The Water Volcano – The new Decade Volcano Program

Image from the 2014 eruption of Gunung Kelud. Photograph by AlexMG.

Image from the 2014 eruption of Gunung Kelud. Photograph by AlexMG.

What is a better starting point for our proposed list of new Decade Volcanoes than a volcano that operates in a way that almost no other volcano does? As the number ten of our list we hereby present a volcano that equally blends being an artesian well and an andesitic tephra producer of note. Without further adue, let us move on to Gunung Kelud.

This image shows the pure scale and brutality of the 2014 eruption. Photograph from Lembaran Tipis.

This image shows the pure scale and brutality of the 2014 eruption. Photograph from Lembaran Tipis.

Within a distance fifty kilometers out from the volcano lives almost two million people and the city of Malang (824 000 people) is inside that distance. As such it more than well has the potential of producing a 1 MDE (MDE = Million Death Expectancy) for a larger event.

Gunung Kelud is already one of the deadliest volcanoes in history with a confirmed death toll that has surpassed 10 000 dead, with half of those during a single eruption.

The modes of lethality include pyroclastic flows, lahars, volcanic ash and widespread scattering of volcanic debris.

Due to the form of its eruptions and way of operation it is likely to also be able to produce pyroclastic base surges and caldera forming events.

Before we move on to the details of the volcano I would like to state that this volcano is relatively well studied and monitored, but the way the volcano works needs to be better understood since it is the archetype of its own class of volcanoes. What we learn at Kelud can be applied to several other artesian crater lake volcanoes, and this type of volcanoes are among the deadliest known to mankind.


Indonesian army rescue workers. Photographer unknown.

Indonesian army rescue workers. Photographer unknown.

Kelud is an inconspicuously small volcano that is only 1 731 meters high. The edifice is craggy and unevenly formed; one could even go so far and say that this is certainly not one of the more beautiful volcanoes of the world.

The edifice contains several lava domes and spatter cones that gives evidence that the volcano has not always erupted through the massive wide bore crater that usually contains a crater lake.

The current crater formed during a Colossal explosive eruption that took place around 230 BC and subsequent Cataclysmic eruptions has further enlarged the crater. Especially the Paroxysmal VEI-5 eruption of 1586 deepened and widened the crater.

Gunung Kelud is well known for its highly explosive eruptions, in the last one hundred years the volcano has suffered no less than five separate Cataclysmic VEI-4 eruptions. The most deadly of the series was the 1919 eruption when the brim full crater lake was disgorged in seconds during the onset of eruption. This caused massive lahars that travelled through canyons and rivers a distance of 37 kilometers. The total death toll of this eruption exceeded 5 000 people, mostly through drowning or crushing as the lahars inundated them.

This led to the Dutch (colonial times) ordering an active mitigation scheme. A series of nine narrow tunnels was drilled through the mountainside so that the water could run off, thusly greatly diminishing the volume of the crater lake.

The latest of the Ampera Tunnels at Gunung Kelud. Photograph is taken prior to the destruction in 2014. Photograph from Pixoto.

The latest of the Ampera Tunnels at Gunung Kelud. Photograph is taken prior to the destruction in 2014. Photograph from Pixoto.

As the 1951 VEI-4 eruption happened the level of the crater lake had been lowered 50 meters and the Lahars produced were minute compared to in 1919. This eruption lowered the crater floor 70 meters and destroyed the old tunnels. This led to a new large crater lake forming, and during the 1966 VEI-4 eruption 200 people died from the lahars.

This led to the construction of new lower tunnels and as the 1990 VEI-4 eruption happened the crater lake only contained one million cubic meters of water.

During the 2014 massive VEI-4 eruption the tunnel was damaged and it is also reported that the crater floor once again has been lowered. It is as of now no news that new tunnels will be constructed, of if the old ones will be restored.

To date the tunneling at Gunung Kelud represents the only effective active mitigation campaign of any volcano on the planet. We therefore believe that it is imperative that the active mitigation program will be continued and that the effects of active mitigation of crater lakes can be further studied using Kelud as a natural laboratory.

Even though the 2014 VEI-4 eruption was very powerful only two people died as their roofs collapsed on top of them. The reason for the low death toll is to be found in the active mitigation as well as with the highly skilled work performed by the Indonesian authorities prior to, during, and after the eruption.

The upcoming eruption was tracked weeks in advance and evacuations were performed continuously as the authorities received new data from local scientists. At the same time the military prepared for the upcoming cleanup process and moved heavy equipment into the region. Within hours after the eruption roads was started to be cleared from ash and debris and within a month most of the 12 304 destroyed or damaged buildings had been repaired or resurrected. This is a feat most countries would be well advised to study. Never had a large natural catastrophe been handled better.

The two cycles of Kelud

Photograph by WB Wilapa.

Photograph by WB Wilapa.

What sets Kelud apart from its explosive andesitic arc brethren is that its eruptions are driven in a bi-modal fashion. It is generally too small to produce repeated eruptions on this scale. So what makes this inconspicuous volcano so powerful? Let us take a look and marvel at the answer.

First we have to take a look at the types of eruptions that occur at Kelud. The primary type is highly explosive and contains juvenile magma that enters from depth. This magma is high in volatiles such as water from the melting of the subducting plate and rich in volcanic gasses. This combination is what makes andesitic volcanoes so explosive compared to mantle derived basalts.

Kelud normally produces a series of explosive eruptions until the magma is depleted of volatiles. This leaves a “dreg” of stale but still hot magma in the magma reservoir and conduits. As that is pushed up it produces a gentle effusive dome building eruption that emplaces a volcanic dome. In the last one thousand years those domes has been emplaced inside the crater at the center, but there are several emplaced domes outside of the crater that has been left untouched.

The latest cycle of new magma intrusion probably started in 1951 and ended with the extrusion of a lava dome in 2007. This dome was most likely extruded partially because new fresh magma had started to enter from the magma acreation zone at depth. We will soon come back to what happened in greater detail.

Most crater lakes in volcanoes in the world forms as rain or snow falls inside an enclosed crater. These lakes are normally fairly cold and rapidly turn acidic from sulphur leaching into the water over time.

In Keluds case the water comes up from the ground. This is evidenced in that there is no connection between water entering the lake and the rain season. Directly after an eruption hot water comes up through the fractures left by the explosive eruption.

This initial water is high in chloride and holds a ph of 6 and contains potassium and sodium. The chloride content points to a deep source and the water is driven upwards by heat convection. There is a shallower source of water that has a slightly more acidic component in the form of a magnesium calcium sulphide.

Both types of waters start to emerge directly after the eruption, but over time the deeper source is gummed up and most likely squeezed shut as new magma comes up from the depth. This changes the water and makes it slightly more acidic (ph 5).

During the last stages prior to an eruption the water changes again as Boron is released and the acidity falls again.

Here we find the answer to the explosivity of Kelud. Not only is the magma high in water as a volatile, we also have pure water introduced directly into the volcanic system at depth. The amount of water introduced into the magmatic system is probably in the order of 250 liters per second.

In 2007 the lava dome was extruded into a very small crater lake due to the tapping of the water through the tunnels. It is easy to believe that the hot extruded dome did away with the water due to cooking it off, but this is just a small part of the truth.

Instead it plugged the natural passage ways of water up through the volcanic vent. Then the water pressure started to rise rapidly as the natural outlet was hindered. Basically the extruded lava dome made Kelud into a gargantuan pipe bomb.

The eruption in 1919 probably had seen a dome extrusion into the crater lake a few years prior to the big eruption. But then the lake was sufficiently large to quench the extruding dome making the process go even more rapidly. And as the dome was blasted out it dumped out all of the water in a matter of seconds.

Kelud and noise

In an innovative study sonar was used to measure sounds emanating from the volcano. The sound is produced as water boils at depth, and as gases are ejected into the crater lake. The ambient water noise level was measured at 130dB.

During the last month before the 1990 eruption the noise level increased to 140dB as more gases was pushed out as magma started to rise and fracturing increased. This actually makes it possible to predict eruptions at Kelud by measuring sound.

The Future of Kelud

The extruding lava dome of 2007. Photograph by Tom Pfeifer.

The extruding lava dome of 2007. Photograph by Tom Pfeifer.

There are two separate dangers of Kelud that has to be taken into account.

First of all we have to look at the crater lake itself. For every eruption the floor of the crater is lowered and the potential volume of the lake increases. If the tunnels had not been operating during the 2014 eruption the water content would have been quite a lot larger than during 1919. Sufficiently so that the lahars would have travelled in excess of fifty kilometers causing widespread destruction, killed people, and taken away the livelihood of many subsistence farmers.

Due to the active mitigation this did not happen. But, let us play with the idea that the Indonesian government decides at some stage that the cost of constantly rebuilding tunnels is to daunting and leave the volcano as it is. Or even worse, also diminishes the resources for the volcanologists guarding the volcano.

In this scenario we would be waiting for an eruption that would kill between 10 000 and 100 000 people as large scale lahars barreled down valleys, rivers and canyons.

I do though not think that the Indonesians will do such a thing. They know but to well what Kelud is capable of and they also know that the cost of the tunnels is low compared to the lives they save.

The second risk lies in the nature of Kelud itself and its eruptive history. The system is gaining strength as the system evolves and it is already capable of producing a large VEI-5 to a small VEI-6. This is not guesswork; this is what the volcano already has done.

An eruption of this severity that has a short run up would kill everyone within a ten kilometer radius from pyroclastic flows, ash fall, lahars and volcanic debris beating people to death. It would also kill people up to 30 kilometers out, but leave a possibility for surviving. Expected death rate for such an eruption is expected to be between 10 000 and 100 000 casualties. This number will obviously be lower due to the Indonesian authorities being adept and fast at performing evacuations as the warning signs come in.

There is though the problem that there is evidence that the eruptions are ramping up in power over time. The frequent powerful eruptions that we are seeing are a sign that the magmatic reservoirs are evolving and growing, and sooner or later the eruptions will become too powerful to contain and a caldera event will most likely happen at Kelud.

If, or perhaps more to the point, when it happens the volcano is almost perfectly constructed to produce the most deadly of the primary killing factors, the pyroclastic base surge. Other andesitic volcanoes around the world that has the same shape and wide bore crater has often produced such surges, and if that happens nothing will survive within fifty kilometers out from the volcano. The ash fall would kill even further out as several meters of ash fall over villages and cities. And tertiary effects such as famine and diseases spread from contaminated water will kill even more people.


Kelud and the proposed new Decade Volcano Program

Kelud has a known history of explosive and dangerous eruptions. It is situated in a densely populated part of Indonesia. It has several modes of lethality that has been evidenced historically. From a scientific standpoint it bears to study, both how it is driven and also due to the unique attempt at active mitigation.

Due to this we have chosen Gunung Kelud as the number ten on our list.


65 thoughts on “The Water Volcano – The new Decade Volcano Program

  1. An excellent start, a volcano that most of the ‘guessers’ (including me) ignored because of its fairly frequent eruptions. Just one tiny correction, perhaps; it wasn’t the Indonesian government which first initiated the tunnel building, but the Dutch colonial authorities (Dutch East Indies) – Indonesia didn’t exist as such until it gained its independence after WW2

    • I assume even, that the meticulousness of the Indonesians concerning monitoring their volcanoes, was at least partly inherited from the Dutch engineering traditions that are concerned with the constant battle against the sea (dikes, wind mills, polders, etc) …

  2. A very nice write up! Having access to water makes the base surge threat all too real. Sorry about the “Bozo” commentary the other day, I did not know Kelud was coming up. Kelud’s last blast was about the equivalent of a one megaton detonation in strength.

    … at least this time it wasn’t me being a jerk and blurting out a correct answer before the question was raised.

    Anyone who has been around here for any length of time, has probably seen me using the Mastin et al formula to calculate the mass ejection rate of a volcano based on the plume height. Mastin et al built that formula based on previous work by Sparks. Mastin has another paper specific to the unique hazard posed by volcanoes such as Kelud.

    The hazards of eruptions through lakes and seawater USGS – L.G. Mastin, J.B. Witter (2000) Journal of Volcanology and Geothermal Research 97 (2000) 195–214

    And the paper that gives the formula for mass ejection rate estimates: A multidisciplinary effort to assign realistic source parameters to models of volcanicash-cloud transport and dispersion during eruptions
    Mastin et al (2009), Journal of Volcanology and Geothermal Research VOLGEO-04231.

    • I am still amazed at your psychic powers.
      Time and again you come up with the answers before the question is raised.

      That, or you are very good at deduction. 🙂

    • I’m gonna go with the deduction thing. Subconsciously the human brain is always processing data. It makes sense and doesn’t require some unknown force of nature to explain. And, just another “hit” that is subject to confirmation bias. We tend to overlook the misses. Charlatans use this phenomena every day. Oddly enough, Miss Cleo never saw the 2002 Federal Trade Commission charges coming.

      Here is a mental exercise for you. In New York, they are scrambling trying to locate two convicted murderers who (pay attention now) broke out of a maximum security prison using power tools… that noone noticed were being operated.

      Reportedly, they even left behind a note stating “Have a nice day.”

  3. Awesome!

    That picture by WB Wilapa looks like the nose of some gigantic beast being lit by a fire, dare I say a Dragon?

  4. Seen anything on the base surge?

    The physics are probably completely different, but nuke blasts are known to have a precursor wave phenomena. (the radiation pulse heats the air and allows the shock front to appear to travel faster than sound)

    An anomalous feature of the blast was the formation of a precursor, a second shock front ahead of the incident wave. This precursor was formed when the shock wave reflected off the ground and surpassed the incident wave and Mach stem due to a heated ground air layer and the low burst height.

  5. Not a ton to argue with here. Definitely a volcano that makes you wonder why it wasn’t included in the original decade volcano list given it’s well-known history, and the fact that it IS a well studied volcano similar to most of the other volcanoes in the decade program, with a known body count.

    My only thought of contention is that it would seem questionable to me whether Kelud could do a VEI-6 or higher within the next 100 years, which is more or less what you would need to reach the 1 million death count (among other things). I don’t doubt it will get there at some point in its geological history, but I wonder if it has a magma chamber large enough to do a VEI 6 or VEI 7 right now.

    • One should remember that it had a medium sized VEI-5 a less than five hundred years ago. It is highly likely to have a large enough magma reservoir.
      The question is about potential, our list all have the potential, and one of them is rather likely to go big in the next 100 years, but obviously not all of them.

  6. No arguments from me, that seems a good summary; would have been on my list.

    My biggest caveat is evacuations and warnings; Indonesians tend to be rather excellent at both. To get a really big number of fatalities, the biggest single factor is lack of the above – which means either a largely or completely unmonitored volcano, OR an eruption with such a fast onset that no amount of monitoring will produce a useful warning. Think Calbuco more recently, Hekla routinely, or – especially – Tarawera historically. Trying to work out which volcanoes CAN do THAT is a rather tricky proposition – but they’re the ones most likely to go MDE, if there’s a population within range.

    Remember, Mount Pelée would have been a footnote in history if only people had had a clue what was going on…

    • “The frog and boiling water”-syndrome. A long as Kelud does it’s usual stuff, the system works but there will inevitably be those who feel that even the modest margin of safety worked into the equation is too much. With time, there will be political pressure to “cut it fine” where exclusion zones are concerned.

      Then the signs from the volcano get ambiguous. Many interpret this as anomalies and stay withing the VEI-4 paradigm (“business as usual”) whereas a minority correctly see this as warnings of a much bigger eruption. Of course the politicians will go with the politically least disruptive option and there’s your recipe for disaster.

      • “recipe for disaster” Assuming this is a reference to the TV program series. The general take from that series is that mondo bad stuff is usually a chain of less disconcerting events or happenings that when coupled together, make for one hell of a mess.

        I think that’s an appropriate way of looking at it. It also fits with the idea that extinction level events are usually the coincidence of two or more really bad things that use up or thwart a species ability to adapt or overcome.

        (such as massive flood volcanic eruptions going on at the Deccan traps and then an Asteroid coming along and smacking carbonate rich rock in the Yucatan area)

    • I agree with you that the authorities in Indonesia is excellent at what they do and that their work diminishes the risk greatly.
      I though weighed in the scientific value of Kelud as being the only actively mitigated volcano into the equation.
      Your reasons are though why Kelud ended up last on the list.

      As always, hats of to Surono and the rest for their work.

  7. Which begs the question… where is the water coming from? Is it possible to drill one or more circles of deep wells, and interfere with the water output?

    Released from the Dungeons, Admin

    • I would assume… from an aquifer. If drilled, it might make a pretty decent hydrothermal plant. But before anything like that is done, the plant operators/funders are gonna want assurances that the thing will work, and that there is going to be an ROI.

      You and I would probably see the savings in live as being a good return on the investment, but most people who put money into something want to at least get a return on their money after it’s said and done. Getting the people who benifit from it to pay for it makes sense… provided they had the money to put into the project. Many in the area are subsistence farmers… and that sort of means that they don’t have a lot of spare capital to invest. Then there is the idea of convincing them that it’s worth while.

      A lot of geology and study has to be done so that the plant operators know exactly what is going on down there.

      • It rains quite a bit on Java too, with the 2014 eruption they were very lucky it did not rain – although it was the rainy season:
        “Total annual Precipitation averages 1755 mm (69.1 inches) which is equivalent to 1755 Litres/m² (43.05 Gallons/ft²).”


    • You would need to drill very deep since the aquifer is deep. On top of that the aquifer is either coming from a large pocket of meteoric water, or the ocean itself. I pondered the though a bit when I wrote the article, and it could be done I think, but the installation would be ruined as soon as an eruption occured.

      At Lughduniense, the scientists have concluded that the water has nothing to do with rain. The chemistry is too different. Leaching rain water crater lakes has a much higher acidity than the neutral water squirting up at the bottom of the crater lake.

  8. Great stuff, cant wait for more!

    I wrote to Dr. Erik Klemetti about his personal choice for a new decade volcano list, and I do expect a post sometime during summer on his blog on this matter.
    Time to get the ball rollin’! 😀

    • Never ask a volcanologist as they will invariably place the one they are currently working on or interested in at or near the top! Dr Boris Behncke’s penchant for Etna is legendary (more power to him for his dedication! 🙂 ), and since Dr Erik Klemetti investigated a volcano in New Zealand (Tarawera?) during his post-graduate years and wrote his doctoral thesis on Volcán Aucanquilcha, Chile, they too might appear on his list. 😉

      • Probably a variation on observer bias. Anyone who has studied a system that intimately is going to know more about it than anyone else. (well, that is if they are competent, which both of your examples are.).

      • Tarawera should be on *everyones* list, as an Awful Warning. Not because it had a large body count, but because of the nature of the eruption.

        A bloody great basaltic dike came up from depth with astonishing speed, in the middle of a rhyolitic province. At 11pm there had been no precursory activity of significance, by midnight there had been strong seismicity and the eruption was starting, by 1am it was in full swing, and by 6am a VEI 5 eruption of extraordinary ferocity had ended.

        If it can happen once, it can happen again. If it can happen in one rhyolitic volcano-tectonic depression, it can happen in others. If it can erupt so strongly at the surface, it can intersect a large volume of near-eruptible rhyolite and precipitate God knows what. I personally believe that one of the most dangerous volcanic scenarios of all is that last one; we largely operate on the assumption that we’ll have a lot of warning before a restless caldera commences a very large eruption. I believe Tarawera is a clue that that may not always be the case; I think these things can sometimes go from zero to VEI 7 in hours or days, if a Tarawera-type event happens in the wrong place at the wrong time.

        • Camping Flagrant inflated 7 meters in 48 hours and then went “boom”. Calderas can do weird things very rapidly.

        • Very good points Mike, I agree with you. Any system capable of serving up ~1 cubic kilometre of very hot, juvenile magma that quickly could indeed precipitate a colossal (VEI 6) or even mega-colossal (VEI 7) eruption within a very short space of time.

          I’ll say no more at this juncture! :mrgreen:

  9. Auker et al have recently looked at a fatality analysis of historic eruptions, and identified high risk regions. For those with an interest in longevity, their Fig. 1 identifies which cities to avoid. Auckland is top on the list. For average fatalities per eruption, the Westindies are head and shoulders above the rest of the world, with the Philippines and the Mediteranean next.

  10. Giving this a bit of thought, I think most of the room for debate in any topic of the world’s most dangerous volcanoes will more likely be related to which volcanoes get left off, not which volcanoes were chosen to be included in the list.

    • Which is as it should be as nothing is ever written in stone! Not only do we encourage debate, Carl distinctly said that he would open up VC to any volcanologist who desires to put forward a different volcanic system. In fact, one of the reasons for the series is to increase awareness of volcanic hazards around the globe, our own included.

      • Curses, “ne’er lead with your chin”. I walked right into that one, didn’t I? 😀

          • One thing that really ticked off my wife is when she found out that her mom had the same profile as I have. Not quite as heavy on the introvert mark, but there. My wife is an ESTJ. You don’t change her mind once it’s made up. (Unless you can drag some convincing evidence kicking and screaming to the table) → The evidence has to be kicking and screaming. Taking a belligerent approach to an argument with her just hardens her resolve. That “J” trait can be a real pain in the neck to deal with.

            And if anybody is wondering what we are talking about. It’s the Meyers-Briggs Type Indicator. Back when I was an instructor, we used it to troubleshoot issues with student training. If you ran a across a student who just couldn’t “get it” is was occasionally helpful to see how they processed new information. Presenting it in a manner compatible with how they look at the world tended to help. About the only time that I saw this fail was with the vampire kid. I think it would have worked eventually, but he kept wandering off at midnight.
            (He had previous pending legal issues back home and never returned from the court appearance.)

    • Makes perfect sense to me. Tho the latest deep quake swarms at 20-30km down do signify a potential medium to long term change. And Katla is known to create eruptions in a fairly short time (up to afew years) once the influx restarts.
      Bottom line, when the actual eruptive deformation occurs, there will be little doubt whether its ice or magma, causing the deformation.

    • I am a bit surprised about the article. I know for a fact after talking to Sigrun Hreinsdottir that the Icelandic volcanologists have known this for at least a decade. Seems like the article stated something obvious.

      • Putting it into the public domain is a good thing though with so many making a living from fear-mongering nowadays.

  11. Reblogged this on Geologically Speaking and commented:
    And we are off to a fab start with the revamp of the decade volcano list. Introducing Kelud (often refered to as Kelut). Little gutted this one didn’t quite make my list (came in about 13!)

    • Thanks Melly! The real fun begins once the entire list has been published and everyone put forward the case for their picks. 🙂

  12. clear and present, fits the bill of this volcano on the previous list very well, I came across it looking for something else, volcano under a city, Goma that is in the Kongo

    • Thanks for posting the link! Nice show, albeit a tad too over-sensationalised for my taste.

      Nyiragonga has already shown its hand, so to speak. In 2002, a fissure opened on its flank which emptied the lava lake (just like your local water tower and tap/faucet). Gravity assisted / pushed, the lava flows entered Goma causing widespread damage. Because it is a Decade Volcano, advance warning was given and 400,000 people were evacuated. “Only” 147 people died even if 15% of Goma was covered by the flows and about 120,000 people became homeless as 4,500 buildings were destroyed. While Goma indeed has over 2 million inhabitants, it is very hard to see this volcano meeting the 1 MDE criterion.

Leave a Reply

Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s