It began 2 April 1991, 35 years ago this month.
Hidden targets
There are many volcanoes in the world. At any one time, perhaps 20 volcanoes may be erupting. Over a full year, on average around 50 volcanoes erupt. Some of these may erupt several times – as Kilauea has done episodically in spectacular fashion, while Great Sitkin has been quietly and continuously erupting for years. Around 500 volcanoes have had known eruptions in the past 500 years, and another 1000 volcanoes are known as potentially active, in the sense that there are indications of eruptions in the past 10,000 years, e.g., a caldera! Around 30 of these are in Iceland and according to the USGS, there are 170 in the US. Who knew! Volcanoes that haven’t erupted for more than 10,000 years are generally considered extinct – unless it is Yellowstone in which case even 50,000 years is seemingly not enough for a downgrade.
Detailed monitoring is limited to volcanoes which are deemed to pose significant hazards. Extinct volcanoes are generally left alone. Potentially active volcanoes also tend to be given limited attention. Even the 500-odd volcanoes with historical eruptions may be left wanting. Most attention is focussed on a minority of most frequently active volcanoes.
In some cases we may still be too relaxed about well-monitored, active volcanoes. The recent past is not always a good indicator of what is coming: not all eruptions are the same, even when from the same volcano. For instance, Etna is amongst the most frequently erupting volcanoes on Earth, with eruptions that are notable but are not too significant. However, in 1669 Etna destroyed villages with massive lava flows which even reached the walls of Catania. Going back further, before 15,000 years ago Etna had massive explosions with ash reaching as far as the future site of Rome. And it had suffered massive land slides, luckily in the distant past. When monitoring Etna, we do so against the backdrop of those events, not just the recent photogenic eruptions. Krakatau is another example. Anak Krakatau was regularly seen erupting, and photos of its eruptions were used as eye candy on many websites. But the sudden collapse in 2018 came as a surprise and killed 400 people. In hindsight the signs of the instability had been there for many months before. The best observed volcanoes are not necessarily the best monitored.
Volcano monitoring is now an international endeavour. Satellite imaging transcends borders. Expertise in analysing seismographs exists in many countries, but the manpower (or funding) may be lacking: it is something where countries can help each other. As has been proven in the story of Pinatubo.
The beginning
Perhaps it started 16 July 1990 with a large earthquake. The magnitude 7.8 earthquake occurred on a fault zone running through north central Luzon, the Philippines’ largest island. Steam clouds were seen from one of those unmonitored, unrecognized volcanoes: Pinatubo, and small earthquakes were felt. Catholics nuns living close to Pinatubo went to Manilla a few weeks later, 3 August, to report these events to PHIVOLCS. PHIVOLCS investigated and found evidence of a large landslide, but that the steam had come from the known thermal area. PHIVOLCS consulted with USGS but there was not enough to call it volcanic activity, The investigators concluded that the landslide had uncovered more of the thermal area but otherwise they did not suspect any changes.We still don’t know whether this was the beginning.
History
Pinatubo had attracted little attention. Luzon has over 30 known volcanoes, of which three are frequently active: Taal, Mayon, and Bulusan. Of the 30, 12 are considered extinct. Pinatubo is located in a mountainous region on the west of Luzon. There are two other volcanoes in this mountain range: Natib and Mariveles. If you have never heard of them, before the 1990’s you would not have heard about Pinatubo either. Mariveles has a 4-km caldera, but has not erupted for 4000 years. Natib has a large caldera, 6 km across, but has not erupted for perhaps 50,000 years. The region was not high on the list of volcanic hazards! It was clear that the rare eruptions here were caldera-size, but there are many places in the world where that is the case and those eruptions were long ago.
Pinatubo consisted of a set of separate domes located across an old caldera, within a densely forested region. The tallest dome was around 1700 meters high, but stood only 600 meters above the surroundings and 200 meters above the next tallest neighbouring mountain. Even the USGS described it as ‘inconspicuous’.
Those surrounding mountains were all part of Pinatubo. They were relics of a stratovolcano which had self-destructed, after which modern Pinatubo grew up on the ashes. Volcanic tuffs are widespread in the region. Although it was clear that they had come from this general direction, Pinatubo had not been identified as the source, and one researcher even described Pinatubo as ‘non-volcanic’. Only in the 1970’s was Pinatubo itself studied in more detail, as part of a site study for a potential nuclear plant. (The plant was built in the 1970’s, about 50 km south of Pinatubo, but it was never commissioned. That is an interesting story.)
Radar image of Pinatubo, March 1991, showing a region around 30 km across
Warren Smith explored the region over a century ago and wrote in ‘The geology of Luzon’ (1913) ‘The rocks are volcanic extrusives [..] there are no active volcanoes along this line, and the old volcanic stocks are pretty well eroded.’ He added a footnote ‘It was reported by Mr. Snyder of the Bureau of Lands that smoke was seen issuing from the top of one of these peaks.’ Sadly, he does not say what he meant with ‘these peaks’!
In 1983, Wolfe & Self (Geophysical Monograph Series. The Tectonic and Geologic Evolution of Southeast Asian Seas and Islands: Part 2) summarised what was known about the mountain:
Mount Pinatubo. The highest volcano on the Bataan Lineament is a composite structure of calc-alkaline affinity rising to 1745 m. It is located at the point where the Iba fault zone, a graben, intersects the Bataan Lineament. There have been no historic eruptions of Mount Pinatubo, but there are reportedly two roaring vents on the summit and sulfur is actively depositing. Dates of 635, 2330, and 8050 years B.P. have been obtained by the I4C method [Wolfe, 1981] from material within mudflow (lahar) deposits. Since there has been no thorough study of this volcano, there may have been younger eruptions. One of the most recent eruptions resulted in a voluminous plagioclase-rich crystal tuff (ignimbrite), which filled stream valleys southwest toward San Marcelino, northwest into the Iba graben and north-easterly toward Tarlac and Angeles.
The Bataan lineament is the series of volcanoes on the west side of Luzon, stretching from Taal northwestwards. Pinatubo is 100 km beyond the end of this lineament but the authors state that it is related. The ‘roaring vents’, presumably the geothermal area, seem a bit of a concern. There is no source given for this information, but it would be unexpected from a volcano that had not erupted for centuries! Something similar in, say, Auckland would set alarm bells ringing. It is worth noting that the local Aeta people did mention small explosions having occurred during previous generations.
Pinatubo was known since the 1960’s to have a geothermal field, on the northwest flank within a 150-m wide depression. The field had weak steaming with some sulfur depositions and hot springs. Plans had even been made for a geothermal energy facility. A few exploratory wells were drilled in 1989 and 1990. Results were presented in 1990, by D. Michels who stated ‘A fumarole near the summit emits gases with magmatic characteristics’ (https://www.osti.gov/biblio/888499). In view of later developments, this may have been telling! The high acidity of the fluids made them unsuitable for exploitation, being too corrosive. They provide a geological cross section of the mountain:
Screenshot
The wells revealed a temperature of 330C at 1 km depth, seemingly higher than expected. They also give the cryptic statement of ‘increased boiling due to rock heat that increased during later stages of testing’. (The testing was done between 18 Dec 1989 and 26 Jan 1990.) They also found excess CO2, indicative of magmatic fluids. This was all interpreted in terms of the eruption of 500 years, with high temperatures attributed to dry rock which had cooled only slowly. In hindsight, was this instead an early indication of a re-awakening of Pinatubo?
To the east of Pinatubo are parts of an old, 4-km wide caldera wall. The remnants of ancient Pinatubo are outside of this wall. This old Pinatubo existed 1 million years ago, and includes several of the nearby peaks. After a long phase of quiescence, self-destruction occurred more than 35,000 years ago, in a VEI-7 eruption which formed the old caldera. It is known as the Inararo eruption. After that, a new Pinatubo grew up in the same place. It had a number of ‘eruptive episodes’ (4 or 5?) separated by quiet phases. The last activity before the 20th century was around 500-600 years ago. it left deep ash beds in various places. It should be noted that it can be difficult to distinguish old pyroclastics flow deposits from those of lahars, so it is not always clear what is caused by what, and this can leave the size of prehistoric eruptions unclear.
Awakening
This much was known before the initial warnings in 1990.
Nothing happened for the next 7 months. Starting on 15 March 1991, local residents began to feel earthquakes. The steaming from the thermal area had not changed and no rockfalls had been noted. It is not clear whether this information made it to PHIVOLCS. The local area was inhabited by Aetas, from the indigenous population, who perhaps counldn’t easily pick up the phone to Manilla! This information later came through members of the Negrito People’s Alliance of Zambales (LAKAS).
The earthquakes increased in the morning of April 2, and large phreatic explosions started at 4pm that day, lasting into the evening. A strong sulfur smell was noticed. Again the nuns went to Manilla to report to PHIVOLCS, two days later. The PHIVOLCS Director immediately sent an airplane to carry out an aerial survey. It found nine thermal vents along a line close to the original thermal area, and a line of deep craters which had formed in the explosions.
Location of the 2 April 191 phreatic explosions. The original thermal area was at the top of the Maraunot river, in the same general area.
Pinatubo in April 1991, showing some of the new steam vents. The gash on the left is from the 2 April explosions. Source: Wikipedia
The investigators still did not think this was unusual. It was just steam activity from a known geothermal area. The PHIVOLCS Director, Ray Punongbayan, disagreed and felt it should be investigated further. PHIVOLCS installed the first seismograph already in the afternoon of the next day. It was located downstream along the Maraunot river and radio-linked to their headquarters. The result immediately moved Pinatubo way up their list of priorities. 223 earthquakes were detected in the first 24 hours. Four further stations were installed during the following week, but now with data written to a floppy disk (remember those?) to be retrieved manually. Daily volcanic updates were issued. The seismic activity decreased somewhat during April but did not cease. Two of the vents remained strong and even carried some ash.
Assistance
This was when the USGS became involved. It had a program to help developing countries with exactly this. The program was funded through USAID (recently abolished), and was called the Volcano Disaster Assistance Program – VDAP in short. It kept a supply of equipment ready for just such a crisis. The Clark Air Base was located close to Pinatubo, and they asked VDAP to come with the next available flight saying “you have $20,000, and do not feel compelled to spend it all.” On arrival, the first thing they saw was an advert for the TV series ‘Last days of Pompei’!
By the end of April, a ‘Pinatubo Volcano Observatory’ (PVO) had been set up in empty bed rooms at Clark Air Base, making use of the on-site infrastructure. The air base had good reason to assist: built on deposits from a previous Pinatubo eruption, it could consider itself in a potential line of fire!
The Pinatubo Volcano Observatory in action. Source: USGS
A variety of instrumentation had to be installed, in a mountainous, sometimes forested and inaccessible area. The US army knew the area well as they used it for training exercises. They provided helicopters for transporting seismographs, tiltmeters and one gas spectrometer, linked to the PVO office by radio. In these ancient times, the most important piece of equipment was possibly the printer. Some of the lessons learned from the St Helens eruption came in handy, such as, don’t station your observatory people within reach of the eruption! This became an issue for both teams: PHIVOLCS observers stationed at Poontabo, 20 km northwest of Pinatubo, and USGS at Clark Airforce Base at a similar distance east of the mountain. In the end, both of these observatories had to be abandoned.
A map of the area with both observing sites. This is after the eruption: at the time the riverscape was quite different and Lake Pinatubo still was a mountain
The PVO was a collaboration between PHIVOLCS and USGS/VDAP, helped by the fact that the two lead scientists knew each other well. That was important: there was a danger that warnings given by a US institution would be perceived as aimed solely on the Clark Air Base. The local population might receive them with delay or might ignore them: PHIVOLCS would know best how to reach them. Vice versa, the US army would be more likely to heed warnings by a US institution! (In the end, they almost didn’t.)
Clark Air Base was enormous, extending over some 40km2 and located as close as 15 km to Pinatubo. The main housing of the site was in the hills east of Pinatubo, even closer to the mountain. The base was used during the second world war (mainly by the Japanese), the Korean war and the Vietnam war. But Pinatubo was a war it lost: after the eruption it was abandoned, the US lease was not renewed and the base remained empty for years.
Warnings
The first task of the combined group was to investigate the volcanic hazard map. That was easy – there wasn’t one. Instead, the group drew out the largest extent where deposits from the previous eruptions were found (using a set of new radiocarbon age determinations) and designated that as the warning area. This turned out (in hindsight) to be both remarkably accurate and a major underestimate. Evacuation plans were made for the quarter of a million people within the hazard area, of whom a third later actually evacuated. The volcanologists found that the flat area of the air base (ideal for landing strips) had been created by Pinatubo pyroclastics, which put themselves (unwisely) in the hazard area.
The seismographs showed that the mountain was rumbling but no clear pattern emerged during the next weeks. There was however a tenfold increase in SO2 emissions, a warning sign of the presence of fresh magma. The earthquakes were centred about 5 km NNW of the mountain, in a region with young-looking faults, and were at a depth of around 5km. This was not near the steaming vents: that area also showed earthquake activity but shallow and much less intense. The large extent of the earthquake region indicated this was more than a minor plumbing issue.
Alert levels
An alert system was created involving the local authorities. The system of numeric warnings worked very well: it was easy to comprehend for the public, who were often unfamiliar with volcanic threats and technical language such as ‘tephra’ ‘magma’, etc. Nowadays, different countries use different system for warning levels. PHIVOLCS still use this same system. New Zealand uses 6 levels. The US, perhaps more visually oriented, uses colour coding. But the basic purpose remains the same: to reduce the information to a single word. The alert level was raised to 2 (straight from zero) on 13 May. This was the first official forecast of a chance of an eruption.
A video made by Maurice Krafft on volcanic hazards became very helpful in communicating the dangers. Nowadays we know well how effective videos are in convincing people, but this was still experimental in the 1980’s. (In a sad coincidence, Maurice and Catherine Krafft died on 3 June that year, in a pyroclastic flow on Mount Uzen.)
Run-up
The steaming intensified during this period, and contained increasing amounts of ash, turning the white steam brown. On 28 May, the ash cloud reached 5 km high. The steam acidified the Maraunot river which reached a pH of 2.45 in early April, having been pH neutral before. Plants withered and aquatic life died.
After this, the situation began to change, with more signs of intent. The earthquake activity shifted to the region of the steaming vents, and grew more shallow. Harmonic tremor was seen. At the same time, the SO2 emission declined tenfold, a sign that a plumbing conduit had become inactive or blocked. On June 3, an evacuation was ordered for 20,000 people within 10 km of the summit. The alert level was raised to 3 on 5 June.
On 6 June, after an initial quiet period, a shallow earthquake swarm in the venting area, combined with inflationary tilt on the east flank, heralded that an eruption was becoming likely. At this time PHILVOCS moved their seismographs to Poonbato, further from Pinatubo, as they were becoming dangerous to reach. The PVO office was moved to the far side of the Clark Air Base almost 10 km further from Pinatubo – in hindsight, still too close.
A lava dome began to form on 7 June, just north of the venting area. It grew quickly, leading to an alert level increase to 4 on that day. The evacuation radius was now set at 18 km, covering 120,000 people (later increased to 25 km). The evacuation area also included the air base, but the army was not yet convinced and stayed put. On 9 June, ash clouds were seen rolling down the Maraunot river. This was interpreted as the onset of an eruption, and the alert level was raised to 5 – perhaps a little premature. Only on 10 June did the Clark Air Base evacuate. That was just in time: the first real eruption started June 12. (The 20-km high ash cloud was called the ‘appetizer” by the volcanologists,)
On June 15, 2am the eruption went for the cataclysmic main course. Pyroclastic flows nearly reached the Air Base, stones fell like rain (which also fell) and the volcanologists in their more distant office had to shield deep inside the structure not entirely sure of their own survival. Ash blanketed a region far outside the perceived danger zone. In one mountain valley, the ash was later measured at 200 meters thick.
In the end, 58,000 people evacuated, many ending up in refugee centres. The PHIVOLCS/USGS warnings had been spectacularly effective. Without them, the death toll would have been in the tens of thousands. Now, ‘only’ 300 people died in the eruption itself, and 100 more died later in lahars. (The oft-quoted death toll of 847 includes deaths that occurred in the evacuation centres, from dirty water and from measles to which the Aeta had little resistance.)
The forecasts of the areas at risk of pyroclastic flows and lahars had been accurate. But a major factor had been missed: the area of ashfall extended much further out, and combined with the heavy rain from a typhoon that hit during the major eruption, caused roofs to collapse in cities and places well beyond the danger zone. Even the US navy base to which the Clark Air Base evacuated was badly affected – the evacuees should have gone further.
There are many vivid descriptions of this eruption, still the largest eruption since Katma in 1912.
It seems amazing, in view of the damage and area covered, that this was ‘only’ a low VEI-6! A VEI-7 would have obliterated the land out to a three times larger distance, affecting millions.
Afterthoughts
The common thread in discussions at the time involved uncertainty. An eruption is forecast, not scheduled, and the forecast comes with a probability. But the affected people can’t handle probabilities. They won’t go until there is certainty. (Even the air base was not fully evacuated until they saw the volcanologists panic.) And if an eruption does not happen – there is always a chance – the people won’t listen a second time. It is a seemingly unsolvable conundrum.
How about the VDAP program? The outcome was a great success. But the role of PHIVOLCS in the reports seemingly diminished over time as the volcanologists focussed more and more on the impending eruption. There were separate activities: most detailed observations of eruption plumes came from PHIVOLCS, while seismic modeling and gas monitoring were done by USGS. Alert levels appear to have been decided by USGS. It feels like the role of PHIVOLCS at later times in the development was a bit underplayed. But it may have been different at the time: this impression comes from reading the reports, each written for a different audience.
I do not know whether VDAP will continue, at a time when the USAID program seems largely terminated. But I hope it will. There are 1500 volcanoes in the world that could potentially erupt. The major eruptions seem to come from volcanoes that had been quiescent for a long time: Pinatubo, El Chicon, Hudson, Katmai, Krakatau, Tambora. The next one may well have been missed from the monitoring programs. An organization that can step in at short notice to provide help where needed is sorely needed. At Pinatubo, it helped prevent a volcanic disaster. It can do so again.
As one of the scientists later noted: “What you think is the worst case, can actually happen“.
Albert, April 2026
Sources used for text and images:
https://www.usgs.gov/news/featured-story/remembering-mount-pinatubo-25-years-ago-mitigating-a-crisis
https://pubs.usgs.gov/pinatubo/sabit/
Other reading:
https://www.volcanocafe.org/when-pinatubo-turned-the-tide/
And of course our very own VDAP April-1 spoof of 2017:
https://www.volcanocafe.org/the-usgs-volcano-observatories-can-we-save-vdap/
A good read is the book Volcano Cowboys covering Mt St Helens and Pinatubo by the USGS. I still remember reading that a scientist sent to Pinatubo discovering a 200m thick volcanic tuff and realizing how big eruptions could be. Interestingly the above write up by Albert mentions exactly that thickness of a similar one left by the “recent” eruption.
https://www.goodreads.com/book/show/178244.Volcano_Cowboys
Reminds me of the 1993 NOVA documentary about the Pinatubo eruption. There’s a point where John Ewert talks about how him and Dave Harlow were looking at the geography surrounding Pinatubo when they first arrived after activity started and thinking “Wow! This looks like Katmai!”
Always good to revisit this eruption good article.
When it comes to regions most negatively affected by the loss of VDAP. Africa and South America is on the tip of my concerns
Also Central America. I don’t know whether VDAP will continue without USAID. The USAID funding has been taken back into government and sits in the same department as USGS. And is seems to me that any program that is both cheap and has saved the lives of thousands of American soldiers has a strong case.The world has changed since the 1980’s and the kind of help countries need will be different now. But volcanic risk is international and still needs cooperation across borders
I’m not so sure that VDAP is needed any longer. Many countries have stepped up and are not only putting out good research, but boots-on-the-ground monitoring and civil defense coordination.
I can’t recall any USGS personnel at Tajogaite, Soufriere (both of them) and come to think of it, the agencies in the Philippines seem to be self-sufficient.
Needs have changed and nations are more self sufficient. But techniques have also changed. At Pinatubo, the USGS brought software to locate earthquakes, something that wasn’t available locally. Nowadays, there are many more earth-monitoring satellites but the data is not always open-access, and specialist knowledge is needed to analyse the data. And different countries tries have different resources. Indonesia is well organised. How about Guatemala or Peru? Congo? Goma is at high risk and monitoring is limited.
I would compare it to earthquakes: many countries have local resources but still ask for help after an event. For volcanoes, the important difference is that eruptions can be forecast – in principle. That is where programs like VDAP can still make a difference.
The world would be more resilient if more wealthy nations or the United Nations developed volcanology missions like VDAP. Sadly the EU volcano programs haven’t grow big enough to play a role like USGS or VDAP, although Italy, Spain, France and Iceland (linked to EU by EFTA) have experience with volcanoes and volcanology. https://www.epos-eu.org/tcs/volcano-observations/services
At https://phys.org/news/2026-04-super-magma-reservoirs-beneath-tuscany.html “Super magma reservoirs discovered beneath Tuscany”.
Thank-you for you analysis and chronology about Pinatubo Albert!
The eruption of Pinatubo was not a sudden volcanic explosion, but one that showed clearly before the big bang that it was going to do something awful/awesome. This is unlike the classic Plinian Eruption 79 AD, when Vesuvius exploded out of nothing in the historical record. Maybe we should distuingish between Alkali explosive volcanoes like Vesuvius with Phonolite or Tachryte magmas on one side and classic Silicic volcanoes with Dacite and Rhyolite on the other side. The historical Plinian eruption was an Alkali magma eruption, while the majority of big explosive eruptions has Dacite or Rhyolite.
interesting point. I expect that the signs of the impending eruption at Pinatubo were there long before they were spotted. Perhaps the presence of the geothermal area (first mentioned in the 1960’s) had something to do with it. The large earthquake may have accelerated a process that was already on-going. We don’t know much about the build-up toe the 79ad eruption. There was a significant earthquake 17 years earlier. But earthquakes are common.
Pintubo erupted Dacite. The question concerning the evolution before the eruption depends on whether the eruption was caused by the slowly growing Dacite alone or whether it was caused by a new and relatively sudden injection of hot magma (Basalt, Andesite) into the Dacite magma chamber.
Dacite is in relation to Rhyolite a relatively intermediate magma. So there must be a starting basaltic magma that evolves over long time into Dacite.
Hi, friends, I am struggling with the Tambora page in italian wikipedia; I’m putting the observations of Hector and others about the first caldera of Tambora (43,000 years ago) and its remnants in the comments here: https://www.volcanocafe.org/tambora-making-history/comment-page-1/
Furthermore, I want to report this article to you; a new model for the formation of explosive calderas: not a simple collapse, but a disintegration of them by giant explosions: https://www.scielo.org.mx/pdf/geoint/v61n1/0016-7169-geoint-61-01-5.pdf
this model can be applied to other calderas, Aira, etc..
Very well written chronology describing a perfect evacuation plan saving many lives.
Pinatubo had a chronology of an escalating eruption over some days. It started with Pyroclastic flows and minor Plinian eruptions. Day for day the Plinian eruptions grew in size and strength until the “king of events” on 6/15. After the cataclysmatic eruption a number of Plinian eruption followed that decreased in strength and volume.
St. Helens had a different chronology. An important reason for this was an Andesite plug/cork on top of the Dacite cryptodome. St. Helens is an example for a volcano with a broadly varying magma composition. It can do the whole spectrum from Hawaiian fluid basaltic magma eruptions until viscious Dacite Plinian eruptions. This varyiation shapes the volcanism of St. Helens and created the relatively hard Andesite plug. As a purely Dacite volcano the 1980 eruption of St. Helens would perhaps have followed the chronology of Pinatubo on a smaller scale.
St Helens was a flank collapse. Rather than an eruption clearing the conduit, the collapse exposed the eruptible magma. There were more differences than the magma type
Chaitèn (Rhyolite volcano) 2008 had a similar, but much shorter escalation chronology compared to Pinatubo. It began to erupt on May 2 and escalated within four days to the main eruption on May 6. It had a VEI4, so was up to 100 times smaller than Pinatubo. This may explain why the path from the first Plinian eruption to the biggest one was short.
Is this the typical behaviour of Dacite/Rhyolite volcanoes that they have growing Plinian eruptions until the big one?
Tambora also did it in 5 days. But do remember that Pinatubo’s early explosions were phreatic, caused by ground water. In a drier environment, perhaps the June 10 eruption would have been the first with 5 days notice for the main event.
Thank you for another very enjoyable article, Albert!
Kilauea is thinking about a rerun. Hydrogen flaming is on-going.
Pu’u O’o is constantly deflating since the summit eruption began in December 2024:
Does this mean that magma is leaving the ERZ towards the summit?
After 2018 Pu’u O’o first had positive deformation until late 2020, when the first summit eruption began. Since the end of 2020 Pu’u O’o is predominantly constantly deflating, although this trend was interrupted by the short phase surrounding the Napau eruption 2024:
OUTL is down by 30 cm over the same period and AHUP by 20cm, while NPOC is down by 10cm. The magma is certainly not heading in that direction! UWD tilt is edging down. There are not a lot of GPS stations to go on, but it looks like magma withdrawal from Kilauea. As it isn’t going into the rifts, it is instead moving to the surface. The eruptions are depleting the deeper magma.
That last episode was really small so recovery has been quick. Looking just at instruments and previous eruption starts, I’d say we could be as little as one week away.
HVO thinks that it could be even sooner; their daily update forecasts an eruption between Apr 21 and Apr 26.
There seems to be some deformation at the southern edge of the 2018 caldera…
Image removed because of being oversized. Smaller version shown below – admin
collapse slide along the flank?
HVO commented in the April 16 Volcano Watch that uplift occurred (possibly up to 12½ cm) during an earthquake swarm which began during the episode and continued for a short time thereafter; some were felt. However, they don’t know what its nature is, but a collapse slide seems to me out of the question considering it went up. They did comment this was the same region where earthquakes had been recorded the past weeks to months.
The fringes are on the upper side of the fault scarp. In a collapse, this side of the fault would move up, so there’s actually no contradiction. There should be matching fringes indicating downward movement on the other side of the fault, but there’s a lot of freshly erupted material there which I assume could cause a lack of coherence in the measurement. Compare with images from Iceland, where you often se a lack of fringes in the fresh lava fields.
I see, thanks. So it could just as well be a further crater collapse due to the weight of the lava emplaced relatively rapidly over the past 1½ years. That leaves me with another question, though, and that is how I should envisage the relation between the swarms and subsequent subsidence noted on the summit tiltmeters. Should I view it as the whole caldera block going down and thereby “flattening” the magma chamber or as a chunk of that block subsiding and acting like a wedge? The former would need quakes all over the caldera (and they certainly have occurred when looking back to the beginning of the year), though individual swarms tend to only be a limited area (the remainder slips aseismically or the quake’s location merely represents the starting point?). With the latter I would expect some form of uplift detectable elsewhere.
Whatever it is, it is remarkably focussed. Wait and see what happens! The earthquake underneath the vents is interesting. The north vent has been quiet since
2.10 M earthquake below the SW corner of Kilauea’s caldera, very close to the vents and 280m above sea level (~700m below the surface).
HVO predicts that “episode 45 lava fountains are likely to occur sometime between Tuesday, April 21, and Sunday, April 26.”
Ongoing earthquake swarm at Katla. Shallow depth.
I feel for sure there will be an eruption here, or at Eyjafjallajokull, or in the saddle in coming years.
Another large earthquake and tsunami warning in Tohoku.
The whole boundary between there and Kamchatka seems very unstable recently.
M7.5
The highest tsunami was observed in the city Kuji with a height of 80cm. Not high, but perhaps with a dangerous current.
JMA is mentioning a 10x over background chances of a bigger quake with 250km. Though the chance is still ~1%, there have been several precedent sequences/events in the last 80yrs or so that had strong precursor events/foreshocks with similar faulting mechanisms that ended with a mega-thrust nearby. Of note, there is a seismic shadow north of this M7.7 event off the SE-E coast of Hokkaido which experts in the past had been mentioning as an area of concern following the 2011 Tohoku mega-thrust.
I’m sure I’ve mentioned this before, but I wonder what effect this seismicity is having on Mt. Iwate. It’s been slowly inflating for about 2 years, and that was before the nearby earthquake activity.
Albert,
catching up some. Your piece about Teide is very interesting. I think Teide can easily be compared to Fuji with its history with several Las Cañadas although Fuji seems to be older by some meagre 200k years and is located near a triple junction.
S.th. is left from the Guanches, and that is parts of the language. We see Gara like in Garachio more often on the Canary Islands. Gara was a Guanche female name.
Thank you for the Teide piece. I do love Teide as it was the first impressive volcano I saw (from La Palma). Before I had only been to Eifel and Auvergne, nice, but eroded of course.
Teide is a beauty from far when the sun is setting in the West and we can see it from the east coast al the way down to Fuencaliente.
Kilauea’s north vent overflowed at least 3 times overnight; so, the next episode is approaching fast.
IMO have a new hazard assessment out.
Grindavík and Hveragerði in highest hazard zone (RÚV, 21 Apr)
The hazard zone stretches all the way from the tip of Reykjanes to well to the east of Reykjavík, which is just about in the middle of the zone. My impression is they really have no idea where or if it’s going to erupt…
Says a lot about volcanoes, they absolutely refuse to do what you expect them to.
There was a series of earthquakes in the realm of Hrómundartindur NW of Ölfusa laguna (Selfoss). Hrómundartindur last erupted during early Holocene. There was a period of unrest 1994-1998, but without any eruption.
I asked Google’s Gemini AI about this article. Here’s the details
Randall: Please scan this article https://www.volcanocafe.org/eruption-and-cooperation-a-pinatubo-story/ and analyze the interaction between the normal people and the professionals. Are we prone to keep repeating these same mistakes? Mt St Helens was a huge wakeup, no one expected the flank collapse. How can the experts and ordinary people work better together?
Gemini’s response:
The interaction between professionals (scientists, government authorities) and “normal” people (the public, local communities) during volcanic crises—such as the 1991 eruption of Mount Pinatubo and the 1980 eruption of Mount St. Helens—represents a pivotal, often tense, negotiation of risk.
The article Eruption and Cooperation: A Pinatubo Story highlights that successful outcomes depend less on technical precision alone and more on the social infrastructure of trust.
Analysis of the Interaction
The Power of Grassroots Reporting: At Pinatubo, the initial alerts did not always come from top-down monitoring networks; they came from local community members (like the Catholic nuns) who reported changes to the landscape. This demonstrates that “normal” people are often the first sensors in a crisis.
The Burden of Technical Language: A recurring friction point is the gap between expert data (seismographs, gas emission rates) and public perception. Professionals often struggle with “throwing a hazard map over the fence,” assuming that data alone compels action.
The Fear of “Crying Wolf” vs. Catastrophe: Experts often fear the economic and social fallout of an incorrect evacuation order (causing panic or unnecessary economic disruption). The public, meanwhile, often normalizes danger until it is too late. The 1980 Mount St. Helens eruption was a watershed moment because it exposed how little even experts understood about flank collapses, forcing a radical shift toward integrated, multi-disciplinary, and collaborative monitoring.
Are We Prone to Repeat Mistakes?
Human psychology and institutional memory are notoriously short. While technical monitoring has improved exponentially, the “mistakes” are rarely technical; they are sociological.
Normalization of Deviance: When volcanoes remain quiet for long periods, communities often ignore hazard maps, treating them as abstract warnings rather than urgent realities.
The “Expert Silo” Trap: Scientists are trained to be objective and cautious. However, in a crisis, the public needs someone to interpret that uncertainty. If experts speak only in probabilities and technical jargon, they lose the ability to guide human behavior.
How Experts and Ordinary People Can Work Better Together
To bridge this gap, the relationship must move from instruction to co-production of safety.
Translate “Risk” into “Lived Experience”: Experts must stop using technical terms (like lahar or pyroclastic flow) in isolation. They need to map those terms to the local vocabulary and specific landmarks that matter to the community.
Establish Trust Before the Crisis: Crisis management cannot begin when the volcano starts rumbling. Trust is built in the years between eruptions through consistent, non-alarmist public outreach and education.
Validate Local Knowledge: Scientists should actively incorporate the observations of locals into their formal monitoring. When a community feels their “boots-on-the-ground” observations are part of the scientific process, they are far more likely to trust the final evacuation orders.
Manage Uncertainty Transparently: Instead of presenting a “go/no-go” binary, experts should be honest about what they don’t know. Communicating, “We are seeing X, which usually means Y, but there is a Z% chance of it being something else,” allows the public to be partners in the decision-making process rather than passive recipients of orders.
The goal is to shift from the expert being the “authority who knows” to the expert being the “trusted advisor” who works with the community to make the safest possible decision in an environment of high uncertainty.
That is pretty good but falls itself in the ‘technical language’ trap. The only thing it really missed is the fact that the community at risk involved two very different groups, the local population (itself diverse) and the US army, which required building up trust separately.
It is frighteningly good, just missing small details like Albert said.
I gave AI this q the other day: Could the Sperm Whale that sank the Ann Alexander in 1851 have been the same that sank the Essex in 1820?
I got a precise NO. Then I asked the next q an hour or so later:
As Sperm Whales reach an age of 60 to 70 years, single ones 80 years it cannot be excluded that the same Sperm Whale sank The Ann Alexander and the Essex.
I didn’t really ask, but made a statement (same area South Pacific btw).
AI: Yes, it is technically and biologically possible that the same Sperm Whale sank both ships.
However etc.
So, to take advantage of AI, we have to debate it.
The only proven fatal accident of that whale is the sinking of the Ann Alexander as that whale was caught five months later and had pieces of the Ann Alexander in his head. This doesn’t speak against him sinking the Essex (inspiration for “Moby Dick”) 31 years earlier as piece from the Essex might have disappeared over time.
It is clear that the AI used other sources of information as well, and terms like ‘expert silo’ come from elsewhere (and it is not really used in the normal meaning). One thing AI cannot do at the moment is correctly citing sources.
– One thing AI cannot do at the moment is correctly citing sources.
Hope it never can. It would be the rescue of all scientists.
For who’s interested:
https://thehistorypress.co.uk/article/leviathans-revenge-whales-that-attacked-ships/
I became intrigued.
It is only the Sperm Whale who occasionally becomes aggressive.
The larger Blue Whale is peaceful.
This is five stories, and there might be some more in Asia that we have no knowledge of. By far the worst story is the one about the survivors of the Essex, all the others were saved.
Off topic question:
Please see https://www.usgs.gov/media/images/electronic-tilt-kilauea-summit-past-three-months for the electronic tilt data. Was there any physical signs when the deflation occurred around April 9th or 10th? It seems like we’re building back up for another dual fountain eruption.
“The USGS Hawaiian Volcano Observatory (HVO) is detecting shallow earthquakes beneath the south rim of Halemaʻumaʻu crater, as well as localized ground deformation, in the south caldera region of Kīlauea summit”
Inflation has reached the level, when E44 started. Precursory activity is hesitant compared to previous episodes. Is this related to the earthquakes in the southern summit area? The earthquakes move towards two directions: 1. Koa’e fault zone, 2. Keanakakoi Crater. 1974 was a three days lasting eruption at Keanakakoi Crater. Is it a possible escape location for the current eruption?
The N vent currently does ongoing spattering. I believe that it began yesterday, first with sporadic lava bomb eruptions, later with more conitnuous spattering. The growth was slowly.
This shows that gas containing magma is in the vent. That’s different to the precursory activity of E44 and others which had overflows of degassed magma. It seems that partially gasrich magma has reached to the surface without the need to push out fluid degassed magma first.
Too wet to erupt.
This volcano that ‘slept’ for 100,000 years was never truly quiet (Phys.org, 22 Apr)
I think Methana is one of the most interesting volcanoes on Earth. It just might be why democracy exists, between giving agricultural surpluses to free up thinkers for thinking, plus producing vast amounts of silver via hydrothermal deposition, which allowed Themistocles to beat Xerxes at Salamis. Well the latter would’ve been from a previous Methana since the Laurion mines are Cretaceous, but it was a volcano that allowed Athens to sink the Persian fleet!
is this deep enough for free water to be supercritical?
Using the definition that supercritical the two phases show a negative latent heat of fusion.
I am not sure that calling the phases liquid or solid make sense here, but converting the less dense phase to the more dense phase consumes energy.
In principle yes. The paper mentions a pressure of 2-4 kilobar, which is a depth of ~ 10km under rock. But I doubt that water at this depth can be called ‘free’. The energy goes into the surface tension of the bubble.
Overflows now from frequent from one vent while the other is flaming.
The view on the V3cam is stunning. I was expecting a few more days before the next episode, but the way this thing is going right now, it might as well just go straight into fountain mode within a few more overflows.
M1.4 (still tiny) 10km below Hekla
That’s one of the most interesting Hekla quakes since the last eruption.
Is it Hekla or SISZ?
Hekla. The SISZ runs further south. Hekla has grown on a rift that runs almost perpendicular to the SISZ and the quake was at the bottom of at rift. Very small though!
North vent is making an attempt.
South vent joined in with more gas rich magma than the previous dome fountains. Will we get a double fountain this time?
It stopped before I clicked send, but north vent is still going.
Looking good
… and it looks like Ep. 45 is starting.
No turning back now 🙂
and thar she blows. V1 and V2 are nicely framed
Note how one vent fountains and one vent flames. Gas and magma take different routes.The speed of interchange between them at the start of the episode suggests the link is not that far underground.
Still single-vent and fairly wide. The north vent is acting as a safety valve and bleeds off the gas. That may be limiting the south vent fountain. But very pretty to watch
Two volcanoes in one shot
Updated to episode 45. Keen observers may note that the volume per µrad has decreased by 7300 m³ compared to E40’s graph I shared then, which is because the discrepancy between what I get from basically “pixel-counting” UWD and what HVO states for µrad lost has been ≤0.5 µrad – i.e. the maximum difference I allow between either number for the episode’s effused volume they state to count towards the average used here -, and HVO consistently noting a lesser volume effused than what using the volume/µrad gets me at.
Additionally, for episode 44’s total volume recharged in µrad I applied my own offset to UWD based on the summit tiltmeters and the start of overflow. This lands it at 31.2, rather than about 25 the graph indicates.
To answer my own curiousity (and perhaps of a few of us here, too), I’ve additionally added the volume in µrad lost or gained in the eruption overall. This is, up to episode 45, a deficit of 56.2 µrad, 20.8 million m³, or, in perhaps more understandable terms, about two slightly smaller episodes’ worth where both north and south vent are active. That doesn’t seem a lot, given that the south caldera chamber (which does the bulk of the heavy lifting) contains about 10 km³ of magma, according to the proposed cross-section of Kilauea HVO published January 2024.
Anyway, this is the new data with this update:
* Episode 41: Ahead was recharged 6.9 million m³ (18.6 µrad) at 6.8 m³/s on average. Effused during was 11.4 million m³ (30.8 µrad) at 381.1 m³/s on average.
* E42: Recharged 11.7 million m³ (31.6 µrad) @ 6.2 m³/s; effused 12.5 million m³ (33.8 µrad) @ 354.9 m³/s.
* E43: Recharged 11.4 million m³ (30.9 µrad) @ 5.9 m³/s; effused 12.6 million m³ (34.0 µrad) @ 385.9 m³/s.
* E44: Recharged 11.6 million m³ (31.2 µrad) @ 4.5 m³/s; effused 6.6 million m³ (17.9 µrad) @ 216.3 m³/s.
* E45: Recharged 7.0 million m³ (18.8 µrad) @ 6.1 m³/s; effused 5.9 million m³ (16.0 µrad) @ 194.8 m³/s.
Thank-you for your update J.O.!
E44 and E45 together had a significant excess of recharged magma over erupted magma. Is some magma building an intrusion or dike anywhere? The recent earthquakes in the southern summit region look like “magma at work”. On the other hand the longterm deformation trend was negative during winter 2025-2026. Is the magma chamber getting exhausted?
Perhaps a return to the situation before August. This is the first time we have had two ‘small’ episodes in a row since the start of the big events in August. My impression was that there wasn’t;t enough magma pressure to supply two vents simultaneously, and that this allowed the second vent to vent the gas needed to drive the eruption. It ran out of gas. Even Kilauea isn’t immune.
I haven’t checked the SO2 output during E45. Did it release relatively much gas without lava? Maybe there was more dissolved gas from the magma that left the magma chamber(s) without the force to put out an amount of magma. The south vent was during E45 and E44 mainly a gas erupting vent like a big fumarole.
Is it possible that with the evolution of the eruption there begins a process of differentiation between magma and gas? I imagine that this process can initiate a shift of the eruption to a steady behaviour like Pu’u O’o 1986 or the later Mauna Ulu phases. A degassed magma needs more time to get to the surface, but when it gets there, it will likely run without interruptions.
HVO writes about SO2: “During episode 44 on April 9, HVO was able to measure a sulfur dioxide (SO2) emission rate of 208,000 tonnes per day as lava fountaining was ongoing. Emissions during episode 45 were likely similar.”
On the website concerning the overall eruption HVO writes “Episodes in the ongoing eruption have produced emission rates up to 75,000 tonnes per day of SO2.”
This is less than the last two episodes. This means that the eruptions erupted a lot of gas without magma, and only ~half of magma.
Volcano Watch concerning the recent earthquakes and deformation: https://www.usgs.gov/observatories/hvo/news/volcano-watch-shallow-earthquakes-and-ground-deformation-kilaueas-summit
I think that the relatively gas poor magma that wasn’t erupted during E44 and E45 slowly intrudes in the southern caldera area. It is to heavy to get to the twin vents and slowly growths towards a news direction. The earthquakes occur only 1-3 km below the surface. We are likely witnessing a process towards a usual Hawaiian eruption unlike the super-Strombolian episodes from 2024 to now. Hawaiian eruptions need magma with less gas content than Strombolian eruptions of all size.
I imagine the possibility that we get a repetition of the 9/2023 eruption on the down-dropped block (now buried by recent lava flows). The location of the newest earthquakes tend towards the down-dropped block. It has a deeper elevation than locations outside the caldera. This map shows the 2023 eruption:
I think this was rather unusual. After the eruption ended around 10:00 Hawaii time, the wind shifted direction and volcanic ash rained down on the Kilauea V3 cam a few times around 12:20 to 12:22, especially at 12:22:15. This means that the ash was suspended for well over 2 hours. Is that possible?
Oops, I just noticed that the wind picked up briefly after it changed direction; so, the ash was probably blown by a gust toward the camera from the ground deposits to the west. Not so unusual.
There were quite a few dust devils active around the end of the eruption. They should probably be called ash devils
And elsewhere in the world, a failed attempt at winning a Darwin award
https://www.yahoo.com/news/videos/volcano-erupts-daredevil-tourists-race-153840534.html
The Pinatubo eruption resembled the Jurassic Park Scenario of a double disaster with a dramatic (like in a Spielberg movie) volcanic eruption or dinosaur threat and a typhoon combined. That’s an evil version of Murphey’s law that things which can happen together, will happen together once. The double disaster of Pinatubo’s eruption combined with the typhoon was more lethal than the eruption alone.
The chance for events like this is high in regions which can produce hurricanes and typhoons. Usually it’s the west Pacific Ocean from Japan to Indonesia. But during El Nino, as it is predicted to begin in summer 2026, the probability for hurricanes at the west side of Mexico is also elevated.
Was it the 2012 where the continents somehow shifted massively,, at the same time causing massive tsunamis worldwide, multiple category-6 size hurricanes all over the planet (including at the equator and poles which is near enough impossible). As well as Yellowstone erupting which somehow caused the planet to start rifting apart with magnitude 9 earthquakes in DC. And they escaped the pyroclastic flows from Yellowstone by going about 70mph in a camper van. Incredible really.
The Day After Tomorrow was another wildly realistic one.
I wants an Io specialized probe, I wants this until the point of crazy very sickly and badely. Maybe Elon Musk coud finaly one day get intrested in Io ?
I have been following Home Reef submarine volcano (well actually we have a real island now, about 3 years old) and the March 22nd 2026 picture from the Copernicus browser seems to suggest that we have an underwater flank eruption, to the west of the main volcano. See https://browser.dataspace.copernicus.eu/?zoom=16&lat=-18.99004&lng=-174.76586&themeId=DEFAULT-THEME&visualizationUrl=U2FsdGVkX1%2BTePrjTnvEsPHee5rxUtsP91qMaC3m%2FZ5B09o%2By%2FOtN6tBgXS5yEVb48RAkC7sQhcgL1J40n63S3j9KXsFMQSnmIdEzmHdjGwz7iLUWaj83w6SZiXslchp&datasetId=S2_L2A_CDAS&fromTime=2026-03-22T00%3A00%3A00.000Z&toTime=2026-03-22T23%3A59%3A59.999Z&layerId=1_TRUE_COLOR&demSource3D=%22MAPZEN%22&cloudCoverage=30&dateMode=SINGLE
July 1986 the 48th episode changed the episodic style towards a continuous eruption of Pu’u O’o. We’ve had 45 episodes since 2024 at the summit, so we’re close to the 48th episode. Will Pu’u O’o be in part a role model for our eruption?
https://volcano.si.edu/volcano.cfm?vn=332010#bgvn_198607
Thank you for the article, Albert. My wife is from the Philippines. From Pangasinan, which is about 126km from Clark Airport, she had said that ash fell in her town. It is good to read about it, but it’s not the same as experiencing. But reading does help with the comprehension of it.
Do you think any other volcanoes will erupt in the Philippines in the foreseeable future?
Thank you for the message! It is hard to know whether it could happen again. Eruptions like this are not common and happen perhaps twice per century.They come from volcanoes not previously known to be dangerous (or even active). The Philippines have many such volcanoes but so do other nations. I think that over our life times, the chance is small. Over a millennium, it gets more significant – to give a rough guess, I’d say 50%. We will have much more advance warning though than we did for Pinatubo. Monitoring has vastly improved and the PVO is an able (though perhaps not fully funded) organisation.
Mayon and Kanalon are both erupting right now,
New post is up! The fiery birth of volcanology – in latin
https://www.volcanocafe.org/pliny-vesuvius-and-volcanology/