Kilauea is a basaltic volcano, driven by magma from a deep hot spot. Over time it has build a summit on top of the older Mauna Loa lava pile, and a one-sided eastern ridge extending far into the Pacific Ocean. There is a rift on the west side as well, but it is short, has no ridge and bends towards the south west under the pressure from towering Mauna Loa. The landscape around Kilauea is mixture of green forest, dark lava plains and ashy desert. Things here can change without warning. A stark lava-scape may once have been a forest. One settlement here was called Royal Gardens, the name of a lush site. It now lies buried by a 30-year long eruption. The historic town of Kalapana suffered the same fate. Leilani was hit by the surprise eruption of 2018, and although much of it survived, the coastal settlement of Kapoho, now a ghost town, did not.

Kilauea is unpredictable. It can do just about any kind of eruption imaginable: effusive, explosive, summit, rift, mini and mega. We have tried to find patterns in its eruptions, but the volcano takes little notice and continues to do whatever it felt like. It is not a nice neighbourhood volcano. People living on its slopes have found themselves at the whim of an unruly ruler. Paradise here can suddenly and easily be lost. Visitors see the beauty and the excitement. The locals know the dark side of paradise.

But over the past two years Kilauea has changed character. Eruptions are now predicted weeks in advance and are almost carbon-copies of each other. They begin with upwelling and spattering, quickly grow in the voluminous fountains, and cease once deflation has reached a certain level. Inflation quickly recovers, and the next eruption begins at the same peak level of tilt as before. There have been minor changes, such as the recent dike on the caldera rim, but the pattern has continued for 55 episodes so far, with number 56 well on the way. Even the end seems predictable: at the current rate of magma depletion, in another year, the pressure needed to initiate an episode is gone.

The reason for the sudden predictability is mysterious. Yellowstone’s geysers are perhaps the closest in volcanic predictability, but there the reason is volcanic inactivity, allowing a stable water circulation to feed the geysers. Kilauea is not like that. We cannot predict how long its predictability will last.

The other mystery comes from the fountains. How do they form, why are they so stable over so many hours but start and end quite suddenly? Perhaps we can say something about that.

Lava fountains

Lava fountains may well be the visually most impressive type of eruption. The record for the tallest lava fountain (at least as observed) is held by Mount Mihara at Izu-Oshima in Japan, where in 1986 a fountain reached 1500 meters tall. Etna is particularly well known for its tall lava fountains. One of these is reported to have reached as high as 3km. However, it is not always clear where the lava fountain ends and the updraft of hot tephra takes over.

Eruptions at basaltic volcanoes often begin with lava fountains, before the flows become effusive. They may come from long fissures, which are also characteristic for basaltic volcanoes. But lava fountains can also appear later in an eruption, as happened during the first Reykjanes eruption. In May 2021, two months into its eruption, it suddenly showed impressive fountains which reached 450 meters high and which were easily visible from Reykjavik.

Volcanoes with more evolved magma can also produce fountains, but rhyolitic volcanoes tend not to.

Kilauea is known to do episodic fountaining, where its long-lasting eruptions can contain a phase of on-off lava fountains. This happened during the Pu’u’O’o eruption which started with three years of cone-building. During this phase, in 1983-1986, there were 44 episodes of fountaining. The fountains ended only when in 1986 the eruption moved to a different location. The eruption would continue for another 32 years (see https://www.usgs.gov/volcanoes/kilauea/science/puuoo-eruption-lasted-35-years), but the fountains did not return.

Before Pu’u’O’o was Maunaulu. This eruption lasted 5 years, from 1969 to 1974. The fountains occured early in the eruption, with 12 episodes over a 7 month period, coming from a fissure. The tallest fissure reached 540 meters – taller than Empire State Building, as HVO pointed out, although their image showing both together would now be seen as an early example of AI hallucination!

Before Maunaulu was Kilauea Iki. During its initial eruption in 1959 there were 17 episodes of lava fountaining. The tallest was just before the sudden end, with the fountain reaching 580 meters high. As in the other two cases, after the fountains ended, an effusive eruption began, but in this case there was a longer delay and is started at a different location, along the east rift.

Kilauea 2024-2026

And now we are in the fourth series of Kilauea lava fountaining in 57 years. The summit episodes started on 23 December 2024. During the first year, there were 38 episodes. We are now at episode 55. Like Yellowstone, Kilauea has found stability.

In this fourth series, the episodes became more regular over time. From April 2025, they increased in volume and occurred at longer intervals. At that time, the lava output and magma input was approximately in balance. From December 2025, the output became more than the input.

A sudden change in April 2026 caused faster, smaller episodes. Most recently, the episodes temporarily ceased, when new openings alomng the caldera wall opened safety valves. A new dike along this path closed these off again, leading to episode 55. What happens next should become clearer soon!

Each episode starts with rising magma levels in the conduits, followed by a series of overflows, each ending in a drain back. Eventually the overflow become faster and lava begins to form an umbrella. This umbrella grows over an hour or less, and becomes a tall fountain, most recently around 200 meters high. After several hours the height begins to decrease but it does not go back to the umbrella shape. The end is quite sudden, with magma retreating deep in the conduit and not coming back up. In some cases, a gas jet spurts out for 1 or 2 minutes just after the end.

Health and safety

The most obvious difference between the overflows, the umbrellas and the fountains is in the speed of the lava. During the overflows, the lava level rises above the rim, but there is little or no excess speed. The lava comes up at a leisurely rate, perhaps as low as 10 cm/s. The umbrellas of 10-30 meters tall require much faster upflow, of 10-25 m/s. Finally, a 200-meter fountain requires speeds of 60-70 m/s, and to get to 400 meters, a speed of up to 100 m/s is needed. To put this in context, that is the speed of a Japanese bullet train. The lava flow rate during a fountain episode is as high as 250 m³/s. It dropped to 100 m³/s before the dike formation on 30 September 2026, but for episode 55 in October 2026 it was back at the high value. It amounts to 500 tons per second on mass! Imagine being hit by a 500 ton lava bomb moving at 200 miles per hour – every second for 10 hours!

This should make it obvious why HVO’s main task seems to be to keep people away from the fountain spectacle. The episodes are best viewed from a large distance, or with a live camera – and even one of those cameras we have seen being destroyed by the high speed lava bullets.

Volcanoes can kill in many ways, but HVO has managed to ensure that the main risk of fatalities from Kilauea is human stupidity error. HVO deserves congratulations and respect for this. But no volcano can ever be made entirely fool-proof, sadly.

(As an aside, I used the numbers for the Kilauea flow rates from https://www.usgs.gov/volcanoes/kilauea/science/eruption-information which are preliminary and subject to change!)

Volatiles

But how can magma be accelerated to these kind of speeds? And where does it get the energy from? This is not an easy question to answer. We have to look at something that is very common around Hawai’i – water.

Magma is more than just molten rock. There are volatiles dissolved into the magma, which were present in the original rock as well. There are three main volatile species : H2O, CO2, and sulfur. How much of these are dissolved in the magma can vary a lot between volcanoes.  For Kilauea, the CO2 content of the parental magma (the original abundance in the melt, as it comes up from the mantle) is around 0.7% by weight. The parental water content is between 0.5% and 1%. Finally, the parental sulfur abundance is around 0.15% by weight.

But these are not the amounts found in magma close to the surface. The magma loses volatiles on the way, and especially the CO2 may be as low as 0.1% when reaching the surface. The rest exsolves on the way, meaning the opposite of dissolving: it becomes a gas. For CO2, this happens at a depth of typically 5 km and so by the time the magma approaches the surface, most of the CO2 is already gone. Water and sulfur exsolve slower. In hot magma, water also reacts with the molten silicates and this releases some hydrogen. The water abundance in the magma near the surface may be down by half. But this is difficult to measure because there is so much water in the environment! Separating water that originated from lava from water already in the atmosphere is often not possible.

The exsolved gasses rise through the magma and through any cracks in the rock that it can find. CO2 can hang near the ground and suffocate the unwary. Escaping water vapour causes the steam clouds which hangs around the eruption site before and after eruptions. And sulfur causes the bad smells emanating from active volcanoes.

Why does it exsolve? The maximum amount of a volatile in magma goes down as the magma rises. Magma under high pressure keeps the volatile dissolved, but under lower pressure some may come out. The picture above shows this for water. At 8 km depth, where the pressure is 200 MPa, basalt can contain about 4% (by weight) of water. Rhyolite is more forgiving and can contain almost 6%. But at 2km depth (50 MPa), these amounts are down to 2% and 3%. If the basalt contains 2% originally, this is still fine, but a bit closer to the surface and some of the water has to come out of the solution: let the exsolution begin. Once the magma becomes supersaturated, small bubbles begin to form. And the closer to the surface, the more bubbles form.

For Kilauea magma, CO2 begins to exsolve at 8 km depth but water only at 1 km. Thus, the deeper magma (say at 5 km) contains CO2 bubbles. Magma at this depth tends to be located in fairly stable magma chambers. The bubbles have time to rise to the top of the chambers and escape through the rocks. In this way, volcanoes breath out CO2 even when not erupting. Water vapour bubbles form much closer to the surface. A stable magma chamber close to the surface can lead to a steamy volcanic breath, in the same way as a deeper chamber causes a CO2 breath. But during an eruption, when magma is rising from the deeper magma chamber towards the surface, water vapour bubbles have no time to escape and remain entrained in the erupting magma.

(This is not the only source of water. Ground water can be found at the same depth and may be heated by deeper magma. This tends to be the cause of the steam in the fumaroles around Kilauea.)

Bubbles in action

It is not easy for a bubble to form in magma under high pressure! Compared to water locked away inn the liquid, it needs a sufficient number of molecules to come together, it needs to create space, and a surface to keep it separate from the liquid. The latter is a stumbling block. There is tension in the surface, and that tension tries to shrink the surface. The smaller the bubble, the larger the tension force per molecule. The initial bubble is only a few nanometers across. If the bubble is too small, the surface tension just forces the molecules back into magma – whether the magma wants it or not. The magma stays supersaturated, and it may take as much as 100 MPa overpressure for these mini-bubbles to become stable. Spontaneous formation of bubble is possible, but it is slow.

It works much better if there is a solid structure, preferably a bit irregular, for the molecule to adhere to. The solid side protects against the surface tension, and it gives the bubble space to grow. Once it becomes a bit larger, the bubble doesn’t need the solid surface any longer and it can detach and become independent. You can see this in a beer glass. The bubbles form on the surface of the glass, detach and begin to rise. (I tried to find a picture of bubbles in beer but couldn’t find one that wasn’t AI generated. You’ll have to pour it yourself. Cheers! You may want to try to foamy-looking one below.) With such a surface, the bubbles quickly become a micrometer across rather than a nanometer.

What solids are available in magma? First, like to beer, there is the surface of the conduit (or the glass). But crystals in the magma also work, and they are much more helpful as they are scattered throughout the magma while the side of the conduit is only found at the edge.

Now a secret law of physics spring into action. As gas exsolves from magma, the bubbles take up space that wasn’t needed before. The magma needs to expand. If that space isn’t there, as in a closed magma chamber, the gas will be driven into minute cracks in the rock that the magma itself doesn’t fit in. The gas begins its arduous escape. But if there is room, for instance in a conduit, the magma expangs into the larger volume. The density of the magma, the total mass divided by the total volume, begins to decrease. As the bubbles grow, the magma becomes less and less dense.

This happens in a stable situation. But if there is an eruption in progress, then this happens while the magma is coming up in the conduit. The higher it moves, the lower the pressure and the more gas comes out of the solution. This has a strange consequence. The amount of magma passing through the conduit per second must be the same at the top and the bottom – otherwise a magma traffic jam would occur. But the density at the top is lower. To keep the same flow rate, in kilograms per second, in a conduit of constant size, the speed must increase. The amount of magma passing through a certain level is given by the cross section of the conduit, times the density, times the velocity. Thus, velocity times density is constant and as the density decreases, the velocity must increase. It is a simple law of physics, but one with dangerous consequences. The traditional rocket-in-a-bottle experiment shows the speeds that a supersaturated, evaporating liquid can reach.

The bubbles are initially tiny. They grow in several ways. The first is straightforward decompression. For an ideal gas, physics tells us that at constant temperature, the pressure times the volume is constant. If a bubble forms at 200 meters depth and is carried upward to the surface, the pressure drops from 5 MPa to the surface pressure, 0.1 MPa (also known as 1 atmosphere, for obvious reason). The volume of the bubble will increase by a factor 50, and the radius by a factor of 3.5.

In reality the bubbles grow much faster than this. This is caused by the second mechanism: more gass is added to the bubble. It enters by diffusion from the liquid, which expels more and more gas as the pressure decreases.

The final mechanism comes from coalescence, also known as cannibalism. Big bubbles tend to eat small ones. Worse, they even chase them! It comes from the fact that the bubbles are buoyant. They have a much lower density than the liquid, and therefore try to rise through the liquid. The buoyancy depends on the volume of the bubble. It is countered by the drag force from the liquid which only reluctantly lets go of the bubble. The drag force depends on the viscosity of the magma, the radius (r) of the bubble and the rise speed of the bubble. Because the buoyancy depends on the volume, which goes as r³, and the drag force only goes as r, larger bubbles will rise faster. They catch up with the slower smaller bubbles – and swallow them.

Because of these three effects, bubbles increase dramatically in size. As the magma rises from 200 meters depth to the surface, the bubbles increase a thousand fold in radius. They can reach sizes of a meter or more. We can see that in the Kilauea lava: the spattering often seen before an episode happens when such a 1-meter-sized bubble reaches the surface of the lava.

Why is the spattering strong before an overflow, but becomes less or ceases while an overflow is in progress? While the magma is sitting stable inside the conduit, the bubbles have time to rise and grow. But when the lava is flowing, the magma is moving up and this reduces the amount of time the bubbles have in the magma: they don’t grow as large, because both diffusion and coalescence need time to act. The spattering comes from the largest bubbles, and they have less time to grow. Typically, when the magma rises at 1 m/s, the bubbles grow only to around 1cm, with little coalescence. To get to 1 meter in size, coalescence is essential, and that requires that the magma takes its time and sits and waits.

Magma foam

But how much gas is there in the magma? It can become considerable. If the magma originally was 1% water by weight, and all of it becomes gas, then the volume of the gas will become ten times larger than that of the magma itself! This is no longer a bubbly beer. It is almost like the foam sitting on the top.

And this is in fact what can happen. The picture below show a piece of scoria, thrown out by an eruption. This is magma that solidified while flying through the air. It is pockmarked by spherical holes. Each of these holes was a bubble, in this case typically of centimeter-size. This scoria is a swiss cheese, with more hole than material! The gas volume exceeded that of liquid. Think magma sponge – although you probably wouldn’t want to use it as such.

Scoria typically comes from basaltic eruptions. Pumice is similar but has smaller holes which are often not round but elongated. Pumice comes from evolved magma, which was much more viscous. This makes it much harder for the bubbles to travel through the magma, and therefore they can’t grow well. The tug of war between the buoyancy and the drag also can make the bubbles more elongated. The rock can be just as spongy as the scoria, though, and you can find pumice stones with very different holes.

Lava doming

At the start of a Kilauea ‘episode’, there is a phase where an overflow turns into a lava ‘dome’, an umbrella. There is little spattering, but the lava forms a round shape which can be ten meters tall, but which continues to feed the overflow. The dome is quite stable (for a while at least) but it can also disappear very quickly. The lava seems to form a sheet around whatever is holding it up.

This phase develops out of an overflow. During the overflow, the upward speed is limited but the bubble growth is suppressed so that the meter-sized bubbles don’t develop. Even a small upward velocity of the magma can make a large difference to the bubble growth. Even in low-viscosity magma, a 1-cm bubble will move upward at a speed of no more than a few cm/sec. If the magma is stationary than it takes a few hours for the bubble to reach the surface – there is time to eat and grow. But during an overflow, the magma itself may move at say 0.1 m/s. Now the bubble reaches the surface in 20 minutes. Even a very slow upflow dramatically reduces the time available for bubble growth.

The upper levels of the magma in the conduit have had time to degas. The exsolved gas moved up and escaped, leaving a partially degassed magma behind. During the overflow, this layer is removed. The layer below is now under less pressure (the weight from above is gone) and it has more gas to come out. The situation remains the same as before, but as more gas comes out of the solution the density decreases and the upward velocity increases. The lava comes out at higher speed. This continues until the gas content is so high that the speed reaches 10 m/s. The bubbles are now small (having had only minutes to grow). What comes out is a smooth flow of low density magma which forms a beautiful umbrella shape. The gas escapes and the lava falls back to the ground. For an exit speed of 10 m/s, the umbrella will reach a height (if ballistic!) of 5 meters, and for 20 m/s it is around 20 meters.

But the situation is not fully stable. If the flow velocity reduces a little, the gas bubbles can grow larger, move faster, and escape. Their faster velocity reduces the gas content of the magma (the bubbles are further apart), the density increases and the velocity declines further. More gas will come out of the solution, but this takes time. In the mean time, the lava level goes down and once it is below the level of the exit, surrounding lava may drain back – forming a degassed lid. The domed overflow ends quite suddenly and we have to wait for new bubbles to form below, before a new overflow can happen.

Lava fountains

But it can also change in the opposite way. At some time, the lava dome begins to grow taller. It quickly develops further, and the dome becomes replaced by a much less regular looking fountain. Clearly the outflow speeds have increased. But how does this happen?

It comes from how the bubbles and the magma interact. You can see the effect in your (fresh) glass of beer. The bubble rise to the top and at the top it forms a layer of foam – the head. There is a sharp dividing line between the beer and the head.

This also happens in the magma in the conduit. Once the gas content is high but the bubbles are small, the bubbles begin to touch each other. There is now a dramatic change where the magma fragments into blobs and filaments. The magma beer turns into a head. It is like a poorly poured beer which is all head and nothing to drink. Somewhere in the magma column, there is a sharp dividing line. Below it, it is a magma with bubbles, Above, it is a gas containing liquid magma drops, a moving magma foam. The magma has turned from a flood into an upward rain.

These dramatics have dramatic effects. The upward velocity increases fast,, beyond this point of no return. The density approaches that of gas, which is 2000 times less than that of magma. Where the initial velocity was 10 cm/sec, it now grows to 200 m/s. The magma rain comes at half the speed of sound, and at the same time is maintains its original flow rate in terms of mass per second, at hundreds of tons per second. These are not numbers within our experience. It is out of this world, and best to stay well away from. But this speed is the reason why lava fountains can reach heights of a kilometre or more. Kilauea is not quite as extreme as this. The fountain heights have been limited to around 500 meters, or an outflow velocity of around 100 m/s.

To get these extreme speeds, magma has to fragment. Without this fragmentation, outflow speeds are limited to typically 10-30 m/s.

(The largest velocities requires a particular shape of the top of the conduit, where it first narrows a bit and then gets larger again. It is called a De Laval nozzle. The cones that build on top of the conduit have a tendency to evolve to this shape.)

But where does the energy come from to reach these kinds of velocities? The answer lies in the gas bubbles. As they expand, the inside pressure acts against the restrains and pushes out. The energy locked up in the highly pressurised bubbles accelerates the gas. But it comes at a cost. As the gas expands, it also cools down. The flying lava remains hot, but the surrounding gas is now very much cooler. And the surfaces of the lava fragments solidify while they are still flying through the air.

When the fountain runs, it is quite stable. The fast rise of the magma causes the gas to come out quickly at the same depth, the magma accelerates and spouts out. It can go on for hours.

What causes it to end? The problem is that the flow rate at the top of the conduit is too fast for the inflow from the bottom. The magma chamber can’t quite keep up. Initially, this is no problem as the pressure there is high enough. But once that pressure drops, the rate at which magma gets into the conduit reduces. This also reduces the flow velocity – after all, the output rate is cross section times density times velocity and the latter is the only number that can go down. After 4 or 5 hours, we see the height of the fountain decrease. Once the flow rate reaches a critical value, the fragmentation stops, perhaps because the bubbles now have too much time to grow. The magma column collapses, and within minutes is below the exit level. Instead of becoming a slow, measured flow, it just turns off altogether. At the very end there is a short period of time where the lava has stopped but bubbles still are coming out. Without the constraint to carry magma, the gas shoots out as a brief jet, visible because the expansion makes them cold enough to condense the vapour back to water.

Explosion

It can get worse. If the magma is entombed, the volatiles can’t get out of the solution and the magma becomes severely oversaturated. It becomes a battle between whatever is entombing it, and the over-pressure of the magma. Eventually, something gives. The ground cracks and the pressure goes down a tad. Instantly, all the excess gas comes out of the magma. The volume increases hundred-fold or more and the mountain explodes. We have just seen Tambora blow up, and a deep crater appear where once there was a magma chamber. Now the velocities can exceed the speed of sound. But the sonic boom is the least of your worries.

Flames

One more question. Before the episode, and at a time when there is no spattering or overflow, some of the vents show strong flaming. What is that? Neither water nor CO2 can burn!

The answer is hydrogen. Water dissolved into hot magma can react with silicates to for OH, leaving one hydrogen free. The hydrogen forms H2 and this is a very small molecule, hard to restrain. It can escape more easily. While the build-up is on-going, hydrogen goes early – and when meetin atmospheric oxygen, catches fire – and forms water in the process.

In the past, people used fires on hills to warn of approaching danger. Apparently, Kilauea does it too.

Water is an amazing substance. The fountain of life is made of water. Even volcanic fountains.

Albert, November 2026