It was the largest volcanic eruption since the start of the world-wide web. The invention of telegraphy in the 1850’s had made long distance connections instantaneous. It changed the world. Newspapers were the most obvious beneficiaries, being able to bring
gossip news from far away places. And in this landscape, Krakatau exploded. 36,000 people died in the ensuing tsunamis – the town of Anjer was completely destroyed. The news exploded around the globe. Strangely coloured suns, moons and skies followed.
The eruption had an unintended side effect: an editor at The Times in London read out the telegraph wrongly and swapped the ‘a’ and the ‘o’. As The Times, in common with all newspapers, is never wrong, reality had to be adjusted. The name of the volcano changed from ‘Krakatao’ (the Portuguese spelling) to the one used in English even today, ‘Krakatoa’. Such is the power of the press in our webbed world.
Before the eruption, Krakatau was a decent-sized but not particularly tall island. There were several volcanoes on the island. The main peak, Rakata, may have been dormant: the descriptions indicate that the 1883 eruptions were from multiple (simultaneous) vents along the Danan-Perboewatan line. These are the lower cones in the background of this pre-1883 drawing. The main explosion, after four months of activity, may have been at Danan. In the last few weeks before the main event the volcanic activity had been steadily increasing. A few months earlier, a picnic party had come from Jakarta to view the eruption. Two weeks before the final flourish, the main geologist (Verbeek) had refused to go on land in view of the danger.
This eruption sequence seems to have been opposite to that of many other eruptions, which start big and decline over time. Krakatau started with a good-sized bang (VEI-4?), declined, but increased again building up to the cataclysmic event. Why? We may never know. A possibility is that during the four months of eruptions, water had begun to slowly enter the shallow magma chamber, decreasing the melt temperature and increasing the melt fraction. In the end, the main explosion left a hole in the sea floor 250 meters deep and 5 kilometers across, killed every single person on nearby islands and wiped the Sunda Strait coast line clear, not only by tsunamis but also pyroclastic flows. The newly discovered world-wide web of telegraphy lapped it all up.
This was certainly not the first time the mountain had exploded: the ring of islands around the explosion dated from one (or more) previous caldera-forming eruptions. But before the rise of the web, the documentation was a bit more sketchy. The stories about a 416 AD eruption come from the Pararaton (Java book of kings):
A thundering sound was heard from the mountain Batuwara which was answered by a similar noise from Kapi, lying westward of the modern Bantam. A great glowing fire, which reached the sky, came out of the last-named mountain; the whole world was greatly shaken and violent thundering, accompanied by heavy rain and storms [‘rain of stones’ in another translation] took place, but not only did not this heavy rain extinguish the eruption of the fire of the mountain Kapi, but augmented the fire; the noise was fearful, at last the mountain Kapi with a tremendous roar burst into pieces and sank into the deepest of the earth. The water of the sea rose and inundated the land, the country to the east of the mountain Batuwara, to the mountain Rajabasa, was inundated by the sea; the inhabitants of the northern part of the Sunda country to the mountain Rajabasa were drowned and swept away with all property. The water subsided but the land on which Kapi stood became sea, and Java and Sumatra were divided into two parts.
This was written up in the 15th or 16th century, and is mythology. Three volcanoes are mentioned: Batuwara, Kapi, and Rajabasa. (In another translation, Kapi is a god of fire rather than a mountain: reading it like this makes a bit more sense of the text, but means that the eruption came from Batuwara.) The last volcano is on Sumatra, at the northern end of the Sunda strait and did not erupt. Batuwara is a presumed extinct volcano on Java but it is not clear that this one is actually meant as the story says that the flooding was east of this mountain while the Sunda strait is to its west. Kapi is normally assumed to be Krakatau but the story places it on dry land. The Sunda Strait consists of pull-apart grabens, 20-40 meters deep and has not been dry land in recent times. The story provides many questions! Looking at the detail, it describes a tsunami which impacted the northern and eastern area of the Sunda strait, and is attributed to Kapi falling apart. No evidence for a major eruption around this time has been found. However, it is not described as a major event: it seems to have been observed (by surviving reporters!) from nearby (volcanic eruptions can’t be heard from tens of kilometers away because of the way sound waves bend: people who saw St Helens erupt comment on the total silence). This was not a caldera-forming eruption. The island ring must therefore have formed much longer ago.
Part of the story makes sense after the 2018 event. It describes a moderate eruption from a volcano (which could have been Krakatau but this link is not stated), which slid into the sea and caused a devastating tsunami. A flank collapse, larger than the recent one, could fit the descriptions. We can take from this that tsunamis from collapse event are a recurrent, and reported, hazard in the Sunda Strait.
(It has been proposed that this eruption took place in 535 AD. This was based on indications for a major eruption somewhere in the world around this time, but there is no evidence (at all) which ties this to Krakatau.)
Why is Krakatau here? The whole region is highly volcanic and almost every feature of significance here or on adjacent Java and Sumatra is volcanic. It is near the eastern edge of the old (Asian) continental crust, lying on top of old subducting ocean crust. Rotation of Sumatra has opened up the Sunda Strait. Krakatau lies near the northern tip of one of the basins, the Krakatau graben. Krakatau’s main magma chambers are at depths from 3 to 7 and 23 to 28 kilometers (the moho) and are located at boundaries in the crust. The Krakatau complex produces basaltic, andesic and dacite lavas. It is an interesting and perhaps understudied region.
The child of Krakatau was not its first born. The first island showed up in 1927, approximately where Dana had existed before 1883. Presumably it had taken some time before the new cone had build up but this is not recorded. The first island did not last long. In mid-1929 the eruption resumed, and two further islands came and went, until in late 1929 the final island raised its head and managed to grow faster than the sea could take it down. It was probably standing on the shoulders of its demised siblings.
It is interesting that the new cone formed here, on the old line of vents. Such a large event should have destroyed the conduit. After a major eruption, a replacement volcano tends to grow along the perimeter of the caldera, away from the caldera centre. Krakatau is made of tough stuff. (However, the funny part is that the 1883 caldera appears to be to the side of Anak Krakatau. It is not clear why the 1883 caldera would be away from the line of vents. Perhaps Anak Krakatau has already filled in a bit of the caldera.)
Frequent eruptions caused fast growth, by as much 10 centimeters per week. (As an aside, this is about the rate needed to get from 250 meter depth to the surface between 1883 and 1927, but this is speculative.) By 2000, the new cone had reached 300 meters. The height prior to the 2018 collapse appears to have been a bit over 400 meters, after a growth spurt in the preceding decade. However it was still very far from rebuilding the old island. The lava flow rates are not excessive: the island has grown by about 0.002 km3 per year. Presumably only a small fraction of the new magma is making it to the surface. Assuming a fraction of 10%, the magma supply rate is 0.02 km3 per year, and it will take 500 years to recover from the 1883 explosion. That still seems fast, and it is likely that the current rate is above the long-term average, perhaps because of some decompression melt. The lavas are currently basaltic andesite, and probably have had this composition since the birth of Anak Krakatau.
Half of Rataka had survived the 1883 explosion but no life had survived here. Much of the flora and fauna may have died already during the four-month precursor phase. The return of an ecology to Rataka is a case study of how an island can come back to life. Ferns started growing from 1886, gras from 1897, trees shortly after and a forest reformed at the lower slopes after 1920. Nowadays it is dense rain forest. Anak Krakatau did not do nearly as well because of the frequent eruptions. The eastern side became forested, but in recent decades the lava flows had steadily reduced this oasis. Nothing appears to have survived the 2018 events: the current images show devastation.
The 2018 collapse
The collapse event has been well published on our new and improved world wide web. The first sign of trouble was the tsunami which came on Saturday evening, 22nd of December. There was no warning: no earthquake (although there was a minor earthquake in the Sunda Strait earlier that day), and no significant larger eruption (although there was on-going activity with lava entering the ocean). The event was not witnessed.
It took some days for the truth to come out. The thick clouds had to disperse first. Radar images (above) showed the first evidence of a major collapse towards the west. Subsequent optical images showed that, if anything, these first indicators had understated the case. The cone was fully gone. The west side of the island was smaller than it used to be, and the east side had acquired a new cover (obliterating the trees, therefore many meters thick) and extended the coast outward. In effect, the whole island had slightly moved east. For your enjoyment, here are the main images.
(A slider comparison can be found on the excellent BBC coverage, https://www.bbc.co.uk/news/science-environment-46743362)
In terms of volume, 160 million m3 has gone, of an original 200 million m3. It is a major set-back in the Krakatau recovery.
It is interesting that the radar images differ from the later optical images. The reason is not clear to me. The collapse may not have been a single event, or the dust was so thick on the west side that radar interferometry did not work well, or (my guess) the radar view was taken from low in the northeast and the southern and western parts were hidden behind the remaining peak from the satellite. The final images show that a bay has formed, open to the west, where the cone existed before. This cannot have just come from the collapse as the material evacuated from the bottom of the bay would have nowhere to go. The central hole may have formed in a subsequent explosive event.
It is interesting that the collapse and tsunami had been predicted in a paper from 2011, including the direction of collapse, the approximate volume, and the height of the tsunami. This new mountain was known to be unstable.
How do volcanoes cause tsunamis? There are three main mechanisms. The most obvious is that of an underwater explosion. In fact the explosion itself is likely to be only a part of the story, and the subsequent collapse of the caldera is the main cause. Water falls into the excavated hole, the sea surface drops, and a wave equal in volume to the new caldera begins to move out. The tsunami that follows will begin with a withdrawal of the water, before the main wave comes in. The best example of such an event (probably) is that of the explosion and collapse of Krakatau in 1883.
The second mechanism is that of a flank collapse, where the landslide pushes the water out of the way, and the wave begins to travel. Here, the tsunami will normally begin with the water rising up. As the flank collapse is on one side of the volcano, the tsunami will mostly travel in that direction. The most dramatic example is that of the collapse of Unzen in Japan in 1792.
The third mechanism is very different: it is the fall-back of ejecta from a nearby volcano, especially as pyroclastic flows. These can cause tsunamis even where the volcano is not directly on the coast. The familiar example is the tsunami that followed the (in-land) Tambora eruption.
These events have one thing in common: casualties. Tsunamis are the dominant source of fatalities caused by volcanic eruptions. Ignore the risk at your peril. In the NDVP list, the top two are both tsunami risks.
What causes a flank collapse? Ultimately, gravity is to blame. Without gravity, nothing would happen. But obviously there is more to it. A flank becomes unstable if it becomes too steep. The easiest way to see is by building a sand castle. Loose sand can only support shallow slopes: the sand grains start to tumble down, until they make a slope of a certain angle. For the same type of sand, that angle is always the same. Add more sand, and the castle gets bigger, but the slope does not change.
What follows is a battle between gravity, friction, and cohesion. The force of friction keeps the sand grain in place. As the angle steepens, friction grows less: it is equal to the weight pressing on the surface times the friction coefficient, and as the angle steepens, less weight is pressing on the surface. (Try this by weighing some object on your kitchen scale while holding the scale at an angle.) At the same time, the force which tries to pull the grains down the slope increases as the angle steepens and gravity more and more pulls it along the surface rather than into it. Try walking on an inclined icy surface and you will see this in action (I remember driving behind a truck trying to get up an icy incline: evasive action was needed. I made it – the truck didn’t.)
But use sand with rougher grains and the angle can become steeper before the cascade starts. Make it wet, and for the right kind of sand (i.e. rock) the slope can become near vertical. The cohesion of the material is thus very important.
The angle you end up with is called the angle of repose, and for many materials it is around 30-40 degrees.
How is this for volcanoes? It strongly depends on how the volcano was made. The sand castle above is most like a scoria cone: material thrown out by the volcano, solidifying in the air and coming down as fragments of varying size. Such a cone may become fairly steep, but it will always be at the angle of repose. Loose tephra may do the same thing. Lava, on the other hand, is a liquid and it laughs at friction. It flows easily, and therefore build shallower slopes, of perhaps 15-20 degrees. Basaltic volcanoes have lava with low viscosity, and this builds even shallower slopes of perhaps 10 degrees. When the lava solidifies, it freezes in position. This mountain is now quite stable, with an angle well below the critical angle.
This makes it sound as if only scoria cones could collapse. In fact, it is the opposite. Scoria cones (and similar) form at the critical angle and can (badly) cascade but rarely fully collapse. The risk comes from volcanoes that grow at too steep an angle. How can that be?
In the case of basaltic volcanoes, the instability happens some distance along the slope. Lava is fine as long it can flow, but after some time it cools and becomes solid – and stops flowing. At this point, the slope will become much steeper. Hawai’i is a point in case. Look at the profiles: you will see that they steepen a lot under water. Once lava reaches the sea, it cools almost instantly and stops flowing (or flowing far), so here is where Hawaiian volcanoes become really steep. The flow rate also plays a role, as that determines how far lava can get before cooling (higher flow rates give thicker layers which cool slower and can therefore reach further). The Holuhraun flows reached a certain distance – and no further. As the eruption waned, the flows no longer reached the edge of the flow field. Look at Olympus Mons: it is surrounded by a steep cliff: this shows how far the summit lava could reach.
Another way of steepening the slope is by having a magma intrusion, especially if this is along the flank. This is what killed St Helens. It is a slow hazard in inflating, pre-erupting volcanoes.
So now we have a steep slope near the bottom of the volcano. How does a collapse happen? Say the whole thing just about hangs together. Near the bottom, a small collapse occurs. This makes the slope just uphill slight steeper. It also begins to collapse, and so a small collapse travels uphill. But now a new law of physics kicks in: the coefficient of friction is smaller for moving objects than it is for stationary ones. As soon as the slide begins, friction reduces and the critical angle becomes shallower. And suddenly the whole mountain is over-steep and unstable. The collapse has begun.
Another way for reducing friction (best avoided) is by adding water. We all know how dangerous aquaplaning is. Fracking can cause earthquakes by injecting water into faults and so reducing the friction that keeps the fault locked. Steep slopes can suffer the same problem when they become water-logged. Typhoons cause landslides just from their rain, leaving slopes too steep to support themselves. Physics can be an unforgiving bastard.
Shaking can also reduce friction, The Hokkaido earthquake did this most effectively. The top layer of the soil was ejecta from an ancient eruption, a thick layer of eroded tephra. The earthquake caused it to slide at every steep surface.
The collapse of Anak Krakatau
So what happened? A hint comes from the location of the new island. The map below, from Deplus et al. 1995, shows the 1883 caldera, viewed from the west. Anak Krakatau grew up on its edge. (This is a bit funny (this feeling inside) as it puts the 1883 explosion (implosion?) well away from the erupting vents.) But in consequence the slope was far steeper to the west, into the caldera, then in the other directions.
Look again at the satellite images before the collapse. There is a small bay visible on the west. This was already present before. Comparing the 2014 and 2018 images, the island had expand to the south, making this small bay more prominent. The volcanic cone reaches the bay, as shown by radar image above.
Sometime (probably 2010/2011), a lava flow reached the ocean a bit north of the bay. On the image slider one can see that the area that collapsed extended from the bay to just north of this lava flow. Perhaps the lava entering the ocean prior to the collapse pushed the slope beyond the limits. A section of coast gave way, possibly at this bay where the steepest slope occured, and the collapse progressed towards the adjacent caldera. This steepened the slope, and more began to slide in. A runaway process followed, with the collapse engulfing the cone itself. The massive slide caused the tsunami.
The collapse exposed hot material to the elements and sea water. This was the cause of the clouds, and the clouds carried ash up with it. It reached to high up in the troposphere. But little new ash was generated: what came was from a limited amount of magma entering the water, causing a phreatomagmatic explosion. The clouds, in contrast, came purely from the exposed heat.
But it is hard to stop magma. The conduit was close to water level, but worse, was sideways exposed. The main explosion now took place here, caused by water coming into contact with the conduit. It was this event which caused the semicircular bay to form, excavating the cone now to below sea level.
This scheme is speculative: we have no evidence other than the images, and it is also possible that the collapse began with a volcanic explosion at the summit. But I consider the speculation here as more likely.
Note the curve enveloping part of the new bay? It is not clear what caused this, but a plausible suggestion is that is material collapsed to the south, and transported by wind and water to here. The map above shows the dominant wind and current direction this time of the year (left). The water was always shallow to the south, outside the 1883 caldera, and much of the material will be close to the surface. I expect that this transport will begin to fill in the new bay – whether it can close it off depends on how deep the new channel is. Further collapses along the edge of the new cone are likely to also fill in the bay, perhaps quite quickly.
Eruptions will also continue and may even intensify. A new cone is likely to form very rapidly, as the magma chambers have been left untouched and the confining weight reduced. How quickly it reaches the surface depends on the depth of water. That is not known. The width of the waves inside the bay is similar to that outside. That makes the water much deeper than the wave height. I would guess it is at least 10 meters but may be much more. But at this depth, the cone could break the surface within a few years – and earlier if infilling from elsewhere continues.
Full recovery will take longer. It may take 50 years before Anak Krakatau reaches its pre-collapse height.
And will it collapse again? Assuming that 100 million m3 was deposited in the old caldera, that is still less than 10% of the volume of the subsea caldera. The hole next to the child is still there, waiting for a wrong movement. But it won’t be as deep close to the island, and therefore the new cone will not be as unstable. That means it may grow larger before history repeats itself. This will not have been the last tsunami to hit Anjer. The world wide web of those days will be ready.
Albert, January 2019