wild guess, I’d suggest it might be connected to the fact that – within wide bounds – calderas are of roughly similar orders of size, and thus the area of the ‘slip patch’ that moves in each event will be of a similar order.
Thanks Hector. I agree that the “typical” CTVD quakes are volcano-dependent. Seems like the magma type is a critical variable. Also, the slope of the ring faults must be involved…caldera walls are often not perfectly vertical.
For a very long time I also thought Kidnappers is the most expansive ignimbrite sheet, but then I came across Nine Hill Tuff and Tuff of Campbell Creek in Great Basin…in terms of areal coverage, truly monstrous
Whoa, 300 km, that’s a lot of distance to cover from a caldera.
Interesting! I co-authored a paper on the 2022-2023 Taupō unrest:
https://seismica.library.mcgill.ca/article/view/1125/1554#toc
A lot of links to interesting caldera papers in the references there. The entire Taupō sequence was VT-type events, except for the largest, M5.7 event, which had considerable CLVD, and was interpreted as an explosive/inflationary signal.
Thanks Mike Ross! The article looks interesting, and congratulations on the co-authoring. I think the Taupo structural caldera block may have shifted upwards slightly, though I’m surprised the magnitude wasn’t larger.
That’s the interpretation, and it’s supported by the evidence of the small tsunami around the lake following the earthquake.
(the specific mechanism of the event was eventually characterised as ‘trapdoor’ FWIW)
Thanks for the link (and paper).
Very interesting article Hector, thanks. I ll try to post some ignimbrite (related) pictures from my collection later today. There is really a story to tell about some of the samples!
Thanks, Rob! Do share.
Fascinating read, Héctor. Thanks for the hard work you put into this!
Thanks, Clive. Hard work indeed.
Tibesi is one of those places that I always wondered about on google arth. Its pretty obvious its a volcanic area and huge scale but exactly when everything happened has been pretty mysterious, though its not dead obviously.
There are lots of presumably basaltic cones on the side of Emi Koussi, including one that looks very recent on the southwest side bear the caldera, it might not have a shallow magma chamber or active caldera anymore but the volcano isnt extinct from the look of it.
Theres also a volcanic field going north from it that at least has some uneroded but weathered vents. There are similar looking vents just north of Tarso Tousside, pre-Holocene but not eroded.
Yes, the youngest flow of Emi Koussi has a fairly well-preserved tephra fallout and no sand or soil on the flow; so I think it’s Holocene. But it’s very small, and the only thing the volcano has done in this epoch. Almost all the magma seems to be going to the Yirrigue complex.
Interesting thing about TarsoTousside is that is looks like much of its lava is pahoehoe, though the peak is more typically internediate blocky lava. Presumably different eruptions of an evolving magma chamber, but if the pahoehoe stage is still trachyandesite/trachyte that is probably the only truely flujd lava of such composition.
Its also something to wonder how old it is, I havent looked in a long time to be fair but last I read it wasnt dated, but clearly Holocene and a vague reference to an eruption in Roman times. But I long since forgot the source now.
If you look closely, Dalafilla volcano in Afar has tiny bits of ropy pahoehoe rhyolite lava in places. So it wouldn’t surprise me if some of the Tousside pahoehoe is trachyandesite, at least the rough-looking one is, though I suspect there’s also trachybasalt in there, I doubt it has been very studied; there’s only a handful of samples from the entire province. The aa lava without sand puddles is possibly Holocene (20,000 year old, dated Arabian lavas have plenty of sand puddles on the aa), which is most of the flow field actually.
The area north of Tarso Tousside is the volcanic field of Tarso Toh, and none of the lava is dated Holocene but one of the maars is early Holocene about 8500 years ago. So its probably still active but intervals are wide and eruptions very big it seems.
It also seems like there is actual confirmed Holocene or very latest Pleistocene activity in Trou au Natron, as sediment buried by the cones in it somewhere was dated to 15000 years old. Tarso Tousside has no dates listed anywhere still, but a lot of it has basically no sand cover at all which is saying something in a famously sandy desert…
Apparently a very similar thing has happened at Emi Koussi too, with a latest Pleistocene sediment being dated under a lava flow in its caldera, of which the only option up there with any preserved flow features at all looks more eroded than most of the mafic flank cones so perhaps it isnt as inactive as it looks. Its more likely a mistake in some way, but its also possible the lava here weathers relatively fast compared to the rift-related volcanism far south and east, and it might be expected sand cover is a bit faster too with more of it around. Or both
These are lithophysa formed in welded ash from St. Egidien in Germany. You can see the structure is lumpy.
It is hard to say if its from a pyroclastic event. Ash may fall in an even, persistent way. Don’t pay attention to the cavity and its filling. Its all about what is surrounding it….
These two are from Nesselhof in the Thunringian Forest also Germany.
Actually I think these are nor from an ash flow, but is has nice flow structures in the rhyolite.
This one is from an Ash flow, might have been quite explosiv. It bears multiple xenoliths, pieces of stone that were included in the ash flow. First Creek, Washington/USA.
This one has a particularly evident fiamme. Nice specimens, btw.
Ash flow lithophysa from Cherry Creek, OR/USA.
May be from an ash flow, Little Naches, WA/USA. Lithophysae still caught in its matrix.
And the last bunch from near the large McDermitt Caldera, just in OR/USA.
These stones include many clasts taken in the flow. Many clasts are fractured, this must have been a violent event.
Ah, yes, to be noted, the stones are cut and polished to get a clear view! 🙂
Great article, wonderful to learn new things, specifically regarding CLVD quakes!
Thanks, cbus! I remember when I first read about CLVD eqs not that long ago, probably with Sierra Negra, and it was a huge jump in my understanding of calderas.
Where can we draw the border between a caldera volcano and a classic stratovolcano?
A caldera volcano must have had some kind of subsidence in a large area. A classic stratovolcano has only explosion craters or small subsidence craters (f.e. Etna).
A difficult class are the Somma volcanoes. They have a large crater or caldera in which a new volcano grows. Are they calderas or stratovolcanoes? Vesuvius has a prominent example: The Somma, in which the new Vesuvius grew. Was the Somma created by an explosion/ejection of material or by a collapse/subsidence? Is a somma rather a caldera or a crater?
Another difficult class are the Complex volcano. As the word says, they are complex. They can have a caldera, but also everything else.
Do Calderas always occur because of the eruption of magma or can they also occur without an eruption? I imagine a great magma chamber, that is created, but fails to erupt. If after a while the upswing of the magma chamber collapses, the ground may do subsidence without an eruption. Grindavik’s sill formation with a graben showed in November 2023, how subsidence can happen without an eruption. In Grindavik the “no eruption” was only temporary, but it seems possible that subsidence can happen without an eruption.
Realistically there is probably a fairly low limit to how big a basaltic caldera can be without any associated eruption at all along the dike. As in, a 1 km3 intrusion could go hidden in the right places (like in the East African rift, as we saw late last year) but going above that is probably a steep increase in chance of eruption until theres a point its impossible to contain all the magma even if most of it stays. At 10km3 its likely there isnt anywhere that can keep that hidden. Although, Hawaii would put that underwater so still out of view 🙂
Just my guess but its unlikely silicic magma calderas can form without erupting. Its a bit of a mystery why the option of a caldera forming with no eruption is even considered at all when the lowest pressure is up…
Suswa somehow must have done a 5 km3+ dike intrusion in the Holocene. It has a caldera at least that big, which cuts lavas without soil, but there is no explosive deposit or lava flow associated.
I had a look and there is a very long graben system south of Suswa, maybe in this area rift filling is more dominated by lateral intrusions from a few central volcanoes instead of upwelling along the whole rift, so there is more space available. Still it seems unlikely a 10km3 dike intrusion can stay entirely underground, even at 10 meters wide and 10 km deep it would be 100km long, and to be honest it sounds easier to erupt than force the crust apart that far. Or be twice as deep or long and 5m wide. Suswa doing 5 km3 might be pushing the limit or even an outlier.
“Where can we draw the border between a caldera volcano and a classic stratovolcano?”
It’s hard to draw a boundary; many volcanoes have both behaviours. I think the stratovolcanoes that are least related to calderas are high magnesium andesite volcanoes; these are magmas that, even when primitive, already have close to 60 % in mass of silica, so I think that makes them buoyant and unlikely to pond in the shallow crust on their way up. The Mexican stratovolcanoes from Colima to Pico de Orizaba are, for example, volcanoes that erupt high-magnesium andesite or their evolved forms, and that’s probably why Mexico has the calderas in the back-arc and not the arc itself. In general, I’ve noticed these volcanoes to often come without calderas, but it’s complicated and I can think of exceptions like the Katmai Group.
Volcanic craters are usually created by the explosive ejection of the summit. Mount Somma (Vesuvius) is said that it was created by collapse. So it’s rather a caldera than a volcanic crater.
Unlike this the 79 AD eruption included the ejection of the summit rocks. Also St. Helens 1980 is an example for the explosive ejection of a volcano’s summit. Contrary to St. Helens Pinatubo did a caldera collapse in 1991.
There have been a lot of tiny earthquakes at Kilaueas summit, as well as a more sizable one 10km deep under the east side of the caldera, at the bottom of the magma chamber. No change in deformation but it looks like change might be on the horizon.
If a vent opens under the lava lake, its pretty likely to be overpowered, and might well cause the lake to drain down and pressurize one or both rifts, probably causing an eruption there. Although it could also just as easily open more vents along the ring fault, which would probably have different outcome, likely resulting in more continuous filling again. Or its like 1823 and the entire fill of the last 5 years drains out at low elevation in a couple hours leaving an enormous glowing hell pit and maxing out the MIROVA scale… 🙂
Probably a great example that Kilauea is also a caldera and does caldera things, maybe not a resurgent caldera but its not as simple as a one time hole that fills back in and restarts.
“If a vent opens under the lava lake, it’s pretty likely to be overpowered, and might well cause the lake to drain down and pressurize one or both rifts, probably causing an eruption there.”
Now that would be an interesting turn of events. I think Keaiwa 1823 is also still on the table, though the way the caldera is filling may not be very ideal for that outcome.
https://sustainability.stanford.edu/news/ai-model-reveals-hidden-earthquake-swarms-and-faults-italys-campi-flegrei
Speaking of calderas, there is an article about Campi Flegeri (luckily, not the fear monger kind). Rather, they use AI to show all the earthquakes at the caldera and shows a ring fault there.
Naples is well and truely screwed, supervolcano it may not qualify but it can probably erupt as powerful as a VEI8 so the name might not be totally unapplicable.
There have been 4,732 earthquakes so far in 2025 at Campi Flegrei. She continues to break records year after year with 4,900 in 2024. That record will be shattered in 2025 and we have blown way past 1984 seismic crisis. This is not just monitoring bias because of a wider array of seismic instruments as the number and frequency of 4-5 magnitude quakes have jumped on top of the heap pile that could become the greater Naples area. Although I feel Tallis’ torment regarding CCN is warranted, there has been alot of talk about its 8km deep magma chamber being too deep to cause a bigger VEI eruption. CF has a shallow chamber that is also feed by a deeper chamber and plenty of CO2 and SO2 bring it up. The ring fault is shows up clearly on GOSSIP website. https://terremoti.ov.ingv.it/gossip/flegrei/2025/index.html
I am not concerned with Campi Flegrei in terms of it’s chances of producing a caldera-forming eruptjon. First issue is that the magma degasses too quickly as it ascends, second is that it’s unlikely to build the strain needed considering how frequently it erupts. As someone has already noted these quakes are the result of inflationary pressure, not something indicative of an immediate popoff
Yes its not right at the limit, and an eruption of any sort would have the same risk factor. But its clearly on the way for a round 3.
Though that ring fault is relatively small compared to the last two ones. Its only about 6km wide. I’m not sure if the northern seismic mess represents the northern edge, if so its only about 3 km in that direction. Even if its merely incomplete, probably not *that* much bigger. By comparison the second one was 12 X 8.5 km. Or about 5 times the size. The first caldera bigger yet. #1 was comfortably a VEI 7, #2 barely, #3 would be a mere mid-range VEI 6 at best.
The north edge is a bit different, maybe a trapdoor structure.
Thing is too, there are two ways to get the caldera smaller than the predecessor. One is a declining system, but it could also just be a system that recovered faster and reached tge same stage in less time.
That also needs to be considered from the time resurgence actually begins too, not from when the last collapse happened, if I remember correct Campi had more eruptions earlier in the Holocene, only one in the last 3000 years or more, so might have only been resurging significantly for not that long.
Could that ring fault signal that the ring fault shows where the next caldera of Campi Flegeri will be?
According to the article, that doesn’t seem to be the case. More like business as usual somewhat.
“The research suggests overall inflation of the caldera is driving earthquake activity through pressure. The study authors did not observe any evidence for the upward migration of magma, which reduces concern over the short term that the area will experience a magmatic eruption, according to the study.”
Sorry, misread the question. I would give it a maybe, although it could be the place for smaller eruptions in the long run. Note that Monte Nuovo is on that ring fault…
I think it’s likely to be the ring fault used to rebuild the magma body after the collapse 15,000 years ago, and the same fault that the caldera will go down with (eventually).
Though that ring fault is pretty small compared to the previous two calderas. Mid-range VEI 6 is probably the most you could get out of it. Its like 1/5th the footprint of the 2nd caldera which was barely a VEI 7.
Yes, it seems to be shrinking. A VEI 6, I think, is the most likely outcome when it does the caldera-forming eruption (but still will devastate the Campanian Plain). Though at the same time, the full caldera will usually be larger than the ring fault because, as the central block subsides, the sides cave in.
I’d assume that Campi Flegrei has a long way to go until next caldera eruption.
But a phreatic eruption is relatively likely depending on the location of a possible dike.
Should we call Campi Flegrei a Complex volcano with a caldera? There are many different locations and types of vents that happened there. Cinder cones, maars, craters. There is not a stratovolcano inside a caldera that we see f.e. in Santorin or Krakatu. Campi Flegrei rather has multiple random vents. Solfatara is 4-5 km away from Monte Nuovo that erupted in 16th century.
I wanted to write “There is not a dominating stratovolcano …”
“I’d assume that Campi Flegrei has a long way to go until the next caldera eruption.”
Not necessarily.
Yeah, a VEI 7 almost overkill from a Naples prospective. More long-distance ash and tsunami effects sure, but having a VEI-6 caldera-forming eruption 10 km from city center is going to pretty well toast everything in pyroclastic flows except the eastern suburbs (and ash from the prevailing winds probably deals with that). Pinatubo was an extremely well-managed event, but at least not *that* many people lived in the core devastation area. Even if Naples was similarly well-managed, you still have stupendous property damage (and still a lot of dead, for every 25 people evacuated, 1 died for Pinatubo).
Its just a reminder location is so important in terms of effect. The 20th century saw multiple VEI 6s. The deadliest eruption was a mere poorly placed VEI 4. We’ve been lucky not to have such a missplaced eruption, especially since a lot of developing countries have far higher populations near volcanos than they did in 1902
Pinatubo had the disadvantage that it was both a volcanic VEI6 and a Typhoon. Both together caused great Lahars and wet heavy ash that killed many. Sometimes bad things don’t happen alone …
To be fair, typhoons/cyclones happen multiple times a year in the triangle between New Zealand/India/Japan, and as probably the majority of volcanoes likely capable of a VEI 5+ are in this area its not exactly unlikely to coincide. Maybe not fully blow up while its in the eye of the storm but within a month of a storm or vice versa.
VEI 7 is also realistically probably only dangerous for people who think they are far away enough but arent, many people will ignore a VEI 4 risk but tell anyone their volcano is about to do a Krakatoa x10 they arent going to be so brave.
Its probably ucky the only likely chance of a high VEI6/VEI7 massive caldera eruption in the 21st century is Ioto, which only has the tsunami risk to consider, a hazard that is far more respected and literally no one will take a chance for. Its a massive area of affect from the wave with a billion people, but no one lives in the incineration zone.
I absolutely agree that a that a VEI 6 is not on the table anytime soon. But at CF you have crumbling Roman concrete for cap rock that has three times more gas emissions than Kilauea during a 2025 eruption pause. Combine that with the potential of seawater interactions, and a deep magma chamber feeding a shallow chamber. Accelerating earthquake activity is already causing real damage to structures even before any potential eruption begins. The next eruption doesn’t have to be a VEI 6 to cause massive pyroclastic surges and ashfall deposits similar to Vesuvio in 79 AD.
Having visited the Naples area, Campi Flegrei is my source of torment like CCN is for Tallis. I have already decided my hell will be watching earthquake, deformation, and gas emission activity at CF continue to accelerate for all eternity without eruption…
Not a caldera volcano, but “il vulcano” has had some unrest recently: https://www.vulkane.net/blogmobil/vulcano-erdbeben-und-anstieg-des-gasausstosses/?fbclid=IwY2xjawMoAspleHRuA2FlbQIxMQABHjfRCIvuh-xlpZNSj-GzdvU460sNBVje7G5f7VBCJ-GteIEnuxIsl4pbbH6K_aem_Eyeb7fEa1N6kurNM1150tg
A peak of CO2 emissions (30,000g/m² daily) and of Fumarole temperatures at 291 degrees. It is unclear whether Vulcano only wants to play or is going to do more serious things.
Zach, very interesting post. The news article I caught did NOT have the ring structure. Selective editing perhaps?
Could be, but it was appearently published here.
https://www.science.org/doi/10.1126/science.adw9038
Unfortunately, it is not “free to download”.
Is it just me or is the icelandic met office getting worse with their updates? I havent even seen a final volume estimate for the latest eruption. Theyve also given very little info regarding april 1st, tho i guess back in 2023 they could just quote some papers considering flow rate and dike volume, there havent rly been any new papers about svarts in the last 6 months or so. Bit of a drought.
Denden:
I think using the word “worse” is probably not the best choice of wording here. If you are asking that a lessening of website material has occurred, I concur, but we need to encourage the IMO, and avoid being critical.
just my $0.02 worth.
What can Kama‘ehuakanaloa’s (Loihi) Caldera do? The caldera is 3.7 km x 2.8 km and ~1km below sea level. 1996 there was a collapse by 300m, but without tsunami threat. Was the caldera collapse more gradual like Kilauea 2018 instead of a sudden “all in one” collapse?
Its probably too deep for tsunami risk. Most likely caldera collapses are same magnitude and speed as at Mauna Loa and Kilauea. Maybe more like Mauna Loa (faster) given it is steep.
It also indicates there might be times Kama’ehuakanaloa gets a similar magma supply as Mauna Loa and Kilauea, 0.1 km3/year. Maybe if there are any points where both Mauna Loa and Kilauea have a relatively low output is when Kama’ehuakanaloa is more active, but this also requires dating of all 3 to high accuracy over at least 1000 years and only Mauna Loa has that far back this way.
Either that, or current Hawaii output is higher than estimated. Particularly considering that just like magma supply is high at both Kilauea and Mauna Loa, there is at least seismic evidence Kama’ehuakanaloa is getting supplied too.
I’ve read that Kama’ehuakanaloa is a long volcano like Mauna Loa and Kilauea. It seems possible that there can happen deep flank eruptions like Mauna Loa 1950 or Kilauea 2018. How well could HVO observe an eruption like this at this depth?
Given there is only one historical eruption known, and still not observed actively erupting, chance is pretty low. Far as I can tell Mauna Loa in 1950 opened too fast to actually make any significant earthquakes, although termor was very intense. Its also not super common, its likely Kama’ehuakanaloa just hasnt done that in the time it has been monitored.
One thing that is important, its probably actually very hard to locate earthquakes that are very shallow in the volcano, because it is topographically prominent and seismometers are on land, so there is no direct line of sight through the crust, ocean gets in the way. The Hawaiian volcanoes erupt extremely fast, and with true warning short enough to make Hekla look slow, the only warning is tiny quakes at the top of the magma chamber that are poorly located even at Kilauea, and probably undetectable with the above complication. Then at some point after days, weeks, sometimes months, intense seismicity starts and erupting begins in an hour or two. But we cant see that well at Kama’ehuakanaloa. I think 1996 was only discovered because they went down to see it after a swarm that was probably a bigger intrusion, and found new lava, but the eruption happened before the swarm so was itself not really noticed, abd there night have been further eruptions along that intrusion. Its not really unlikely at all that small eruptions have occurred since given what we know about Mauna Loa and Kilauea, which start to refill collapses pretty much immediately.
Most likely, Kama’ehuakanaloa erupts about as often as Mauna Loa, but less volume. If anything it would be more often, but small still. Its actively growing like those two abd I dont think I have ever seen reference to anything other than its summit studied well.
Kilauea seismometer showing a short tremor about 4 hours ago as I write this.
Theres a slide comparison of Kilauea on December 23 2024, and today, showing the 2018 caldera filling up.
https://www.usgs.gov/observatories/hvo/news/photo-video-chronology-september-3-2025-kilauea-summit-fieldwork-after
Its hard to see how much it changes per week or per episode but the change in absolute is extraordinary. Vertically the 2018 caldera has filled at least 1/3 of the way above its late 2024 level, much more at the vents themselves but growth rate there is a lot less constant. That being said E33 might return to more of a vertical fountaining compared to the diverted ‘lavabow’ fountain, and build the tephra cone again.
Its still unclear whether the east rim would overflow north into the greater Kaluapele depression, or if the vent area builds up high enough to overflow south from there, which is more likely to go first.
The lava field, that is covered repeatedly with new lava during each episode, is in fact a lava shield. The east side of the lava field is the foot of the lava shield, where only thin layers are added with each episode. The western 50% of the lava field grow faster and higher.
As long as the eruption continues with episodes, I’d expect that lava won’t overflow towards the greater, flat Kaluapele depression. It looks more likely that the cone will one day reach to the level of the caldera rim.
Héctor:
I compliment you on your clever use of coloring in the graphics, so as to bring out the tephra layers. I appreciate this very much and it adds clarity to your whole presentation. Thank you for caring for us enough to put this in, as it just doesn’t appear by magic 😉 There is consider work to add the coloring.
Thank you!
Just read a couple of interesting papers on Katla and then she does star? I think she might’ve read my piece on her and is throwing a fit. Of course, she’s patient one but she’s got a temper on her, and waiting just makes it worse.
2 green stars for Katla this am.
https://en.vedur.is/earthquakes-and-volcanism/earthquakes/myrdalsjokull/
Very shallow. I expect these are related to the summer ice melt
September is the ice minimum in the Arctic region, maybe this is a moment when melting water is prone to meet Katla’s hydrothermal system with subglacial hydrothermal explosions.
Would Katla without ice look like Haukadalur (Great Geysir)? https://en.wikipedia.org/wiki/Haukadalur_(Bl%C3%A1sk%C3%B3gabygg%C3%B0)
Katla without ice would look like Okmok, a massive shield volcano or at least shield shaped volcano, with a deep breached caldera.
https://www.usgs.gov/media/before-after/filling-kaluapele-kilaueas-summit-caldera-a-result-episodic-lava-fountaining
Using this slide comparison, the elevation of the vents now is slightly higher than a prominent ledge in the pre-eruption wall, now buried. That ledge on a map of the eruption on Day1 is about 3350ft elevation, or about 1020 meters. The north vent might be about 30 meters aboce that, and south and middle vents similar again, so actually the highest E32 vents are about at least 1050m elevation maybe nearer 1100m, which is in the range it was before 2018 though circumstances arent exactly comparable.
It is growing up nicely! But the recovery after the last episode, as shown by the tilt, is a bit slower than after previous episodes.
I think the early recovery after E31 was partly the I part of the DI event that occurred right before. And after the little flat line it recovered up to E32 at a pretty constant rate. The rate now is the same as that, not reduced overall.
https://www.usgs.gov/media/videos/timelapse-video-showing-changes-kilauea-summit-eruptive-vents-a-result-episodic-lava
A timelapse video too 🙂
Pretty much since the tilt stopped falling in early January its been flat. But the tiltmeter also doesnt necessarily pick up inflation if it magma is directly under it, for UWD this likely doesnt apply much but it does for SDH.
ESC moved up a bit during last episode. Did some magma flow towards uper ERZ?
If magma was actuallly moving into the area uplift would be much more pronounced, probably totally dominating the daily solar heating cycle, so this is probably just a sensitive instrument picking up the summit pressure change.
Thank you, thank you for the wondrous tour of Tibesti region’s caldera zoo.
For too long, the area was ‘Here Be Dragons’, variously spears, muskets, AKs…
Mars was better known !!
I suppose one reason some big pyroclastic flows can travel so far is they may ‘lay their own roads’, initial ejections filling local terrain relief, so later blasts get a clear run towards the horizon…
Be NOT There !!!
Clearly identifying the Campi Flegrei ring-fault has unambiguously mapped *that* Monster.
Upside, the ‘Gulf of Naples’, oval from CF clockwise via Vesuvius, Pompeii, Sorrento, Capri and Ischia, is apparently sedimentary ‘To Great Depth’, ‘Native Strata’, not a vast caldera.
Down-side, never mind CF’s unsettling bradyseism and Maar potential, Ischia continues to inexorably extrude its volcanic plug. How that and, yes, Somma / Vesuvius, are connected to CF seem almost ‘Icelandic’…
How is the current eruption cycle of Taal going to continue? After the big eruption 2020 the activity became last years more moderate. Is there still a big final eruption possible or is Taal rather going to slow down step for step? 1977 the final eruptions were phreatic, likely comparable to the recent phreatic eruptions in July and August 2025. https://www.gmanetwork.com/news/topstories/regions/952905/minor-phreatic-eruption-monitored-at-taal-volcano-phivolcs/story/
1903-1911 developed the opposite way with a big VEI3 eruption in the final year (tsunami, Pyroclastic flows, destruction and fatalities)
1960s/70s had lava effusion, but no big final blast. Now though the lava effusion iis instead refilling the magma chamber after the 2020 dike, and that might take a long time. Pretty likely to just go back to sleep and recover eventually but in the 2060s or so might be when it does something more powerful. Just a guess though
Looks like Iwo-Jima erupted this week, phreatomagmatically on the west side of the island, creating quite a large crater right next to several buildings. I wouldn’t want to be stationed on there at the moment!
By the way what a great article (and the follow up comments). Always learning something new.
https://www.youtube.com/watch?v=Oa0IU5ddx7I is the commentary on this eruption. I have kept NASA-Firms satellite photos from the past 2 or 3 years for Iwo-Jima and there are several hot spots which periodically show up. This eruption came right up where one of the most persistant hot spots is.
Try https://zoom.earth/ you can see eruption plumes in progress and when eruption starts most of the time, provided there isn’t too much cloud.
Looking at the map provided in the video, would that be a developing ring structure around Ioto?
There has always been a ring structure there surrounding the main island but running through it on the west side. The eruptions are indeed occurring mainly on this ring.
The Copernicus browser shows this eruption from the Sentinel 2 L2-A Satellite taken Sept 2nd 2025. See https://drive.google.com/file/d/1qpuQJ–JeDek0R68cpjFLMCtRyf8Lk_P/view?usp=sharing
Oh and Laguna del Maule is doing some CLVD earthquakes right now!
? Context?
It has a very recognisable ring fault and the quakes dating back to 2020 all envelope the complex in a circle so I’m assuming this is what it is, as we know it is quite rapidly resurging. The info came from the geologyhub video/sernageomin.
5000 earthquakes a day on september 2nd, 3rd & 4th. He comments that the nature of the earthquakes suggests slippage due to uplift but I haven’t yet seen the moment tensors.
Thanks.
Any reason why Kilauea is outgassing extensively after E32 ended? See https://www.youtube.com/watch?v=fiyttmA7YkA as both V1cam and V2cam show this. This seems heavier than its ever been, since these episodes started. I did notice the progression of the outgassing the last 3 or 4 episodes as it gradually changed location of maximum vapor release, but it seems to have increased now.
It looks pretty normal, until there is actually lava at the surface the plume is mostly condensation from tbe atmosphere, probably onto H2SO4 or other sulfates as the SO2 oxidises in the atmosphere. As with all clouds this varies with humidity and its overcast there now.
If it was doing this much but in direct sunlight and with lava at the surface already that could be different. But its been seen now countless times that visible onset or increase in degassing only happens within seconds of the lava surfacing in a new spot. In E32 there were other streamers who caught some of the vents opening and even through the tepgra cone this applies.
Also, back in 2020 Kilauea went from quiet background to erupting in an hour, green to red on the watch scale. It doesnt give warnings like that. HVO speculated it might heat up the lake or SO2 increases but no it is 0-100 immediately.
As we saw with recent episodes, the volumes of the episodes is growing. So there is obviously a big magma body waiting for the next episode. This magma body is degassing now. The big magma volume releases obviously big gas volume. Otherwise the episodes would have erupted more explosive, gasrich magmas.
It is, but the visibility of the plume is weather related. Also most of the fume sources are going to be passive degassing of the thick and probably welded tephra build up behind the vents on the wall, its probably over 150m thick in most of that area. Some could also be degassing of lava intruded into the tephra from the vent area as this did happen in E32.
https://www.facebook.com/share/p/1B93rYAc2j/
Pretty much perfect comparison angle of Kilauea in April 2018 and August 2025. Shows both how big the 2018 collapse was, but also how fast it actually filled in so far too. Theres still quite some way to go, 90-100 meters in the east caldera, and about the same at the vents with the higher rim and more so with the cone growing.
The first flows outside the caldera from this eruption will probably be from a satellite vent of the main one like E30, or maybe spatter fed if the fountain is diverted west over the rim. That may happen basically now actually, if the fountains are more vertical for E33, as seems to be more likely after E32 looked at the end and the south/middle vents getting stronger. But the seeming incease in seismic activity under the caldera is bringing in some uncertainty now too.
So according to Geologyhub, Laguna del maule is having seismic crisis, 5,000 quakes/day. Peak quake just happened 4.2. Pretty impressive for the system.
The surface of the lava field is now 480m above the bottom of the crater lake 2019-2020:
This is more than 50% of Stromboli’s altitude (924m) above sea level. Since December 2020 Kilauea has erupted vertically more than half of Stromboli.
E32 was wie 9 million m³ three times bigger than the Napau 2024 eruption (2.9 million m³). The episodes have the size of many single historical eruptions.
Albert, would it be prudent to say that the uplift on the J605 station is not caused by magma entering the sill beneath Motoyama? The validity of the main point in my upcoming post depends on this.
https://geodesy.unr.edu/NGLStationPages/stations/J605.sta
Motoyama seems pretty dead. The movement shows that the source of uplift is underneath the main island, and Motoyama is uplift in the main inflation, not specifically from its own reservoir.
The uplift was dead up until the recent eruption and now it’s even higher than before! It’s not clocking on these graphs.
Check out the August report.
https://www.data.jma.go.jp/vois/data/report/activity_info/329.html
I think the uplift shown on the J605 station isn’t from the magma intrusion into the sill but from magma entering the chamber.
The uplift isn’t centred in Motoyama though.
I am talking about the long term trend not the current uplift spike. Basically, I am asking if magma entering the sill beneath Motoyama would cause uplift at the J605 stationm
A mogi model. The deeper the chamber, the wider the uplift. If the magma is 500 meters deep, then it would need to be underneath Motoyama to cause inflation there. If it is 2 kilometres deep, then it can be underneath the centre of the island and still cause uplift at Motoyama. If the magma is in a wider sill underneath the island, then the inflation also becomes more spread out. With only three stations, we can’t easily tell the difference. But we do know that there are eruptions around the ring fault and it makes sense to place the magma sill inside the ring fault. That would inflate Motoyama, as it is not too far outside the ring.
Ok, thanks.
It was my interpretation that these were being the cone-sheet intrusion and not the ring-fault, could this still be possible?
Certainly. The ring fault gets its magma from inside the ring, after all.
I think you mean Suribachiyama, Motoyama is the main Island? Best as I can tell, Surbachiyama averages about 20-25% the inflation of the center of the island. Which still is a stupendous amount of uplift, there are beaches many meters above sea level, but shows it is sort of off to the side of the center of uplift.
Sorry! You are right. It is where I think this particular gps is located but I got my naming wrong!
Slightly off-topic, but a coworker recommended a hot spring to me earlier, and when I looked it up, I discovered that it’s in-between Aso and Kuju. Talk about a spicy location, no wonder it’s hot.
Just an observation that I have made before. HVO is now marking quakes with less than 1 mag on the main map you pull up when opening the website. I know that they mark them on the map below, but when they start marking them on the main page, it makes me think they are paying close attention to them?? Cal me crazy.
Seismic Data
Earthquake Hypocenters Map and Cross Section
025-09-08 18:08:09
Earthquake
Magnitude:0.8M
Depth:0.6mi
2025-09-07 23:38:12
Earthquake
Magnitude:0.9M
Depth:1.7mi
2025-09-07 11:23:40
Earthquake
Magnitude:1.6M
Depth:1.4mi
2025-09-07 11:23:33
Earthquake
Magnitude:0.9M
Depth:0.9mi
Mac
Ice been wondering about it too. These abundant small quakes very closely resemble the high pressure quakes that occur before intrusions.At the very least, it could be that the current eruption only has a couple weeks, my guess based on the location is that another fissure will open in the caldera probably around the ring fault of the 2018 caldera. This may or may not be a permanent change.
If one or both rift conduits starts flaring up this changes things a lot.
Magnitude 3.3 earthquakes near Hekla.
From RÚV:
Biggest Vatnafjöll quake for four years (8 Sep)
Multiple quakes (9 so far) since 11:00 PM last night on the southern end of the Juan de Fuca plate near the Gorda Ridge:
https://earthquake.usgs.gov/earthquakes/map/?extent=40.7431,-129.4519&extent=43.50872,-123.29956&listOnlyShown=true
Most of them are between lower 3 to lower 4, but the first quake was a 5.8 165 miles WSW of Port Orford.
There was also a 4.9 about 2 hours later close to the site of the 5.8 quake.
Most recent quake at the time of this comment was a 3.6 at 11:45 AM just barely on the opposite side of the Gorda Ridge from where the other 8 quakes occurred.
2 more quakes in now in the same general area:
– a 4.9 around 2 PM
– a 5.1 about half an hour later
Don’t quote me on this but isn’t this a normal? It’s not on the cascadia zone proper.
The activity is within the Blanco Transform Fault Zone, and is riddled with faults…mostly strike slip like the quakes in this ongoing swarm. Activity here is frequent with M4-M5 mainshocks happening regularly (every year or so) on different faults.
Two things that’s unusual is what appears to be triggered shocks (not within the immediate aftershock zone) on both sides of the Juan de Fuca ridge…as well as how large the aftershock zone is. Pretty large for a quake not even an M6.
Some unusual stress transfer must have taken place?
Increasing CO2 emissions of Vulcano are probably caused by an increasing inflow of magma to the magma chamber: https://www.youtube.com/watch?v=CJQogWIrlRw
An early sign that the volcano is awakening slowly?
While Etna and Stromboli erupt Strombolian, Vulcano usually erupts Vulcanian as its name says, so an order bigger and more dangerous.
About Ioto. From what I’ve been able to find out, there were 10-15 cm of uplift over the course of 5 days leading tp the eruption. The eruption on Sept 1 came from a short fissure in a long-lived fumarolic center; it produced pyroclastic surges that destroyed a desalination plant and buried a radius of a few hundred meters around the vents. The explosions had incandescent material in the daylight, so they must have erupted fresh trachyte magma. Since then, I’m not too sure if anything else has happened,
Thanks, that’s a bit concerning if eruptions of fresh magma are happening on land now as opposed to erupting out of the ring fault.
Out of curiosity, do we have any good up-to-date models on potential tsunami risk of Iwo Jima in the event of a caldera formation event? I thought I had seen that some of the initial estimates on tsunami risk in earlier Iwo Jima posts at Volcanocafe were somewhat overestimated due to not modeling the tsunami as a point source?
I have to imagine that Hunga Tonga is an useful case in being able to model this type of event, but I also know that the scaling and intensity is not linear here.
I’ve been silently lurking a bit more than usual lately. With too much other stuff to do, I haven’t had time to write comments here, but I do read the articles (and most of the comments) and I did notice the mention. Thanks for a nice article!
Speaking of Bardarbunga, it’s not showing any signs of slowing down. If anything, the KISA GPS seems to be accelerating away from the caldera. That means that inflation keeps on going and that we can expect another trademark CLVD quite soon, maybe around October. In the meantime, South Iceland Seismic Zone is stirring a bit. One star at Vatnafjöll a couple of days ago and now another one smack in the middle of the zone. I’m keeping my eyes on that area for now.
New post is up! The ugly duckling sea
https://www.volcanocafe.org/the-north-sea-and-the-zuidwal-volcano/